MIDDLEBURY

COLLEGE

THE EGBERT STARR LIBRARY

 i RNA
y
,

aol2.
Ate

t Satie eva gE?)ents

Wes SO Stat regi
Byes
cua 5
Se ce
An

hy, Nesms
ee

eee ee LN
’ fi hie a
x
 COERCIVE
FAMILY PROCESS
Other volumes in the Social Learning Approach series:

VOLUME 1: Families with Aggressive Children

VOLUME 2: Observation in Home Settings

ISBN for the series: 0-916154-10-6

 A Social
Learning Approach
VOLUME
3
COERCIVE
FAMILY PROCESS

Gerald R. Patterson
Oregon Social Learning Center

MIDDLEBURY
COLLEGE
LIBRARY

-“U. DOX
Eugene, Oregon 97440
Library of Congress Cataloging in Publication
Data (Revised)
Main entry under title:

A Social learning approach to family intervention.

Vol. 2. edited by J.B. Reid.

Includes bibliographies.
CONTENTS: v.1. Families with aggressive
children.—v.2. Observation in home settings.—
v.3. Coercive family process.
1. Problem children. 2. Juvenile delinquents.
3. Mentally ill children—Family relationships.
4. Family psychotherapy. I. Patterson, Gerald
Roy. II. Reid, John B., 1940-
RJ500.S65 618.92’ 858206 75-27000
ISBN 0-916154-00-9 AACR1

Copyright © 1982
by Castalia Publishing Company

All rights reserved. No part of this book may be

reproduced by any means, nor transmitted, nor
translated into a machine language without the
written permission of the publisher. Excerpts may
be printed in connection with published reviews in
periodicals without express permission.

ISBN: 0-916154-02-5
Printed in the United States of America
Copies of this book may be ordered from the pub-
lisher.

Illustrations by Gerald R. Patterson

 Dedication
To Marion S. Forgatch, Richard R. Jones, and
John B. Reid, friends and colleagues in a very per-
sonal enterprise.
 Table of Contents
Page
SCS ERIR2 CUE
SET
SLSCLo0s ct ee neeee ix
PCM IOWICGGINICNTS | 0 cence wees eeenesencseensetenes xi
RCO org fo ne ha on ae oe seats ata, 04sApe Se ra te gee xiii
Chapter 1
IP MEtareliAVIOTOfChESCIENEISIS:...
2.5 oes tae wileeo he eeeen eg eahnee eee volewoeeeeeee 1
Chapter 2
eR ORR RETO EOP hs oy istic ie Syd sabi
sowin! a Salsvhstyeisga areswile ayKo Cee hwswebinar nsee 11
Chapter 3
ORAS ELELOCOSSoi. eet ae ccc ehhBais) 90.5owowe Seco Ain;
uel99. WA)
mS) GiB-SweAdeweeid 41
Chapter 4
Averseewents: Phe innocuousDeterminants... 2... 0.6 ee ce tweet ees eweeees 67
emo amramRODEGs.Waller . oi.oie is i diedie ee lee ep as eeviolele eswa ewe bieeles +Keeeore82
Chapter 5
Poeencemrorcement for Aggression. . 2.66... ee eee weeee eee rnc ese aennns 85
Chapter 6
EER ETC SSO. Gn SC TS ees nc EES SOG pedecae ew alee Selleeuies 111
Chapter 7
ee eeeeremmorcementand Escalation,. oi... cee ee ees See sew eeeewewemelceee eee 141
Remeere enya. John EF) Knutson... ce ee ee ee eee ee cee eeeenews eee 165
Chapter 8
Aenerocecial Analysisof Structureand Process... 2... sce cee neces terre ewe eeneenen 169
Chapter 9
Turn-Taking,Synchronicity,and Reciprocity............ 00. er eer 199
Chapter 10
ae eanegement and Disruptionof Families... 2... 0. ce ce be ce ee tec eeeee means 215
Chapter 11
Similaritiesand DifferencesBetweenSocialAggressorsand Stealers ........... 0.00 cceceeee eee 239
Chapter 12
Beemer ervoneLovesThen, But... 6... eee ceeeece ene mentee tweeenonemenees 269
Chapter 13
Simeeencontactsaud Ireatment Outcomes... 0... ee ce ented eee beenneeees 293
as vinlh Go ings vpnSpeweedee ee ew 309
RTE TIOWHe BR) 2 WE 8 Be le 323
neo coo. eM ey gee a 325
I Ree es vowsain «Heok creehiesitoeeh acemaemee
se slaereeeees s 353
Tc 363

vil
e ee aed ore
iPi “an4
et el: %wisi"
wi
a mana
if aw
ts edn
%, ¢
.. Taf J [ io Paar
@me “ie ye,
wy “hw lesier

Pots wll gheee

if

yeh ened hh
=i
E: a
vie Pert ‘t yuy?“tT éi i
ant 8 wa

. 4 I toaknwaiter? be aiey Onceie

Kar | irae oon a

> a
rere ali EL I suraetyeinne? an.
¥

ar bm+TOewegALpAe
mca
afods,
rent ntiCEl
‘*
wD

% we Lie F 28 4 a
' Ce ee

. ide - ‘ ; :
Oye . een belie

. i> ‘ is . wyPater dye Lad : ug

*e ‘ vs Pe het > lal a)

j : vs
‘ * aod Pe “oy >a, wees i 4 ce iarie
sé \ , Pst
Pe ye ;
na
" : / pe J pate oh & a r= yer a.
eros ; 2a ml wea ' baal *y vos py
ee > aee ie het eae My rey fee «* IEE
Woe 2a ea kes

‘ re, i) aaa y
o \ - on A
q Pep ts io, AZ
ie
 Preface

This book is about antisocial children and their search on complex clinical and developmental
families. It is based on the study of over 250 fami- processes. Many features of this volume were de-
lies referred for treatment to the Oregon Social signed with this group in mind. I would also like
Learning Center (OSLC). For comparison pur- to communicate to them the sense of enjoyment
poses, we have also studied several hundred fami- and satisfaction that I have received from a life-
lies with normal children. Our efforts to treat the time commitment to clinical research.
families of social aggressors, stealers, chronic de- The second audience for which the book was
linquents, and abused children took us into designed is comprised of counselors, case workers,
homes, schools, courtrooms, and eventually back and therapists who have devoted themselves to
into the laboratory. This commitment to treat- working with antisocial children and their fami-
ment provided us with an intensive exposure to lies. Some of these clinicians may find that the
the family processes that are associated with vio- chapters on methodology (3) and theory (5, 6, 7,
lence among family members. 8, 9) represent areas for which they have only lim-
The picture of family structure and process that ited training and/or interest. However, the key
is presented here is based on multilevel assess- hypotheses about antisocial behavior come not
ment. Our studies included a minimum of six ob- from theory, but from clinical experience. The
servation sessions in the home. Each observation practitioner should find his or her clinical experi-
session provided over 1200 behavioral events de- ences reflected in several of the chapters (2, 10,
scribing the interactions among family members. 13). Several other chapters reformulate these ideas
This was further buttressed by extensive assess- into a social-interactional perspective of family
ment batteries including self-report, interviews, process (4, 11, 12). Most clinicians should find
achievement tests, laboratory procedures, and tel- this to be a useful way of thinking about these
ephone interviews. The perspective that has complex families. This perspective does, in fact,
emerged has both a clinical “feel” to it, plus some serve as the basis for our own treatment proce-
hard-headed components based on objective data. dures for these families.
There are many reasons for writing a book like It may be that one volume cannot serve two
this. On the one hand, I wished to communicate such disparate audiences. However, I think it is
our ideas about aggression and family process to possible. The central focus upon family members’
young graduate students. I would like to convince interactions has a protean quality to it that should
them that it is possible to carry out rigorous re- interest clinicians and researchers alike. The con-

ix
cepts in this book have very little to do with ab- is, therefore, not finished; I have simply decided to
stractions such as drives, perceptions, or ego de- stop changing it. Analyses completed since that
fenses that are typically encountered in books decision have already put some parts of this vol-
about family aggression. Coercion theory is based ume out of date.
on performance (i.e., what people do) as viewed There are four major sections in this book. The
from a bilateral or interactional perspective. The first section (Chapters 1 & 2) introduces the gen-
answer to the question of what causes family eral research strategy and the topic of the antiso-
members to do what they do seems to be found in cial child. Chapter 2 is essentially a review of the
the behavior of the other person. To understand research findings regarding the incidence of ag-
the aggressive child, you must understand—on a gression in children by age, sex, social class, fami-
moment-by-moment basis—how other people re- ly size, and birth order. A discussion of the differ-
act to him. Describing such analyses is the central ent kinds of antisocial behavior, and a brief intro-
task for this book. duction to concepts such as arrested socialization,
Originally, it was a third audience that social skills deficits, and deviancy-drift is also pre-
prompted me to begin this volume. I had intended sented. The second section (Chapters 3-7) begins
to write a single chapter as an introduction to a with a discussion of field observation procedures,
new book about family therapy. It was to be a including a review of the research studies relating
brief statement for the OSLC staff and the rather to observer agreement, observer bias, reactivity,
large cadre of consultants who have assisted us test-retest reliability, and validity. Chapter 4 ex-
during the last several years. The purpose of this amines the utility of aversive events in shaping
single chapter was to pull together the main ideas family interaction. This includes a consideration
and findings for a performance theory of family of their effectiveness as punishment for children’s
aggression. The “chapter” was neither brief nor behavior. In Chapters 5 and 6 the concepts of pos-
simple; it has now taken four years to prepare. itive and negative reinforcement are examined as
The therapy book has not yet been written. they apply to interaction sequences. The third sec-
It became apparent that, as a group, we had not tion (Chapters 8-10) introduces the reader to both
carefully thought through the complex issues asso- microsocial and macrosocial analyses as they re-
ciated with our attempts to apply concepts such as late to antisocial child behavior. The idea is that
positive reinforcement, punishment, reciprocity, disruptions in child-rearing practices can be
or negative reinforcement to sequential data. brought about by crises, marital conflict, and oth-
When applied to social interaction, these concepts er types of stress impinging on the family. This
take on a meaning that is different from the mean- section discusses our efforts to describe and mea-
ing that they had acquired in laboratory settings. sure these child-rearing practices. The last section
Redefining these concepts required an exploration (Chapters 11-13) describes the contribution of the
of several different bodies of literature. My studies procedures and the concepts to understanding—at
were enormously facilitated during a year spent at an empirical level—families of aggressive children.
the Center for Advanced Studies in the Behay- In Chapter 11, Social Aggressors and Stealers are
ioral Sciences at Stanford University. After my shown to be the outcome of different familial
return, the Oregon Social Learning Center staff processes and data differentiating clinical from
systematically protected my time so that I could normal samples are presented. In Chapter 12 the
complete what seemed to them (and to me) to be empirical definitions for the role of the mother and
an endless task. The process of writing this book father are reexamined. Chapter 13 provides a clin-
led to a continuous demand for yet more experi- ical description of the antisocial child and a specu-
ments, and for new analyses of existing data. lative typology of the families referred for treat-
This, in turn, led to the generation of new hypoth- ment. Finally, a brief review of the OSLC studies
eses which required new types of data. This book evaluating treatment effectiveness is presented.
 Acknowledgments

It became apparent in the mid-1960’s that in or- ception of the project. In addition to an unusually
der to study the problems of treatment and family high threshold for ambiguity, he has an uncanny
process that are associated with antisocial chil- ability to focus his attention on a problem and
dren, we needed the full attention of a sizeable then provide a useful (albeit often nonlinear) reac-
staff. These problems required a well-trained pro- tion. Over the years, our discussion of clinical
fessional staff of observers, therapists, and experi- cases and of the data formed a precipitate. I chose
menters who could work closely together for long to label the result coercion theory. It might well
periods of time. These exigencies led me eventual- have been a joint task, but John is currently occu-
ly to leave the university setting and design a sanc- pied with other problem areas such as child abuse
tuary that would be suitable for the long-term and observation methodology. As a result, the
study and treatment of aggressive families. We task has been undertaken by me.
were fortunate in being able to compete success- The demand for expertise in solving a wide
fully for a long series of research grants from the range of problems covered by our studies easily
National Institute of Mental Health. Most of them. exceeded my own personal competencies and
were from the section for Studies of Crime and De- those of the staff. For this reason we have relied
linquency with Saleem Shah as director. heavily on consultants from outside the group. In
Since the mid-1960’s, there have been a number the area of reinforcement theory, this included W.
of staff and colleagues who have contributed to Bricker, F. Kanfer, J. Knutson, R. Littman, W.
the development of a performance theory of coer- Sheppard, H. Shoemaker, and J. Straughan; in the
cive family process. I would particularly like to ac- area of punishment, R. Parke; in the area of devel-
knowledge the contributions of Gary Birchler, Pat- opmental psychology, R. Bell, E.M. Hethering-
ti Chamberlain, Joe Cobb, Tom Dishion, Hy ton, and E. Maccoby; in the area of sociology, R.
Hops, Richard Jones, Marion Forgatch, David Burgess, D. Elliott, R. Sparks, and M. Wolfgang.
Littman, Rolf Loeber, Rudy Lorber, Roberta Ray, We required “hordes” of statisticians and metho-
David Shaw, Magda Stouthamer-Loeber, John dologists such as W. Carlson, R. Dawes, R.
Vincent, Mark Weinrott, Robert Weiss, Tom Freund, D. Hartmann, P. Holleran, G. Sackett, R.
Wills, and Robert Ziller. There is also a shadow Sparks, and E. Thomas. My old friend R. Buehler
figure whose presence is felt throughout this book, should see reflected here the outcome of many
and who should be acknowledged early on—this is hours of discussions of social factors as they relate
John B. Reid, my close collaborator since the in- to family life. I am sure that each of these consul-

Xi
tants feels privately that we needed many more as- ministrative chaos. I was also given a series of
sistants. first-rate research assistants to help follow the
There is another group that has made an impor- leads that were generated during the writing of
tant contribution to this volume—the professional this book. These people included Marion For-
observers who not only collected the data but, in gatch, Sandy Kronsberg, Deborah Toobert, and
part, constructed the code as well: Betty Brum- Katie Whalen. To Trish McNeil, who typed the
mett, Rachel Condon, Peggy Gabrielson, LaVella endless revisions, I owe a very special debt. Final-
Garber, Vicki Halper, Jonnie Johnson, Shannon ly, there is that rare privilege which attends having
McCarthy, Debbie Toobert, and Irene Troup. a son take his father’s words seriously enough to
During the last four years, I have been granted publish them.
special status as a protected species by the OSLC In a very real sense, this book is the outcome of
staff. Chief among the protectors has been Will a group effort. I consider myself fortunate to have
Mayer, guardian of the castle keep, and Gary been a participant in such a process.
Morse and Mary Perry, who brought order to ad-

Xii
 Editor's Note

During the editing and production of this book I it was my task to hold the author back from feel-
have had the pleasure of working with several very ing the relief that comes with setting the pencil
talented people. Their efforts were an integral part down for the last time. In a book of this size and
of the improvements in this book that have taken complexity there were an enormous number of de-
place during the past year. I would like to give a tails to be worked out. This was accomplished
special credit to my associate editor, Cheryl Bru- during regularly scheduled sessions with the au-
nette, for her attention to the subtleties of the Eng- thor which took place at his home in the early
lish language that haunt us all. I would also like to morning hours. At times it seemed as though the
thank John Macioce at the Willamette Valley Ob- book would never be finished. But now, after see-
server for his patience in typesetting the text and in ing the page-proofs of the book, there is no ques-
constructing the numerous tables and figures. tion that the project was worthwhile. After all the
Many people other than myself have been in- stress that this book has generated for the Patter-
volved in making this book as accurate as possi- son family and our loved ones, it is our hope that
ble. In so doing, I am sure that we have disturbed the concepts in this book will help others to reduce
the dust from every book and file at the Oregon the stress and aversiveness that is found within
Social Learning Center. My thanks to Sandy many families.
Kronsberg and Trish McNeil for their assistance
in checking the data and the references that were Scot Patterson
used in the book. Editor
In pushing toward the completion of this book

Xi
 i a wy
x , ty?) ee Pere #4

SEL f . a vou 4
is) i _
' ° ci) br) au »
a] » t = ' Ww i”
“ "i oA

sth
ne

( a "a8

oa] BY ce af
; 6 ey ee eB
Vihi i . : y 1 2 -
d a. F y , ee: 4 9
{ 1
\ her he i \

ye

2*

‘ ' v

‘ iy LPR

; ba) we

age

“4 § ‘ ‘ i 4

| f }
t if ry '

= j’ ; ui ; : bdr Pow

} ' ' Pere7 }ts ee te? aft|

tt ‘ j ft aye REEL

‘an saad 4! weds satin

i tie vite i 4
; ee ; h, iM Cpie 1 11ft, i
, } fg PM)pd! Lai sou? iar
} eine re bai i wit so yer ;
 Chapter 1

Changing the Behavior

of the Scientists

Although the development of a coercion theory erly applied each new technique to the challenge of
is still in its beginnings, we have developed a way helping aggressive children. First nondirective play
of going about things that differentiates us from therapy and then psychodrama were tried as treat-
others who study aggressive children. Cairns ment strategies. But each technique, in turn, failed
(1979b) noted the similarities in perspective and to have an impact. At that time it seemed that I
procedures among investigators such as W. Har- needed advanced training in more sophisticated
tup, J. Gottman, and the Oregon Social Learning therapies; from 1953 to 1955 I interned at the
Center staff and labeled it “social interactional” Wilder Clinic in St. Paul. It was one of the best
(see Appendix 1.1). It is a pleasant surprise to re- psychoanalytic outpatient facilities for the treat-
ceive such a label, an indication that one has ar- ment of severely disturbed children. As part of a
rived at some station. But such a label also signals team effort, the children received intensive individ-
that one has moved from some other point. This ual and group therapy. The parents also received
chapter is my retrospective account of the forces treatment as a means of coping with their contri-
that governed this shift in perspective. bution to the process. This was the best there was.
Aggressive children are very dramatic in their In our enthusiasm, we wrote a paper about it (Pat-
behavior. This makes the phenomenon easier to terson, Schwartz, & Van der Wart, 1956). How-
study because one can see and measure what the ever, the outcome evaluations for such therapies
aggressive child does. Furthermore, there is an were in general agreement that the procedures
abundant supply of aggressive children in any were not effective (Levitt, 1971). The best was not
community. These children often constitute the good enough.
bulk of the case load for both child guidance clin- By the early 1960's, the writer’s clinical experi-
ics and residential treatment centers. Finally, the ences suggested that the most highly trained thera-
linkage between early antisocial problems and lat- pists were simply not effective in dealing with the
er careers in crime is such that it is possible to re- aggressive child. At the time it was not clear
ceive support for programmatic studies. These whether the problem lay in the theories of aggres-
pragmatic considerations suggest that the aggres- sion or in the treatment technologies. This was
sive child is a prime candidate for research. particularly puzzling because the therapy experi-
ences provided for the families seemed closely tai-
The Best Was Not Good Enough lored to fit what we thought aggression was all
As a young therapist in training, this writer eag- about. Aggression was thought to reflect the in-
adequacies of the child’s “defenses” against under- and bizarre social behaviors. The results presented
lying anxiety. Therapy with the child was meant in Ullmann and Krasner’s edited Case Studies in
to remedy this problem. The child’s anxiety and Behavior Modification (1965) did not prove the
crumbling ego defenses were thought to be distant new techniques worked. Rather, it was a bold
reflections of parental neurosis. Intensive therapy statement of what clinical psychology could be
for the parents was designed to correct their neu- like, i.e., it could become scientific. This earlier
roses. Generally, the senior staff spent their time work also suggested some alternative ways of
treating the anxiety reactions, parental neuroses, treating antisocial children. Why not reinforce the
or childhood schizophrenic reactions. The aggres- child for engaging in nonaggressive behavior? Al-
sive child was given to the student or the trainee. though this idea did not work, it was so novel and
We could only go on doing the best we were trained testable that it was an impetus to our work. Ru-
to do and observe our own and our students’ treat- mor had it that Ivar Lovaas had already tested the
ment failures. idea in a laboratory setting and shown it to be
There were, of course, occasional successes in feasible (Lovaas, 1961b).
which one could detect real changes in the child.
Furthermore, regardless of which treatment was Moving Into the Field
used, the majority of the parents reported that The operant format for social engineering had
“some improvement” had occurred. The solace several characteristics which gave the new work
brought by these successes was usually short-lived; an interesting twist. For one thing, operant psy-
follow-up reports generally showed that the child chologists insisted upon collecting data about the
was still out of control. The occasional successes observable behaviors of their clients! This led to
and the positive feedback from the parents sup- the birth of a whole new set of dependent variables
ported several generations of therapists, despite in clinical psychology. Who had ever heard of dai-
their failure to effect permanent changes in child ly counts of the number of towels hoarded as be-
behavior. ! ing one operational definition for bizarre institu-
tional behavior; or daily counts of “number of
Reinforcement Concepts bites” as relating to obesity; or “out of seat” as it
At the same time, there were some powerful related to classroom hyperactivity? The ingenuity
forces within psychology that had a bearing upon of investigators such as O. Lindsley, I. Goldia-
how we would proceed. Concepts from the rein- mond, N. Ayllon, M. Wolf, and N. Azrin seemed
forcement theories of Thorndike, Hull, and Guth- without limit. Their contributions to clinical as-
rie found their way into both developmental psy- sessment may turn out to be even more important
chology and personality theory (Miller & Dollard, than their contributions to the technology of be-
1958). These theories translated concepts that havior change. As therapists, most of us had been
were previously nontestable tautologies into an trained to attend to patient self-report. It was not
empirical language. This reformulation of clinical only our vehicle for behavior change, but also our
concepts led to a series of studies summarized in a primary means for measuring change. In contrast,
book which had an explosive impact on the field, the “behavior modifiers” were in the school, on
Bandura and Walters’ Social Learning and Person- the playground, and in the home counting events.
ality Development (1963). The new area was called By coincidence, they were learning to do what
“social learning”; it attracted a group of young in- their revolutionary counterparts in biology had
vestigators who produced an enormous number of been doing for decades. The ethologists had also
empirical studies. By applying social reinforce- been going into the natural environment to ob-
ment and modeling concepts to a wide range of serve child behavior (Blurton-Jones, 1966). Un-
theoretical and applied problems in developmental fortunately, this development had been largely ig-
psychology, the studies provided the basis for a nored by psychologists until the late 1960's.
new statement of the processes determining social Parent report had served for decades as the
behavior. mainstay for constructing theories about child
There was a parallel revolution in clinical psy- rearing and its impact upon child behavior. But we
chology as operant psychologists applied the prin- were looking for data which would give us better
ciples in B.F. Skinner’s prestigious Science and purchase on these questions. We began to see field
Human Behavior (1953) to investigations of clini- observation as a different scientific enterprise from
cal problems. Their work was based solely upon laboratory and clinical interview techniques. It
reinforcement concepts. The first studies provided was not necessarily a better view; it was simply a
data that described observable changes in behav- different one. Fiske’s (1974) conclusions were sim-
iors such as stuttering, hyperactivity, anorexia, ilar. He noted that there seemed to be two ap-
proaches to psychological phenomena. One was family and appeared at the door as planned. How-
to judge a person’s characteristics, and the other ever, the seven-year-old “out-of-control” child had
was to observe his actions and the relationship be- not been prepared for the sight of two burly men
tween them. Cairns and Green (1979) were also in horn-rimmed glasses appearing in his living
aware of this distinction. They discussed the work room and staring at him. He simply wheeled
by Radke-Yarrow and Waxler (1979) and others about, dashed out the back door and climbed a
showing little correlation between these two tree. My assistant and I stood nonplussed until the
methods; yet they note that this does not mean the mother gave us a cup of coffee and later sent us
new method (observation) is good, and the old home (see Illustration #1).
one (judgment/ratings) is bad. It is, rather, that Another problem with field observation was to
these methods represent two very different kinds learn how to describe what could be seen. What
of information about a person, and each has a dif- form should these descriptions take? P. Schoggen
ferent contribution to make to our understanding and his colleague, R. Barker, had previously con-
of child behavior. For example, rating procedures vinced us that observation data should be cast in a
take into account the ability of human beings to sequential form. The relatively high yield from
make complex judgments based on heterogeneous such a strategy was reflected in their publication,
sources of information. Such ratings represent an Stream of Behavior (1963). The importance of the
integration of new information with the previous- sequential format was also emphasized by the
ly held implicit (or explicit) theories of the rater NIMH researchers studying aggressive children
(Wiggins, 1973). However, as Cairns and Green (Raush, 1965). However, the most compelling ar-
(1979) point out, if the experimenter is concerned gument for a sequential format was that it fit what
with questions relating to stable qualities of the the clinician could see in the home. There seemed
child, then the process of synthesizing these rat- to be a natural flow to family interaction. The be-
ings may have some unique properties to which havior of one family member seemed to follow
observation systems are not well suited. They go that of the others in an orderly fashion. Even at
on to suggest that observation systems are well de- this early juncture, it seemed that some events
signed for a different set of tasks, such as analyz- were patterned, and that these patterns were re-
ing those characteristics of the child (a) that are peated again and again. We began to notice that
temporary, (b) that vary as a function of the con- each interaction sequence was not unique. Fur-
text or setting, and (c) that depend on the quality thermore, the repetitious quality of familial inter-
of social interchange: action seemed particularly descriptive of unpleas-
ant episodes. It seemed that it should be possible
“. . direct observations of behavior can be the to analyze such patterns.
key for identifying how actual behaviors are elic- Simultaneously, a number of investigators, such
ited, maintained, and organized. Such informa- as R. Wahler at Tennessee, developed a complex
tion may be critical for explaining the processes observation code system tailored to families of ag-
that regulate social patterns, as opposed to de- gressive children. At OSLC the development of a
scribing the social patterns. Both goals—descrip- primitive behavioral coding system required three
tion of individual differences and explanation of years of pilot work (Patterson, Ray, & Shaw,
how differences arise—are worthy goals, but they 1968). It was then used routinely to collect base-
are different. Accordingly, observation techniques line data in the homes of problem and normal chil-
are useful (or even indispensible) for identifying dren. The fact that the code data described se-
the dynamic controls that operate during the quences of events led eventually to a fundamental
course of social development. shift in our perspective regarding familial aggres-
“Conversely, ratings would not be useful in sion. But this happened very gradually. We intro-
yielding information about contextual, interper- duced first one term, then another, into our dis-
sonal, and developmental processes. Why not? Be- cussions, e.g., stimulus control, antecedent, prob-
cause these sources of variance are typically elimi- ability, structure, process, interactional unit.
nated at the first stage of data recording” (Cairns Eventually journal editors began commenting that
& Green, 1979, pp. 221-222). our shift was perceptible to them; in many cases it
was not clear to them why we were asking certain
However, there were no handbooks detailing questions. In retrospect, it is clear that we were
how to make this shift to field observation. As a continuously re-examining traditional assump-
result, our first attempts at field observation were tions and concepts. The definitions of concepts
not very successful. For example, in one of our kept shifting to accommodate this new data. Ideas
first home visits, we made an appointment with a about observer reliability, the trait of aggression,
Illustration #1: A High-Level Beginning
the definition of setting and the temporal and situ- analogous to the changes which took place in
ational stability of aggressive behavior gradually many other aspects of our thinking. It was not that
took on very different meanings. the old concepts were discarded, it was that their
definitions had been altered.
Current Perspective
These transitions were never easy. There were A New Dependent Variable
no boundaries showing when a shift began or Most of us in the mid-1960’s were convinced
ended, and no staff meetings to decide when it had that a simple frequency count was the preferred
become so perceptible that it deserved a label. dependent variable. It was the perfect means for
Looking back, it seems that the times of greatest showing that one event occurred more often than
movement were accompanied by the most humor. another. Given that it was necessary to demon-
For example, our preoccupation with moment by strate changes in frequency over time, then these
moment shifts in what a child was doing led to a data could be transformed to rate measures. It was
profound interest in the impact of the immediate simple and direct. For most questions these meas-
social environment on the child’s behavior. We ures served very well. However, rate or frequency
had been trained to believe that the environment concepts are a clumsy way of describing the rela-
had an effect, but hardly on a second by second tion between events. The interactional view sug-
basis. We wondered if it was reasonable to break gests that most social interactions are structured in
environmental impact into six-second units. With the sense that one event is correlated with another.
much amusement research seminars were held in In fact, some methodologists such as D. Hart-
the early 1970’s to consider the possibility of de- mann (personal communication) make a persua-
veloping a “six-second personality theory.” How- sive argument for lag and cross-lag correlations as
ever, over the next several years the realities of an- a reasonable statement for what these relations
alyzing interaction sequences that were based on are.
these data led us to accept the idea. Gradually, we became used to thinking of the
The findings from our longitudinal studies com- structure of social interactions in terms of condi-
pleted during the 1970’s reiterated the fact that the tional probabilities. Given that one event occurred,
trait of aggression was stable over long periods of what was the likelihood it was preceded by this or
time. But when viewing the same trait at a micro- that behavior of the other person? If the child
social level (as sampled by the code categories) the whines and the mother gives him a hug, what is
boundaries of that concept were becoming the likelihood that the child will continue? With
blurred. At that level the definition of aggression this flexible language we can place behavioral
had been expanded to include “bilateral” effects. events in sequence like beads on a string. We are
As a result, the definitions now included measures limited only by the quantity and quality of the se-
of initiations, and measures of reactions to initia- quential data itself. Probability theory lends itself
tions by both members of the dyad. To understand beautifully to the problem of functional relations.
a child’s aggression we now had to measure how The reader who plans an investigation of this ma-
others reacted to his aggression. terial would be well advised to provide him- or her-
Later, we became interested in the possibility self with a solid background in these matters with,
that forces outside the dyad were impinging upon for example, the primer by L.D. Phillips, Bayesian
these bilateral interactions (see Chapters 4 and Statistics for Social Scientists (1973). Many inter-
10). The analyses quickly identified a set of “mac- actional concepts are based on probability theory
rosocial” variables which clearly altered some as- but require some additional effort, e.g., Markov
pects of the microsocial process. Initiations and re- chains, signal detection, Guttman Scalogram, and
actions reflected, in part, long standing disposi- information theory. Because these concepts are
tions. But it was becoming clear that they could such useful tools for the new questions being
also reflect, at least on a short-term basis, the im- raised, I find myself reaching more frequently for
pact of such non-trait related variables as illness, Coombs, Dawes, and Tversky’s Mathematical
loss of income, or other crises. Psychology (1970) than for the more familiar texts
The concept of a trait had now become a many on parametric statistics.
splendored thing. It retained its traditional com- Probability theory and mathematical psycholo-
ponent of stability across time. But it now in- gy are not new to the social sciences. They are
cluded new concepts and data from interactional “new” to persons such as myself. Currently, there
psychology which focused upon the process by are a number of individuals who are exploring the
which traits are changed. These changes in our interface between these mathematical concepts
way of thinking about the trait of aggression were and the concepts emerging in interactional psy-
chology. Individuals such as John Gottman, Uni- Generating Hypotheses
versity of Illinois, Gene Sackett, University of
What variables determine shifts in social inter-
Washington, Ewart Thomas, Stanford University,
actions? Social psychology had only indirectly con-
Roger Bakeman, University of Georgia, and Don
cerned itself with such questions. For this reason,
Hartmann at the University of Utah are the “meth-
we proceeded with the study of interaction unen-
odologists” for this new area. Its “theorists” in-
cumbered by very much theory. There were, in ad-
clude many of the above and Ross Parke, Univer-
dition, few “facts” to keep in mind—a comfortable
sity of Illinois, Willard Hartup, University of Min-
or uncomfortable state of affairs depending upon
nesota, and Robert Cairns, University of North
one’s taste for uncertainty. The goal in construct-
Carolina.
ing a performance theory was to express the rela-
Performance Theory tive contributions of the variables controlling on-
going behavior by the amount of variance ac-
Since its inception as a field of study, psycholo- counted for. Within this framework, some varia-
gy has been preoccupied with the general problem bles would account for variance relating to differ-
of human learning. Most of the early theorists, ences between subjects in the performance of ag-
such as Skinner, Hull, and Guthrie carefully differ- gressive events. It is possible that the variables ac-
entiated between learning and performance. In counting for differences across settings would be
that context, what is learned is not necessarily per- different.
formed. With the exception of Hull, the bulk of The investigator begins by selecting a pet hunch
their writings deal with the analysis of variables about variables determining social interaction. Ul-
thought to determine “learning.” As a result, most timately, the score for a favored variable must be
of the variables concerning “laws” of human be- tallied. The crucial test for any variable is the
havior relate to learning; few, if any, are con- amount of variance on any of the three dimensions
cerned with changes in performance per se. (among subjects, among responses, and across
Questions about the structure of social interac- time) accounted for. If it does not contribute in
tion led us away from traditional concerns about this sense, then the variable must be dropped from
human learning. Our major preoccupation was the performance theory.
not with the question of how children learned ag- In constructing a performance theory we did not
gression. Instead, the question became, “Why do examine any particular set of hypotheses or varia-
they perform aggression at different rates?” bles. Because we were more familiar with rein-
If one accepts the persuasive arguments of mod- forcement theory, that literature served as a start-
eling theorists, such as Bandura (1977) and Kanfer ing point for hypotheses conceptualizing aggres-
and Phillips (1970), much of what is learned about sion in families. However, it soon became appar-
social interaction is that which is modeled by other ent that conventional terminology such as stimu-
persons. The intricacies of what the child must lus, response, and reinforcement provided an
learn about complex social behaviors and lan- awkward fit to many aspects of social interaction.
guage are presumably acquired in this manner. If As a result it was necessary to generate a supple-
one accepts this position, and the present writer mentary set of terms that described antecedents,
does, then by the age of five most children in this target events, and consequences in a language that
society have learned most of the garden-variety ag- was as neutral as possible.
gressive behaviors employed by children. With Clinical contacts with families of antisocial chil-
this in mind, the following questions arise: Why dren provided another important source of hy-
does one child perform aggressive behaviors at potheses. As therapists we were committed to the
rates ten times higher than another child? Why task of helping these people. This exposed us to
does one child perform aggressively at a higher some aspects of aggression in family interactions
rate in one setting than another? Why does he use that were not sampled by our observation code.
a particular aggressive response more than anoth- These omissions from our assessment device led us
er? The answers to these three questions constitute to search for new methods for measuring what the
a performance theory. The variables which relate therapist could “see.” As a result we began to in-
to these questions outline what is meant by a coer- clude measures of parent monitoring, setting con-
cion theory. It is a set of statements about pain- sequences, crises, insularity, and sidetracking,
control techniques employed by one or both mem- which fit the clinical phenomena but were not de-
bers of a dyad. Each variable impacts either the rived from any particular “theory.” Ultimately, the
performance of the other person or the perform- constraint operating in selecting a specific variable
ance of the target subject. revolved around the question of how well it could
 be measured. In most instances we had to innovate aggressive?” If the new variable can account for a
our own means of assessment. significant amount of variance in the child’s per-
formance, then it is added to the “theory.” The
- An Applied Mission overall effect is an accumulation of information
In this section, I argue for the need to carry out and a gradual shift in one’s perception of what the
mission-oriented programmatic work on applied phenomenon is all about.
social problems. At present, much of the work in In the final analysis, sequential data can be used
developmental and child clinical psychology is a only to generate functional relations. As noted by
piecemeal affair. The majority of the studies are Parke (1979), functional relations are not proof of
carried out by investigators who publish a single causality. For that reason, the development of so-
paper on a topic and then move on to a new prob- cial interaction theories requires the employment
lem. Quite properly, these investigators select of both observation data and experimental manip-
problems which are amenable to current proce- ulations. In other words, the analysis of sequential
dures. They also tend to select problems which fit data from field observations generates hypotheses,
within some favored paradigm. For example, cog- but the analyses themselves provide only a weak
nitive and reinforcement theories are currently test of the hypotheses. Ultimately, the concepts
possible candidates in psychology for status as generated by our observations require experimen-
paradigms. If one is trained to work within either tal and longitudinal designs to establish causal
of these frameworks, the strategy becomes that of connections. In the work described in this volume,
searching for phenomena fitting the explanation. considerable effort has been made to develop ex-
The phenomena that do not fit the explanation are perimental procedures designed to test hypotheses
set aside in favor of those that will. One is not on coercion processes. Whenever possible, the
committed to a problem, rather, to a single ex- manipulations are carried out in the natural set-
planation of behavior. As a result, one can find ting.
studies using reinforcement or modeling concepts The recent publications by Petrinovitch (1979)
to study aggression, but as a whole the findings and Fiske (1974) alerted me to the fact that we
lack coherence. It is as if one were looking through have not drifted so far. Both emphasize the need
a kaleidoscope with lovely p values and F values for sampling behavior in natural settings. Petrino-
floating within ANOVA’s that are whirling across vitch (1979) also talks about the utility of a con-
the screen. They do not fit together in any satisfy- cept in terms of variance accounted for. From his
ing pattern. Each group of investigators has its Brunswickian viewpoint, relationships between
own language and its own set of publication variables are described in probabilistic terms. So
sources. It is noteworthy that they rarely cite the wheel comes full turn; the emerging perspec-
another group’s publications. tive that we thought was so unique was. antici-
As an alternative, the OSLC group first selected pated several decades ago in the writings of Bruns-
a problem and employed those models that seemed wick.
necessary and most useful to study the problem. It
makes little difference which paradigm one initial- Conclusion
ly selects. Each investigator begins with the para- The focus of our research is to understand why
digm of choice and discovers its inadequacies. The children are aggressive. Because we were most fa-
present author was better trained in reinforcement miliar with reinforcement theory, that literature
than in the research areas relating to human mem- served as a starting point for hypotheses conceptu-
ory, perception, and cognition. For this reason, I alizing aggression in families. Our clinical contacts
began the study of aggression with reinforcement with families of antisocial children provided
concepts and applied them first to studies of child another important source of hypotheses. As thera-
and peer interactions (Patterson, Littman, & Brick- pists, we were committed to the task of helping
er, 1967), and later to parent-child interactions as these people.
well (Patterson & Cobb, 1971, 1973). There was, Early in our efforts to treat families, we found it
however, never any doubt that other models inev- necessary to measure change in families. To meas-
itably had to be introduced to solve some aspects ure change, we decided we should observe in the
of a problem. At OSLC the addition of new varia- home itself prior to and following treatment. Lat-
bles goes on continuously. The process begins er, we added many more assessment devices, but
when one or more people become convinced of the the important decision was to observe. The se-
necessity for introducing a new variable to explain quential form in which these data were cast led to
a particular resistant wrinkle in the data. The goal analyses of family interaction which altered the
is always to solve the problem, “Why are children questions we had about the nature of aggression.
 Gradually, we became used to thinking of the Footnotes
structure of social interaction in terms of the con-
1. As it is ordinarily used, parent global report
ditional probabilities that described the perform-
of treatment outcome constitutes a trap for the un-
ance of aggressive behavior. In constructing a per-
wary therapist. There have been a series of studies
formance theory, variables were selected on the
reviewed in Chapter 3 of this volume analyzing
basis of the amount of variance accounted for.
parent global report data. The studies consistently
These variables described the use of pain-control
showed that parents of deviant children were sig-
techniques in families with aggressive children.
nificantly biased to report improvement even
The label that we have given to the resulting social
when more objective data showed no changes oc-
interactional perspective is coercion theory.
curred. As a rough figure, 60% to 70% of the par-
ents will give such biased reports. The studies do
not prove that parents lie to therapists. They simp-
ly assert that parents are human and wish to retain
Acknowledgments their hopes for the child’s eventual adjustment.
The writer wishes to thank the members of the However, the findings mean that, regardless of
OSLC writing seminar, who carefully critiqued an one’s treatment technique, the feedback to the
earlier draft of this manuscript: P. Chamberlain, therapist will be that he or she is helping the ma-
M. Forgatch, R. Lorber, D. Littman, D. Moore, J. jority of the clients! This, in turn, is apowerful re-
Reid, and D. Toobert. Sandra Kronsberg fulfilled inforcer for continuing what one has always done
a similar role for a later version. in the past; therein lies the trap.
 Satis (FF ta ay rave 10 5 Ga i eee
Sefer 26 yilie Poteet Tee lerae? 1 Fycpeyty

Meese Cn ooh

aid. Bret wwii \

58) 2) gi) OT}
dt sA)oie ' ber pe Pees y

Bijts ii j | 97

«peibdink
j ft

f<
 _ Chapter 2
Abstract
The research literature is reviewed regarding the incidence of aggression in children by age, sex, so-
cial class, family size, and birth order. Support is presented for the stability of measures of aggression
across time and settings.
Traditional factor-analytic studies of children’s conduct-problem behaviors suggest the feasibility
of considering two major patterns for antisocial behavior, Social Aggressive and Stealer. Most chil-
dren referred for treatment for antisocial problems can be classified into one or the other of these pat-
terns. It is assumed that these patterns may require separate explanations as to the kind of family pro-
cesses which produce them. They may also require different forms of treatment.
It is hypothesized that extremely antisocial children represent a form of arrested socialization.
Their development is retarded in two different respects. First, the rate and form of the problem child’s
deviant behavior represent that which is acceptable for 3- and 4-year-olds. It is assumed that problem
children have been allowed to continue this behavior. Second, the problem child is often deficient in
social skills, e.g., work, self-management, and relation to peers.
A deviancy-drift hypothesis is also presented. It relates the child’s relative deficiencies in social
skills and his ranking on antisocial behaviors to his later adjustment problems. The child who is both
extremely aggressive and extremely retarded in the development of social skills is thought to be at risk
for later problems as an adult; i.e., extremely antisocial children do not outgrow their problems.

10
 Antisocial Children

The purpose of this chapter is to draw together skilled, but they are particularly unskilled in their
the research literature regarding antisocial chil- use of punishment for deviant behavior. Their
dren. However, these materials are not being held punishment doesn’t work. Many of them are also
up to view in an altogether dispassionate manner. inept at providing modeling and reinforcement for
Instead, the literature is examined for themes and their child’s prosocial behaviors. As a result, some
trends which would be of particular relevance to of these problem children are not only extremely
coercion theory. This brief introduction is in- aggressive, but they also lack basic social survival
tended to forewarn the reader as to the kind of fil- skills; e.g., they cannot engage their peers, they
ter being applied. cannot work, and they tend to have poor academic
Aggressiveness, such as hitting, is viewed as a skills. Family management skills can be taught; in
subset of the coercive techniques employed by fact, at OSLC training in these skills is the central
family members to alter each other’s behavior. Ex- focus for the treatment of families with antisocial
tremely aggressive children who are referred for children. When these skills are not practiced, the
treatment employ these techniques at very high level of aggression for all family members tends to
rates in a variety of settings. There is a strong case rise. The net result is that the system begins to
to be made for thinking about child aggression as. drift toward a kind of psychological anarchy. As
a trait, a disposition to react aversively which is this happens, the key figure, such as the caretaker,
stable across time and settings. This trait can be becomes increasingly dysphoric. Family members
measured, and the performance score can be ex- also become increasingly dissatisfied with the fam-
pressed in terms of the frequency or probability of ily. As a result, they tend to avoid each other and
an aggressive response. However, the determin- engage in fewer shared activities. Thus, the entire
ants for this score reflect the dispositions of the family is caught up in a process which produces
child and the persons who are targets for this ag- changes in mood and self-esteem. In families of
gression. In effect, the trait score reflects the inter- antisocial children some parental “neuroses” can
actions which occur in that setting. An extremely be thought of as the outcome of a process rather
aggressive child has a supply of “victims” who re- than its cause.
act to him in a predictable manner. Each of these Children who drift into extreme deviancy and
victims contributes to the performance level of the retardation of social skills are thought to be most
aggressive child. at risk for adult maladjustment. Generally, the ex-
Parents of aggressive children are generally un- tremely aggressive child does not outgrow his

11
problems. While the nature of the problem may Textbooks on aggression typically cover an enor-
change over time, his status as deviant does not. mous range of phenomena such as homicides in
These children constitute a major social problem. families, hijacking airplanes, schoolyard bullies,
It is from their ranks that the chronic delinquent is marital conflict, muggings, riots, and wars. Some
drawn. As Garmezy (1976) points out, the antiso- ethologists (Eibl-Eibesfeldt, 1974) even include
cial child is at even greater risk for major. adjust- Eskimo song competitions as instances of aggres-
ment problems as an adult than is the child of a sion! All of these phenomena might satisfy that
schizophrenic parent. Thus it is important that we part of the definition which includes inflicting pain
understand who these people are and where they upon victims. However, this writer assumes that
come from. each of these phenomena should be studied sep-
arately. It seems unlikely that, in the near future,
What Is Aggression? there will be any general theory of aggression
The generally accepted definition for the term which covers all of these events. It is more useful
aggression involves one person inflicting pain to establish an empirical foundation for a small
upon another. Some writers stipulate that the act subset of aggressive phenomena, e.g., those events
must have the goal of injuring the other, i.e., the which occur in family interaction. Understanding
aggressor intended to hurt his victim (Berkowitz, the variables relating to family aggression may or
1973a; Feshbach, 1970). Others, such as Buss may not contribute to an understanding of inter-
(1961), Bandura (1973), and the present author, personal aggression in other settings, such as those
do not include intention in the definition of ag- found in football “games,” ghetto riots, or bar-
gression. room brawls. Each of these phenomena will prob-
Regardless of which stance one adopts, there ably require its own techniques of study and data
are difficulties. There are many aversive events analysis; each probably involves very different de-
that impinge upon us during our lives: for exam- termining variables.
ple, a dentist drills your teeth; you are awakened Within the family those events will be labeled as
by loud noises from a jackhammer being operated aggressive which can be shown to be both aversive
on the street beneath your hotel room early in the and contingent. Assessment of either characteristic
morning; or your children have turned the TV up is acomplex process. Field studies in the homes of
very loud. In all of these instances someone is do- aggressive children produced 14 categories of
ing something which, to you, is aversive. Are all to events which seemed aversive and could be coded
be classed as instances of aggression? Writers such reliably (Reid, 1978). A priori ratings by parents
as Berkowitz and Feshbach would be quick to showed surprising consensus about which events
point out that the dentist, jackhammer operator, were perceived as aversive and which were not
and the children did not necessarily intend to in- (Jones, Reid, & Patterson, 1975). They agreed
flict injury. Therefore, none would be labeled as that if these events were experienced by a victim,
aggressive acts. At this point some psychoanalytic they would be perceived as painful. A functional
writers might also raise the issue that dentists se- analysis showed that these aversive events effec-
lect their occupation because of an unconscious in- tively suppressed ongoing prosocial behaviors;
tent to inflict injury. The difficulty in using this i.e., these aversive events functioned like punish-
concept lies in the complex issues involved in ment when contingent upon ongoing prosocial
measuring intentions. The idea itself has great ap- interactions (Patterson & Cobb, 1971). The
peal. It is the focal point for much of our great lit- details of these analyses will be presented in later
erature, and it is an integral part of our legal struc- sections (Chapter 6). For the moment, it is suffi-
ture. However, we have no ready means for meas- cient to say that there was a modest convergence
uring intention. Even those writers who include it between the a priori ratings by parents and the
as a variable in their definition of aggression sel- functional analyses of the suppression effect.
dom introduce it as a variable in their empirical Once an event is classified as aversive, how can
studies. we determine if it is being used contingently? To
There is, then, the problem of distinguishing be classed as contingent, it would have to be
aversive events that are aggressive from others that shown that an aversive event reliably follows some
are not. There is no definitive resolution of this is- behaviors and not others. In natural settings this
sue. However, we have adopted two strategies can be done by first collecting data which accu-
which we think facilitate empirical research on ag- rately describe interactional sequences. If Sister-
gression. The first is to focus upon only a small Tease is followed by Brother-Hit in a repeated pat-
subset of the total spectrum of human behaviors tern, then it seems likely that the two events are
which could potentially be classed as aggressive. contingent. To determine whether the sequence is

12
 random or contingent, one can compare the base- the termination or withdrawal of the antecedent
rate value of Brother-Hit to the likelihood that he constitutes a reinforcement for the response. In
_ will hit when he is teased. Given a significant dif- family interaction about two-thirds of the child’s
ference between the two values, then one can as- coercive interchanges can be classified as instru-
sume a contingent relation; i.e., Brother-Hit is mental aggression. The antecedents for the behav-
more likely to occur given a specific behavior of iors are neutral or positive, and the most likely
the sister. Aggression in families does not consist outcome is a continuation of attention or some
of fortuitous events. This is a key assumption in other prosocial behavior by other family members
coercion theory. (see Chapter 5). The remainder of the child’s ag-
The dentist does not apply the drill contingent gression is in reaction to attacks by other family
upon any particular behavior of the client. Simi- members. This is analogous to Knutson’s irritable
larly, the aversive events provided by the jackham- aggression.
mer operator and the children watching TV are A wide spectrum of aversive events are employed
not contingent upon a specific behavior. They in a contingent fashion by family members. It is
would, therefore, be categorized as aversive but assumed that all of these events reflect a common
not necessarily aggressive. process. From this perspective the code categories
One might object that an increased understand- Whine, Tease, Disapproval, Yell, Humiliate, Non-
ing of microsocial events, such as teasing, scold- comply, and Hit all differ in topography yet are
ing, and whining is an unwarranted concern with members of the same set. Members of this set
trivia; usually aggression refers to high-amplitude share several characteristics: (1) they are aversive;
behaviors, such as hitting, stabbing, or shooting. (2) they are used contingently; (3) they produce a
However, a second key assumption in coercion reliable impact upon the victim; and (4) the reac-
theory is that the innocuous, garden-variety avers- tion of the victim has both a short-term and a
ive events, such as teasing and scolding, can, un- long-term effect upon the aggressor. The rates at
der certain conditions, escalate to high-intensity which these events occur differentiate normal fam-
aggression. One of the goals of the present formu- ilies from families of antisocial children. The func-
lation is to understand how this process operates tional relations between these events and the reac-
within families. Under what conditions will esca- tions they provoke also differentiate these samples.
lations in intensity come about? Throughout this volume I will sometimes refer
Buss (1966a), Bandura (1973), and Berkowitz to this process as “pain control” and at other times
(1978) distinguish between instrumental aggres- as “coercion.” The term coercion, however, has a
sion which produces an external reinforcer, and somewhat broader coverage. It includes hypothe-
hostility for which the reinforcer is a pain reaction ses which relate family management disruptions to
by the victim. The cross-cultural observation the performance of aggressive behaviors. In addi-
study by Lambert (1974) showed that 80% of the tion, there are functional relations between micro-
time children’s aggression was instrumental, and social structures and mood changes. There is also
5% of the time it was hostile. Hartup (1974) a relation between macrosocial stressors which
showed that while object-oriented aggression impinge from outside the family and their assumed
characterized a large proportion of the aggression impact upon both family management skills and
of the preschool-age child, it decreased from ages microsocial processes. Finally, coercion includes
4 through 7 years. He also speculated that threats the escalation hypothesis, which describes the pro-
to self-esteem produced hostile attempts to injure cess by which a dyad moves from the exchange of
others, i.e., person-oriented aggression. Other rather innocuous aversive events to the higher
writers, such as Rule and Nesdale (1976) have amplitude aggressive behaviors which characterize
pointed out that anger is more likely related to child and spouse abuse.
person-oriented aggression and therefore a better The term coercion also includes the child prob-
differentiation might be between anger and non- lem behaviors usually labeled “aggressive.” We
anger. will examine the literature to see what is, in fact,
John Knutson has made a differentiation which normally included in the definition of that term.
neatly summarizes our findings from analyzing However, coercion includes some behaviors that,
family interaction (Knutson & Hyman, 1973; Fol- while aversive, would ordinarily not be included
lick & Knutson, 1977). He distinguishes between in the definition of aggression, e.g., Noncomply,
irritable aggression and instrumental aggression. High Rate, Cry, Ignore, Whine, and Yell.
Instrumental aggression is controlled by its positive In early drafts of this volume I used terms such
consequences. Irritable aggression, on the other as conduct-problem and out-of-control to label
hand, is elicited by its aversive antecedent; that is, the identified problem child. M. Rutter (personal

13
communication) pointed out that the latter term tortions even for such simple tasks as estimating
includes in its history implicit assumptions about a event frequencies. These studies will be reviewed
breakdown in inner controls, with the symptoms in detail in Chapter 3; suffice it to say here that in-
representing uncontrolled motivational forces. To terview data lack the precision necessary for the
avoid that kind of confusion, I deleted the phrase study of functional relations between one event
from this volume and substituted in its place the and another.
term “antisocial.” However, D. Olweus (personal These considerations led us to seek a new kind
communication) has pointed out to me that this of data for the study of familial aggression. The
term also has a history which gives it a meaning sample to be studied were families referred for
that does not altogether fit my present purpose. treatment because one or more of their members
He notes that the term has been traditionally ap- had been labeled as aggressive.! After conducting
plied to older delinquent youths. I agree with his an interview and administering the usual self-re-
point but believe that the term antisocial can serve port questionnaires, the writers took notebooks in
equally well when applied to younger children. hand and went into the homes of these families.
From the perspective of coercion theory, behaviors The code system described in Chapter 3 was de-
such as Whine, Yell, Hit, and Steal, all involve signed while sitting in their kitchens and living
other persons as targets. In a sense, all coercive be- rooms. Three years of trial and error led to the
haviors involve a victim and the term antisocial construction of a 29-category code system which
would seem to be appropriate. could reliably document some of what could be
As we pursue these studies, I believe it will be seen in these families. Fourteen of the categories
shown that there are a number of different pat- sampled aversive behaviors occurring in family in-
terns of coercive behaviors. When a child displays teractions. The methodological studies on reliabil-
one member of such a subset, he is very likely to ity, stability, observer bias, observer presence ef-
perform the others as well. Furthermore, there is fects, and validity are summarized in Chapter 3.
enough consistency in these interrelations among It is literally impossible to devise a code that de-
the symptoms that the patterns recur regularly in scribes all aversive events found in all families. It is
referrals for treatment (e.g, Achenbach & Edel- also true that most families employ some aversives
brock, 1978; Patterson, 1964). These patterns are which are so subtle that they can only be detected
important because they reflect different familial by other family members. It seems reasonable,
processes. In the present volume we explore only however, to begin by designing a behavioral
two of the most salient subsets, Social Aggressors coding system which samples the most obvious
and Stealers. However, we are convinced that events. Such a system would accurately reflect
there are other more subtle forms of coercion, what is going on in most families most of the time.
e.g., Immature and Hyperactive patterns, where The 14 aversive events, their definitions, and base
the victim is less clearly identified but contributes rates of occurrence are shown in Table 2.1. The
nevertheless to the coercive process. data were based on a sample of distressed families
Throughout this volume the term antisocial referred for treatment to the Oregon Social Learn-
child will be used as a synonym for the coercive ing Center (OSLC). Six to ten home observations
child. In this context, it will ordinarily refer to the were conducted in which each of the family mem-
two typologies which we have studied, Social Ag- bers was sampled for a minimum of 60 minutes
gressors and Stealers. By implication, if later (Reid, 1978).
analyses were to show immature and hyperactive It may be that our culture assigns children and
children to be coercive, then the meaning of the parents different types of aversive events as part of
term antisocial would be expanded to include their roles. Disapproval by mother occurred, on
them as well. the average, once every three minutes; mothers
used Destructive and Physical Negative less often.
The Assessment of Coercive Events Some categories, such as Cry, Whine, Yell, Non-
The interview and the questionnaire are the tra- comply, and Destructive (of property) were used
ditional techniques for gathering data on aggres- primarily by children. Others, such as Command
sion in families. While such data have a place in Negative and Ignore were used primarily by par-
the present volume, they have only limited value ents. Some aversive events were used often by all
for the study of interaction. As R. Cairns (1979a) family members.
pointed out, the study of process requires sequen- For many purposes the data from the 14 cate-
tial data. Data from interviews and ratings are ill- gories can be summed to form a single score, Total
suited to such a task. There is good reason to be- Aversive Behavior (TAB). The score serves as a
lieve that parental interviews provide serious dis- general measure of level of coerciveness. Its test-

14
 Table 2.1
The Definitions and Mean Base Rates (per minute) of
Aversive Behavior for Members of Distressed Families
(adapted from Reid, 1978)

Code Problem Mothers Fathers Siblings

Categories Definitions Boys N=27 N=18 N=54
INE=27
Command Command in which immediate compli- -008 0.046 0.023 .008
Negative ance is demanded, aversive consequences
threatened, sarcasm or humiliation
directed toward receiver.

Cry Whining or sobbing sounds. .019 0.000 0.000 .024

Disapproval Verbal or gestural criticism of another 134 0.314 0.182 .120
person’s behavior or characteristics.

Dependency Request for assistance when person is .007 0.003 0.000 .008
obviously capable of doing task himself.

Destructive When a person damages, soils, or breaks .031 0.000 0.000 .011
something.

High Rate Physically active repetitive behavior .044 0.000 0.000 .042
which is likely to be annoying.

Humiliate Embarrassing, shaming, or making fun .020 0.011 0.015 .015

of another person.
Ignore Intentional and deliberate nonresponse .005 0.023 0.019 .010
to an initiated behavior.

Noncomply A person does not do what is requested .092 0.011 0.009 .064
of him in response to a command, com-
mand negative, or a dependency within
12 seconds of the request being made.

Negativism A neutral verbal message delivered in a 1 LS 0.019 0.012 .059

tone of voice which conveys an attitude
of “Don’t bug me.” Also included are
defeatist statements.

Physical Physical attack or attempt to attack .042 0.019 0.003 .021

Negative anotherperson.
Tease Act of annoying, pestering, mocking, or .050 0.001 0.014 .028
making fun of another person.

Whine A slurring, nasal, or high-pitched voice. .036 0.001 0.000 .052

The content of the statement is irrele-
vant.
Yell Shouts, yells, or loud talk. .057 0.009 0.000 .036

SUM OF MEANS FOR ALL

FOURTEEN AVERSIVE BEHAVIORS -660 .457 LUT) 498

MEAN TOTAL AVERSIVE* .047 .033 .020 .036

*This mean is obtained from the sum of the fourteen means. Typically, this score is obtained by summing up fourteen scores for each
subject and calculating the mean of that distribution.

15
retest correlation for a 12-month period was .74 which give data on the likelihood of crying by age
df= 7; p<.05): level of infant. Given an awake condition, the
These codes catalog the aversive events used probability of such an aversive behavior seems
by family members to control each other’s behav- very high during the first few weeks and then grad-
ior. Their employment constitutes an operational ually diminishes. Presumably, there are important
definition of the components in the pain-control individual differences among infants in both the
process. It is one thing to construct a system which base-rate values and variability of this behavior. It
can be used to code aversive family interactions; would be of interest to know if fluctuations in cry-
however, to prove its worth as a measure of family ing covaried with changes in the mother’s mood.
aggression, it is necessary to demonstrate the val- In addition, longitudinal studies are needed to de-
idity of these categories in at least three different termine if mothers who quickly learn to meet the
ways. First, the aversive categories must signifi- child’s needs are also more adept at child manage-
cantly differentiate between identified aggressive ment when the child is 2 or 3 years old.
boys and matched nonproblem boys. Antisocial In his review of preschool aggression, Cairns
boys were shown to perform these behaviors at (1979b, p. 206) makes the interesting observation
significantly higher rates than normals by Patter- that half of the interchanges among 1- and 2-year-
son (1976) and by others using comparable coding olds could be thought of as “aggressive.” He cites
systems (Bernal et al., 1976; Lobitz et al., 1976; the Holmberg (1977) study, which showed that
Snyder, 1977). Other analyses have shown that this amounted to little more than taking a toy
mothers and siblings of aggressive boys differed from another child. However, even at this early
significantly from comparable members of non- age, they had learned not to deal with adults in
problem families (Patterson, 1980b; and Chapter this manner; only 5% of their interactions with
10). Second, it must be shown that these 14 code adults were aggressive. By the time the children
categories which assess aversive events covary were 2/2 years of age, only 20% of their interac-
with measures of what parents mean by the term tions with peers were aggressive. This trend is in
“antisocial.” To test this, the parents of families keeping with our assumption that the socializing
referred for treatment were asked to give daily re- process increases the child’s prosocial interaction
ports on whether or not certain problem behaviors and steadily decreases that which is coercive. As
of concern to them had occurred. The mean daily we shall see, parents seem to accept 3% to 5% of
reports for a two-week baseline were then corre- the preadolescent child’s coercive behavior as be-
lated with the mean TAB score based upon the ob- ing within normal limits. Lambert (1974) found in
servation data. In three different samples the two summing data from six cultures that children play-
measures of general deviancy correlated .69 ing outside the home, primarily with peers, aver-
(df=14, p<.01), .46 (df=21, p<.02), and .59 aged about 10% coercive interaction.
(df=31, p<.001). Other studies relating to the With the exception of the investigation by Hen-
validity of the coding system are reviewed in ry and Sharpe (1947), studies in nursery school
Chapter 3. In the final analysis, the validity of the settings generally indicate a decrease in children’s
coding system as a measure of familial aggression coercive behavior as a function of age (Green,
relates to the extent to which these data facilitate 1933; Dawe, 1934; Hartup, 1974; Ricketts,
our understanding of the process. This cannot be 1931).2 For example, Dawe (1934) studied chil-
summarized in a single index. It is, in fact, the dren in a free-play situation. The average quarrel
function of this volume to demonstrate that anal- was a brief 23.6 seconds; it tended to last longer if
yses of interactions involving these 14 aversive the setting was outdoors (35.1 seconds) rather
events provide a proper empirical base for under- than indoors (17.4 seconds). For younger chil-
standing aggression in families. dren, 78% of the quarrels were possession related;
this dropped to 38% for the 4- and 5-year-olds.
Demographic Data Relating to Attempts on the part of the teacher to intervene
Child Aggression had relatively little impact upon duration. The
“attacks” were usually directed at children of the
Age same sex, and were usually centered around con-
It is assumed that infants instinctively employ flicts over possession of an object. The coercive
aversives such as crying. This may have survival behavior usually consisted of pushing, pulling,
value in that an infant can use the techniques of and physical negatives (pinch, hit, kick). The cor-
punishment and negative reinforcement to train relation between age and frequency of quarrels
the mother in more effective parenting behaviors. was —.41.
To date the writer has been unable to find studies Hartup’s (1974) review of his and others’ find-

16
 Figure 2.1
Age and Sex Differences in Frequency of Outbursts
(fromGoodenough,
1931)

35%

— Boys
bg] Girls 30%
Sexes combined

25%

of
outbursts
observation
hour
Frequency
per 20%

15%

10%

5%

0%

Age in years

ings suggests that the kind of aggression employed there, the 18-month-old child performed at the
also changes as a function of age. His data showed highest level, averaging an outburst about every
proportional decreases in conflicts concerning ob- seven or eight hours. Holmberg (1977, cited in
ject possession or blocking (instrumental aggres- Cairns, 1979b) found the peak frequency of child
sion) but only slight changes in person-directed, aggression toward adults in the nursery school to
retaliatory, and hostile outbursts. In Lambert’s occur at 24 months. Goodenough found a steady
(1974) six culture study, the mean figure for the decrease in conflicts through age 6. These findings
range of age groups was 80% instrumental aggres- are in keeping with the observation study by Fawl
sion and 5% hostile. (1963), who studied less intense conflicts. He
Studies of preschool children in the home showed an average of 5.4 disturbances per hour
showed a comparable decrease in aversive interac- for children aged 2 to 11 observed in the home. In
tions as a function of age. In the classic study by his study, the correlation between age and conflict
Goodenough (1931) 45 mothers kept daily diaries frequency was —.76, demonstrating that younger
for children ranging in ages from 7 months to 7 children performed aggressive behavior at higher
years 10 months. The diaries were kept for a mini- rates. Again, a majority (80%) of the outbursts
mum of seven days and a maximum of 133 days. lasted less than one minute.
The data for the frequency of the children’s anger The OSLC code was used to collect observation
outbursts are summarized in Figure 2.1. As shown data in the homes of 44 children from normal fam-

i
 Table 2.2
Changes in Total Aversive Behavior Scores by Age
Mean TABScoreby Age |
OSLC F
Samples 2-4 5-6 7-8 9-10 11-12 13-15 Values df
Social
Aggressive 5.205 4:135
N 41 22 29 Mf 11
Mean 223 1.18 99 82 .68
S.D. WE! .86 53 85) 18
Stealer TAT S372
N 4 7 21 22. 19 5
Mean 1.83 .80 40 ay) .60 56
S.D. .93 .66 Bip 27 .60 44
Normal 329. 3:48
N 12 15 10 US: 6 4
Mean .56 2 il gle} .22 mull
S.D. 50 46 ape 516) sls} .06
Mean TAB Score
at 90th percentile oD 393 “S)8) All — =
*p<.05
***p <001
— indicates insufficient N for calculation.

ilies. This was essentially a working-class sample indicated that these behaviors occurred, then the
(see Chapter 3 for details). As summarized in Ta- family was categorized as a member of the Stealer
ble 2.2, the findings from six to ten observation sample. The details for the classification proce-
sessions showed a significant decrease in TAB dures and sample characteristics are summarized
scores as a function of age.* The children aged 2 to in Chapter 3. If the baseline observation data in
4 performed at an average rate of .56 responses the home provided a TAB score of .45 and none of
per minute. By the time the child was of school age the referral symptoms included stealing, then the
his rate was down to .32 responses per minute, family was placed in the Social Aggressor sample.
and then further decreased to about one aversive The data in Figure 2.2 suggest a significant de-
every eight minutes. The results for the analysis of crease in the level of aversiveness as a function of
variance showed these changes with age to be reli- age for both of the clinical samples. Younger re-
able. ferred children were significantly more coercive
The level of coerciveness for normal 2- to 4- than the older children. Children who were Steal-
year-olds is comparable to the mean baseline TAB ers were intermediate in their level of coerciveness,
score of .75 reported for the first sample of antiso- performing more coercively than Normals but less
cial children referred for treatment at OSLC (Pat- so than the Social Aggressors.
terson, 1974a). This finding is reiterated in Figure These findings for age differences relate to the
2.2. It can be seen there that the average 10- to general hypothesis that antisocial children repre-
11-year-old antisocial child referred for treatment sent a form of arrested socialization. It is assumed
is performing at the level of coercion which our that a longitudinal study would show that the to-
culture finds acceptable in 2-, 3-, and 4-year-olds. be-labeled deviant preschool boy may perform at
The children from distressed families were gen- only slightly higher rates than nonproblem boys
erally referred to OSLC for treatment because the prior to age 3. However, as he matures he contin-
parent or a community agency had perceived sev- ues to perform at about that same level. If he con-
eral behavior problems. If this included “stealing,” tinues at this level long enough, someone in the
and the baseline Parent Daily Report (PDR) data community will finally label him as “deviant.” Pre-

18
 Figure 2.2
Mean Aversiveness by Age and Sample

ce Normal N=44
eal Steal ers N = 78

V Social
. Aggressors N=140

Note: The data for Stealers

!OEE
SG_
RA
a nd
SSF
NSS
NE
Age Grouping in Years
Social Aggresso rs have been c ombined for the 2-3-4 age group.
 Table 2.3
Percentage of Out-of-Control Behaviors by Age
(from Rutter, Tizard, & Whitmore, 1970)

Age

Behavior 10 years 11 years 12 years

N=(1018) N=(996) N =(1050)

Parents’ Data
Often fights 202, 11.0 8.0
Temper loss at least once a week 4.7 3354 2.6
Wet bed at least once a month 4.1 2.6 1.8
Restless and overactive 34.6 30.2 PudN
hes

Teachers’ Data
Wets pants 1.0 0.5 0.1
Fidgety 18.9 14.7 18.1
Steals 3.0 3.1 1.2

sumably his parents are unable and/or unwilling ders. This subgroup was asked to participate in
to impose the sanctions which parents of normal more intensive assessment procedures. Those
children impose for these same behaviors. questionnaire items which described out-of-con-
Survey data for preschool- and school-aged chil- trol behaviors and which showed significant
dren show a general trend toward reduced levels changes with age are summarized in Table 2.3. Of
of antisocial behaviors as a function of age. Grif- particular interest were not only the age-related
fith (1952) studied children ages 6 through 14. changes but the relatively high base rate for some
Parent ratings of child aggression showed a de- of these behaviors. According to the parents’ per-
crease with age for such behaviors as impudent, ception, about one in eight of the 10-year-old chil-
noisy, and anger outbursts. (It is interesting to dren fights often. The base rate for temper tan-
note that the child’s self-perception data showed a trums and stealing were about one in 30. It would
curvilinear relation between age and aggression.) be useful to have correspondingly high-quality
Similar decreases were reported in the longitudinal data for a U.S. sample of children of all ages.
study of maternal reports of quarrelsomeness and In sum, the data indicate a general trend toward
temper tantrums by MacFarlane, Allen and Hon- reduced levels of antisocial child behaviors as a
zik (1962). Their small, highly selective sample function of age (2 through 12 years). These trends
was followed from age 21 months through age 14 are fairly consistent in home and school settings
years. Some problems, such as negativism and ir- and across cultures.
ritability, showed large fluctuations but no regular
trends as a function of age. Observations in six Sex of Child
cultures showed a decrease in frequency of as- Reviews of the literature on childhood aggres-
saults when comparing ages 3 to 5 with ages 6 to 8 sion consistently note higher rates of physical ag-
(Whiting & Whiting, 1975). However, when this gression for boys than for girls (Mischel, 1966;
measure was combined with measurements of Feshbach, 1970; Maccoby & Jacklin, 1974). For
horseplay and insults, the combined score did not example, observation studies in preschool settings
covary significantly with age. showed boys generally performed at higher rates
The Isle of Wight study by Rutter, Tizard, and than girls (Dawe, 1934; Henry & Sharpe, 1947;
Whitmore (1970) is perhaps a classic in its own Hartup, 1973). The large-scale study by Smith
right. The total population of 2,199 ten- and elev- and Green (1975) showed that in 13 of 15 nursery
en-year-olds was assessed by a multiple-screening schools, boys tended to be more aggressive than
procedure. First, parent and teacher question- girls. Dawe (1934) found a mean of 13.5 quarrels
naires were used to identify a subgroup of 268 for boys and 9.5 for girls. While the differences
children who might be at risk for psychiatric disor- found in other studies were generally of this order,

20
 they were not always significant (Henry & more likely than girls to be referred for treatment
Sharpe, 1947). of conduct-problem behaviors such as stealing, ly-
These overall performance trends are of inter- ing, and fighting (Wolff, 1967).
‘est. There is reason to believe that even at the pre- It is not only adults who perceive more boys
school level, boys and girls may differ in the kind than girls to be extremely aggressive; the children
of aggression which they display (Maccoby & themselves perceive these sex differences. The
Jacklin, 1974). While the findings are not always cross-cultural studies of adolescents’ self-report
consistent, many of the studies point to a greater data showed that girls described themselves as
use by girls of verbal attacks. Furthermore, the more conforming to adult values (Devereaux,
two sexes may be differentially responsive to set- 1970). The careful review by Hood and Sparks
ting constraints within the nursery school. For ex- (1970) showed consistent sex differences in self re-
ample, D.S. Patterson (1976) found the overall ported crime for high school students in both
trend was for boys to be more assertive than girls. America and Scandinavia. The self-report fre-
However, when only boys were placed in the Art quency of delinquent behavior for boys was two to
Center of the room, they quickly escalated to high- four times higher than that for girls. Comparable
intensity aggression, whereas girls did not. For sex differences characterize the self-report data for
boys, art defined a setting which related to escala- adult men and women: e.g., the rates of assault
tion in aggression; for girls, this same setting was were 49% and 5% respectively; for car theft, 26%
apparently defined very differently. Studies which and 8%; for burglary, 17% and 4%; and for rob-
systematically sample settings, kinds of aggres- bery, 11% and 1%.
sion, age, and sex of the child could tell us impor- The findings from the Isle of Wight survey de-
tant things about how our culture programs ag- tailed parent and teacher concerns for preadoles-
gression for children. cent children (Rutter et al., 1970). All of the find-
These fragments of data suggest that sex differ- ings shown in Table 2.4 portray significant differ-
ences in some kinds of aggression appear very ear- ences between boys and girls. The categories de-
ly in the nursery school. The data from Good- noting high rates of activity indicated that boys
enough (1931) showed that in the home the mean were significantly more involved than girls. In the
rate of outbursts per hour for boys was consistent- home, these differences were modest in magni-
ly higher for all age groups (2 through 7 years 11 tude; however, in the classroom, they were exacer-
months). Incidentally, at ages 12 through 23 bated. Teachers perceived that almost twice as
months it appears that girls have slightly more an- many boys as girls presented problems which in-
ger outbursts than boys. However, the Holmberg volved restlessness and fidgeting. Parents and
(1977) study found no difference by sex for chil- teachers agreed that two or three times more boys
dren 24 months and younger. than girls were involved in stealing, truancy,
Maccoby and Jacklin (1974) review the evi- fighting, lying, bullying, and destructiveness.
dence for biological variables which may account These findings reiterate the theme that boys are
for some of the variance in these early sex differ- perceived as more out of control; but they also
ences. The evidence is sufficient to give pause to suggest that there is a complex interaction be-
any who might wish to take a strictly environ- tween setting, sex of child, and kind of out-of-con-
mental stance on the issue of early sex differences. trol behaviors.
Hopefully, the next generation of studies will take Reid (1978) provided data for base rates of var-
these variables into account as they systematically ious responses by age and sex for an extremely
explore the process by which these differences oc- small sample of 27 normal families observed in
cur. The question is, at a molecular level, how their homes. The data are summarized in Table
does a culture arrange for 5-year-old boys to learn 2.5 separately by age groups. Note that the same
how and where to be physically aggressive? How family may have contributed to more than one age
do we teach 5-year-old girls mot to use physical group. Of the 30 comparisons, ten were signifi-
force? cant. With certain interesting exceptions, the gen-
Sex differences among older children. Typically, eral drift of the findings re-emphasized that boys
teachers and parents refer more boys than girls to are more out of control than girls. Again, it was
child guidance agencies for treatment of conduct the younger boys who were more likely to hit,
problems. This tendency was noted in the classic push and shove. This theme was noted earlier in
study of several thousand referrals during the the studies of nursery school aggression. In the
1920’s by Ackerson (1931). Studies carried out home, young boys were also significantly more
three decades later in a different country showed a likely to Disapprove and Yell. Young girls were
similar pattern. British boys were significantly significantly more likely than boys to be Negative,

pai
 Table 2.4
Percentage of Out-of-Control Behaviors by Sex
(from Rutter, Tizard, & Whitmore, 1970)

Parents’ Report Teachers’ Report

Problem Boys Girls Boys Girls
Behaviors N =(1564) N =(1500) N =(1743) N =(1683)
Poor concentration Dal 18.2 S951 25.0
Restless/ overactive 6220 25.5 16.8 8.1
Fidgety/squirmy 14.4 9.8 233 10.9
Steals S14) 2.6 3.4 1.6
Truant 1.8 0.2 Del 0.7
Fights diSe2: By) 13.6 4.4
Lies 16.1 OF, 9.4 29
Bullies 6.7 4.0 7.4 2.1
Destructive 7A 1.4 3.0 0.7
Tempertantrum (at leastoncea week) 4.4 PIES) — _
Irritable 34.5 27.9 11.7 6.5

Tease, and Cry. The longitudinal study by Eron et al. (1971) actu-
The mean TAB scores for the older children ally showed higher rates for the high-status boy.
showed a general decrease in rate for both sexes. The dependent variable was a peer sociometric
Older normal girls were significantly more likely measure of aggressiveness. The Isle of Wight study
than their male siblings to Cry and Whine. These showed few parent-reported deviant-child beha-
behaviors hardly seem like attack behaviors; they viors related to status differences for boys or girls.
are more likely to be a victim’s technique for deal- However, in that study the teacher data indicated
ing with attacks by other family members. Older that low-status boys and girls had a significantly
male siblings were, on the other hand, significant- higher incidence of poor concentration and ab-
ly more likely to Noncomply, Tease, and High sence from school for trivial reasons. Fighting for
Rate. boys, and stealing and truancy for girls, showed
similar correlations.
Socioeconomic Status
Olweus’ (1978) study of bullies in schools also
In the research literature there is no consistent failed to find a correlation between social status
relation between socioeconomic status and aggres- and bullying or peer ratings on aggressiveness.
sive behavior in children. While the early work on The distribution by status for the OSLC sample
delinquency suggested that the working class con- suggests a disproportionate sampling of working-
tributed a disproportionate amount to official de- class families. It is likely that these differences re-
linquency rates, the review by Hood and Sparks flect the dynamics of local referring agencies more
(1970) showed that a significant confound existed than an estimate of the incidence of aggressiveness
in these analyses. Using official delinquency rates in Eugene’s general population. Presumably, mid-
(offenses known to the police), the early studies dle- and upper-class aggressive children are re-
showed an average ratio of about 5:1 when com- ferred for treatment to local physicians rather than
paring low to high socioeconomic status groups. to OSLC.
If, however, the estimate was based upon the ado-
lescent’s self-report, then the ratio would have to Family Variables: Size, Father Absence,
be revised to about 1.5:1.0. Birth Order
If one moves away from delinquency rates as a Family Size. It seems reasonable to suppose that
measure of antisocial behavior, then the difference as the number of children in the family increases,
among social classes diminishes further. The re- there would be a proportionate decrease in the
view by Feshbach (1970) showed no consistent re- amount of time parents spend with any given
lation between child aggression and social status child. The number of crises and parental responsi-
when teacher and parent reports served as criteria. bilities would also be likely to increase to the point

22
 Table 2.5
Sex Differences in Normal Sample
(from Reid, 1978)

6 years and under 7 years or over

Beys Girls Boys Girls

Behavior N=(15) N=(7) N =(37) N =(22)

Mean S.D. Mean S.D. Mean S.D. Mean S.D.

Total Aversive Behavior | 581 (.543) 514 (.256) .221 .149) 52815) (.214)
Command .043 (.036 .022 (.021) .040 .036) 043 .028)
Command Negative .005 (.007 .000 (.000) .003 .006) 002 .005)
Cry .016 (.024 114* = (.125) .000 (.000) 018* .037)
Disapproval fil lye (.088 .054 (.045) .083 .054) 083 .054)
Dependency .006 (.009 .008 (.016) .003 .008) 001 .002)
Destructive .009 (.013 .007 (.008) :003 .008) 001 .002)
High Rate .042 (.063 .020 (.024) O12 .027) .001 .003)
Humiliate .004 (.007) .000 (.000) 002 (.005) .004 (.009
Noncompliance .088 (.070) .078 (.048) .044* .051) .021 (.033
Negativism .013 (.030) a0Sil n(GOS2)) .024 .024) .032 (.050
Physical Negative 0292 (E0277) .013 (.013) .009 .016) .008 (.017
Tease .020 (.040) .040* = (.038) 028% .039) .005 (.011
Whine .124 (.176) 37 (.169) 009 (.013) WAS (O55
Yell HOS (1.98) .013 (.015) .005 .010) .008 (.019

* p value for “t” was < .05. However, in that the same family may have contributed a child to each age group, the assumptions under-
lying the “t” statistic may have been violated. For this reason, the reader should interpret these findings with caution.

where many tasks ordinarily carried out by the covariations between family size and the follow-
mother might be delegated to an older sibling. ing: destructiveness, fighting, disobedience, lying,
From a social learning viewpoint, this would im- bullying, stealing, and temper. Girls showed com-
ply an increasing contribution of siblings to the so- parable relations for destructiveness, fighting, ly-
cialization process. ing, and bullying. Rutter goes on to point out that
The analysis by Patterson (1980b), and those there were significant correlations between IQ and
presented in later sections of this volume, indict these same variables; furthermore, IQ correlated
the siblings as major determinants for aggressive inversely with family size. Although the data were
behavior. In large families the parents are less able not shown, he claimed that partial correlations
to track and punish their children’s out-of-control which took IQ into account showed greatly re-
behavior. These hypotheses are in keeping with the duced relations between family size and out-of-
findings from the study by Zajonc (1976). Zajonc control behaviors reported by parents and teach-
found an inverse relationship between family size ers:*
and intelligence. The assumption was that in The observation study by Burgess, Kimball, and
larger families members spent increasing amounts Burgess (1978) provided a direct test of the hy-
of time interacting with children rather than with pothesis. They observed interaction in the homes
adults. Presumably this results in less stimulation of normal and abused children. The data showed
and enrichment, which in turn is related to lower a significant correlation (.43) between family size
IQ scores. and physical aggression. Furthermore, with in-
A number of writers, such as Douglas (1966), creasing family size, the relative amount of inter-
have commented that delinquent youths tend to action with siblings increased while the relative
come from larger families. The Isle of Wight study -amount of interaction with parents decreased.
provided additional support for the relationship The general tenor of the findings suggests that
between antisocial behavior and family size (Rut- family size and rates of deviant behavior covary. It
ter et al., 1970). For boys there were significant would seem appropriate to begin investigating the

26
 Table 2.6
Mean TAB Scores for Intact and Father-Absent Homes

(from Horne, 1980)

Distressed Sample Nondistressed Sample

Family
Member Father Absent Intact Father Absent Intact

Mean N Mean N Mean N Mean N

Target Child 981 (9) .412 (15) 223 (9) PANETT (16)
Mother .681 (9) .362 (15) "329 (9) E235 (16)
Older Sister .988 (4) 129 (8) .148 (5) .132 (6)
Younger Sister .754 (6) FoyIs} (7) — .414 (7)
Older Brother 544 (6) 165 (6) .134 (5) 128 (4)
Younger Brother — — 420 (5) .627 (7) }

— indicates that there were insufficient data for analyses.

determinants for the covariations. For example, child caused the divorce. However, the Hethering-
the interesting study by Circirelli (1976) showed ton et al. (1976) and the Wallerstein and Kelly
that mothers’ reactions to their children in a labor- (1976) studies showed that the increase in antiso-
atory setting varied as a function of whether the cial child behavior followed the divorce. These
children had older siblings at home. In large fami- studies do not tell us why the break-up of the
lies, the mothers delegated part of their maternal home produces child aggression. The hypotheses
role to an older sibling. The differential reactions detailed in Chapter 10 would have it that crises
presumably reflected this shared role. such as divorce or illness can disrupt the practices
Broken homes. Traditionally, this variable has of family management skills. It is suggested that
been thought to be associated with antisocial chil- even a temporary abeyance in monitoring and set-
dren. In our culture the likelihood of living in a ting consequences for violating house rules is like-
broken home is now rather high. As shown in the ly to be accompanied by increases in antisocial
studies reviewed by Hetherington, Cox, and Cox child behavior.
(1976), approximately 10% of the children in our There is another aspect of the broken home
culture are being reared in single-parent families. which is not clearly identified in the existing litera-
Most of the early studies on the effect of broken ture. Our clinical experience suggested that father-
homes did not control for the contribution of so- absent homes were associated with the more ex-
cial class. The carefully designed survey of Rutter treme forms of antisocial behaviors. Two analyses
et al. (1970) did provide such controls. Their find- by Horne (1980) demonstrated that a case could
ings showed a general relation between child psy- be made for this hypothesis. First, he compared
chiatric disorders and broken homes. However, the TAB scores for each of six members from dis-
within these disorders the incidence of broken tressed and nondistressed families that were either
homes for the antisocial child (25%) was not sig- intact or father-absent. The findings are summar-
nificantly greater than the incidence reported for ized in Table 2.6. Only the ANOVA values for the
the other psychiatric subgroups. A broken home, intact and father-absent comparisons were in-
then, seems to relate to an increased likelihood for cluded. In the clinical sample, the effect of father
a disturbed child (neurotic or antisocial). absence was to produce TAB scores twice as high
A study of divorced families by Hetherington et as for members from intact families. In the nondis-
al. (1976) showed that following the absence of tressed sample, father absence was not related to
the father, there were significant increases in out- significant changes in TAB scores. This suggests
of-control behavior, especially for boys. The study that it is not father absence, per se, which deter-
also demonstrated the causal relation between mines the increase in coercive performance. I think
broken homes and antisocial behavior. For exam- it is the interaction between father absence and the
ple, one could assume that having an antisocial disruption of family management practices. In

24
 families where these management skills are mar- contradictory to the well known findings pre-
ginal, the father’s leaving may tip an already pre- sented in Mischel’s (1968) early formulation. The
_ carious balance. This assumption is supported by studies he reviewed at that time demonstrated lit-
the next step in Horne’s analysis. The fathers from tle consistency across time and no consistency
intact families were asked to absent themselves across settings for traits such as aggression. Devel-
from their homes on alternate evenings. A com- opments since then have caused Mischel (1979) to
parison between the three sessions in which he was modify his earlier extreme position; but his dra-
present and the three sessions in which he was ab- matic (1968) statement served a useful function as
sent showed no consistent effects for the 16 non- a goad for research in this area.
distressed (intact) families. Some family members The samples upon which Mischel based his
showed nonsignificant increases in TAB scores (1968) formulation were essentially normal. The
during absent nights (mother and older brother), present writer would argue that children at the ex-
while others showed decreases. Similar compari- tremes (95th percentile or greater) on traits such as
sons for the distressed sample showed that, with withdrawal or antisocial do not outgrow these be-
the exception of the problem child, all family haviors. I also believe that these children are con-
members were more coercive on father-absent sistent across settings in their practice. Their style
nights. of coping with social situations is such that people
The findings from the Horne (1980), Hether- react to them in a highly predictable manner.
ington et al. (1976), and Wallerstein and Kelly These reactions, in turn, “maintain” the with-
(1976, 1980) studies lead us to assume that the drawn or antisocial behavior across time and set-
presence of the father contributes to the effective tings. The traits constitute a kind of “life style,” or
practice of family management skills. A broken at least a consistent orientation towards social in-
home places the family at risk for a disruption in teraction. It appears that the pattern of the “diffi-
these practices. If such a disruption occurs, then cult child” is discernible at a very early age. For ex-
the likelihood of the broken home producing an ample, the MacFarlane et al. (1962) longitudinal
antisocial child appears to be very high. study showed a correlation of .73 (df=32, p<
Ordinal Position. Coercion theory implicates .001) between the total number of problems at age
siblings and peers as important agents for teaching 6 and at age 14.
social aggression (Patterson, 1980b; Patterson, The issues raised by Mischel (1968, 1979) re-
Littman, & Bricker, 1967). The ideal training sit- quire answers. The sections which follow review
uation has a supply of vulnerable victims, such as the studies related to the stability of the trait of ag-
younger siblings, and some effective models, such gression across time and settings.
as an antisocial older sibling or parent. This
would be compounded if the parent is inept in Stability of Coercive Behaviors
managing coercive interchanges. If this scenario Across Time
is correct, then the antisocial child would most It is assumed that the coercive child becomes
likely be a middle child, the second most likely be- highly skilled in the use of aversive events to alter
ing either the youngest or oldest. Only-children his immediate social environment. In this fashion,
should be the least likely. These post hoc specula- he alters the ongoing social behaviors of other per-
tions are consistent with the findings from reviews sons with whom he interacts.° With maturation,
of the literature. The highly aggressive child tends the mode of coercion might change; but his com-
to be the middle child (Rutter et al., 1970; Ander- mitment to using these techniques at high rates
son, 1969). Rutter’s data showed that 48.6% of (above 95th percentile) remains relatively con-
antisocial children were in the middle ordinal stant.® As he acquires verbal skill he might, for ex-
rank; a comparable figure for that position for the ample, replace whining, crying, and yelling with
control group was 22.2%. As compared to the sarcasm and argumentativeness. A later section
control group, youngest and only-children were will present data which shows that the antisocial
underrepresented in the antisocial group. child is retarded in the development of many social
skills. This further locks him into the necessity of
using coercive means to impact his environment.
Aggression as a Trait Gradually, an increasing number of people report
As used here, the term “trait” refers to a disposi- experiencing him as unpleasant. At this point he is
tion to respond which is somewhat stable across finally labeled “deviant.”
time and settings. The studies reviewed in this sec- It is assumed that extremely coercive children
tion represent a substantial body of literature will persist in using coercive techniques over time.
which supports this contention. This is somewhat Presumably these techniques maximize short-term

2)
 Figure 2.3
Stability Correlations as a Function of Age at First Test and Follow-up Interval
(fromOlweus,1979)
Interval in years (T, —T,)
0S 1.0 LOSE 2EO 22532035, 4.0) 45S: (5:0 10912 RIS als eelWe
Gel 7h eis 19 20 21

(in
T,
Age
years)
at

payoffs and are unlikely to be given up unless the these time intervals. The mean stability correla-
child is trained in delay of gratification and in pro- tion across these studies was .63; the mean inter-
social techniques for bringing about changes in val for the retesting was 5.7 years. Figure 2.3 sum-
other persons. After the age of 9 or 10, special marizes the data from these studies. Some studies
training by the family and/or community is very employed measures of aggression at several differ-
unlikely. The extremely coercive child, therefore, ent intervals. All of these findings are included. As
is very likely to become an extremely coercive Olweus points out, the evidence for test-retest sta-
adult. bility for measures of aggression is on a par with
D. Olweus (1976, 1979) has performed the use- comparable measures for children’s intelligence.
ful service of pulling together 16 studies of chil- As a general case, I think an argument can be
dren’s aggression which employed a longitudinal made for the long-term stability of measures of ag-
design. His review covered samples ranging in age gression after the ages of 5 or 6. For example, the
from kindergarten to high-school-aged students. early study by Kagan and Moss (1962) showed a
The follow-up periods ranged from 6 months to correlation of .61 (df=87, p<.001) between as-
10 years. The measurement procedures varied sessments at age 6 and at age 14 of child aggres-
from observation, peer sociometric, to parent and sion toward mother. The comparable correlation
teacher ratings. The findings consistently demon- for behavioral disorganization (cry, temper, de-
strated the stability of children’s aggression across struction, rages) was .52 for boys and —.03 for

26
 Table 2.7
Stability of Extremely Aggressive Subjects

Given a peer sociometric aggressive Then at ten year follow up the proportion of subjects at or above:
score at Grade 3, at or above the:
the same percentile for Grade 13 the median for Grade 13

95th percentile 38.5% 100%

90th percentile 32.0% 88%

85th percentile 32.3% 79%

girls. A study by Gersten et al. (1976) also sup- children do not “outgrow” their problems. The ex-
ports this hypothesis. In their study, 732 children tremely deviant child may train others to react to
ranging in age from 6 to 18 years were followed him in such a way as to maintain his deviant sta-
for five years. Data were obtained from interviews tus; but I think there are also several other factors
with their mothers. The findings again showed im- which contribute to this continuity. First, there are
pressive stability for cluster scores on aggression the previously noted social skill deficits thought to
(correlation .50) and for delinquency (correlation characterize some antisocial children. These defi-
.44). Of even greater interest was the finding that cits have implications for one’s ability to form last-
there is greater stability for these conduct-problem ing relations, and for one’s ability to make a liv-
clusters than for the neurotic clusters. In fact, the ing. Given extreme deviance plus social skill defi-
authors suggested that neurotic disabilities were cits, then the extremely antisocial child would be
best left untreated, letting spontaneous remission likely to grow up to be an adult who, at best,
receive due credit as the treatment agent. makes a marginal adjustment. The findings from
We would expect to find that the more extreme the large-scale, 14-year, follow-up study by West
the child’s aggression score, the more extreme the and Farrington (1973) were in keeping with these
child’s aggression at follow-up. This hypothesis is speculations. A boy rated by teachers as extremely
contrary to the idea that extreme scores tend to re- aggressive at age 9 had a .14 chance of being a vio-
gress toward the mean. The reason for believing lent adolescent delinquent! In fact, 70% of the vi-
this hypothesis lies in the assumption that ex- olent adolescent delinquents had been rated as ag-
tremely deviant persons, whether antisocial, de- gressive at age 13; 48% had received this rating at
pressed, or schizophrenic, reprogram their social age 9. The findings are being replicated and ex-
environments. There are two lines of support for tended by the programmatic studies of children at
this idea. risk at UCLA (Rodnick & Goldstein) and at Min-
The first set of findings are from the ten year nesota (Garmezy). M. Shea (1972) described a
follow-up study of Eron et al. (1971, 1974). They large-scale follow-up study by A.J. Hafner and W.
obtained peer sociometric ratings at Grade 3 and Quast of clients referred to the psychiatry section
again at Grade 13. The stability correlation for ag- of the University of Minnesota Medical School. A
gression was a modest .38 (p<.001). I assume sample of 284 adolescents referred for general
that the extremely antisocial children were the medical conditions were also followed up. Ninety-
main contributors to this covariation. I reanalyzed two percent were located and interviewed. The
their data to test the hypothesis that the upper 5% follow-up interview as adults (mean age 38 years)
or 10% of the coercive boys would be highly sta- covered areas of family life, social, educational,
ble across time. The alternative hypothesis would sexual, marital, and occupational experience. In
be that children with the most extreme scores addition, records were obtained from the schools,
would be the most likely to demonstrate the phe- military, court, prison, FBI, Veterans Administra-
nomenon of regression toward the mean. As tion, and hospitals. The data available from their
shown in Table 2.7, boys at the 95th percentile at initial hospital contacts were categorized by the
Grade 3 had a .38 chance of being at, or above, system described by Achenbach (1966) into Exter-
the same percentile ten years later! Ten years later, nalizers (antisocial) and Internalizers (excessive in-
all of the children rated at the 95th percentile in hibition, anxiety, psychosomatic). The externaliz-
Grade 3 were perceived as being above the sample ing male adolescents tended, as adults, to have be-
mean on the trait aggressive. havior or personality disorders. They were also
These studies assert that extremely coercive more likely to be committed to correctional hos-

pas
pitals, have a lower social status, and lower occu- “Although behavior patterns often may be sta-
pational level. ble, they usually are not highly generalized across
The first follow-up study at OSLC was reported situations.” (Mischel, 1968, p. 281)
by Moore, Chamberlain, and Mukai (1979). They Somehow the effect of this stance has been
studied over 200 families referred to OSLC since translated into an adversarial either situational or
1968, in which the problem child was age 14 at personality disposition. The writer’s own perspec-
follow-up. These families were followed each year tive on the “debate” which ensued is stated by Laz-
after termination of treatment to determine arus and Lanier (1978) as follows:
whether a police contact occurred. The data
showed that referred boys (mean age 9.8 years) “In response to Mischel’s (1968) challenge that
who engaged in high-rate stealing tended to be situational rather than personality factors by and
greatly at risk for being picked up by the police by large determine behavior, there has been much
the age of 14; the likelihood was .84. In fact, 67% current debate (cf. Endler & Magnusson, 1976)
of these young Stealers became chronic offenders. about what seems to us to be a pseudo-issue that,
one might have thought, had already been dis-
Early in the 1960s, there was a series of retro-
posed of long ago. Aside from mobilizing analytic
spective studies which related childhood antisocial
thought and serving as a refresher course on the
patterns, especially those leading to poor peer re-
complexities of human behavior, the debate mere-
lations, to later adult problems such as dishonora-
ly capitulates extreme ideological positions long
ble discharge from the military (Roff, 1972, 1961;
ago referred to by Murphy (1947) as the ‘organism
Roff, Robins, & Pollack, 1972). Retrospective
error’ and the ‘situational error.’ We would have
studies of adults in psychiatric hospitals, with pri-
supposed that most personality and social psy-
or admission to child guidance clinics, focused the
chologists, regardless of their own particular inter-
findings more clearly. ‘Pritchard and Graham
ests, had by now implicitly or explicitly accepted a
(1966) showed that conduct problems in child-
multifactorial and interactional view of the deter-
hood were associated with adult problems ranging
minants of human action and reaction (see also
from depression and antisocial character disorders
Pervin’s discussion of the centralists and peripher-
to inadequate personality. The antisocial child
alsts)).. (p. 2or)
was most likely to be identified as an antisocial
adult. The classic study by Robins (1966) com- The author feels that the early researchers were
pared 524 children studied at a child guidance misled by investigating a pseudo-issue. The ques-
clinic with a matched sample of nonreferred chil- tion of stability should be cast in interactional
dren from the same area. As adults, 71% of the terms. From that position a child’s coercive perfor-
antisocial group had been arrested, 11% had been mance score is a joint function of what the child
diagnosed as psychotic, 22% were sociopathic, and brings to a setting and the manner in which the
70% were divorced. Only 20% of the antisocial setting is programmed by adults.
group showed adequate adjustment as adults. In From the social interaction perspective, it is the
passing, it should be noted that if the father was adult who determines what /evel of child aggres-
antisocial, there was a considerable increase in the sion is allowable, e.g., usually very little in the
likelihood of the child exhibiting antisocial beha- classroom but more of it on the playground.
vior as an adult. Furthermore, when these anti- Adults control the aggression level for a setting by
social children later had families there was an in- effectively imposing negative sanctions for proto-
creased likelihood that their offspring would, in types of deviant behavior.® For the eleven class-
turn, display high rates of antisocial behavior. rooms observed in the Patterson et al. (1972)
study, the mean likelihood of inappropriate beha-
Stability Across Settings” vior for normal peers ranged from .41 to .11! (The
It wasn’t until the 1960’s that systematic obser- mean was .25). These considerable differences
vation data were collected in settings other than suggest that there are very different levels of con-
nursery schools. The classic field studies by R. trol being exerted by teachers. Mothers also vary
Barker and his colleagues left little doubt that chil- in the limits that they set for aggressive behavior in
dren behaved differently in different settings their homes. As a result, if one observed 50 antiso-
(Barker, 1963; Barker et al., 1962). In the late cial boys in each of their classrooms and homes
1960’s Mischel’s challenging statement called at- and then calculated the across-settings correla-
tention to the fact that most contemporary person- tions, it would almost certainly be close to zero.
ality theorists had not properly addressed questions Fifty mothers set the performance limits in the
concerning the impact of settings upon behavior: home and SO teachers set the limits for their class-

28
 rooms. There is no reason to expect that a given The second hypothesis is concerned with intra-
mother /teacher pair will set similar levels. From a individual consistency across settings among re-
social interaction point of view, performance level sponses. Coercive response hierarchies exist be-
- is primarily “in the setting” and only secondarily cause they produce predictable reactions from oth-
“in the child.” Observation data collected in the er people. Given that these reactions are consis-
homes and classrooms for problem children dem- tent, then the ordinal rankings among coercive re-
onstrated low-positive to zero-order correlation in sponses should be stable across settings. A simple
rates of deviant behviors across settings (Patter- means for testing this hypothesis would require
son, Cobb, & Ray, 1972; Bernal et al., 1976; that the same child be observed in both home and
Johnson, Bolstad, & Lobitz, 1976). school settings; a ranking of the coercive responses
Correlating performance scores from class- actually used in the two settings should demon-
rooms and homes generates little or no evidence strate significant correlations. Thus, if a given
for stability across settings. These findings are in child whines at very high rates at home, this should
keeping with Mischel’s (1968) position. However, also be his “favorite” technique even in a well-run
if coercive performance level is primarily deter- nursery school. While he may whine very little in
mined by the adults present, then how is one to the nursery school, it should nevertheless be his
test the claim that extremely aggressive children most favored coercive ploy.
are “stable” across settings? The interactional In the first study, two deviant boys from the
viewpoint suggests two hypotheses: 1) the antiso- OSLC sample were observed at home and at
cial child’s rank among his peers with respect to school. For one child the rho for the rankings of
his performance of aggressive behavior is stable 20 coercive responses across the two settings was
across settings; and 2) there is stability among re- .73 (df =19,p< .01) and for the other .87 (df=19,
sponses across settings. p<.01). These data suggest that the coercive re-
To test the first hypothesis one can calculate the sponse that was most preferred in one setting
child’s relative ranking among his peers for aggres- tended to be the one most frequently employed in
sive behavior in each of two settings. It would be the other. However, findings from two subjects
expected that the extremely aggressive child would hardly constitute a confirmation of the hypothesis.
rank high in both settings. Adults in both settings It is, therefore, heartening to report that A. Harris
should label the child as deviant because of his and J. Reid (in preparation) have replicated the ef-
high status within his own group. Most epidemio- fect for a sample of 53 normal children observed
logical studies find that about 10% of a sample of in the classroom and on the playground. The
school-aged children are “disturbed.” By chance, a mean correlation was .68 (range from .042 to .96).
child would be labeled as deviant in two settings While the performance scores for the aggressive
about 1% of the time. In point of fact, the data child may vary, he receives the same ranking
show that a child /abeled as deviant in the home among his peers and is consistent in his use of par-
has a good chance of receiving the same label in ticular coercive responses in setting after setting.
the school. The converse is also true. For example, The data relating to these two hypotheses lend
of 27 boys actually shown to have conduct beha- support to the idea that aggressiveness is a trait
vior problems in the home, 52% were also that is relatively stable across settings.
claimed to have social and/or academic problems The more general hypothesis that extremely ag-
in the classroom (Patterson, 1976). Johnson et al. gressive children show consistency across settings
(1976) observed the classroom behavior of a small and time received indirect support from the analy-
sample of children identified as having conduct sis of OSLC data files by Kirkpatrick (1978). Her
problems at home. The observation data indicated analysis of the community follow-up data showed
that 50% of these children were deviant in the that children who committed police offenses by
classroom as well. Forty-five percent of the chil- age 14 had, at age 9, been labeled as deviant in
dren referred for conduct problems in the class- both the home and school settings. M. Rutter
room were also shown to be deviant in their inter- (personal communication) also reported that chil-
actions with members of their families. This over- dren who were identified as antisocial in both
lap is considerably more than would be expected home and school settings had the poorest prog-
by chance. An even higher percentage was found nosis for later adjustment.
in a comparable study by Moore (1975). Howev-
er, in another study, only one in five disruptive Some Dimensions of Antisocial Behavior?
children in the classroom were observed to be de- Ordinarily, a wide spectrum of behaviors are
viant in the home (Walker, Hops, & Johnson, subsumed under the terms “antisocial” or “con-
197575 duct problem.” For example, teasing, firesetting,

29
 Table 2.8
Results from Correlational Analyses of Referral Symptoms for Two Samples

Studies UnsocializedAggressive (Socialized)Delinquent

Jenkins & Hewitt, 1944 Mean Correlation Mean Correlation
Destructive at school 41 Stealing 33
Violence sii Truant from home ay
Fighting 35 Truant from school .48
Quarrels 34 Police arrest 48
Destructive 733 Stays out at night 44
Incorrigible 32 Bad companions .42
Boastfulness By? Gang running 139)
Teasing SY Loitering Rey)
Expulsion from school 31 Lying 36
Unpopularity Bil

Aggressive Delinquent
Achenbach, 1978 Mean factor loading Mean factor loading
Argues avi Steals outside home .70
Disobedient at home .66 Steals at home .67
Temper tantrums .64 Destroys things
Fighting .61 belonging to others ARVs
Cruel to others .60 Vandalism 54
Teases 56 Sets fires .50
Shows off a5b) Truant 48
Loud rod Runs away 48
Disobedient at school soil Bad friends 44
Screams 49 Lies, cheats 44
Swears 46 Swears OT
Poor peer relations 45 Disobedient at school 31
Brags 45
Lies, cheats 43
Moody .43
Demands attention 41
Unliked .40

disobedience, stealing, fighting, truancy, and tem- Michigan child guidance clinic. There were a
per tantrums would all fit into these categories. number of clusters of problem behaviors identified
The parents of the children referred for treatment by correlation analysis. Two were of particular in-
at OSLC reported an average of three to four of terest. One, Unsocialized Aggressive, described
these “symptoms.” The pattern seemed to vary 10.4% of the cases and was comprised of abrasive
widely from one child to another. From the begin- events found in social interaction. The other de-
ning, it became apparent that children labeled as scribed rule-breaking behaviors; it was labeled So-
antisocial really constituted a very heterogeneous cialized Delinquent. Table 2.8 gives the mean cor-
group. This was of immediate concern because it relation of each symptom with other members of
seemed likely that different subgroups would re- the cluster for the Jenkins and Hewitt study. The
quire not only different “explanations” but differ- referral problems are ordered by the magnitude of
ent treatments. the correlations.
The traditional literature was examined for in- The behaviors defining the Social Aggressive
sight into these issues. In the classic studies by Jen- Cluster have two very important characteristics in
kins and Hewitt (1944) and Jenkins and Glickman common. Almost all of them would be experi-
(1946), the data consisted of the referral problems enced by others as aversive. Furthermore, they
abstracted from case files for 500 referrals to a have both a direct social impact and a victim. Sec-

30
 ondly, they occur at sufficiently high rates so that are defined by the available factor scores. Of par-
they could be observed in the home, e.g., quarrels, ticular interest are the children high on Achen-
teasing, and boastfulness. However, the behaviors bach’s Aggression and Delinquent factors. They
‘listed under the Delinquent Cluster were not nec- seem to describe two very different kinds of cases
essarily events one could observe as the child’s im- referred for treatment at OSLC. In fact, the many
mediate reaction to another person. In addition, antisocial children referred to OSLC could be or-
stealing, truancy, staying out late at night, and dered along these dimensions.
loitering are all low base-rate events. They may When the family comes to OSLC for the intake
occur once a week, or at most, a few times each interview, the parents are asked to select from a
day. Certainly they did not lend themselves to the checklist those problem behaviors which are of
usual observation formats. most concern to them, i.e., those which they wish
In their replication study, Jenkins and Glickman to change during treatment. The checklist is pro-
(1946) used behavior ratings from 300 institution- vided in the book by Patterson, Reid, Jones, and
alized delinquents. The findings extended and es- Conger (1975). It was modified somewhat and
sentially replicated the original work. The work standardized by Chamberlain (1980). In her ex-
stimulated many other investigators to become tensive study, 65 parents of normal children iden-
committed to the problem and to search for con- tified problems each day for a 12-day baseline
vergence across age groups, settings, and various from a 34-item checklist. For the ten families
modes of data collection. This literature has been where the mothers and fathers gave independent
carefully reviewed by Quay (1965), who was also reports on problems of concern to them (targeted),
one of the most prolific contributors to this field. the correlation between parents was .89 (df=8,
His review underscored the general consistency of p <.001). The hierarchical-cluster analyses of the
the findings in this area. The two factors (Social reported frequencies produced four clusters. The
Aggressive and Delinquent) appeared in study first three clusters are presented in Table 2.9. They
after study. are consistent with the earlier findings which em-
Achenbach’s programmatic work began in 1966. ployed clinical samples. Her two clusters, Aggres-
It constitutes the most rigorous and large-scale ef- sive and Immature, intercorrelate .50 (df=63,
fort to date. He carried out a careful analysis of p< .001) and define what I mean by Social Aggres-
case study material for 300 male and 300 female sion. Her third cluster, Unsocialized, defines what
patients from a child psychiatry department. Sev- I mean by Stealer. The children studied at OSLC
enty-four symptoms were intercorrelated and the are, on the average, 8 to 9 years of age, a bit young
data subjected to a principle-factor method of to be called (Unsocialized) Delinquent. For this
analysis. The analyses produced a bipolar princi- reason, the term Stealer!° is used throughout this
ple factor, Internalizing-Externalizing, and a uni- volume.
polar second-principle factor labeled Severe and These rather consistent differentiations into
Diffuse Psychopathology. This pilot study led, in types of antisocial behavior are in keeping with
turn, to the development of a parent checklist our clinical experience. There, too, the treatment
which samples social competence, school perform- of Stealers proved to be a very different matter
ance, social activities, and problem behaviors than did the treatment of Social Aggressors (Reid
(Achenbach, 1966, 1978). Systematic work has & Patterson, 1976). The first look at family inter-
been carried out for different age groups and sep- action data also convinced us that families which
arately by sex; further studies are in progress. produce Stealers may be quantitatively different
For the 1978 study, data were available for 450 from both Normals and from Social Aggressors.
problem boys. Of the nine factors derived, two The analysis by Reid and Hendricks (1973) showed
were more germane to the present discussion. The them to be more coercive than Normals but less
factor loadings for these two are summarized in than Social Aggressors in their interactions with
Table 2.8. Not only were these current findings re- family members. The mothers of Stealers were
lated to the narrow-band factors in his earlier shown to be less friendly than mothers of either
work, they also look surprisingly consistent with normal or socially aggressive boys. This pilot
the work carried out 30 years earlier by Jenkins study was replicated and extended by the findings
and Hewitt (1944). The systematic work by Ach- summarized in Chapter 11.
enbach and his colleagues lends itself nicely to the It appears that the Stealer and Social Aggressor
identification of “types” of antisocial children. patterns share a common base. In fact, the base
The distribution of scores for each factor has been occurs so often for all of these children that it may
normalized and converted to T-scores. Thus, one not show up consistently in correlational analyses.
can construct typologies based upon profiles that The base is that the parents allow all of the antiso-

a4
cial children to disobey, that is, to be noncompli- murderers, he found that as children they showed
ant. For example, .89 of the 142 antisocial chil- a combination of firesetting, bed wetting, and
dren were described by their parents as being diso- hyperactivity!
bedient (from Table 11.1). In their observation However, the assumption is that regardless of
study of problem children, Griest et al. (1979) which antisocial pattern(s) the parents allow, all
found a correlation of .50 (p<.05) between the antisocial children will be described as basically
likelihood of Noncomply and the number of prob- disobedient, i.e., out of control. For example, in
lem behaviors. one study of referrals to the OSLC clinic, the like-
Given that the child is disobedient, most fami- lihood of Steal as a complaint given the occurrence
lies insist that he specialize in one antisocial pat- of Disobedient was a conditional p of .56.. Better
than half of the children who disobeyed were also
viewed as Stealers. Given Disobey as a complaint,
Table 2.9 the likelihood of Whine was .62; the likelihood of
Symptom Clusters Hit was .64; and the likelihood of Temper Tan-
trums was .61. The apprenticeship for becoming
and Reliability Coefficients
an antisocial child seems to be a process of learn-
(fromChamberlain,
1980) ing disobedience.

Chamberlain’s
The Drift Toward
Items Loadings
Marginal Adjustment as Adults
Aggression Alpha=.88 Beta—e7, As noted earlier, children with severe conduct-
Yell 159 problem disorders can be described by their posi-
Runaround .743 tion on two different dimensions. One is a devi-
Talkback 721 ancy dimension; and the other dimension refers to
Aggressive .687 the acquisition of social skills. I assume that these
Noisy O77 variables are negatively correlated; for purposes of
Argue .674 discussion, I would think of the magnitude of the
Defiance 621 covariation as being around —.4 to —.5.
Noncompliance .618 The deviancy dimension describes the child’s
Immature Alpha= .80 Beta al position on some measure of antisocial behavior
relative to peers of the same age. In the present
Complain -748 context, the measure would be parents’ reports of
Whine .696 persistent stealing or the TAB score based upon
Irritable .670 observation data. The greater the deviation from
Pout .568 the mean, the more likely the child is to be labeled
Tempertantrum .567 as deviant. However, even if his rates of coercive
Negative .523 behavior exceed the 95th percentile, there is little
Unsocialized Alpha=.65 Beta=.46 chance that he will be labeled as deviant before the
age of 4 or 5. I suspect that the majority of these
Hyperactive 765
coercive preschoolers at the 95th percentile are not
Parenthits .580
thought of as deviant by their parents. However,
Stealing Or
as the problem child penetrates community agen-
Lying .428
cies such as the nursery school, the pediatrician’s
Bedwetting 345
office, and finally the elementary school, the par-
Schoolcontact 294
ents are increasingly likely to be given negative
feedback about the behavior of the child. Eventu-
tern or another. Parents of problem children will ally the school, the court, or neighbors may force
vehemently reject some kinds of antisocial behav- the parents to contact community professionals.
ior, e.g., “No kid of mine would get away with The OSLC referral data suggest that by the age of
stealing.” However, the same father may accept 9 or 10 many of these extremely coercive children
hitting and temper tantrums with equanimity. have been so labeled.
Very few parents allow all of the symptoms to oc- Being labeled deviant is just one stage in the
cur. In that regard, J. Tuppin, University of Cali- drift towards marginal adjustment as an adult. La-
fornia at Davis (personal communication) men- beling, per se, does not determine the drift; it is
tioned that in his retrospective studies of multiple simply an indication that it has occurred. Elliott’s

a2
 (1979) elegant study showed that for delinquent relationships as an adolescent and adult. These
youths, contact with police or the court (which omissions create more situations in which coercive
constitutes being labeled) did not alter their self- skills may be employed to maximize short-term
‘perception, nor did it alter cther people’s evalua- gains.
tion of them. However, being involved in counsel- It should be noted that not all socially unskilled
ling did produce negative changes in evaluation. children become antisocial. Programmatic studies
As the child develops, there are certain social by H. Hops (in preparation) have shown that
survival skills which parents must teach. In the many unskilled children are withdrawn. Inciden-
preschool years these are self-help skills, such as tally, some of the studies by Walker and Hops (in
feeding and dressing, toilet training, and solitary preparation) showed that effectively increasing so-
play. If the child is delayed in the acquisition of cial skill behaviors did not reduce the level of devi-
these skills, it is probably of little significance. It ant behavior for extremely antisocial children.
means that either the parents are not very effective There seem to be two systems, each with its own
socializing agents, or the child is, by temperament, set of determinants.
difficult to train. These skill deficits are not likely, Not all extremely coercive children are socially
by themselves, to be highly correlated with the unskilled; conversely, not all children labeled as
likelihood of later deviancy labeling. If, however, unskilled are antisocial. For example, young Steal-
the child is also extremely coercive, then the writer ers or Social Aggressors from some middle-class
assumes that the child has begun a perceptible families might attain academic proficiency. The
drift toward marginal adjustment as an adult. writer believes that it is the drift toward extreme
There are two reasons for this assumption. positions on both the deviancy dimension and the
First, as the child becomes increasingly out of con- skill deficit dimension which predicate poor adult
trol, both peers and adults will reduce the amount adjustment. It is the socially unskilled young Steal-
of contact with him. They avoid and escape his er who will most likely continue his offenses against
presence whenever possible because he is thor- property and may eventually be represented in of-
oughly unpleasant to be around. As a result, the ficial police records. The socially skilled Stealer or
extremely coercive child becomes a “loner.” He is, Coercer is less likely to be labeled as deviant; his
therefore, spending less time in his peer group social competencies presumably function as a
learning the subtle social skills necessary for ade- moderating variable. At present, there are no data
quate survival in the forthcoming adjustment available which could test these hypotheses about
phases. His explosive response to negative feed- the interactive nature of coercion level and social
back is a further block to learning both social and skill deficits. There is, however, extensive litera-
academic skills. Reduced interaction together with ture demonstrating that antisocial children are less
his abrasive style makes it increasingly difficult for skilled.
him to acquire the survival skills he will need later
on. The large-scale survey by Achenbach (1978) The Early Deficits
indicated that parents of deviant children perceived The general position taken here fits well with
their children to be significantly less competent the author’s recollection of the stance presented by
than did parents of normals. The parents re- John E. Anderson just before his retirement as the
ported less participation in sports, hobbies, jobs esteemed Chairman at the University of Minneso-
and chores; these children had fewer friends and ta Child Welfare Station. He repeatedly pointed
failed more often in school. out that children diagnosed as “emotionally dis-
Limited exposure to peer socializing influences turbed” were really “just socially unskilled peo-
may be one determinant for deficits in social com- ple.” It was not then, and is not now, a popular
petencies. Another determinant has to do with re- position. However, that kind of thinking has re-
duced responsiveness to positive and negative cently been applied to a number of problems in
sanctions. The studies reviewed in Chapters 6 and “psychopathology,” e.g., depressed adults, and
11 suggest that aggressive boys were less responsive withdrawn children. Anderson’s common-sense
than nonaggressive boys to both social reinforcers position was supported by the Achenbach (1978)
and to punishment provided by adults. This study reported earlier and by a questionnaire
would suggest that even when parents do try to study (Ferguson, Partzka, & Lester, 1974). In the
use reinforcers to shape prosocial skills, they may latter, an item analysis of questionnaires filled out
be relatively ineffective. The coercive child fails to by a sample of parents of disturbed children and
learn simple work skills at home (chores), then parents of nonreferred families showed that the
fails to learn academic skills. Later, he fails to former perceived their children as lacking social
learn skills which are necessary to form intimate competencies.

38
 Table 2.10
Studies Relating Social Skill Deficits to Antisocial Behavior

Early Management Phase

Investigators Findings
1. Douglas (1966); Robins (1966); Schlesinger (1978); . Antisocial child reported to have had difficulty in
West and Farrington (1973); Fagan et al. (1977) training for bowel/bladder control.
2. Kaffman and Elizur (1977) . Enuretics more likely than non-enuretics to be rated
as deviant, including more aggressive.
3. Wahler and Moore (1975) . Inverse correlation between observed nonsocial toy play
and occurrence of problem behavior in home.

4. Saxe and Stollak (1971) . No correlation solitary play and aggression, mother-
child interaction in laboratory.
5. Werner and Smith (1977) . Social immaturity on the Vineland during preschool
correlated with later delinquency during adolescence. |
Infant activity at year one and later hyperactivity also
correlated with later delinquency.

Peer Relationships
Investigators Findings
1. Moore (1967) . Nursery school children tended to reject aggressive
child.

2. Jennings (1950); Olweus (1974); Roff (1972); . Aggressive children have low acceptance by peers in
Rutter et al. (1970) elementary school.
3. Roff (1961) (1972); Eron et al. (1971); Hirschi (1969); . Delinquents have or had poor relations with peers
Jenkins and Hewitt (1944); Cureton (1970); West and and/or socially immature.
Farrington (1974)
NOTE: McCord (1977); Elliott (1979); Elliott and
Voss (1974) did not find this relation.

Work Skills
Investigators Findings
1. Minton, Kagan, and Levine (1971) . Very few antisocial children complete chores at home.
2. Cobb and Hops (1973); Hops and Cobb (1977); . Aggressive children tend to be low in classroom
Cobb (1970); Hops and Cobb (1974) survival skills, such as attend, volunteer, stay in seat,
work at desk.

3. Hirschi (1969); McCord (1977) [trend but not signifi- . Delinquent children spent less time doing homework
cant] Conger and Miller (1966); Wadsworth (1979) or generally have poor work habits.
4. Rutter et al. (1970); Maccoby (1966); Werner and . Antisocial boys read less well, also perform less well
Smith (1977); Benning, Feldhusen, and Thurston on IQ tests and achievement tests, grades.
(1968); Glairn and Annesly (1971); Graubard (1971);
Camp, Zimet, Doomineck, and Dahlm (1977); Wads-
worth (1976); Roff (1972); Shea (1972); Cobb (1970);
Eron et al. (1971); Scarpitti (1964)

5. McCord (1977); Glueck and Glueck (1968); Scarpitti . Antisocial children lower than normals only for
(1964) nonverbal IQ test. Delinquent children more likely to
have dropped out of school and/or be truant.
NOTE: Hirschi (1969) found no relationship.

It seems that the skill deficits may be apparent at them as constitutionally different and, in some
a very early stage in development. As shown in Ta- sense, more difficult to train? Many parents who
ble 2.10, children diagnosed as antisocial are fre- bring their children to OSLC would have us be-
quently described as having difficulty with bladder lieve so. Or, does this lag in training underscore
and bowel control training. Does this identify the effects of rather inept parenting practices?

34
None of these studies provides a base for choosing class children it was skill in nonverbal reinforce-
one or the other of these two alternatives. At this ment. Incidentally, other variables related to peer
point, it can only be said that children who are acceptance included the amount of off-task behav-
slower in achieving self-control over bowel and ior and the amount of punishment given.
bladder may have a greater likelihood of being la- There is another more subtle sense in which
beled as antisocial. The large-scale follow-up these children do not fit in. The observation of
study by Werner and Smith (1977) also showed nursery children by Strayer (1977) showed that
that extremely active and socially immature be- aggressive children tend to ignore the metastruc-
havior during the preschool years correlated sig- ture in groups. They behaved as if the dominance
nificantly with adolescent delinquency. These chil- hierarchy which influences their peers did not ex-
dren were at risk only if the mother was poorly ed- ist. This, in turn, would be a constant source of
ucated and living under rather chaotic conditions. discomfort and annoyance to others.
Active and immature children from intact middle- Until he enters nursery school, the child has
class families were not at risk for later delinquen- minimal contact with his peer group. Then he
cy. must learn how to initiate interactions and how to
respond to those of others. He learns to share, to
Later Deficits follow rules in playing games, and to lead without
Peer relations. There is a general relation be- being bossy. With increasing age, these social
tween the child’s status as deviant and the likeli- skills become increasingly subtle, e.g., how to
hood of his being rejected by peers (Gottman, form friendships, how to be a leader, how to be-
Gonso, & Rasmussen, 1975). The sociometric come intimate, how to give and receive affection,
study by Sells and Roff (1967) showed that low and how to give and receive criticism. This is the
peer acceptance was associated with parental re- time when he must defend himself against attack
ports of disruptive coercive behaviors at home. by others. Without being devastated, he must
This general relation has been noted by several learn to deal with another child grabbing his toy.
writers (Gottman et al., 1975; Lott & Lott, 1965; Apparently a certain amount of skill in aggres-
Taguiri, 1952). Peer rejection was shown by Gott- sion is necessary to survive in the nursery school
man (1977) to be unrelated to the amount of inter- environment. In our culture, the normal preschool
action with peers; rejection was, however, signifi- child who has developed skills in sharing, coopera-
cantly related to negative interactions with peers. tion, and leadership will also have learned to be
Hartup (1969) also identified the child’s use of mildly aggressive. However, I think that his rates
aversives as being related to peer acceptance. The of aggression are well within normal limits, i.e.,
child who was accepted also gave more positive re- less than the 95th percentile. For example, Mur-
inforcement. As a consequence, the extremely co- phy (1937; cited by Radke-Yarrow & Waxler,
ercive child is very likely to be rejected. In a study 1976) found a consistent positive correlation be-
of classroom interaction, H. Hops (personal com- tween nursery school children’s aggression and
munication) found that children who were disrup- their ability to express sympathy. Henry and
tive of classroom routines were accepted. The Sharpe (1947), Feshbach and Feshbach (1969),
child who was extremely aggressive on the play- Friedrich and Stein (1973), and Barrett and Radke-
ground was likely to be rejected. It seems, then, Yarrow (1977) all found positive correlations be-
that rejection implies that the peers experience the tween observed aggression and prosocial behav-
behavior as being aversive. The study of school iors such as comfort, sharing, and sympathy. The
bullies by Olweus (1975) showed that it was not observation study by Radke-Yarrow and Waxler
aggression per se which related to rejection. The (1976) clarified the problem by demonstrating
bully directed most of his aggression to preselected that these positive correlations were obtained only
targets. His popularity was only slightly dimin- for nursery school children who performed at low
ished by his role as bully. levels of aggression. In their study, for boys below
The study by Gottman et al. (1975) further the mean in aggression, the correlation between
demonstrates that the nature of the skills involved aggression and sharing-comforting was .50 (p<
in producing peer acceptance may vary as a func- .05) in one setting and .48 (p<.05) in another.
tion of social class. His elegant multiple regression However, for boys above the mean on aggression,
analyses of observed interactions showed that skill the comparable correlations were —.45 (n.s.) and
as a reinforcer accounted for most of the variance —.19 (n.s.). One might label lower rates of ag-
in peer acceptance for both middle- and lower- gression as “assertiveness” and reserve the term
class children. However, for the former it was skill “aggression” for extremely high rates, i.e., the up-
in verbal reinforcement that mattered; for lower per 5%. Presumably it was the latter who were

35
identified in the peer sociometric study by Moore self to the point where one throws away degrees of
(1967). That study showed that the aggressive freedom, e.g., the type of businessman (or schol-
child tended to be rejected by nursery school ar) who becomes so immersed in the work process
peers. that he shortens his life.
Clinical experience with the several hundred A priori, there would seem to be a sequence
cases referred to OSLC underscores the theme that through which most children in middle-class fami-
these are not highly skilled children who are effec- lies move. First, there are childhood experiences
tive competitors. They are not the “occasional ag- which provide on-the-job training for work re-
gressors” popular among children. On the con- lated skills. These fortuitous and/or planned com-
trary, the problem child is usually unskilled at fist ponents of the socialization process generally re-
fighting or wrestling; he tends to be an inept com- flect the middle-class commitment to the Protes-
petitor. He is usually isolated and rejected by his tant work ethic. First the child is taught to pick up
peer group. The studies cited in Table 2.10 attest his own personal effects and put them away. Later
to the consistency of this finding in the research lit- he is expected to keep his room clean and orderly,
erature. Roff (1961, 1972) found that poor peer then to assist in minor household chores. Later
relations characterized adults who were given dis- still, he may accept responsibility for major tasks.
honorable discharges from the armed forces. Inci- By elementary school age, this might include part-
dentally, Roff and his group also found poor peer time jobs for pay. In adolescence it might include
relations as a consistent precursor to a wide spec- summer employment. The coercive child meets all
trum of adult psychopathologies. of these requests by learning how to escape and
A number of recent major studies on delinquen- avoid them.
cy have also found that youths officially labeled as The classroom functions as another setting for
delinquents were characterized by poor relations training in work skills. The very earliest clinical
with their peers (Hirschi, 1969). This theme was classroom observations of families referred to
noted previously in juvenile delinquency research OSLC showed that these children lacked the skills
(Jenkins and Hewitt, 1944). The covariation is so to adjust to academic demands. Cobb (1970) ob-
reliable that it is now a regular feature in longitu- served a sample of normal first-grade children in
dinal studies designed to predict later delinquency. the classroom. He found significant correlations
West and Farrington (1973) found that less popu- between achievement test scores and “survival
lar 8- and 9-year-old boys were significantly at skill” scores for boys. These correlations were .71
risk for /ater delinquency. Roff (1972) found that for reading and .63 for arithmetic. The compara-
low peer acceptance was associated with later ad- ble correlations for girls were lower. The child
olescent delinquency for middle-class but not for who was disruptive, nonattending, and noncom-
lower-class boys. In the same study, low peer pliant did less well on the achievement tests. In the
status was correlated with later status as a school next study, Cobb and Hops (1973) trained the
dropout. teachers to shape survival skills for first-grade
Work skills. Two skills required prior to adoles- children who were deficient in these areas. The
cence are critical to adult adjustment. We must training effectively increased both survival skills
learn to relate, and we must learn to work. There and achievement test scores for this subgroup of
is no single source for acquiring either of these children. They then replicated the effect for an en-
skills. The way in which we relate to casual ac- tire first-grade classroom (Hops & Cobb, 1974).
quaintances and to intimates must be some joint In their final study, Hops and Cobb (1977) com-
function of earlier relations among family mem- pared training in survival skills, direct instruction,
bers and later peer relations. However, the details and a no-treatment control. Both experimental
of this most important process are only beginning groups showed significantly greater gains on read-
to receive the attention of empirically oriented in- ing achievement tests than did the contro! group.
vestigators. As might be expected, the gains for direct instruc-
The world of work is not much better under- tion were greater than those for training in surviv-
stood. In emphasizing this as a crucial skill, the al skills. This important series of studies empha-
writer is perhaps falling into our cultural stereo- sized the need to consider noncognitive survival
type about which values are central; but if one skills as they relate to academic achievement. It
does not solve the problem of income/work, then seems very likely that the conduct-disordered child
as individuals our degrees of freedom as decision lacks the skills necessary for adequate perfor-
makers are extraordinarily constrained. Poverty mance at even the first-grade level.
seldom offers freedom of movement or thought. There is another thread running through these
On the other hand, one can be trapped by work it- studies which would be interesting to examine

36
more carefully. The survey study by Rutter et al. aggressive behavior. Several follow-up studies
(1970) showed a relation between reading difficul- have also shown very little relation between the
ties and antisocial behavior. The follow-up study child’s self-concept and his later delinquency, e.g.,
by Werner and Smith (1977) showed a correlation Benning et al. (1968). In her intensive follow-up of
of .24 between a need for remedial education at the Cambridge-Somerville sample, McCord (1977)
age 10 and later delinquency. found that low self-esteem in children did not pre-
The remainder of Table 2.10 reads like a litany dict later criminal careers. These findings should
for failure. As the aggressive child moves through not surprise us. Questionnaires and interviews
school, the studies reiterate a theme of lower given to adult criminals showed that rapists, em-
grades and lower scores on achievement tests and bezzlers, and auto thieves do not see themselves as
IQ tests. As might be expected, there are related criminals (Elliott, Ageton, & Canter, 1979)!
findings of early truancy and tardiness and finally West and Farrington (1973) reviewed four
dropping out of school. studies which showed that delinquents have more
negative self-images than nondelinquents. In one
Self-Perception study, delinquents were more prone to endorse
Coopersmith (1967) and others affirm that as such adjectives as bad, dirty, cruel, dangerous,
the child matures, he becomes increasingly capa- dishonest, and worthless. Several of these studies
ble of differentiating himself from others. The an- found a significantly greater loading for delin-
tisocial child receives more aversive input than the quents on the bad-evaluation dimension for the
other members of his family. He is very likely to be Osgood semantic differential. However, in the ex-
rejected by his peers and to fail in his academic tensive survey study by West and Farrington
subjects as well. Given such an accumulation of (1973), while delinquent-prone youths described
negatives, one would expect that midway through themselves as more aggressive, even these findings
the elementary grades the aggressive child would were not greater than one would expect by chance.
conclude that he is clearly different from other These aggressive self-concepts did not predict later
children. This differentiation should become delinquency.
clearer with age. At some point, the antisocial It seems that in focusing upon self-esteem, we
child would presumably develop a negative image may have been asking the wrong question. There
of both self and others. are two alternatives that seem promising. Garmezy
The research literature is far from complete in and his colleagues at the University of Minnesota
its quality or quantity. However, the literature have been studying children who are at risk for
that does exist suggests that the extremely coercive psychopathology (Garmezy & Streitman, 1974;
child does not assign blame to himself, nor does he Garmezy & Nuechterlein, 1972). Their analyses
see himself as less worthy than others. What he repeatedly underscore the importance of the
does develop is an extremely negative view of oth- child’s feelings of efficacy in arranging his own
ers, his family, his peers, and eventually the police life. For the ghetto child, this sense of being in
and the social system itself. His attitudes towards control correlated more highly with achievement
others are antisocial; his self-esteem is unim- test scores than did that venerable variable, socio-
paired. economic status! One wonders if rephrasing the
Two studies (Magee, 1964; Silverman, 1964; question in this manner wouldn’t also differentiate
both cited by Feshbach, 1970) showed that ag- the normal child from the child with a conduct dis-
gressive boys had lower self-esteem than did non- order. In clinical situations many antisocial chil-
aggressive boys. However, in three studies of nor- dren convey a sense of events moving outside of
mal children, there was no correlation between ag- their control as if they were almost “forced” to
gressiveness and children’s self-esteem (Cooper- strike that other child or take that purse.
smith, 1967; Eron et al., 1971; Olweus, 1975). In a similar vein, the longitudinal study by West
Olweus (1975) studied two samples of normal and Farrington (1973) strongly implicates the es-
boys ages 13 to 14. They filled out a self-report sentially antisocial attitude toward others, partic-
questionnaire which included a scale, Positive Self- ularly toward the police, as being an important
Report. The scores on this scale correlated .06 concomitant of early antisocial patterns. This mis-
with peer sociometric data scored for Starts Fights anthropic view is reiterated in the large-scale sur-
and —.07 with the score on Verbal Protest. The vey by Olweus (1975). A factor analysis of a large
corresponding correlations for the second sample pool of self-report items generated three which re-
were —.05 and .14 respectively. These findings lated to aggression. There is a self-righteous quali-
from this large-scale study showed the tendency to ty to these answers, a commitment to violence as a
view oneself in a positive light did not covary with solution to problems. In some ways the antisocial

or
child is a caricature of the cultural emphasis upon
the rugged individualist. There is no sense of un-
Footnotes
certainty or doubt. He stands against the tide, re- 1. The label deviant or aggressive is not, of
sisting all onslaughts upon his individualism. His course, always earned. Analyses of the first sample
victims deserve what they get. For him there is no of 27 families referred to OSLC for treatment
existential anxiety. He is a hero. Failure in school showed that 11% were false-positive errors (Pat-
is not his doing: “It is a lousy school.” Being re- terson, 1974a). Referral sources cited problems
jected by peers is not his fault: “Those kids are occurring either at home and/or school. However,
snobs; they don’t know anything.” There are con- neither observation or Parent Daily Report data in
flicts at home: “They pick on me all the time.” It is the home, nor observation in the classroom dem-
not him; it is their fault. He has no control over onstrated above normal rates for problem behav-
what teachers do or over what victims do. These iors.
victims are the objects of his contempt, but they 2. Hartup (1973) and others have noted that
are not a matter for his concern. while overall measures of aversive behaviors may
decrease, what this really reflects is a reduction in
Conclusion the frequency of physical aggression. The same
Children’s antisocial behaviors were shown to data showed increases in frequency of verbal ag-
be a heterogeneous set of symptoms. For conven- gression with age. Green (1933) reported a similar
ience in studying these problems, the focus in the increase in verbal aggression for preschool chil-
present volume is upon two patterns thought to be dren.
“homogeneous.” These groups are labeled as Steal- 3. For the present analysis, only one child in
ers and Social Aggressors. These patterns describe each family contributed a TAB score. Wherever
children who maximize’ immediate payoffs at an- possible, the score was selected which described
other’s expense. In either case, it is assumed that a the youngest child in the normal family. This al-
labeled deviant child is an example of “arrested so- lowed a somewhat better description of normal
cialization”; e.g., his abrasive and/or thieving be- preschool children than we would otherwise have
havior is simply something which his family ac- obtained.
cepted and allowed to persist. The children at the 4. Rutter and others also noted that neurotic
extremes do not outgrow these problems. These reactions in children were associated with smaller
are personality traits which are relatively stable families. There was a tendency for these children
across time and across settings. to be an only child. The latter finding was also
For many antisocial children there is an accom- noted by Anderson (1969).
panying deficit of social skills. The relative social 5. It is hypothesized that when a child coerces,
incompetency increases with age. Deviancy and the prosocial behavior of the other person be-
social skills are thought to interact in defining mar- comes more predictable. The findings relating to
ginal adjustment as an adult. This interactive pro- this issue are reviewed in Chapters 6 and 7. Pre-
cess is labeled “deviancy drift.” sumably the prosocial reactions of family mem-
The family processes relating to stealing are bers are more uncertain, or at least their sustained
thought to be different from those which deter- positive reactions are less likely. If this is true, then
mine socially aggressive behaviors. The remainder the function of a child’s coercive behavior is to re-
of this book will focus upon how family processes duce this particular uncertainty. Ironically, his ef-
relate to social aggression and to a lesser extent, forts to make others more predictable in the long
stealing. run insure the stability of his own traits to be coer-
cive!
6. The question is, how should one determine
which 6-year-olds are above the 95th percentile?
McCaffrey and Cummings (1967) introduced a re-
finement in the assessment that bears upon this is-
Acknowledgments sue. They interviewed 164 teachers to identify in
The writer wishes to acknowledge the contribu- each classroom those considered to be problems.
tion of the writing seminars in which earlier ver- Twelve percent were identified as having problems
sions of this chapter were critiqued. My colleague, of some kind. Thirty percent of these were again
Rolf Loeber, made numerous suggestions for im- identified as problem children when teachers in
proving later drafts. I particularly wish to thank the next classroom were interviewed in the ensu-
Professor D. Olweus whose detailed comments ing two years. A sequence of two probes separated
produced many changes in the manuscript. by a year might be a more powerful way of identi-

38
fying children at risk for later adjustment prob- reporting concerns about lying were 15% and
lems as adolescents. 16% respectively; disobedience was 31.5% and
_ 7. Many of the ideas here are the outcome of 22%. Such consistency may be coincidental but
extended discussions with the contumacious Alice seems worthy of note.
Harris. Her persistent refusal to understand my 9. Roach (1958) reported that antisocial prob-
perfectly lucid position led to extensive revisions lems constituted the largest single basis for referral
of earlier hypotheses. to child guidance clinics. Similarly, cases referred
8. Martens and Russ (1932) showed that par- by teachers for treatment consisted of assaultive
ents of normal and problem children report the (30.9%) and delinquent (30.6%) (Rogers, Lilien-
same kinds of concerns. In fact, the rho between feld, & Pasamancek, 1954). The extensive survey
rankings of frequencies for various child behaviors by Rutter et al. (1970) showed that for the dis-
was .90. The samples, however, differed in the turbed 9- and 10-year-old group, 41% were diag-
overall rates with which problems occurred: a nosed as conduct problems; another 27% con-
mean of 3.63 for problem and 1.3 for nonproblem sisted of a mixed pattern of neurotics plus acting
children. Incidentally, Shepherd, Oppenheim, and out. Thus the majority (68%) of psychiatric prob-
Mitchell (1966) excluded severe acting out cases lems involved conduct problems as a major, or at
and showed no difference between frequencies in least secondary, consideration.
listings of parental concerns for normal and prob- 10. The covariation between hyperactive and
lem children. Steal, Lie, Bed Wet, and School Contact was a sur-
Survey studies showed surprising consistency prise. The children in Chamberlain’s sample were,
even across cultures in the incidence of certain on the average, 7 years of age. Is hyperactivity in
conduct problem behaviors among normal fami- the very young 4- or 5-year-old a precursor for
lies. For example, stealing in 10- to 11-year-old later problems relating to Steal? P’ve not encoun-
boys was 5.7% for Rutter’s sample in the Isle of tered this relation before. However, I notice that
Wight study (1970) and 4% in the MacFarlane et in Achenbach (1978) the item Hyperactive loads
al. (1968) California sample. The rates of parents upon a variable of the same name.

4
 Se

>
i

nr

ro Mlcd ee glerreya ive wow tm wee se"

reg ious ’ Be ae Meare
saseiyslie!’ <4: |; heap aaa elma dal Al
ee pot ma pore} ae A 4piw. Gri
bpp pty Le Vs ia a Miles whey wal
1 - i cTe pi aii i.
tte ZA
: ‘am
Se f sh 618

4 Prtya | ’ i ’ : }
! i i} by iy * t
s Lilt A '
eiiiain, |b Le ‘ wh
\ » Sa « ! ; wing
beroen hd apy f
his CM
pWenaatm ves

tay elerd coli

Salem
‘ ee,
er eink eS
2a eran ing ‘i Us wae. aths
} vt et «
(i ‘Missa

a4
q

i asat }
 Observations of
Family Process
The observation devices currently used at OSLC has been used to measure family interactions in
represent almost two decades of evolution. These over 200 families with aggressive children and 60
data collection systems were developed in response families of matched normals.
to a need for an assessment methodology which As we began to use the new code it quickly be-
could accurately measure changes in family inter- came apparent that it was necessary to assess fami-
action. In the 1960’s we began with exhaustive ly interaction at many different levels. The FICS
narrative accounts of family interaction, but this seemed uniquely adapted to the description of the
data collection technique was extremely cumber- molecular structure of interaction sequences; but
some. The next step was to develop an item pool it was a clumsy and inefficient method for sam-
for a standard coding system that sampled discrete pling low base-rate events. Given that the key re-
units of deviant and coercive behavior. To accom- ferral problem for half of our sample was stealing
plish this, we spent time in the homes of both nor- and/or fire-setting this quickly generated a crisis.
mal families and families with antisocial children. This exigency led to the introduction of the Parent
In terms of clinical impressions, these two groups Daily Report (PDR); data was collected by con-
were sharply delineated; being an observer in one tacting the parents daily for a report on specific
home was a very different experience from observ- problem behaviors that had occurred during the
ing in another. The function of the code that previous 16 hours.
emerged was to reflect these differences in family The purpose of this chapter is to review the de-
interaction. Eventually, there was a consensus velopment of the FICS and PDR as assessment de-
among staff members that the code categories ade- vices. Particular emphasis has been given to the
quately described aversive behaviors in family in- methodological problems which are inherent in
teraction. However, we knew at the time that our these data collection systems. The psychometric
coverage of prosocial interactions was far from properties of both devices are also discussed.
complete.
After three years and six revisions, the Family Prior Observation Approaches
Interaction Coding System (FICS) was ready for Reviews of early applications of observation
application as an assessment device (January, methodologies with children (e.g., Wright, 1960)
1968). By that time, the preliminary psychometric reflect a narrow conception of the scope and scien-
evaluations had been completed and the results tific usefulness of observation procedures. Not
were promising (Reid, 1967). Since then, the FICS only were early observational studies limited pri-

41
marily to preschool subjects, but the settings sam- haviors was expanded far beyond the limits of the
pled were, for the most part, restricted to nursery nursery school classroom. Barker and his group
schools. In addition, the data consisted of simple emphasized the necessity of collecting data relat-
frequency counts. As Wright (1960) pointed out, ing to such macro-units as Behavior Episodes, Be-
such gross descriptive data simply did not allow havioral Settings, and Environmental Force Units.
for precise hypothesis testing. In fact, these early Observation data were collected in narrative form.
studies generated few hypotheses about children’s It was assumed that discrete units of “behavior”
social behaviors. The main contributions of these occurred in nature and could be identified by ex-
early observational investigations were their so- amining these narrative accounts. The hypotheses
phisticated studies of the psychometric problems generated by these data were of sufficient quality
involved in the collection of observation data. that by the mid-1960’s, reviews of the literature
Analyses by Thomas, Loomis, and Arrington were optimistic about the utility of this approach.
(1933) of observer reliability, stability of event Furthermore, the large number of hypotheses tested
sampling, and observer error were models in this by these data seemed to attest to the power of this
regard. Although crude by today’s standards, the approach.
data collected in these early studies merit closer re- Recent observational studies (e.g., Bobbitt et al.,
view than they have received. Some of the findings 1969; Caldwell, 1971) differ from the pioneering
relate to current issues and theories. For example, efforts of Barker and his colleagues. Rather than
an incidental finding reported in the classic study stressing exhaustive narrative reports of interac-
by Goodenough (1930) showed that situational tion, the emphasis is upon tailoring the code sys-
changes produced differential effects on behavior. tems to test hypotheses about limited aspects of
Early work in the development of observation behavior. Rather than simply analyzing event fre-
techniques, except in the area of child psychology, quencies, behavior is examined for interdependen-
was astonishingly unsophisticated regarding the cies with environmental events. The theories which
measurement problems inherent in the collection emerge are interaction theories. Our experience in
of observation data. This is reflected in a review developing the FICS suggests that a coding system
by Heyns and Lippitt (1954), which points out must be tailor-made for each setting and perhaps
that the problems of event sampling, observer for each purpose. Asa result, it appears unlikely
presence, observer bias, observer drift, sequential that it will ever be possible to develop an omnibus
dependencies, and validity had hardly been con- coding system that could sensitively measure all
sidered, much less investigated. Certainly, the gen- child behaviors in all settings. The FICS primarily
eration of investigators trained in the 50’s and 60’s samples coercive behaviors that are used among
in the complex problems of response bias, social family members. The data that are generated lend
desirability, test-retest reliability, internal reliabili- themselves to the analyses of changes in rates of
ty, itemetrics, and construct validity could hardly specific behaviors that are of clinical interest.
be expected to take seriously a branch of science They also lend themselves to analyses of more
that ignored these realities. Recent reviews of ob- complex units such as the frequency with which a
servation techniques by Wiggins (1973), Johnson parent responds punitively to a specific kind of
and Bolstad (1973), and Jones, Reid, and Pat- child behavior (e.g., Taplin & Reid, 1977). Of
terson (1975) take the position that all of the tra- even greater interest are the data showing changes
ditional psychometric puzzles are found within the in the behavior of dyad members as a function of
observation methodologies. Jones (1973) suggests extended interaction chains (e.g., Patterson,
that there are, in fact, additional problems raised 1974c).
by observation methodologies for which tradition-
al assessment literature has no answers.
Parent Report
In the early 1950’s, Roger Barker and his col- as an Assessment Device
leagues devised a methodology that included de- Our approach to family therapy stresses the
scriptions of the environment as it interacted with need for change in intervention procedures (Pat-
the child (e.g., Barker, 1951, 1968; Barker & terson, Reid, Jones, & Conger, 1975). The reason
Wright, 1954; Willems & Raush, 1969). This was for this emphasis lies in the fact that the best exist-
a significant development in that Barker was one ing procedures meet only the minimal require-
of the first to adopt an interactive position. In the ments for effective treatment of problem children.
two and a half decades following the publication Although outcome studies have shown that many
of Barker’s One Boy’s Day (1951), data were col- severely aggressive children can be helped by the
lected over a broad range of settings in both the present procedures (Patterson, 1974a, 1974b,
USA and England. The range of subjects and be- 1975b; Reid & Patterson, 1976), a significant mi-

42
nority realize no measurable benefits. In addition, in deviant behavior. In this context, the term glob-
two follow-up studies suggest that some parents al refers to attempts by parents to synthesize infor-
slip back or retain many of their earlier modes of mation from extended time periods (e.g., more
interacting with their children (Patterson, 1976; than 24 hours) and/or covering a range of broad-
Taplin & Reid, 1977). Although a perfectly effec- spectrum variables which are defined only by con-
tive treatment for aggressive children is unlikely, ventional usage (e.g., such terms as “warmth,” or
the odds for therapeutic success can always be im- “destructive”). Such global judgments have typi-
proved if treatment procedures are continuously cally been made in the absence of prior systematic
modified on the basis of outcome effectiveness. observations by the parent.
The possibility of improving therapy practices Child-rearing. One commonly used method for
comes not from the therapist’s good intentions, obtaining data from the parent is the clinical inter-
but from the therapist’s use of frequent inputs of view. In some instances, the parent uses rating
relatively high quality data. This contingency ar- scales to make judgments of the child’s adjust-
rangement requires that the therapist receive at ment. In other cases, the interviewer summarizes
least weekly assessments of the behavior of the the information provided by parents. Regardless
child, and perhaps of his own therapeutic behav- of format, there are studies which suggest that dif-
ior as well. Given feedback of reasonably reliable ferent interviewers obtain very different informa-
and valid data, then the therapist is in a position to tion from parents, and that the same interviewers
continuously change his own behavior to achieve receive different information from parents from
optimal effectiveness. Over time, favored tech- one interview to the next (e.g., MacFarlane, Al-
niques which are not supported by the data must len, & Honzik, 1962).
be discarded. The void created by discarded tech- Much research on socialization has been based
niques must be filled by innovations which in turn upon parents’ recollection of childrearing pat-
will be evaluated. Over time, one would expect terns. The reliability of such recall, however, has
the treatment procedures not only to look differ- repeatedly been called into question (e.g., Robins,
ent, but also to become increasingly useful. 1963; Radke-Yarrow et al., 1964, 1970).
Therapists in traditional treatment settings have These data which show that parents are not al-
almost always received verbal feedback from par- ways accurate in reporting information about
ents and sometimes from teachers. However, the their children raise doubt about the utility of in-
reviews of child therapy studies by Levitt (1957, tensive efforts to get developmental histories using
1971) suggest that over a period of two decades, traditional child guidance interview procedures.
there have been xo increases in the efficiency of The validity of parent global ratings has also
traditional therapies being applied by well-trained been checked by comparing such ratings with re-
personnel. The present writer assumes that this ports from the child, school records, and system-
impasse occurs because the feedback data given to atic observations. In one study (Burton, 1970),
therapists have been consistently biased. The bias mothers’ judgments and school records showed
is a relatively consistent tendency on the part of low correlations. A study by Schelle (1974) showed
parents of disturbed children to report improve- no correlation between changes in. school atten-
ment in the behavior of the child when in fact no dance reported in school files and parent reports.
real changes have occurred. Data reviewed in the Burton (1970) reported little convergence between
section which follows suggest that roughly two- parent and child judgments about trait measures
thirds of such parents will report improvements describing the child. This finding is consistent
when asked for global judgments. This means that with the low level of agreement (40%-60%) be-
even if the therapy isn’t working, the therapist will tween parent and child for even relatively well-
receive supportive comments from the majority of specified symptoms such as fears, overactivity,
the families with which he works. Given such rein- and temper tantrums found by Lapouse and Monk
forcement, there is, of course, little reason for the (1958).
therapist to change his behavior. This lack of support for the validity of parents’
global judgments is also found in studies using ob-
Reliability and Validity of servation procedures as a criterion measure. These
Parent Global Report Data studies show only low level correlations between
The following is a review of the research per- ratings based upon interviews with the mother and
taining to three different types of parents’ global observations of mother/child interaction (e.g.,
judgments: (a) parents’ descriptions of child-rear- Antonovsky, 1959). Many of our “facts” concern-
ing practices, (b) parents’ descriptions of their ing the functional relationships between mothers
child’s behavior, and (c) parents’ reports of changes and children may be determined primarily by the

43
method of data collection, e.g., interview versus nonoccurrence of a list of behaviors for which they
observation (Bing, 1963). Although. some studies had come for assistance. Both sets of criterion data
show modest agreement for molar variables across showed significant decreases in observed rates of
methods (e.g., Smith, 1958), the majority of targeted deviant behaviors for families in the ex-
studies show discontinuities (e.g., Baumrind & perimental group. There was a nonsignificant in-
Black, 1967; Burton, 1970; Honig, Tannenbaum, crease in rates for the families in the placebo
& Caldwell, 1968; Sears, 1965). group. However, parents’ global descriptions of
In examining the problem further, it seems that improvement in their children revealed that all of
persons living within the same system do not nec- the parents in the experimental group and two-
essarily agree in their global perceptions of child thirds of the parents in the placebo group believed
behavior. Mothers’ and fathers’ trait ratings of that the child was “improved.” The performance
their children show only modest positive correla- of the placebo group is of interest because it exact-
tions (Becker, 1960). Similarly, Novick, Rosen- ly matches Levitt’s (1971) estimate of the propor-
feld, and Block (1965) showed only 36% agree- tion of disturbed children who will improve with-
ment between parents in identifying problem be- out treatment.
haviors in their own child. The conclusion to be The tendency of parents to overestimate treat-
drawn from these studies is that parents are unable ment efficacy was also shown in two other studies.
to provide reliable or valid global reports of their Patterson and Reid (1973) reported that observa-
children’s behavior. tion and parent daily report data agreed in assign-
Therapy change. Several studies suggest that ing 63% of a sample of treated families to an “im-
parents have a bias to report improvement in the proved” status. However, parents’ global judg-
behavior of problem children when no observable ment data suggested 100% improvement. The au-
changes have occurred. This is of some impor- thors would obviously like to accept the latter fig-
tance because of Levitt’s (1957, 1971) claim that ure, but our clinical judgment suggests that we
better than two-thirds of non-treated emotionally were really helping about two out of three fami-
disturbed children improve over time. His reviews lies. Johnson and Christensen (1975) carried out a
were based upon studies which used parent report similar treatment study, using comparable criteria.
as the primary outcome criterion. Given the pres- Their observation data showed that 38% of the
ence of parental bias his estimates for base rates of treated children showed significant improvement
spontaneous change would seem considerably in- in their behavior, while parents’ global judgments
flated. suggested that 93% had improved! A study by
In a study by Collins (1966), the treatment of Schelle (1974) evaluated the validity of parental
children with problems had been delayed and it estimates of improvement on an extremely specific
was therefore necessary to re-establish a baseline behavioral dimension: school attendance. He re-
measure of adjustment. The parents did not know ported that for the group in which parental ques-
of this situation. When asked for new ratings of tionnaire data showed the greatest improvement,
their children, they assumed the children had been the actual school attendance data showed that the
receiving treatment. Their ratings showed signifi- children were worse! While much more work must
cant improvement, even though treatment had not be done on this problem, it is the author’s opinion
yet begun. A similar finding was obtained for ‘a that pre- and post-test measures of parental global
no-treatment control condition in a study by Clem- estimates of change in problem child behavior will
ent and Milne (1967). It is possible that the chil- tend to overestimate treatment effects.
dren’s behavior improved simply as a function of Improvements in Parent Report Data. In retro-
the passage of time. However, the findings also spect, it seems likely that the wrong questions
support the hypothesis that parents perceived have been asked of parents. Global judgments
change when none, in fact, occurred. have been elicited which require memory of events
A study reported by Walter and Gilmore (1973) over long time spans. For example, some investi-
showed that parents’ global ratings of changes gators have obtained information from mothers
were unrelated to the actual behavior of the child. about child behaviors occurring over a twenty-
In that study, families were randomly assigned to four hour time period. In addition, parents have
experimental and placebo groups. Multiple criteri- been required to use complex and/or poorly de-
on data were collected during baseline and again fined variables. A study by Douglas et al. (1968)
five weeks later to assess changes in child behav- showed rather substantial correlations for parent
ior. Observation data were collected in the homes. and observers’ data when the child behaviors were
In addition, each parent, during baseline and in- well defined. It would seem more reasonable to
tervention, made daily reports on the occurrence/ make more modest demands upon the parents and

44
in the process perhaps obtain higher quality data. sumed that asking parents to make binary deci-
A laboratory study of three mother/child pairs by sions (occur/nonoccur), covering only the preced-
Peine (1970) showed promising results. Each ing eight to ten hours, will minimize distortions in
mother/child pair participated in a series of five memory and judgment. The current practice is to
laboratory sessions in which the mother was asked express this score as mean frequency of symptoms
to count well defined behavioral events as they oc- per day. The reliability, stability, and validity of
curred (e.g., such behaviors as “touch toys,” “fol- PDR data will be discussed in the psychometric
lows directions,” and “aggressive”). Many of the properties section of this chapter.
intra-subject correlations between observers and The PDR criteria may well turn out to be the
mothers were in the .80’s and .90’s. However, the first choice as an evaluation measure. Assuming
mothers consistently underestimated the /evel of that it survives further psychometric analyses, it
deviant behavior by as much as 600%! This find- has some definite advantages. First of all, it is a
ing is consistent with a study reported by Herbert means for obtaining data describing low base-rate
(1970) in which low absolute agreement was found events which are often foci for treatment but sel-
between observers and parents. It seems that un- dom seen by observers (e.g., fire-setting, truancy,
der the best of conditions, parent data may pro- stealing). Second, PDR is much less expensive to
vide an accurate estimate for the ordinal rankings obtain than the observation data.
of subjects but may underestimate actual mean In current clinical practice, the PDR score con-
level. stitutes one of our two major criteria for evaluat-
In our treatment research, parent report data ing treatment. Its long range utility will be further
are consistently collected at three different levels defined by its correlation with police offense rates
(Patterson, Reid, Jones, & Conger, 1975). The as the child moves through adolescence. In lieu of
levels vary in the specificity of the behaviors and these data, which are now being collected, it might
the time intervals involved. First, parents are asked be noted that the PDR score at termination is the
at termination to fill out rating scales which re- Oregon Social Leaning Center’s best predictor of
quire global estimates of improvement such as status during follow-up. Using six months’ follow-
those employed by Patterson and Reid (1973). In up data, the termination PDR score was in 83%
accord with our own findings and those of John- agreement with the follow-up PDR score and in
son and Christensen (1975), it is expected that 58% agreement with the Total Deviant follow-up
better than 90% of the parents will perceive some score. Again, success was arbitrarily defined as
areas of improvement. At this level, parent global 2 30% decline from baseline levels.
judgment is viewed as a necessary, but not suffi-
cient, criterion for status as a success.
Development of the Family
Second, a method is used for obtaining more Interaction Coding System (FICS)
differentiated data from parents, which systemat- In the mid-1960’s a social learning group was
ically samples centroid factors describing child formed at the University of Oregon (G. Patterson,
personality. A significant change between pre- and D. Anderson, W. Bricker, M. Ebner, A. Harris, R.
post-ratings would be viewed as a necessary, but Littman, J. Reid, and J. Straughan). Their concern
not sufficient, criterion for success status. The with disturbed children led them into classrooms,
scales and scoring keys are to be found in Patter- homes, and institutions. Several methods were
son, Reid, Jones, and Conger (1975). tried as a means of collecting data in these settings.
Third, the Parent Daily Report (PDR) criterion At first, narrative accounts were written in long-
has been developed to provide a more powerful hand describing the target subject’s behavior and
criterion measure based upon parent data (Patter- the environmental consequences supplied for it
son, Cobb, & Ray, 1973). As part of the intake (Buehler, Patterson, & Furniss, 1966). To speed
interview, the parents are asked whether each of up this process, aggressive events were described
the list of 31 “symptoms” are of sufficient concern on magnetic tape using the face-mask microphones
to warrant changing. The list is published in Pat- and portable tape recorders developed by the
terson, Reid, Jones, and Conger (1975). Parents Barker group (Schoggen, 1964). These tapes could
are asked to collect data on the occurrence of these then be transcribed and coded (Patterson, Mc-
symptom behaviors during two weeks of baseline Neal, Hawkins, & Phelps, 1967); but the cost was
and at the end of treatment. These data are col- very high. In addition, the data analysis required
lected each day; the number of problems picked by endless tabulations. As a result, it was decided to
each family usually ranges from three to nine. The construct a coding system that could be used in a
parents indicate the occurrence or nonoccurrence field setting. Furthermore, the coding system was
of each of those events during a given day. It is as- designed so that the data could easily be punched

45
 Table 3.1
Behavioral Code Definition
AP Approval HU* Humiliate PP _—Physical Positive
AT Attention IG* Ignore RC Receive
CM Command IN Indulgence SS Self-stimulation
CN* Command Negative LA Laugh TAR dial
CO Compliance NC* Noncompliance Ege elease
GR> Gry NE* Negativism TH Touch
DI* Disapproval NO Normative WH* Whine
DP* Dependency NR _ No Response WK Work
DS* Destructiveness Pi eePlay: YES? Yell
HR* High Rate PN* Physical Negative
*indicates categories that are included in the Total Aversive Behavior (TAB) score.

onto cards and stored for computer analyses. tial form were effective in generating many hy-
As an opening strategy, it was decided that potheses about aggressive behaviors occurring
progress could be maximized by focusing upon the within families. The preliminary psychometric
assessment of a single type of problem, the severe evaluations were promising (Patterson, Cobb, &
conduct problem child. This meant giving up in- Ray, 1973; Reid, 1967).
vestments of clinical time already made to the as- The FICS was designed to describe aggressive
sessment and treatment of autistic, phobic and behaviors and the antecedents and consequences
withdrawn children. From the late 1960’s on, associated with them. About half of the code cate-
most of the time and energy of the reconstituted gories in the FICS describe aversive behaviors. The
social learning group (G.R. Patterson, J. Reid, R. other half of the categories include various proso-
Ray, D. Shaw, J. Cobb, H. Hops, and K. Skin- cial behaviors. Currently, the FICS is composed of
drud) was invested in three complex problems: (a) the 29 code categories listed in Table 3.1. A defini-
how to conceptualize the process of aggression in tion is provided for each behavioral code in Ap-
families; (b) how to treat the families in which it pendix 3.1.
occurred; and (c) how to measure changes in ag- Family interactions were categorized into dis-
gressive family interactions. crete units as they occurred. The observer alter-
The first task in developing this observation sys- nately coded, in sequence, the behavior of the sub-
tem was to come up with an item pool or set of ject and then the person(s) with whom the subject
categories which could constitute an observation interacted. Each event was described by code let-
system. The “items” in this case were specific be- ters referring to the category to which it was as-
haviors thought to be relevant to either the clinical signed, together with the number identifying the
or theoretical purposes of the assessment task. family member with whom the target subject was
Clinical experience and discussions with parents interacting. If required, several subject numbers
of socially aggressive children suggested some rele- and code categories were used to describe a single
vant behaviors. These clinical impressions provided interaction. For example, a child could be crying
a general perspective from which to view family and hitting at the same time.
interaction. The investigators ventured into the The data were recorded continuously, and pro-
homes of families with antisocial children. It was vided a relatively complete, sequential account of
this experience which led to the development of the interaction of a target subject with all other
the specific code categories which define the FICS family members. The observer received an audi-
(Reid, 1967). Three years of study were required tory signal every 30 seconds from a device built
for the development of a workable field observa- into the clipboard. At the beginning of each 30-sec-
tion coding system. After six revisions, the data ond interval the observer shifted to the next line of
showed that the observation procedures provided the protocol sheet. On the average, observers were
a reasonably complete description of family inter- able to record five interaction units (both members
action (Patterson, Ray, Shaw, & Cobb, 1969; of a dyad) every 30 seconds. For each observation,
Reid, 1967). In addition, it seemed that the field the order in which family members were selected
observation data collected in continuous sequen- as target subjects was randomly determined. Each

46
 Figure 3.1
Diurnal Variations in Frequency of Outbursts
(from Goodenough, 1931, p. 107)

mieZ.

alti

10

.09

08

hour
Frequency
per .07

06

-05

-04

03

02

01

.00

7:30 8:30 9:30 10:30 11:30 12:30 1:30 2:30 3:30 4:30 5:30 6:30 7:30 8:30

Hour of day

family member served as target subject for five Some Practical Problems
minutes; then the whole series was replicated. The During these first several years we had very little
observers keypunched the data shortly after re- information regarding the procedures for observ-
turning from the field. ing family interactions in the home setting. As a
In the chapters which follow, many of the varia- result, many of our early decisions concerning
bles will be expressed as “rate per minute.” Within how these observations were to take place were
the present context, the upper limit of the rate per made on atrial and error basis. We wondered, for
minute is approximately ten six-second responses example, if we should schedule our observations
per minute for any given subject. In addition, be- for some standard time of day. If so, what time of
cause the coding system “chunks” the behavior of day was best suited for observing aggressive family
the target person into six-second intervals, what interactions? The best data that we could find re-
may have been “continuous” events become dis- lating to that question were collected a half-cen-
crete occurrences, e.g., a child crying for 18 sec- tury ago by Florence Goodenough in her classic,
onds would be coded as a sequence of three “Cry.” Anger in Young Children (1931). As shown in fig-
For this reason, it is not possible to differentiate ure 3.1, the diaries kept by mothers of preschool-
recurrence from duration. aged children indicated that there were two prime

47
 \ | ye

©)
oO
\
2

2
©
hos eel ca
ol dee

Illustration #2: Family Interaction in the Age of Technology

times for “conflict.” The witching hours for these problem to our attention. For example, as the ob-
young children seemed to be just prior to lunch server walked in the front door, the family simply
and just prior to dinner. The data were in agree- melted away. Somebody had to go to the grocery
ment with our own hunches, so for the next dec- store, another person disappeared into the bath-
ade most of the families were observed just prior room. The remaining two turned on the television
to, or during, dinner. and remained transfixed for an hour (see illustra-
In the beginning our-goal was to observe fami- tion #2). The staff discussed this problem at length,
lies in “natural” settings. But, as noted earlier, our and finally decided that the data would have to be
very first contact with families demonstrated that collected in semi-structured home settings. This
this was going to be difficult to achieve. The first structuring of the home environment was repre-
“subject” of these attempts at observation chose to sented by a set of formal rules for observation ses-
climb an apple tree instead. The next few families sions as set forth in Table 3.2.
that were observed brought other aspects of the As we proceeded, it became clear that in some
real sense we were not viewing the family in their gan to measure, for example, a variable that we
ordinary environment. Many of the families that called family noncooperation. It appeared that
we were seeing for treatment told us that all of the those families who missed appointments, were not
family members were rarely in the same room at at home, or arrived late tended to be the ones that
the same time. As a consequence, the rules that the therapists later had the most difficulty in treat-
were being implemented for observation produced ing. An empirical study is now being carried out to
an unnatural situation. The wheel had come full determine just how useful this variable might be in
turn, for this was the very reason we had left the predicting successful treatment outcomes for indi-
laboratory in the first place. However, the trade- vidual families.
Observer Effects on the Data
Table 3.2 Structuring the home environment was our
guarantee that the observers would see the events
Rules for Observation Sessions
that we wished to study. But we had to account for
(from Reid, 1978, p. 8) several other variables before we could assume
Everyone in the family must be present. that the data collected by observers represented a

bh
WwWN
No guests.
The family is limited to two rooms.
true
homes.
picture of what
It occurred
was
to us that
really going
the use of observers
on in these

The observers will wait only 10 minutes for all to be present to record behavioral events in the home may gen-
in the two rooms. erate other sources of distortion in the data. As a
Telephone: No calls out; briefly answer incoming calls. result, a series of studies were designed to measure
HN
oN
No TV. effects of observer agreement (reliability), observ-
No talking to observers while they are coding. er drift, observer bias, and observer presence on
Do not discuss anything with observers that relates to your the data. The results of these investigations are
problems or the procedures you are using to deal with them. summarized in the sections which follow.

Observer Agreement
off for bringing the family together in a semi-struc- Reliability is a basic requirement for any meas-
tured environment was that it maximized the oc- urement system. Observer reliability in coding is
currence of the events we wished to study. Studies defined as the degree to which observers code be-
are now being designed for the new interaction haviors in accordance to some predefined criteria
code that will tell us if imposing this type of struc- (e.g., a comparison of one observer’s protocol
ture was, indeed, the best way of achieving this [coding sheet] to that of another; a comparison of
maximization. one observer’s protocol to a precoded videotape of
Another immediate concern was whether or not a family session). In our system, every third home
observers should talk or otherwise interact with observation session during baseline, and every
the families that they were observing. Most of the fourth session thereafter is attended by two ob-
observations were of families referred for treat- servers. The percentage agreement level is then
ment. The OSLC staff devoted an enormous calculated separately for each 30-second line on
amount of time and energy to helping these fami- the protocol:
lies. As a result, the observers quickly became en- number of frames of agreement
meshed in the treatment process taking place in
number of frames of agreement plus number
the families they observed. These developments
of frames of disagreement
led to rule number seven (Table 3.2). Families
were told before the study began that the observ- A frame is defined as a six-second time sample
ers were forbidden to talk with them. The routine of behavior, and is subdivided into two parts: the
for observers was to enter the home at the ap- subject number and antecedent behavior (part
pointed time, greet the family, seat themselves, one) and the respondent’s number and consequa-
and begin their observation procedures. For a tion (part two). Agreement involves all four com-
number of years, the observers also called the fam- ponents. An instance of double-coding (two be-
ilies at regular times every other day to obtain Par- haviors per subject or respondent) would be con-
ent Daily Report data. sidered an additional interaction segment, and
Over the next decade, several spinoffs occurred would add to the denominator. Therefore, while
that were not anticipated. These were primarily one line on a rating sheet contains five six-second
new assessment devices which may well turn out time samples the total number of agreement points
tobe as useful as the coding system itself. We be- may exceed ten points if either the subject or other

49
family members emitted more than one behavioral
response during any six-second time sample (or Table 3.3
frame) on that line. Observer Reliability
Calculating observer agreement on a line-by-line
basis provides a general estimate of the quality of (from Reid, 1978, p. 77)
the data. Typically, these data on interobserver re- Percent

liability are obtained on a regular basis for project Code Rxy' agreement?

cases and posted on a bulletin board. It is expected

AP 76 54
that the values will not fall below a mean of 70%
AT 96 90
for a given observer pair on any given session. The
CM Ae 86
average is around 75%. As pointed out by Cohen
CN .66 65
(1960) and others, the interpretation of this agree-
CR .96 V2
ment score must take into account the level of
DI 92 72
chance agreement. Cohen’s Kappa is such a means
DP .88 91.
of expressing agreement, and at the same time cor-
DS co2 HS
recting for chance level; it is, in fact, a preferred
HR 65 59
method for analyzing observer agreement.
HU 2. 74
As Johnson and Bolstad (1973) point out, there
LG 93 68
are a large number of methods one could use in
IN .86 61
calculating observer agreement. Each has its assets
LA .96 74
and liabilities. For some purposes, the technique
NC .67 61
outlined above might constitute an underestimate
NE Ao 38
of agreement. For example, the stress upon correct
NO ee 95
notation of events and sequence could very well be
NR 195, 96
irrelevant to tasks requiring only an estimate of
PL 1.00 Ts.
mean rate of occurrence. For these analyses, a
PN .94 AWE
more appropriate dependent variable could be the
PP nes 59
mean frequency of a given code category obtained
PX Not observed —
from the entire protocol.
RC .89 59
Table 3.3 summarizes the available data relat-
SS .67 30
ing to observer reliability on 28 of the code cate- TA wey 94
gories. As shown there, the agreement between
TE .86 56
observers present during the same sessions was
TH .88 91
reasonably good. The median correlation of .92
WH aa 49
(across categories) was as high as that found for
WK 94 92
some of the better assessment devices. Note, how- YE 80. 56
ever, that these important findings were based Median ey 72
upon only 11 pairs of observer protocols. This is
hardly a sufficient sample upon which to base im-
1. Based upon entry-by-entry agreement.
portant decisions about the precision of a measur-
2. Based on analysis of 11 protocols.
ing instrument.
It should be noted that structural analyses of the
type described in Chapters 8 and 9 are based upon may obtain very high observer agreement. Reid
statements such as p(A|B) and so forth. As pointed has worked with several methods for calculating
out by Hartmann (1977), global measures of relia- complexity. For example, if complexity was de-
bility hardly apply to such variables. What is fined as N different entries divided by the total en-
needed are estimates of the reliability for these tries for five minutes, then the correlation between
functional indices. To the writer’s knowledge, percent agreement and complexity was —.52 (p<
such an analysis has been made in only one study .001). Those same measures were then applied to
(Patterson & Moore, 1979). The data from a sin- another set of OSLC data; the correlation was
gle case study showed the conditional p values cal- —.75 (p <.01). These results indicate that reliabil-
culated from protocols of two observers were in ity was lower for the more complex protocols.
good agreement. Skindrud (1972) defined complexity as the per-
J.B. Reid has noted that observer agreement is cent of unreplicated interactions within each seg-
largely a function of the complexity of the interac- ment. This measure correlated —.53 with observ-
tion. By selecting simple interaction segments, one er agreement for one sample, and —.65 for anoth-

50
 Figure 3.2
Decay in Observer Reliability
(from Taplin and Reid, 1973, p. 551)

End Training Data Collection

100

No check group

@———— Spot check group

90
Random check group

=E ;80
Spot check
5
vo
A)
iT Spot check
<
70
ae

60

Sessions

Reprinted with permission from “Effects of instructional set and experimenter influence on observer reliability,” Child Development,
1973, 44, 547-554. Copyright 1973 by The Society for Research in Child Development.

er. Reid extended these analyses in the report by covertly assessed as to their accuracy. On the first
Jones et al. (1975). He found, for example, that day of covert assessment, there was a mean drop
the protocols for child subjects were consistently in accuracy of 25%! Romanczyk et al. (1971)
less reliable than those for adults. They were also found that their observers performed at 76% ac-
significantly more complex. curacy during overt checks, and 39% when checked
Future presentations of this kind will probably covertly.
be centered around a different theory of measure- Taplin and Reid (1973) then began to analyze
ment. The theory of generalizability by Cronbach, those variables that might explicate this phenome-
Gleser, Nanda, and Rajaratnam (1972) provides a non. It was now obvious that being a “reliable ob-
very different conceptual base for considering server” was not a Static disposition and therefore
problems such as agreement, stability, and validi- the entire enterprise was in jeopardy. In this first
ty. So far, we have only begun to apply this model study, they compared the performance of three
to OSLC data; this was done by Jones et al. (1975). different groups. The first group was told that
The model will be used to analyze our new obser- their data would not be checked; the second was
vation coding systems. told that their protocols would be compared to the
Skindrud (1972) found that he could train ob- criterion observer on a random basis. The third
servers to be reliable in the use of the FICS in group was told they would be spot-checked at reg-
about 20 hours. This figure seems fairly represen- ular intervals. There were six observers in each
tative of our experience since that time. However, group. The results are shown in Figure 3.2. The
Reid (1970) was convinced that observers became gradual decay from the training level of 80% was
highly variable after they had been trained. He significant for all three groups. The no-check
used a variant of the FICS, and trained his observ- group showed the largest decrement; and, al-
ers until they reached the usual criterion of accura- though the difference is not statistically signifi-
cy in coding videotapes of mother-child interac- cant, the random-check group performed the best.
tion. Following this training, the observers were The next study examined the possibility that the

51
observer group itself could be used as a feedback additional studies from both the Stony Brook and
and support device to maintain reliability. DeMas- the Oregon laboratories using well-trained observ-
ter, Reid, and Twentyman (1976) trained and ran- ers have not supported that finding.
domly assigned 28 observers to one of three Skindrud (1973) described one of the first well-
groups. Group 1 received daily feedback as to ac- controlled field studies testing for the effect of bias
curacy, and discussed the prior protocols with the upon well-trained observers. These observers had
group. Group 2 also held daily discussions about several years’ field experience and regular (re)train-
the finer points in coding decisions, but was not ing sessions in applying an earlier version of the
given feedback as to accuracy. Group 3 was given FICS. Over a period of several years, two different
no feedback and did not participate in group dis- “calibrating observers” had offices separated from
cussion. Group 1 did significantly better than eith- the rest of the project. They were kept uninformed
er of the other groups; in fact, their accuracy im- of the status of the families being treated. The reg-
proved over time. Group 2 observers tended to ular cadre, on the other hand, knew which fami-
drift away from their original decision rules. lies were in the clinical and which were in the
This important series of studies by Reid and his matched normal samples. They also knew for the
students led to permanent changes in the data col- clinical sample whether the family was in baseline,
lection procedures employed at OSLC. Observer treatment, or follow-up. A comparison of the
training sessions are now held every other week. rates of deviant behavior obtained from the in-
All four observers view a tape, code it, and then formed versus the uninformed observers showed
discuss any decision problems that they have en- no significant effects for either family status (clin-
countered. At regular intervals, pairs of observers ical or normal) or for treatment status (baseline or
accompany each other into the field. After their termination). However, the small sample involved
protocols have been tabulated, their agreement in this field study led to the decision to carry out a
scores are posted. large-scale laboratory study.
In this study, Skindrud (1972) enrolled 28 ob-
Observer Bias servers (women from the local community) in an
Rosenthal (1966) has defined experimental bias intensive three-week training program. They
as the extent to which experimenter effect or error learned to use the present code by viewing video-
is asymmetrically distributed about the “correct” tapes of parent/child interaction. The minimal
or “true” value. In the present context, such dis- level of accuracy required to participate in the re-
tortions could take the form of a constant bias on mainder of the study was 70%. The trainees were
the part of the observer. Conceivably, some of then divided into three groups. One group was
these errors could arise because observers inten- given a bias to expect the next series of 12 sessions
tionally distort their data. While intentional errors to reflect a 30% increase in deviant child behavior.
in recording and computation have been reported A second group was given a bias to expect a de-
in the literature (Azrin, Holtz, Ulrich, & Goldia- crease of the same magnitude. A third group was
mond, 1961; Rosenthal, 1966; Rosenthal & Fode, not instructed as to the experimenter’s expecta-
1963; Rosenthal & Lawson, 1964), those reports tions. The data indicated that there was no signifi-
have primarily related to students serving as data cant effect of bias upon the data coded by the ob-
collectors for their instructors. The possibility of servers. Even the trends were not in accord with
observer error, whether intentional or not, must, the expectations the experimenter had given to the
however, be guarded against. The most likely groups. Thinking that perhaps the less reliable ob-
mechanism for creating observer bias would seem servers might reflect the bias, a separate analysis
to be the situation in which the experimenter com- was run for this subgroup. Again, there were no
municates his expectations to the observers, there- significant effects. When tested to determine their
by exerting a subtle influence on the moment-by- understanding and acceptance of the experiment-
moment coding decisions they make. Although er’s expectations, the observers showed that they
few studies have systematically assessed the effects did understand. A power analysis showed that
of observer bias in the natural setting with human groups of the size used in the second study were
subjects, many researchers have taken measures to sufficient to detect a bias greater than 15%. Taken
minimize its potential effects (O’Connor, 1969; together, the studies suggested that if well-trained
Thomas, Becker, & Armstrong, 1968). A prelim- observers are biased, the magnitude of the effect is
inary study by Kass and O’Leary (1970) showed not very large.
that experimenter expectations produced a small Kent, O’Leary, Diament, and Dietz (1974) rep-
but significant effect on the data coded by one licated Skindrud’s findings. Ten well-trained ob-
group of observers, but not for another. However, servers were assigned to one of two groups. One

52
 group was told the next series of videotapes would present versus intervals when the observers were
show a child getting better, and the other group absent, the data indicated that there was no effect
was told that they would show a child getting on the mothers’ ratings. These data supported
worse, Again, there was no significant effect of ex- none of the various hypotheses being tested relat-
perimenter expectations upon the data actually ing to the impact of observer presence on mother /
coded by the observers. In this analysis, they even child interaction.
searched for the effect by analyzing each code cat- One can conceptualize the problem of observer
egory separately. Experimenter expectations were, presence in a variety of ways. However, the studies
however, dramatically reflected in the observers’ completed thus far seem to arrange themselves in
global judgments! Ninety percent of the observers terms of the following sequence of questions: (a)
in the first group thought they detected a decrease do subjects being observed notice the observer?;
in the child’s rate of deviant behavior. Seventy per- (b) does being observed produce increases in
cent in the second group thought they detected an socially acceptable behaviors?; (c) does observa-
increase. tion elicit a set to suppress socially undesirable be-
In his 1972 investigation, Skindrud reported a haviors?; and (d) can families “fake” good and
pilot study in which he reinforced the observers “fake” bad?
for collecting data congruent with experimenter Questions concerning the effect of the observer’s
expectancies. Even with an N of six he obtained presence upon the behavior of the target subject
changes in coded data of borderline significance. represent a central issue in evaluating the utility of
O'Leary, Kent, and Karpowitz (1975) carried out observation techniques. Earlier studies in this area
a study along the same lines that closely paralleled posed the question as an either/or proposition. At
field situations. Four observers viewed videotapes, this time, it seems more reasonable to assume that
and were told to expect changes in two code cate- the presence of an observer produces some altera-
gories. The observers were reinforced when their tion in ongoing behavior, e.g., at the very least,
coded data were congruent with the experiment- the target subject notices the observer. The more
er’s expectancies, and punished when they did not. productive stance is to assume some effect and to
The changes in the data for the two target categor- identify its magnitude and quality. It is suggested
ies were in accord with the reinforcement contin- that the kind of effect and its magnitude must vary
gencies, while code categories not involved did not as a function of setting and target subject. This
reflect these changes. These findings strongly em- seems reasonable enough, but how does one pro-
phasize the necessity for therapists not to provide ceed to measure such effects?
observers with this type of feedback on their ob- The task of constructing an adequate design to
servations. test this question has occupied investigators for
As things now stand, it does not seem that ex- the last decade. Three general approaches have
perimenter or therapist expectancies ipso facto been followed; none have proven entirely satisfac-
bias the data collected by well-trained observers. tory. One approach suggests that the target sub-
However, expectancies can affect observers’ global jects are aware of the observers, and that they be-
judgments about what they are seeing. Experi- have accordingly by presenting the most socially
menters could perhaps bias their coded data by us- desirable facade possible. Therefore, “aware” sub-
ing reinforcement contingencies, but this is not jects would be expected to demonstrate lower rates
likely to happen in most well-run field studies. of deviant behavior and higher rates of socially de-
sirable behavior throughout the sessions. As an al-
Observer Presence Effects ternative, one might expect that over repeated ses-
Each of us has experienced being observed, and sions the subject would habituate to the sensitizing
we know that in some settings such an intrusion effect of the observer’s presence, and gradually ap-
sets constraints upon some of our behaviors. The proximate baseline rates of deviant and socially de-
problem is that it is difficult to delineate these ef- sirable behaviors. The second and perhaps most
fects empirically. An unpublished study by Paul powerful approach has involved the use of experi-
(1963) showed that mothers being observed in mental manipulations in which subjects are re-
their homes “felt constrained.” In that study, ten quested to fake “good” and fake “bad.” The third
mothers and their preschool children were ob- approach involves using unobtrusive technology
served over a ten-week period. The mothers re- (hidden recording devices).
ported feeling very much aware of the observers’ Reactivity to Observer Presence. Pollack, Vin-
presence throughout the study. However, when cent, and Williams (1977) proposed a two dimen-
the mothers’ ratings of their children’s behavior sional model describing the demand characteris-
were compared for intervals when observers were tics involving observers. The first factor, nonspe-

53
cific reactivity, describes orienting behaviors that session was significant. In a study of adults, Moos
indicate the subject is aware of the observer. This (1968) compared the effects of psychiatric patients
may include behaviors signifying discomfort, self- carrying a microphone transmitter to the effects of
conscious “fiddling” behavior, and perhaps a gen- being observed. When psychiatric patients were in
eral increase in variability. These behaviors are a day room setting, the effect of the observers was
thought to habituate over time. The second factor, to elicit significantly higher rates of socially appro-
impression management, does not habituate; it priate behaviors such as smiling, talking with
implies a general attempt to present a socially de- hands, looking at the speaker, playing with an ob-
sirable demeanor to the observer. ject, and arm and foot movements. Zegiob, Ar-
There is ample evidence attesting to the viability nold, and Forehand (1975) studied 12 mother/
of the Pollack et al. (1977) nonspecific reactivity child pairs as they sat in a waiting room. On two
factor. For example, Barker’s (1951) data describes successive visits, they were either informed or un-
attempts by children to interact with the observ- informed of the fact that they were being ob-
ers. Observation data collected in nursery schools served. Under the informed conditions, there was
by Connolly and Smith (1972), showed high rates a significant increase in the amount of play inter-
of orienting behavior among the children, particu- action and in the mothers’ use of positive verbal
larly during the first few observation sessions. It comments and attempts to structure the interac-
should be noted that even though habituation ef- tion. Randall (1975) showed a significant effect
fects were reported over the eight sessions, the ori- for ten of the 27 mother/infant behaviors that
enting behaviors did not fall to zero. Grimm, Par- were examined. During observer present condi-
sons, and Bijou (1972) also noted in a classroom tions, the mothers talked less to their infants and
setting that high rates of orienting behaviors per- gave fewer directions and prohibitions. As might
sisted over a six-month period in which the chil- be expected, the infants’ behavior also showed sig-
dren were regularly observed. Studies of family in- nificant changes.
teraction also showed that parents remained very With children as subjects in a classroom setting,
much aware of the observers. An unpublished there seemed to be no consistent effect of observer
study by Paul (1963) showed that during a ten- presence upon child behavior. For example, a study
week series of observations, mothers’ global rat- by Surrott, Ulrich, and Hawkins (1969) involved
ings indicated a continuing awareness throughout observation of four target subjects in an interven-
the sessions. tion study. Comparisons of observer-present to
Bechtel (1967) devised some fascinating meas- videocamera-only conditions showed greater time
ures of nonspecific reactivity for visitors to an art working for all four subjects under the observer-
gallery. He assigned visitors to one of four groups. present condition. Mercatoris and Craighead
One group was asked to rank their preferences for (1973) extended the design to include all subjects
the pictures while an observer was present. A sec- in the classroom, and an ABAB reversal. The study
ond group was also asked to rank their prefer- lasted for 30 sessions, with the observer alternate-
ences, but remained unaware of being observed. A ly present during consecutive five- to ten-day blocks,
third group was asked to wait, and the fourth then absent during the B condition for an equiva-
group consisted of visitors following the usual lent amount of time. The videocamera was present
procedure. The effect of the overt observers was to during the entire study. There were no differences
reduce the number of seconds spent in that area; between conditions for rates of inappropriate or
their movement within the area was also more re- appropriate behaviors. Weinrott, Garrett, and
stricted. Todd (in preparation) studied six children in a spe-
In the series of studies concerning impression cial class. After manipulating observer presence
management, consistent changes in social behav- and absence conditions, they concluded that virtu-
iors were demonstrated, but the effects did not al- ally none of the variance in the data could be at-
ways fall neatly along a dimension of social desira- tributed to observer presence effects.
bility. The complex effects varied as a function of Neither the habituation nor the variability hy-
the kind of subjects and the setting. Nelson, Lip- potheses are supported by research findings. The
enski, and Black (1976) collected data on college Mercatoris and Craighead (1973) study failed to
students’ baseline rate of face touching. The mean find any changes in mean level for either appropri-
rate was 12.3 responses per session when the stu- ate or inappropriate behavior. As noted earlier,
dents were unaware that they had been observed. Patterson and Cobb (1973) and Johnson and Bol-
An observer was introduced and the students were stad (1973) found no significant changes in the
informed that face touches would be recorded. mean level over six to ten sessions in the home. A
The resulting decrease in rate to 8.4 responses per corollary of this hypothesis would be that the sub-

54
jects would display increased variability during the when comparing “look normal” to “look bad”
first few sessions, and then stabilize as they be- conditions. Ten of the 12 deviant families and nine
came accustomed to being observed, i.e., the esti- of the 12 nonproblem families seemed effective in
mate from the first block of sessions would tend producing this shift. These shifts were accompa--
not to correlate well with estimates from later ses- nied by significant increases in parental commands
sions. Neither the correlational analysis of class- and punishment, and a significant decrease in par-
room sessions nor comparable analyses of family ental positive consequences. The data from the
sessions supported the variability hypothesis. combined sample showed no significant shift in
In summary, the studies carried out to date pro- child behavior from the normal to the good condi-
vide consistent support for the Pollack et al. tions. Seven of the 12 normal families were effec-
(1977) nonspecific reactivity factor. As shown by tive in producing this shift, whereas only four of
their orienting responses, subjects definitely seem 12 in the problem sample were successful. Ques-
to be aware of being observed. There is also some tionnaires given to the parents showed they gener-
evidence for adults that support the impression ally perceived themselves as more effective in ac-
management factor. However, the sparse findings celerating good behavior than in accelerating bad
do not support a simplistic model of deviancy sup- behavior. However, parent perceptions of behav-
pression. Rather, it seems that adult social behav- ior change were not in accord with observations of
iors that are altered by the presence of an observer the same behavioral events.
vary as a function of both setting and sample. Weinrott and Jones (1977) followed a similar
Classroom studies of children offer little support format, but this time in the classroom setting. For-
for the impression management factor. There is es- ty pupil-teacher pairs from grades one through
sentially no support for either the habituation or three participated in the study. Half of these teach-
variability hypotheses. ers had identified a socially withdrawn child, the
Manipulation of sets to “fake.” If one assumes other half a disruptive child. Baseline observations
family members could distort their interactions were collected in all classrooms during all phases
when the observer was present, then it should also of the study. After two baseline sessions, the teach-
be possible to directly manipulate such a set. This, ers were instructed to make their target child look
in fact, was the format adopted in a series of clas- as socially desirable as possible. If the target child
sic studies by Steve Johnson and his colleagues. In was withdrawn, the teacher was to help the child
the first study (Johnson & Lobitz, 1974a), 12 par- appear to be as outgoing as possible. If the child
ents of normal families were instructed to make was disruptive, the teacher was to help him look
their preschool children “look bad or deviant” for as cooperative and as quiet as possible. During the
three observation sessions, and to “look good” on demand conditions, the behavior of the teachers
three alternate sessions. On “bad” days, the rates changed significantly in the expected directions.
of deviant child behavior and rates of negative and The withdrawn children showed significant changes
parent commands were significantly higher than from baseline toward a more outgoing pattern of
they were during “look good” days. The design interacting; but the changes in the disruptive chil-
made it possible to demonstrate that parents of dren were nonsignificant. These results were in
normal children could definitely alter the behavior keeping with the results from the Lobitz and John-
of their children. However, it was not clear that son (1975) study. The author believes that these
parents of distressed families would be equally results demonstrate the control exerted by the an-
skilled in controlling child behavior. tisocial child over his environment. It is apparent
A second study by Lobitz and Johnson (1975) that both normal and withdrawn children can be
involved volunteer parents of younger children. helped to “look good” (at least to a limited de-
One sample of 12 families had “problem children” gree); but it also seems that antisocial children are
as labeled by one or both parents. The children in not so readily manipulable.
the other sample of 12 families were presumably A more recent study involved distressed and
problem-free. On two consecutive days, the par- nondistressed married couples (Vincent et al.,
ents in both groups were instructed to make their 1979). The results again emphasize the constraints
children “look good”; then on two consecutive that coercive interactions have upon those who are
days to make them “look bad”; finally, on two involved in this process. In the laboratory ana-
consecutive days they were to “look normal.” logue situation, after a baseline period, the cou-
Families were randomly assigned to one of the six ples were instructed to fake good or fake bad. The
possible orderings of these conditions. As shown adults in both samples were able to significantly
in Figure 3.3., parents in both samples produced alter their verbal behavior in accord with the in-
significant increases in deviant child behaviors structions. For example, during fake good, the

ip)
 Figure 3.3
Fake Good, Fake Bad, and Normal Conditions for Two Samples
(adaptedfromLobitz& Johnson,1975,p. 72)

(oa ~~ 16
> >
cs he!
Ce Normal
fel
vo 2Vv
mM Sample
=y%D 12
<= o Deviant
Os Sample
= 3 10
TSwy
25
Y vo 8
An
aes
ate
neue
o
oe & Vv
= ca 4
avu1S
lee a?

Fake Good Normal Fake Bad

Experimental Conditions

Parent Behavior: Fake Good Normal Fake Bad

Normal Deviant Normal Deviant Normal Deviant

Negative 2.6 Sail 6.4 5.9 12.6 13.0

Command You 11.0 Abie 11.4 13.4 16.4

distressed couples were able to reduce their nega- more effective in controlling behavior than is the
tive statements and increase their positive state- desire to look good. It also seems that the present
ments relative to their baseline levels. During fake investigators just happened to select a clinical phe-
bad, both groups increased their verbal negative nomenon that best lends itself to observation in
responses. It seems that distressed couples are situ. We remain convinced that the observers do
more effective as impression managers than are set constraints upon the rate and intensity of fa-
antisocial children, i.e., they can alter the quality milial violence. We are also convinced that the
of their verbal behavior if they choose to do so. It families of antisocial children are sufficiently dis-
is, however, fascinating to note that neither sam- rupted so that many elements of the basic process
ple of couples was able to manipulate their non- are clearly visible: for example, Reid, Taplin and
verbal behaviors in accordance with the instruc- Lorber (1981) found that mothers of abused chil-
tions. Under the fake good and fake bad condi- dren were observed to hit their children at ten
tions, the groups remained discriminable. In addi- times the rates of normal mothers even though the
tion, neither sample shifted significantly from mothers knew that they were being observed be-
their own baseline levels. cause they were abusive!
Pending further research in this area, it seems Unobtrusive measures and deception studies.
reasonable to assume that both parents and chil- One would hope to have some unobtrusive meas-
dren from out-of-control families find it difficult ure that provides an absolute base for comparing
to “fake good.” Habitual modes of interacting the effect of observers being present to their not
provide powerful constraints for familial interac- being present. While there are no studies that have
tion patterns. Presumably, these constraints are successfully accomplished this, several inves-

56
 tigators have made innovative approximations. Harris (1969) trained 15 mothers as observers to a
Purcell and Brady (1966) equipped 13 boys in a reliability level of 80%. The mothers were then
residential treatment center with radio transmit- randomly assigned to one of three groups. For the
ting devices. Raters coded the transcripts of their outside observer (OO) group, mothers from other
interactions according to a set of personality trait families served as observers, and they collected
dimensions. Another set of data was collected by data for six sessions. For the mother-observer
observers using Barker’s technique for obtaining (MO) group, the mothers collected data for three
specimen records. The mean correlation between sessions in their own homes. During the last three
the two modes of data collection was .79. Cottage sessions, an outside observer collected the data.
parents rated the boys before and during the study. When the mother collected the data in her own
There was no significant change in their behavior family, she wore dark glasses. The family mem-
as a function of the observers and radio transmit- bers were told she was participating in a psycholo-
ter. This latter finding was in accord with the data gy experiment to determine the impact of social in-
from mothers’ ratings of preschool child behaviors teraction on encoding perceptual symbols. In the
emitted prior to and during the observation ses- third group, the mothers (MM) coded the entire
sion (Paul, 1963). six sessions for their own families. In all three
Bernal et al. (1971) experimented with the place- groups, the subjects consisted of fathers interact-
ment of an audiotape recorder in the home of a ing with their children. The mean proportion of
mother and her three children. Over a six-week social interaction for the groups in the various
period, the observer was present for one session conditions was recorded. It was predicted that the
per week; recordings were made for seven-minute impact of the outside observer would be to in-
intervals four times each day throughout the peri- crease socially desirable behaviors, i.e., social in-
od. Family members knew when the two sets of teraction. However, a 3x2 ANOVA for repeated
data were being collected. For the response “moth- measures showed that none of the main effects nor
er command,” the correlation between the two any of the interaction terms were significant. In
modes of data collection was +.89 (p<.02). It fact, the trends for two of the three relevant com-
was interesting to note that the deviancy rates parisons were in the opposite direction from that
were /ower for the audiotape data than for data which was predicted. A comparable analysis for
from observers in the home. Johnson and Bolstad rates of total deviant child behavior also produced
(1975) replicated and extended this basic design to nonsignificant results.
include 12 families of younger children (ages 4 to However, it is possible that mother observers
8). One hour prior to dinnertime, an audio record- might have dropped markedly in their reliability
er was turned on; all family members were aware following their training program. Because Harris
of the arrangement. On alternate days over the did not carry out any reliability checks during her
six-day period, observers were present while the experiment, the lack of findings could have been
recorder was on. The data from the audiotapes due to a decrease in the reliability of the data. In
showed no significant effect for the observers’ addition, there were enormous differences among
presence or absence on the occurrence of deviant families in their rates of deviant behavior, and in
child behaviors. Correlational analyses also their rates of social interaction. Finally, the Harris
showed observer presence or absence was not re- study used such a small N that it might not have
lated to children’s or parents’ rates of coercive be- had the power necessary to test for subtle effects.
haviors. With these qualifications in mind, White (1972)
The three studies employing unobtrusive meas- used professional observers, a larger N (25 fami-
ures were consistent in demonstrating that observ- lies, SO children), and shifted the setting from the
er presence did not produce significant changes in home to the laboratory. He also used an A,B, A,B,
deviancy as compared to results from audio re- repeated measures design involving four 30-min-
cordings. The possibility remains, of course, that ute sessions. During the observer absent periods,
the mechanical recording devices themselves pro- A, and A,, the mothers and their children were
duced distortions if they could be compared to told they were “waiting” for the observers. During
normal baseline levels. The problem here is to esti- B, and B, an observer appeared and coded their
mate the “real” or absolute level that would exist if interactions while sitting in the same room with
the subjects were unaware of the presence of a re- them. Actually, the family was observed by con-
cording process. cealed observers who coded their interactions dur-
Deception studies. Two deception studies have ing all four trials.
been carried out to get at the problem of esti- The data from White’s experiment can be inter-
mating an absolute level for deviant behavior. preted in several ways, depending on one’s pre-

a7
 Table 3.4
Mean Total Deviant Behaviors for Four Conditions

(from White, 1972)

Experimental conditions
Observers were:
Family
member Absent Present Absent Present F value

Mother .072 .068 .120 .148 2 AWoe

Older child .072 .044 .240 158 3.80*
Younger child .180 .170 .180 305 22

*p<.05
"p< .01

dilection. The data for the family as a whole preparation; Weick, 1968; Wiggins, 1973).
showed nonsignificant variations across the condi- High-quality data is critical to the research and
tions. In that sense, one could say the study repli- clinical activities at OSLC. As a consequence, we
cates the Harris (1969) study. However, an exami- have carried out extensive evaluations of the psy-
nation of the data in Table 3.4 points to an inade- chometric properties of both the FICS and the
quacy in the design. The data clearly showed ef- PDR. The evaluations began when these assess-
fects for trials; note that baseline two is higher ment devices were first constructed and continue
than baseline one for two of the three family mem- into the present. This section summarizes the rele-
bers. A pilot study showed no order effects, but it vant findings.
seems evident in comparing baselines one and two
that White did have a significant order effect (not Test-Retest Reliability
tested). The order effect serves as a confound, (and related questions)
making these data noninterpretable. The study In the context of observational coding systems,
should be redone, using a design controlling for the traditional notion of test-retest reliability be-
the likelihood of an order effect. comes a rather complex affair. One can begin by
asking whether scores obtained during one week’s
observation correlate significantly with scores ob-
Psychometric Properties of the FICS tained a week later, or with scores obtained a year
and PDR later. Such data will be presented here. But within
As a data collection device, a coding system re- the context of six to ten repeated observations in
flects the outcome of a number of arbitrary deci- the home there is the possibility that the family
sions, e.g., the content of what one observes, the members change as a function of being observed.
number and size of the behavioral units, and the This is a variant of the observer reactivity hypoth-
use of frequency or duration measures. However, esis. In the present context it refers to the possibili-
once these decisions have been made, the resulting ty that there may be systematic changes in mean
system is “just” an assessment device. As Jones level as a function of repeated testing. Do family
(1973) points out, an observational system, like members display high rates of prosocial behaviors
any assessment device, should be evaluated in during early sessions and increasing rates of devi-
terms of the various psychometric properties sub- ant behavior during later sessions? The habitua-
sumed under the traditional notions of reliability, tion hypothesis would suggest that this may hap-
stability, and validity. Modern systems of coding pen. If this were a significant phenomena, and
parent/child interaction demonstrate varying de- families were differentially affected, it would con-
grees of psychometric sophistication (e.g., Bales, tribute to lowering the correlation between scores
1950; Bobbitt et al., 1969; Caldwell, 1968; collected early and those collected later in a series
Moustakas, Sigel & Schalock, 1956; Radke-Yar- of observation sessions. Data relevant to the habit-
row et al., 1976). Contemporary reviewers of this uation hypothesis will also be considered in this
literature reflect the increasing demand for careful section.
psychometric analyses of these instruments (John- There is one further query to be considered; it is
son & Bolstad, 1973; Jones, Reid & Patterson, analogous to questions about the internal consis-
1975; Lytton, 1971; Mash, 1976; Wahler, in tency of scores from personality questionnaires.

58
How many items, with what kind of psychometric habituation effects that can be identified by using
properties, are required to establish an internally correlational analyses. As mentioned earlier, Har-
consistent measure of a trait such as anxiety or ag- ris (1969) suggested that observer presence may
gression? In the context of observational coding produce increased variability in family interaction
systems the question is altered somewhat. The in the first few sessions. As family members be-
home is vot constant as a setting. Even within the come habituated to observer presence, this varia-
constraints set by our procedures, there are tre- bility presumably decreases. However, the correla-
mendous differences from one evening to another tional analyses of classroom behaviors by Masling
in terms of setting variables. To obtain a stable es- and Stern (1969) did not support the variability
timate of a trait measured by a coding system both hypothesis. Similarly, the home observation study
the test and retest scores must be based upon rep- by Paul (1963) failed to support it. Ahome obser-
resentative samplings of settings. Settings (and ses- vation study by Harris (1969) showed a trend for
sions) are somewhat analogous to items on a ques- the data from observer-present trials to correlate
tionnaire. How many sessions are sufficient to in- at low levels with data from observer-absent trials.
sure an adequate sampling of settings? Inciden- However, a comparable study by White (1972)
tally, this implies that if one could construct a tax- failed to replicate the trend.
onomy of settings then setting complexity would Event Sampling. Given that the setting varies
covary with measures of test-retest reliability. somewhat from session to session, then how many
However, these are questions which have yet to be samplings are necessary to establish a stable estim-
explored systematically. For the present, we will ate for a particular code category? In the report by
focus on the simpler form of the event sampling Jones et al. (1975) the data base consisted of 60 to
question—how many minutes of sampling are re- 100 minutes of observation data per subject from
quired to reliably predict what a child’s rate will be sessions in the first week of baseline. The scores
over the next six or seven sessions? In the sections were correlated with comparable estimates from
which follow, we will first consider the findings the second week. Roughly two-thirds of these cor-
which relate to the habituation hypothesis, then relations were significant at p<.01 (see Table
those which relate to event sampling, and finally 3.5). For the 14 code categories describing coer-
those which relate to test-retest reliability. cive behavior for boys, the median correlation was
Habituation Hypothesis. If the habituation hy- .58. These findings suggest that estimates based
pothesis is correct, we would expect the mean fre- upon 60 to 100 minutes of observation data pro-
quency of deviant behaviors to increase over time, vide a reasonably accurate score. This much data,
while socially desirable behaviors should show a collected in three to five sessions, serves as an ab-
concomitant decrease. Our first attempts to assess solute minimum for assessing performance. It
habituation effects were based on small samples in should be noted that this does not apply to esti-
which session changes were compared separately mates of low base-rate events. We simply cannot
for mothers, fathers, and problem children (Pat- gauge the accuracy of estimates of events which
terson & Cobb, 1971). The results indicated that occur less than once during a 50-minute interval.
there were no significant changes in mean level Thomas et al. (1933) found that 120 minutes of
across sessions for any of the code categories, for data in five-minute segments distributed over a
any of the family members. The next study com- period of months gave stable estimates. They also
bined the data collected on 14 boys from normal noted that some social behaviors required greater
families and 17 boys from distressed families (Pat- sampling than others. In one analysis, the data for
terson & Cobb, 1973). Again the ANOVA for re- 20 boys were collected in a series of 12 five-minute
peated measures showed no significant changes samples. The correlation based upon odd/even
for any of the categories. Similarly, Johnson and samples from this set was .69 for Talking and .56
Bolstad (1973) found no evidence for changes in for Physical Contact. Increasing the amount of
mean level of family interaction over sessions. time in sampling increased the stability correla-
Mercatoris and Craighead (1973) observed chil- tions to .94.
dren in a classroom setting for 20 sessions and Smith (1931) also examined these questions in
found no significant changes in mean level over detail. He reviewed studies by other writers who
trials. None of these studies offer support for the were concerned with estimating the amount of
habituation hypothesis. As a result, low test-retest data required for obtaining a stable estimate of the
reliability scores could not be attributed to this rate or likelihood for any given variable. Most of
source. these studies showed that 70 minutes, sampled in
Even if there are no changes in mean level for five-minute blocks, gave minimally stable results.
deviant and/or prosocial behaviors there may be Currently, this serves as a rough rule of thumb in

59
 Table 3.5
Stability Correlations for First and Second Half of Baseline!
(fromJones et al., 1975, p. 72)

Family members
Target boys Mothers Fathers
Code categories (N= 54) (N= 54) (N =41)

Approval 02 48 nS.
Attention 36 38 43
Command 23 73 44
Command Negative .68 80 54
Compliance 67 2 38
Cry 90 — —
Disapproval .66 62 48
Dependency .24 _— —
Destructiveness 46 _ _
High Rate 16 = —
Humiliate Hs) 54 33
Ignore if 63 —
Laugh 28} 46 35
Noncomply 63 34 45
Negativism 154 68 op
Normative .44 49 15
No Response valk .20 2
Play Si 59 .48
Physical Negative 38 63 63
Physical Positive — .69 aD
Receive oo 21 .16
Self-Stimulation 29 OS .06
Talk 26 53 38
Tease 35 85 43
Touch 31 68 38
Whine .63 — _
Work 59 50 16
Yell .74 45 —

'The sample consisted of 27 normal plus 27 problem families.

—: Insufficient data for calculations.

designing studies at OSLC. It should, however, be and settings across time. The assumption is that
replaced as quickly as possible with rules based within a physical setting (such as the home or nur-
upon more extensive empirical studies. sery school) the same general interactional settings
Thus far there is little information available re- will repeat themselves. Given enough sampling of
garding the amount of event sampling required to sessions, it should be possible to obtain an accu-
obtain accurate assessments of functional rela- rate estimate at least for high-rate events. However,
tions. Taplin (1974) calculated the conditional these event sampling correlations are, by defini-
probability that a positive parental consequence tion, somewhat confounded. A significant correla-
would follow a particular coercive child response. tion implies two things. First, it suggests that
The means based upon the first three sessions cor- enough behavioral events were sampled to provide
related .56 with the means for the three sessions an estimate whose accuracy is specified by the
that were obtained during the following week. At magnitude of the correlation. However, the same
OSLC, R. Loeber is currently preparing a more correlation implies that the behavior of the indi-
systematic study of this problem. vidual is stable across time. In this manner the
Notice that all of these studies sample behavior problem of event sampling combines elements

60
 from two problems in traditional psychometric procedure for collecting data. In many ways it is
analysis. Our studies were somewhat similar to an inexpensive method for obtaining a general
studies of internal consistency; they are also simi- measure of child deviancy level. The problem
lar to short-term test-retest reliability. In that the which is of most concern to us lies in the reactivity
studies of event sampling were based upon a series of the measure. There is good reason to believe
of sessions over a two-week period, the correla- that it is highly reactive as a measure of treatment
tion of the first three with the last seven also gives outcome. The review of studies by Patterson
information about short-term reliability. Future (1980b) showed that mothers were likely to per-
studies might be better served by using an odd ceive improvement as measured by the PDR score
(three sessions) versus even (three sessions) format even though little or no change was shown in TAB
when studying event sampling. scores.
Test-Retest Reliability. Given that we now have
an estimate based upon at least 60 minutes of ob- Validity
servation data per subject then how reliably can When expressed as a single index, validity speci-
we predict a score for the following month or fies the degree to which an instrument measures
year? We have carried out only a limited analysis what it was designed to measure. The FICS was
of this problem. We were mainly concerned with designed to describe coercive interchanges among
the long-term stability of the TAB score because it family members—how well does it do this? There
is the score most often used in studies at OSLC. In is no single test which will provide the answer to
the first study, the TAB score was calculated for this question. There are, however, a series of tasks
each of 27 boys based on three or more sessions which a valid measure of coercion should fulfill.
during the first week of baseline and again for the First, one would expect a valid code to significant-
second week (Patterson, 1974a). The reliability ly differentiate families of normal children from
score of .78 (df=25, p<.01) showed that the in- families with identified antisocial children. Sec-
strument provided a relatively stable estimate of ond, the observation scores should correlate with
deviancy based upon three or more sessions. In the reports of aggressiveness provided by parents,
next study, a small sample of nine normal children teachers, or peers. Finally, there is the matter of
were studied for six baseline sessions and then the construct validity of the coercion variables that
observed again twelve months later for two ses- make up the coding system. To the extent that
sions. The TAB scores correlated .74 (df=7, these variables contribute to our understanding of
p<.05). This small-scale study showed that the familial aggression they also contribute to the va-
test-retest reliability for an extended time interval lidity of the coding system.
was in the same range as that found for other in- Does the FICS consistently differentiate families
struments used to measure children such as the of normal children from families of children iden-
Stanford-Binet. tified as antisocial? There are a number of investi-
The psychometric studies of PDR have just gators who have employed the FICS, or a revision
begun (Chamberlain, 1980). Recent studies sug- of it, to make such a comparison. In all such
gest that mothers’ reports on PDR may reflect studies, the problem child and members of his
transient mood states (Chamberlain, in prepara- family have been shown to be more coercive than
tion). As yet there has been no event sampling comparable members of normal families (Patter-
analysis for this instrument. However, there have $0n501976; 51981b Reid etral.,198i;. Snyder;
been several studies of the test-retest reliability of 1977; Conger & Burgess, 1978).
composite scores for limited time intervals. Differentiating normal from clinical samples is a
For a clinical sample, over a two-week interval, first step in demonstrating the validity of the FICS.
the test-retest reliability of the PDR score was .60 However, it is conceivable that the differentiation
(df=16, p<.01) (Patterson, 1974b). Chamberlain between samples relates to some dimension other
(1980) obtained a test-retest correlation of .82 than the one which we put forward. For this rea-
(df= 60, p<.001) for a sample of normal families. son, it is necessary to demonstrate that scores
She also found a correlation of .89 between moth- based upon the FICS correlate with some other
ers and fathers for their estimates of targeted devi- measure(s) of antisocial child behavior. The Parent
ant behaviors. Christensen (1976) obtained inde- Daily Report (PDR) criterion was developed to
pendent data from mothers and fathers for week- provide another measure of child deviancy. As
ends. The correlations for reports from mothers noted earlier, PDR is a reliable procedure for col-
and fathers ranged from —.47 to 1.00 with a lecting data. The prediction would be that some
mean of .51. composite measure of deviancy based upon obser-
These studies demonstrate that PDR is a reliable vation data should covary with PDR deviancy

61
scores. In the FICS, the most frequently used com- two methods in measuring treatment outcome. In
posite score is Total Aversive Behavior (TAB). It a treatment outcome study, Patterson (1974b) re-
sums across the frequency of occurrence for each ported that 67% of a treated sample of children
of 14 code categories measuring coercive behav- showed a reduction of at least 30% on both the
iors. This score also provides a reliable estimate PDR and the TAB scores from baseline to treat-
for general coerciveness. ment termination. Weinrott et al. (1979) also car-
The two methods vary not only with respect to ried out an elegant time series analysis of PDR
data source (parents versus observers) and format measures for single subjects. The findings from
(informal versus systematic data), but also in this study suggested that the PDR scores showed
terms of the time intervals sampled (one hour ver- substantial agreement with the observation data
sus an entire day). One might expect, therefore, to for the measurement of treatment impact for a
find low intercorrelations between TAB and PDR replication sample. More recent studies (Cham-
scores. Nevertheless, the mean frequency for PDR berlain, in preparation) indicate that the PDR
scores collected during baseline correlated .69 score may be somewhat reactive. About ten per-
(df= 14, p<.01) with the mean TAB scores derived cent of the mothers in the clinical sample report
from the observers’ data collected during the same that their children are much more deviant than is
baseline period (Patterson, 1976). The compara- reflected in their TAB scores. These “distortions”
ble correlation for a sample of children who steal seem to relate to the mother’s depression and
was .58 (df=31, p<.001). Chamberlain (1980), anger as measured by the MMPI. We are particu-
in her study of normal families, found a correla- larly interested in this phenomenon because it may
tion of .48 (p<.05). relate to the fact that the PDR score also seems
It seems that the two modes of data collection somewhat reactive as a measure of therapy out-
do, in fact, covary in their assessment of the gener- come.
al level of antisocial behavior. This sets the stage In the final analysis, the validity of the code will
for a more definitive evaluation of the two proce- be determined by its contribution to our under-
dures. Each instrument shared in common meas- standing of coercive family processes. Are the vari-
ures of Cry, Destructive, Negativism, Non-Com- ables measured by the code relevant to that pro-
ply, Whine, and Yell. Each instrument also meas- cess? One might think of this question as being the
ured deviant traits that were not shared. First, do focal point for this book. As noted earlier, the pur-
the two modes agree in identifying the occurrence pose for a performance theory was to identify vari-
of the traits that are shared in common? Is a child ables from the code which account for the vari-
identified by observers as high rate on Yell also ance related to individual differences among chil-
reported by the mother to be high rate on the same dren in the level at which they perform coercive
trait? Secondly, there is the question of how pre- behaviors. The multivariate analyses in Chapters
cisely the two instruments measure individual 11 and 12 directly address the issue of the validity
traits. Can each procedure differentiate among de- of the key constructs that are measured by the
viant traits? The study by Waksman (1977) used code.
OSLC baseline data from 47 boys for an analysis
of convergent and discriminant validity using the Sample Characteristics
multi-trait/multi-method procedure. Of the six Since January 1968, the FICS has been used
traits measured in common by the two methods, routinely to collect data on the clinical cases that
significant convergent correlations were obtained have been referred to OSLC for treatment. The
for Whine (.47), Cry (.37) and Destructive (.33). majority of these clinical cases were preadoles-
However, only Whine met the criteria for discrimi- cents that were referred for antisocial behavior. As
nant validity. The across-method correlation for noted earlier, the clinical population at OSLC is
Whine was greater than any of the within-method composed primarily of children who steal and
correlations for Whine with the other five traits. children that are socially aggressive. About one-
The two modes of assessment show consistent fourth of these antisocial children were also physi-
agreement in identifying general deviancy levels cally abused (Reid et al., 1981). Carlson (1979)
for child behavior. There is even modest conver- re-examined the case files for OSLC referrals and
gence in measures of three deviant traits. When found that about one-fifth had been previously di-
taken together, the findings strongly support the agnosed as hyperactive and were medicated prior
claim of validity for the coding system as a meas- to referral to OSLC. None of the chronic delin-
ure of antisocial behavior. quent children treated at OSLC since 1977 are in-
In the context of discussions about validity one cluded in the sample described in this section, nor
might also examine the agreement between the are any of their observation data included in the

62
 Table 3.6
Parent Occupation for Two Samples

Percentage for
Nonproblem Problem
Occupational level* families (N = 40) families (N = 170)
I. Major professional or large business 5) 1.8
II. Lesser professional or business il) S36)
III. Administrative or small business 1255 10.6
. Clerical 15 18.8
V. Skilled laborers Lees! 10.6
. Semi-skilled laborers 10 19.4
. Unskilled laborers or 22S 31.8
welfare recipients
Not classifiable DAS, 1.8
*from Hollingshead & Redlich (1958)

analyses in this volume. that following the careful matching of normal and
Local newspaper advertisements were placed at clinical cases there have been changes in the kinds
intervals offering payment to nonproblem families of cases referred to OSLC. The new referrals in-
for participating in an OSLC study of family inter- creasingly involved welfare and lower-class fami-
action. During the first ten years of the study nor- lies. These families were participating in studies
mal families were selected only if they provided a concerned with children who steal, and more re-
close match to a clinical family on SES, family cently, child abuse.
size, age of the target child, and father presence. The age distribution data show that the majori-
The demographic data for these samples are de- ty of the cases ranged in age from 5 to 12 years,
scribed in this section (the data were tabulated in with a median age of 7 to 8 years. These data are
1979). presented in Table 3.7.
The occupation of the parent was classified ac- During the first few years, referrals of families
cording to the procedure used by Hollingshead with only two members were not accepted because
and Redlich (1958). As can be seen in Table 3.6, we were interested in sibling interaction as well as
about 80% of both samples were working-class in parent/child interchanges. Also, because of ob-
families. It also seems that the matching for the server fatigue, families with more than seven or
two samples was not as carefully done at the lower eight members were seldom accepted. By the early
occupational levels. Forty percent of the distressed 1970’s, both sets of restrictions were lifted. As the
families were from father absent homes; the com- data stand in Table 3.8, they cannot really be
parable figure for nonproblem families was thought of as representative of the total popula-
21.1%. The lack of fit is apparently due to the fact tion of antisocial cases that might have been re-

Table 3.7
Age Distributions for Both Target and Problem Children

Percentage for
Nonproblem target Problem children
Age grouping children (N = 40) (N =170)
2-4 10.6

17
5-6
Sil
7-8
2325)
9-10
19.4
11-12
5)05)
13-15

63
 Table 3.8
Family Size for Two Samples

Percentages for
Nonproblem families Problem families
Number of people in family (N=40) (N=170)
8.2

ee 20.6

30 PLUS:
#35) 24.2

20 ADH
NE

5 5588)

4.7

2)

aS
over 12 D5)

ferred. The epidemiological study by Rutter et al. FICS is that it was not designed to explore the in-
(1970) found that 34% of his control group of tricacies of prosocial interactions. It was tailored
families had four or more children, as compared to measure aversive interchanges in families. But
to 44.4% for his antisocial group. Because of the in most family interactions, these categories de-
manner in which the OSLC data were tabulated, it scribe only 5% to 10% of what is going on. Cate-
is not possible to make an exact comparison. Even gories such as Work and Talk constitute the ma-
as it stands, however, the OSLC sample does seem jority of the prosocial behaviors that are coded.
in keeping with the idea that antisocial children But these categories are really wastebasket cate-
come from large families. gories that obfuscate the subtleties of prosocial in-
Normative Data. Normative data for an earlier terchanges. We were aware of this limitation when
sample are provided in the book by Reid (1978). we designed the FICS, but we did not have the
The data were collected from 27 distressed and 27 time or the funds to develop a code which would
nondistressed families referred to OSLC. The rate provide a more complete picture of prosocial in-
and standard deviations are given separately for teractions. In 1976, a group comprised of Marion
each code category for mother, father, deviant Forgatch, Dennis Moore, Leona Mukai, Mark
child, and siblings. In addition, the sibling data Weinrott, and the author began work on a new
are subdivided into younger and older, male and code that would correct some of these inade-
female. quacies. The revised code was field-tested for sev-
eral years and, in fact, provided a basis for carry-
Epilogue ing out most of the in situ experiments described
Science is only as good as its instruments. It is, in Chapters 5, 7, 8 and 9. In 1979, D. Toobert, D.
therefore, necessary to move on and develop in- Moore, and the author spent a summer studying
struments of greater fidelity and precision than the normal and “super” normal families. This led to
relatively primitive FICS. Undoubtedly, the use of the development of new categories which de-
six-second units distorts some aspects of family in- scribed the socialization processes in these fami-
teraction. It was convenient to construct the first lies. The interactional events were recorded in real
coding system in this manner; it even provided a time.
rough estimate (within a six-second error) of the This new code (the MOSAIC) is now being
duration of events. It would, of course, be better field-tested (Toobert et al., in preparation). These
to use real time in describing behavior. As noted trials will address-one further deficit in the FICS
by Hartup (1979) and others, this would permit which we had to overlook in our earlier work be-
the use of density and duration as dependent vari- cause of time constraints. We are now evaluating
ables; either of these would be preferable to the the generalizability of the data from one “stan-
use of rate or frequency. dard” time of sampling (the dinner hour) to other
The second and more serious problem with the times of the day. A related concern is the generaliz-

64
ability of the data collected in a semi-structured
format to unstructured home settings.
Footnotes
Like its predecessor, the MOSAIC is expensive 1. This chapter is an update of material pre-
to use. It costs an average of about 3.5 hours of sented in A Social Learning Approach to Family
observer time for each session; appointments are Intervention, Volume 2: Observation in Home
missed, the journey to the home takes time, and Settings, by John B. Reid (Ed.), 1978. Readers
the data must be proofed and stored. The cost for who are familiar with that publication may want
six observation sessions in the home is about the to skip this chapter.
same as the cost for the standard write-up for the
intellectual assessment of a child. 2. The Walter and Gilmore (1973) and Wiltz
Wright (1960) reviewed the 1,409 empirical and Patterson (1974) studies showed that four
studies in child psychology for the period 1890- weeks of treatment were sufficient to produce sig-
1960. He noted that 8% of these studies employed nificant changes in behavior. For this reason, any
observation techniques. Bronfenbrenner (1977) case receiving this much treatment was tallied as
carried out a comparable survey for the period treated in evaluating success. This is in contrast to
1972-1974; again, only 8% employed observation many of the traditional therapy studies which do
procedures. It is apparent that this technology is not include data from these dropouts in evaluating
one whose time has not yet come. Perhaps the treatment success rates. For example, two studies
problem with its adoption lies in the complexity of reviewed by Graziano and Fink (1973) showed
the task. uncounted dropouts of 40% and 60%!

65
 Chapter 4
Abstract
Aversive events in family life arise both from conflicts among family members and from sources ex-
ternal to the family. These events serve the dual function of altering mood and shaping behavior.
This chapter reviews the literature describing the relation of aversive events to autonomic reactions,
anger, and mood changes. Conditions of arousal in which negative attributions are made are most
likely to be followed by attacks. The literature suggests that antisocial children may be more likely to
attribute hostile intentions than normals.
Child-rearing years are generally associated with reduced happiness. It is hypothesized that avers-
ive events from conflicts with children and crises external to the family produce lowered self-esteem
and depressed mood for the caretaker. Data are presented that test for the covariation of mother’s
daily fluctuations in mood with frequency of crises, community contacts, and aversive child behavior.
The Multiple R values ranged from .55 to .78, with mothers’ Lubin scores as criterion measures.

66
 Chapter 4

Aversive Events: The

Innocuous Determinants

Coercion theory gives central status to the low- sity from the innocuous exchanges that are com-
key aversive events that are commonly found in monly found in all families to the high-intensity vi-
social interactions. These events are part of every- olence that is found in only a few. Based on their
one’s experience, and are therefore thought to be observation studies of child abusive families, Reid,
of little importance. While day-to-day variations Taplin, and Lorber (1981) formulated the steps by
in the density of these events may be noted, little which the parents’ interchanges with the child may
weight is given to their passage. Extremes in densi- escalate in intensity to include hitting. A prime
ty might be labeled a “bad day,” something to be function of coercion theory is to provide an empir-
expected, somewhat analogous to a case of psy- ical base for understanding the determinants of
chological smog. They are accepted as just anoth- this process.
er facet of everyday life. In the context of family interaction, aversive
In contrast, the present stance is that the rate events are thought to play a dual role. On one hand,
and intensity with which these “innocuous” events these events are the key components in a behavior
occur defines a process which, under certain con- shaping or changing process. Given time, this pro-
ditions, can lead to dramatic outcomes. For exam- cess can produce dramatic changes in behavior for
ple, the replicated studies of G.W. Brown and his families of socially aggressive children. The sec-
colleagues showed that the relapse rate for schizo- ond role relates to the short-term effects these in-
phrenic patients correlated with the suspected rate nocuous events have upon the autonomic system
of low-key aversive interchanges with their family and the alteration in mood or affect that is associ-
members (Brown & Rutter, 1966). The measure ated with their continued presentation.
upon which this prediction was based consisted of Taken individually, most of these aversive events
a scoring of hostile and critical remarks made by are trivial, a psychological mote as it were. The
relatives about the patient prior to admission for fact that they are part of a slowly unfolding pro-
treatment. The relapse rate for patients that were cess means that their contribution to dramatic out-
returned to highly aversive families was 58%, as comes goes largely undetected. This is in stark
compared to 16% for low-aversive families. The contrast to the colorful explanations for the same
study was replicated by Brown, Birley, and Wing phenomena that are offered by traditional psy-
(1972), chology. The metaphors of the latter have become
Chapter 7 of this volume describes a general themes for drama and literature, e.g., the love of
model for conceptualizing the escalation in inten- Oedipus for his mother; the hydraulic forces of in-

67
stinctual aggression lurking beneath the surface in often saw these parents hit. The parents also
Golding’s Lord of the Flies; the relation between a seemed to attribute malevolent intentions to the
sexual love of the father and obesity in Lindner’s child. The combination of negative attribution
classic Jet Propelled Couch. Whatever their even- and anger seemed to be an important part of the
tual status as explanations, the reading public has process leading to physical abuse.
come to expect psychologists to provide formula- Our clinical contacts with chronic delinquents
tions at least as dramatic as the phenomena they began in 1977. The majority of these families
purport to explain. What is offered here will be a seemed to be involved in very intense, long-stand-
divergence from these dramatic explanations. ing “struggles.” The family members were palpab-
Rather than cataclysmic episodes, flood tides of ly angry with each other. They held one another in
rage, or crumbling defense structures, we are con- complex webs of attribution. In case after case,
cerned with coercive family processes that change the impact of this was to seriously impair the treat-
with glacial slowness, a process that is comprised ment process. While we had rarely encountered
of events that are inherently banal. these problems before, it now seemed as though
This chapter details the stressors that are gener- angry struggles characterized every other case.
ated by aversive interchanges among family mem- Finally, there was a growing awareness that the
bers and from sources external to the family itself. treatment process itself was inundated by, flood
The impact of these events as elicitors of changes tides of crises. The parents came in with perfectly
in affect and mood is emphasized particularly as good “excuses” for not being able to carry out
they relate to the caretaker role within the family. their planned programs. They had been caught up
When the density of these aversives, or stressors, in a crisis that would have disrupted the plans of
exceeds some threshold value held by the caretak- even the most organized middle-class family.
er, it is hypothesized that he or she will report feel- However, as treatment progressed, it became ap-
ing “down,” “depressed,” “anxious,” or “tired.” parent there was not just one crisis, but a series of
The related hypothesis is that an individual’s gen- them. Each family seemed to be caught up in its
eral feelings of life satisfaction will be associated own sequence of disasters. These crises were mac-
with some ratio of aversive to positive events. Fi- rosocial events that clearly had an impact upon
nally, it is suggested that the impact these aversive family interaction patterns; they disrupted the
events have on behavior is related to the processes ability of the parents to practice effective family
of arousal, attribution, and anger. management skills. These macrosocial events also
covaried with the negativity of parental reactions
Arousal, Attribution, and Anger to their children.
Our initial attempts to build a performance the-
ory made little use of the concepts of arousal, at- A General Formulation
tribution, and anger. Our primary focus was upon Bandura (1973) viewed the determinants listed
the analysis of structure in family interactions. But in Figure 4.1 as potentiating pre-existing disposi-
even then, these other processes were obviously tions to aggression. The external event, by itself,
present. During our first contacts with antisocial might increase the likelihood of an attack. Arousal
children and their families, anger was a salient fea- would further increase this likelihood. In addition,
ture of the behavior of both the parent and the if the individual attributed a hostile intent to the
child. A considerable percentage of the parents, intruder, this may further increase the general level
especially the mothers, were very angry in their of arousal and add to it the label “anger.” This, in
dealings with problem children. In many cases, turn, would be accompanied by further increases
they seemed angry with the world in general. In in the likelihood of initiating an attack.
fact, even after successful treatment some parents The writer is in accord with this view, but would
continued to make negative attributions about the suggest that the arousal-attribution process is as-
child, e.g., “he is really bad,” or “you can’t trust sociated with /onger coercive sequences. These, in
him.” Their attributions turned out to be a self- turn, correlate with increased likelihoods for high-
fulfilling prophecy; a follow-up of treated families intensity aggression at each juncture in the chain
often showed these cases to be failures. (Loeber, 1980). In this sense the arousal-attribu-
Observers often described angry interchanges as tion process is thought to be an important deter-
being a part of sequences in which someone would minant of high-amplitude initiations of aggression
be physically assaulted. It was about that time we characterizing violent children and adults.
began appreciating the fact that a sizable propor- It would be expected that persons who engage
tion of the OSLC sample was made up of child- in violent assaultive behavior would be more likely
abusive parents (Reid et al., 1981). The observers to describe themselves as angry than would nonas-

68
 Figure 4.1. Processes Relating to Likelihood of Aggression

External event Initiate aggression Escalation in intensity

Attribution

saultive individuals. They would also be more expanded the list of aggression-related determi-
likely to attribute a hostile intention to their vic- nants for autonomic arousal. For example, the
tims. For example, child-abusive mothers might studies reviewed by Bandura (1973) demonstrate
view their children as trying to upset them or get the arousing function of shock, frustration and in-
their goat. If frequent extensive sequences are re- sults. In addition, he holds that viewing the behav-
lated to the arousal-attribution process, then it fol- ior of an aggressive model or an aggressive film
lows that the more violent individual would en- serves as a potentiator for aggressive behavior; it
gage in longer sequences of coercive behavior. also serves an arousing function. He cites four
It is thought that most coercive interchanges studies that show that viewing violent films is
among normal family members involve low-level emotionally arousing. As shown in figure 4.1, un-
arousal-attribution processes. They are what John der some conditions arousal is thought to be facili-
B. Reid has labeled “nattering.” This is also analo- tative for aggression. Anger-related arousal result-
gous to what John Knutson calls “irritable aggres- ing from goal blocking or insult has also been
sion.” A natterer is a coercive dilettante. Although shown repeatedly to facilitate aggressive interac-
John B. Reid has never formally defined this term, tions (Bandura, 1973; Berkowitz, 1973b; Rule &
his usage of it implies that it is ahaphazard avers- Nesdale, 1976). However, the latter point out that
ive reaction. The parent could ignore the situation the state of arousal produced by temperature
and remain uninvolved, but instead becomes a changes is not related to aggression. Similarly,
thoughtless participant in an aversive exchange. Holleran’s (1977) review found that, in and of it-
As will be shown later, the parents’ involvement in self, high-decibel noise did not produce aggres-
no way clarifies or alleviates the situation. Rather, sion; it was associated with aggression only when
it serves to extend these low-intensity inter- the subjects were angered. His own laboratory
changes. In the present context, it is interesting to study replicated the findings of Rule and Nesdale
note that periodically these sequences escalate in (1976).
intensity. Thus it appears that nattering can be- One feature of Bandura’s (1973) work received
come the prelude to events that are much more a good deal of attention in the last decade because
painful. High rates of background nattering are it had major implications for social policy. He
thought to be related to increased likelihoods for demonstrated the possibility that viewing aggres-
becoming involved in high-amplitude aggression. sive films was not only arousing, but also facili-
tated aggressive attacks in later social interactions.
Arousal Berkowitz (1974) also believes that viewing such
The social learning position has considerably films enhances a “readiness to aggress”; i.e., given

69
an appropriate external stimulus, then the likeli- some of Lang’s later studies showed, there was lit-
hood of attack is increased. The issue is still being tle covariation among measures of autonomic
debated, but a substantial number of well-de- arousal, self-ratings of anxiety, and approach be-
signed studies now provide support for the Bandu- haviors.
ra-Berkowitz position. As things now stand, the We would like to begin working at OSLC with
laboratory studies provide a consensus for the idea autonomic measures. In that regard, some of the
that aggressive films significantly facilitate the recent literature suggests that measures of nonspe-
likelihood of attack behaviors for subjects who are cific skin conductance might be most useful. The
already angry. This probably has some bearing on review of Miller and Grant (1979) found that non-
children watching television at home. If further re- specific skin conductance responses were closely
search lends support to this position then the linked to emotional arousal, while tonic skin con-
American people may have to decide if it is neces- ductance level and heart rate more sensitively re-
sary to censor media presentations for violence. flected cognitive and attentional factors. Szpiler
There are a number of studies reviewed in Ban- and Epstein (1976) also found that nonspecific
dura (1973), and in Chapter 7 of this volume, skin conductance was a highly sensitive measure of
demonstrating the not very surprising finding that threat, and alone among arousal measures reflected
a range of aversive events—shock, heat, insults, successful avoidance of threat. As noted earlier,
noise, and humiliations—all elicit autonomic reac- the tendency of family members to make critical,
tions. It might be noted, however, that there is lit- aversive comments about schizophrenic patients
tle consensus as to which physiological measures was related to the patients’ later tendency to re-
are best suited to measure this activation. Lacey lapse (Brown & Rutter, 1966; Brown et al., 1972).
(1967) proposes that one should consider patterns In the study by Tarrier et al. (1979), after a base-
of these measures; it is possible that each individu- line period, critical and noncritical relatives were
al may have an idiosyncratic pattern. Lang (1971), introduced into a situation in which schizophrenic
in his classic review of these and related matters, patients were being interviewed. The presence of
put the problem into perspective as follows: noncritical relatives was associated with declining
scores for nonspecific skin conductance measures.
“Investigators originally hoped that single meas- The presence of relatives who were critical was as-
ures of autonomic functioning would have a sim- sociated with the maintenance of the high arousal
ple indicant relationship to the psychological con- levels found in baseline. Normal subjects showed
structs in vogue. However, it became apparent lower baseline levels, and a steady decline through-
very soon that skin conductance increase, for ex- out the session.
ample, did not equal “anxiety” or “drive,” nor was More directly relevant is the study which showed
its relationship to these constructs simple or fixed. a covariation of arousal with aversive events
In point of fact, a great variety of situations and found in social interaction. In working with small
events will evoke skin conductance changes, rang- groups, Kaplan, Burch and Bloom (1964) used so-
ing from the closing of a door or the sound of a ciometric measures to form dyads with a negative,
friendly voice, through a whole gamut of emotion- neutral, or positive affinity for each other. During
al and physiological stressors. Nearly all physio- a 20-minute discussion, the number and ampli-
logical responses can be generated bya great varie- tude of the GSR reactions were recorded each min-
ty of internal and external stimuli, and it seems ute; then correlations were calculated for each
unlikely that any physiological event could be used dyad. The number of positive correlations above
in an exact substitutive way, as an index of psy- .29 was 75% for the negative affinity groups,
chological state. Thus, by observing the physiolo- 37% for the positive groups, and 26% for the neu-
gy of an organism, we are not able to go back- tral groups. In the replication study, the compara-
wards and reconstruct the stimulus input or the ble figures were 55%, 15%, and 18% respective-
psychological state that contributed to its genera- ly. The groups did not differ in the number of neg-
tion. To assume this kind of reciprocal relation- ative acts or in the amount of participation. The
ship is the classic indicant fallacy.” (p. 99) authors suggested that the greater magnitude of
covariation in the negative affinity group was due
In the context of Lang’s position, a judgment to a greater sensitivity or tracking of each other’s
about an event as being aversive or nonaversive is aversive behavior. It appears that this might be
based on information about the stimulus context, true for family members as well. Recent studies in-
plus interoceptive cues denoting autonomic arous- dicate that families of antisocial children are more
al. Each subject may have his or her own idiosyn- likely to engage in unprovoked attacks upon each
cratic rules for combining this information. As other (Littman & Patterson, 1980). This, in turn,

70
could relate to increased arousal and tracking for strated the dramatic effects of exposure to inescap-
aversive events. able shock. Seligman extended the earlier findings
by demonstrating that both noncontingent reward
Control of Aversives and Arousal and noncontingent aversive events produced a
The lay person and professional would agree state in which the individual was characterized by
that personal control over a forthcoming aversive a reduced ability to escape and avoid future stress
event would be related to reduced arousal. Miller situations. In the case of noncontingent reward
and Grant (1979) expand on this idea by pointing (e.g., the welfare state), the effect, as he labeled it,
out that during conditions in which aversives are was “learned laziness.” In the case of noncontin-
predictable the subject can identify a “safety time,” gent shock, the effect was labeled “learned help-
and let down a bit. The authors also identify vari- lessness.”
ous techniques—such as self-relaxation, denial, The Seligman metaphor seems to provide a very
distraction, and reinterpretation—that might re- close fit to experiences characterizing members of
duce the impact of aversives, whether controlled out-of-control families. To an outside observer,
or uncontrolled. Lefcourt (1973) reviews studies many of the attacks by family members upon each
demonstrating that subjects who were able to con- other seemed “unprovoked.” Our data indicate
trol high-decibel white noise suffered less disrup- that over two-thirds of the coercive chains for the
tion in their work performance than subjects ex- deviant child were apparently “unprovoked” (see
periencing the same amount of noise outside their Chapter 11). In the remaining one-third of the at-
control. Other studies indicated that self-reports tacks, the deviant child is responding to attacks by
of the perceived intensity of aversive stimuli showed other family members. It is assumed that this kind
decreases in the intensity reported when the sub- of uncertainty characterizes the majority of the in-
jects were able to exercise control over the stimuli. teractions within these disrupted families. It is
Lefcourt also cited studies by Corah and Boffa worthwhile to point out that the caretaker role is a
(1970), and Haggard (1943) showing that physio- focal point for such uncertainty. This is particular-
logical measures of stress were reduced if the sub- ly true if the mother is inept at terminating the at-
ject could control the aversive stimulus. In the tacks that are launched against her. The implica-
same vein, he describes the study by Weiss (1971), tion is that the caretaker role, as a result, is accom-
involving pairs of rats in which one member could panied by an increase in stress, and by concomi-
work to terminate a shock. The yoked control re- tant increases in self-reports of anxiety, depres-
ceived the same number of shocks, but the shocks sion, and psychosomatic reactions (Seligman,
were outside of control. Weiss concluded: 1975):
The control or lack of control over aversive
“The present experiment showed that regardless
events seems to relate consistently to disruption in
of whether electric shock was preceded by a warn-
performance on future escape and avoidance tasks.
ing signal, by a series of warning signals forming,
There is some evidence for somatic reactions (duo-
so to speak, an external clock, or by no signal at
denal ulcers) being more likely and more severe for
all, rats that could perform coping responses to
uncontrollable shock (Brady et al., 1958; Weiss,
postpone, avoid, or escape shock developed less
1971). However, Weiss (1971) also found that the
severe gastric ulceration than matched subjects
group that was given the opportunity for a coping
which received the same shocks but could not af-
response also showed the somatic reactions. This
fect shock by their behavior . . . the present re-
emphasizes the importance of the stress character-
sults, in combination with earlier experiments,
istics of aversive events per se, regardless of
serve to establish that the beneficial effect of cop-
whether or not they are controllable. It should be
ing behavior in stressful situations is of considera-
noted that a review of the burgeoning literature in
ble generality.” (p. 8)
this area by Averill (1973) leaves in doubt the issue
Many writers, such as Seligman (1975) have of whether control or perceived control is associ-
emphasized the same points. The lack of control ated with greater reduction in physiological meas-
over aversive events, or the absence of an effective ures of stress.
coping response, seems to be an operational defi- Many of the parents with aggressive children
nition of what is meant by “stress.” As Seligman that are referred for treatment report that they feel
emphasizes, this may, in turn, produce a state of helpless, anxious, and depressed. Many also indi-
conditioned helplessness accompanied by feelings cate that they have a variety of somatic complaints
of depression and anxiety. He describes a series of (Patterson, 1980a). As Maier (1961), Seligman
laboratory studies expanding upon the earlier (1975) and others have noted, direct guidance and
work of Maier (1961) and others who had demon- manipulation is usually required before condi-

Ta
tioned animals can re-engage in effective problem- ses from unpaid bills, or other similar situations,
solving behavior. In the case of parents, specific but label their emotional reactions in very differ-
training in child and family management were ent ways. One mother could see all this unpleas-
shown to produce reductions in these self-reported antness as being an expected “slice of life, as it
stress reactions (Patterson, 1980a). Additional really is,” and experience only profound fatigue
training in self-control procedures may further re- and vague depression. The other might attribute
duce the stress reactions. For example, the labora- the sibling conflicts to a malevolent intent on the
tory studies by Kanfer et al. (1974), and Kanfer part of the children “to get her goat.” Such an at-
and Seidner (1973) demonstrated increased toler- tribution might, in turn, lead her to believe that
ance of extremely aversive stimuli given simple the arousal she is experiencing is anger. This attri-
training in thought (attentional) control and in the bution-labeling process, in turn, might increase
manipulation of internal standards. Training par- the likelihood of a high-amplitude physical reac-
ents in relaxation, meditation skills, or giving tion by her. This dual attribution-anger reaction
them tranquilizers might serve a similar function. seems to characterize many of the mothers of fam-
All of these coping devices would be perceived by ilies seeking treatment at OSLC. Because of its ob-
the mother as resulting in a reduced stress reaction vious relevance to escalations in physical violence,
to aversive family interactions. it is important that we learn to measure and inves-
The other necessary component of treatment tigate both the affective and the attributional be-
would include a means by which the family could haviors. In the clinical studies by Toch (1969) and
reduce the overall level of aversive interchanges. It Berkowitz (1978), assaultive males often described
is not only how one perceives the aversive events themselves as having a “short fuse,” i.e., quick to
in the environment, but their actual density and anger. It seems very likely that these shorter laten-
intensity that determines stress. Mothers with an- cies and higher amplitudes were accompanied by
tisocial children are constantly coping with family increased likelihoods for attributing hostile inten-
crises. One might, in fact, characterize these wom- tions to their opponents.
en as being subjected to chronic stress. Even if Dodge (1980) hypothesized that aggressive chil-
medication or changes in attribution are provided, dren might engage in faulty attribution, partic-
it would be less than humane to leave these moth- ularly in inherently ambiguous situations. A peer
ers in a world in which aversives impinge upon sociometric identified 15 aggressive and nonag-
them at very high rates. gressive boys in each of three grades. In the labor-
atory situation, the puzzle on which boys had been
Attribution and Anger working was rearranged, presumably by the other
Berkowitz (1973b), Bandura (1973), and others child. Three different attributions were made for
have noted that the attributions made by the vic- this episode: hostile, benign, and neutral. In the
tim about the intention of the attacker are signifi- last phase, the subject was given an opportunity to
cant determinants for anger and aggression. The aggress against the other boy. The hostile attribu-
general formulation is based on Schachter’s two- tion condition produced significantly more aggres-
factor theory of emotion (Schachter & Singer, sion for both aggressive and nonaggressive sub-
1962). Emotional reactions are characterized by jects. As expected, there was a significant main ef-
two necessary components: the first is the arousal fect for samples with the aggressive subjects show-
state and the second is the cognitive label. Taken ing more aggression. Much of this difference was
together they define the emotional response. As attributable to the ambiguous attribution condi-
explained by Schachter (1964): tion; here, the aggressive child was significantly
more likely to attack.
“Given a state of physiological arousal for
The findings suggest that aggressive boys may
which an individual has no immediate explana-
be significantly more likely to attribute hostile in-
tion, he will ‘label’ this state and describe his feel-
tent to peers in ambiguous situations. Thus, what
ings in terms of the cognitions available to him. To
may seem to be an unprovoked attack may relate
the extent that cognitive factors are potent deter-
to cue distortions.! This is extremely interesting
miners of emotional states, it could be anticipated
considering the extensive literature (see Chapter
that precisely the same state of physiological arous-
11) showing that-antisocial children are character-
al could be labeled ‘joy’ or ‘fury’ or any of the great
ized in study after study as having “attentional def-
diversity of emotional labels, depending on the
icits.” Those studies show that these children gen-
cognitive aspects of the situation.” (p. 53)
erally do not track carefully, nor do they usually
Two mothers might experience the same high make careful discriminations. The complex inter-
density of quarrels among children, telephone cri- actional flow of the family or the playground

V2
would offer a rich field of possibilities for cue dis- ation of aggression might be entertained. Thus,
tortions or misattributions. Both the home and the rather than distinguishing between hostile and in-
playground are environments in which many brief, strumental aggression, it might be more appropri-
unpleasant experiences occur. They are likely to ate to draw a distinction between angry and non-
be accidental, but can easily be misconstrued as af- angry aggression, the latter encompassing those
fronts or attacks. responses that are directed toward obtaining a
To provide a more direct test of this possibility, nonaggressive goal. Although Buss (1971) has pre-
Dodge (1980) carried out a second experiment. viously drawn the angry versus nonangry distinc-
The subjects who served in the earlier study also tones (p.. 861)
participated in this second one. They were told a
Berkowitz (1973b) emphasized the position of
story in which a peer was involved in a negative
anger as a motivational state intervening between
outcome (from the viewpoint of the subject). The
frustration and aggression. The presence of anger
intention of the peer was left ambiguous. The sub-
potentiates existing cues which, in turn, elicit ag-
ject was instructed to describe how it might have
gressive responses. The experimental studies re-
happened, that is, to reconstruct the intentions of
viewed by Rule and Nesdale (1976) are generally
the peer. Aggressive subjects attributed a hostile
supportive of this position. When subjects were
intention to the peer 50%more often than did the
angered, there was an increased likelihood of ag-
nonaggressive subjects. The aggressive subjects
gressive attacks. Typically, the anger condition
also said they would retaliate when they perceived
consisted of an aversive stimulus (shock or insult)
the peer’s intentions as hostile. Only 26% of the
plus a clear communication to the effect that the
subjects who made a benign attribution said they
other person intended to inflict pain upon the sub-
would retaliate. There was one further outcome of
interest: if the peer was known to be an aggressor, ject.
then the likelihood of a hostile attribution was five
times greater. In the context of family interaction,
Aversive Events in Family Life
Some of the aversive events altering mood and
the antisocial child has already learned that his
satisfaction are generated by interactions within
siblings and/or parents are aggressive, and in
the family. Others impinge on the family from
situations where an affront was unintended or the
sources external to it. Regardless of the source, the
intent was unclear the antisocial child might be
mothers of antisocial children must deal with
more likely to attribute a hostile intent and attack.
more aversive events than mothers of normal chil-
The formulation may relate to the finding by Litt-
dren. The rate of conflicts among children is high-
man and Patterson (1980) that antisocial boys
er for families of antisocial children than for fami-
were eight times more likely than normal boys to
lies of normal children. Similarly, there are prob-
launch “unprovoked” attacks upon other family
ably higher rates of conflicts between spouses. As
members. These findings are parallel to the find-
detailed in Chapters 10 and 12, these families are
ings of Raush (1965) in his study of the interac-
thought to be deficient in the problem-solving
tions of antisocial boys.
skills which would lead to the resolution of these
It seems the two concepts of attribution and in-
conflicts. This same deficit is thought to relate to
tention potentially have much to offer a coercion
the higher frequency for crises external to the fam-
theory. The problem now is to find a means of
measuring them as they occur in ongoing interac- ily. The unpaid bills, and altercations with neigh-
bors, schools and community agencies contribute
tional process.
to a crescendo of pressures. In their aggregate,
Anger and Aggression these crises are a prelude to the frequent moves,
the disconnected telephone, and the isolation
According to writers such as Berkowitz (1973b),
characterizing the families of antisocial children.
the purpose of instrumental aggression is to pro-
As shown in Figure 4.2, these rising tides of adver-
duce a reinforcement (e.g. money or social ap-
sity are thought to be determined partly by a break-
proval), contingent on the response. Hostility was
down in the performance of problem-solving skills.
thought by these writers to relate to anger and an
In some cases the disruption may be only tempo-
intention to hurt the other person. In that conjunc-
rary, as in the case of divorce (Hetherington, Cox,
tion, Rule and Nesdale (1976) write:
& Cox, 1976). In other instances, the parents
“
. . the observation that the goal of injuring have simply never learned how to solve familial
another person can be construed as serving an in- conflicts; for them life crises seem to be due to the
strumental function (Bandura, 1973; Hartup, machinations of fate.
1974) indicates that a more appropriate differenti- The next hypothesis was that these accumulated

a3
 Figure 4.2. The Relation Between Familial Stress and Changes in Mood

Increase in crises external

to the family

Inept Reduction in satisfaction

problem and/or
solving dysphoria

Increase in conflicts
within the family

aversive events are major determinants for satis- among Holmes’ items, e.g., sibling fighting. Simi-
faction with one’s role, family, and with life itself. larly, telephone calls regarding unpaid bills, ex-
It was also thought to relate to changes in affective tended illness among family members, or minor
states such as mood, depression, and happiness. traffic accidents would not be noted. The present
As documented by Holmes and Masuda (1974), writer assumes these “minor” events have a cumu-
there is a relation between physical and psychiatric lative effect directly determining satisfaction and/
illness on the one hand, and a high density of life or mood. Any given event is, by itself, trivial, but
stress events on the other. Marriage, divorce, the cumulative effect can be debilitating. Other
change in employment, death, and similar factors things being equal, the happy, satisfied person is
each take their toll. Some investigators have one who has developed a high order of skills in
shown a higher incidence of such life events for coping with normal life stress; a lack of skill re-
members of lower social classes, and relate this sults in a steady accumulation of crises and con-,
finding to a higher incidence of psychiatric disor- flict.
der among that group (Myers, Lindenthal, & Pep-
per, 1974). Paykel (1974) demonstrated that the The Attrition of Satisfaction
amount and intensity of these events varied across and Happiness
psychiatric disorders. Suicidal patients declared Happiness and satisfaction are related variables,
the greatest density of stress just prior to onset. but they do not measure exactly the same thing.
The next greatest density of stress was declared by As used here, satisfaction involves a comparison
depressed patients and the schizophrenics. Most between what is expected and what is obtained,
of these studies were retrospective in format, so while happiness ratings reflect current mood. It is
their findings must be accepted with some caution. assumed that the cumulative effect of aversive
However, the findings from the predictive studies events from within family interactions, plus crises
of Rae and Holmes, reviewed in Dohrenwend and from external sources, produce both an erosion in
Dohrenwend (1974), suggest that 79% of major satisfaction and a general dysphoria. A lack of
life events were followed by illness. However, in skill in managing these events and/or a series of
that appropriate base-rate data were not provided, unaccountable misfortunes will covary with reduc-
it is difficult to know just how to interpret these tions in satisfaction and happiness. If the condi-
findings. tion becomes chronic, then one might expect so-
The life event studies represent a major contri- matic repercussions and the development of so-
bution to our understanding of illness and stress. matic symptoms.
However, many of the family conflicts to be re- There is a considerable body of literature on the
viewed in this section would not even appear variables relating to the determinants of “happi-

74
ness” among normal adults, e.g., Bradburn and spouses per day; the number of Instrumental Dis-
Caplovitz (1965). It is not surprising to find that pleases received was 2.1 per day. Similarly, the
education, income, good health, and adequate lei- wives reported receiving about three times as
sure time were among the variables significantly many Affectional Pleases as Affectional Dis-
associated with increased happiness (Wilson, pleases. The means were 1.8 and .7 per hour re-
1967). Individual variations in happiness have spectively. Husbands reported that they received
also been related to the number of close friends 5.4 Instrumental Pleases and 1.5 Instrumental
contacted, the frequency of pleasant events, and Displeases from their spouses per day. For hus-
those that are anticipated for the future (Shelly, bands, the means for Affectional Pleases and Af-
1970). Other studies have shown that for women fectional Displeases were 1.7 and .6 respectively.
these variations covary with the quality of interac- These data suggest that husbands and wives ex-
tions with friends, and for men with the quality of change pleasant and unpleasant events in an equit-
work experiences (Wessman & Ricks, 1966). Un- able fashion. However, wives receive far higher
happiness has been related to difficulty at work, rates of aversive events from children than do hus-
unemployment, marital tension (Wilson, 1967), bands. When taken together, these two sources of
poor peer relations, conflict with siblings (Wess- aversive events result in a decrease in happiness
man & Ricks, 1966), and general tension (Shelly, and an increase in stress for mothers in the care-
1970; Wilson, 1967). taker role.
Several writers suggest that happiness with one’s A further modification of the Please and Dis-
lot is best described by the ratio of positive to neg- please list (which combined Affectional and In-
ative experiences (Bradburn & Caplovitz, 1965; strumental events) was then used for a sample of
Shelly, 1970). From this perspective, a person may 12 distressed and 12 nondistressed couples. The
have many positive experiences but may still feel ratio of Pleases to Displeases was 29.7:1 for non-
unhappy because the density of negative events is distressed couples, and 4.3:1 for distressed cou-
much higher and produces a ratio which is out of ples. These results suggest that for nondistressed
balance. Conversely, some persons with a great couples, the ratio of pleasant to unpleasant events
many crises may still perceive themselves as happy was on the order of 30:1. When this ratio falls to
because the situation is more than balanced by 4:1 the couples reported feeling distressed. It
positive experiences from friends, work, and fam- should be kept in mind that the spouses were de-
ily. For example, Dawes (1978) summarized the scribing events from their relationship. If ex-
data from a series of his own studies by writing, changes from children were included, then moth-
“If we love more than we hate, we are happy. If we ers of normal children would probably report al-
hate more than we love, we are miserable.” In his most equal daily totals of Pleases and Displeases,
studies he simply subtracted the frequency of argu- while mothers in antisocial families would likely
ments from the frequency of intercourse. In one report a heavy imbalance in favor of Displeases.
sample of 30 happily married couples, only two Wills et al. (1974) calculated the across-subject
couples argued more than they had intercourse. In multiple R using the mean ratings of marital satis-
contrast, all 12 of the unhappily married couples faction as the dependent variable. The analysis for
argued more often than they had intercourse. In this tiny sample showed that Instrumental Dis-
his studies he finds a consistent covariation be- pleases accounted for 44% of the variance and Af-
tween ratings of marital satisfaction and the fre- fectional Displeases for 21%; the data for Pleases
quency of lovemaking minus the frequency of ar- accounted for 27%. These findings were in keep-
guing. The correlations for various samples ing with the general findings in the literature on
ranged from .40 to .81. marital relations, which indicated that aversive
Survey studies of married people usually identi- events contribute more heavily to marital satisfac-
fy men as being happier than women (Bradburn & tion than do positive events. The Wills study was
Caplovitz, 1965; Gurin et al., 1960; Wilson, replicated by Barnett and Nietzel (1979). In their
1967). To the writer, this suggests an imbal- study, 11 distressed and 11 nondistressed couples
ance for women in the amount of aversives they filled out the daily measure of pleasant and un-
receive while fulfilling the requirements of the pleasant experiences. The distressed couples de-
caretaker role. It is very likely that husbands and scribed significantly more unpleasant (Instrumen-
wives differ on both the numerator and denomi- tal) events than did the nondistressed couples.
nator in a happiness ratio; wives receive fewer Again, Instrumental Displeases accounted for
positives and more aversives. In a study by Wills, most of the variance in couples’ ratings of satisfac-
Weiss, and Patterson (1974), wives reported that tion with their marriage.
they received 6.4 Instrumental Pleases from their The findings for the covariation of aversive

ris)
events with measures of mood and satisfaction are due to abnormalities in the birth process. The
highly suggestive. Studies are needed now that sys- Harvard follow-up studies of premature births
tematically delineate the contribution of aversive compared them to children who experienced a
events from different sources such as conflicts with normal birth (Shirley, 1939). In preschool, chil-
spouse, coercive interactions with children or rela- dren that were born prematurely were character-
tives, and crises from outside the family. These, in ized by a higher frequency of ratings as distracta-
turn, should be systematically related to measures ble during testing sessions (45% versus 13%).
of the mother’s mood and/or satisfaction with her During the play session observed at the center, the
role as caretaker. If the studies could be carried prematures showed a higher incidence of crying
out over extended periods of time we might also (80% versus 57%), rapid shifts in toys played
obtain measures on the relation of these events to with (43% versus 23%), help-seeking (43% ver-
physical and psychiatric impairments. These sus 17%), and nervousness (83% versus 27%). A
studies are badly needed; they are the means by similar study was reported by Parmillee (1962).
which we can understand this little-known side of The two- to five-year follow-up indicated that in-
the “caretaker role” in our society. What are the fants with abnormal signs at birth were more like-
stressors accruing in the caretaker’s role; what is ly to be rated as having a short attention span
the cumulative impact of “innocuous” stressors; is (35% versus 6%) and hyperactive (46% versus
it the same at different phases in the development 12%). Children described as irritable, crying, rest-
of the family? Weiss (1978) provides some insight less, hyperactive, and having a short attention
into this last question. His findings suggest that span directly stress their caretakers.
the contribution of aversive events from the Even normal children may provide a formidable
spouse becomes even more important for mar- source of stress for parents. This is especially true
riages lasting ten or more years. Ultimately, our for younger children. Clarke-Stewart (1973) ob-
success in producing nondistressed families may served normal infants (ages 9 to 18 months) in
depend upon our ability to teach young people cri- their homes. The infants spent 8% of their waking
sis and conflict management skills that will reduce hours crying (range 1.6% to 16.3%). The more
the amount of stress that is associated with the fretful child was associated with mothers who
caretaker role. were viewed as less skilled. It is difficult to separ-
ate out cause and effect here, however, the cross-
Children as Stressors lag analyses showed that the extent to which the
For the caretaker, children are prime generators mothers were rejecting at t, correlated .51 with
of aversive events. This is generally experienced negative infant behaviors at t3. These results sug-
more intensely by mothers of younger children. It gest that the mother’s reaction to aversive infant
is also the case that some young children are more behaviors may exacerbate the situation, i.e., care-
difficult to manage than others. These differences taking skills probably interact with infant charac-
in ease of socialization may be constitutional. teristics in determining the degree of stress.
There is some evidence, for example, that activity It is part of the conventional wisdom to perceive
level is inheritable. Willerman (1973) found a cor- 2-year-old children as “negativistic.” The findings
relation of .93 for monozygotic twins in examin- from a study by Reynolds (1928) are in keeping
ing a measure of activity level. The comparable with this. Children of ages 2 through 5 years were
correlation for dizygotic twins was .52. There are, given a number of requests to solve problems.
however, well-known problems with confounding Two-year-old children gave an average of eight re-
due to identical twins shaping each other to a fusals out of a possible 13. The comparable score
greater extent than fraternal twins; but even so, for 3-year-old children was 4.61 refusals, 3.26 for
these findings are suggestive. Willerman (1973) 4-year-old children, and 2.42 for 5-year-old chil-
also reported significant correlations between rat- dren. In that study, the correlation between age
ings by parents of the activity level of their chil- and negativism was —.53.
dren and recollections of their own activity level Observations made in the homes of normal pre-
during childhood. The present writer assumes that school children showed an average of 3.4 distur-
there are differences in activity level and irritabili- bances per hour-in the interaction of mother and
ty at birth. If such differences persist, it seems rea- child (Fawl, 1963). The preschool child was char-
sonable to suppose that extremely active children acterized by higher frequencies of conflict than
present more difficult child management prob- were older children. For the range of children 2 to
lems. This results in an increase in stress for the 11 years old, the frequency of conflicts correlated
caretaker. —.73 (p<.01) with their age. The general tenor
Differences in ease of socialization may also be of these findings is in keeping with that presented

76
 Table 4.1
Status as Aversive Events

Mean ratings by mothers of:

Stealers Normals

OSLC Jones et al. (1975) Johnson & Bolstad (1973)

Child Behavior (N=15) (N = 20) (N =31)

CommandNegative GS Gee: 1.83

Hit 8.2 Vol 1.07
Dependency 4.9 U2, —
High Rate Tou Hes: 1S
Noncomply 8.1 8.1 1.28
Tease 6.9 Ue) 1.20
Yell isl VE 154.
Disapproval S58) 5.8 1.87

Ignore 73 S23 37

No Response 6.2 —
Whine 6.3 7.8 1.06
Negativism 6.8 Uy 1.69

Cry fail 5.8 1.96

Humiliate 6.4 8.0 _-
Destructive 8.2 8.5 1.20
Command Bee 5.6 2.30

in Table 2.2. As shown there, the observed rate of actions with children result in a dysphoric mood
coercive behavior for normal 2- to 4-year-old chil- for the caretaker. The data that have been pre-
dren was comparable to the rate of 9-year-old an- sented thus far do not demonstrate that mothers
tisocial children referred for treatment. giving commands every few minutes actually ex-
The presence of a young child provides several perience these interactions as aversive. In addi-
kinds of stress for the caretaker. The child tends to tion, evidence has not been presented for the idea
spend only brief interludes in solitary play, then that these child behaviors covary with the mother’s
returns to his mother’s side. In fact, the observa- dysphoric mood or with her dissatisfaction with
tion study by Schoggen (1963) showed that one of the caretaker role. The data that is relevant to
the most frequent goals of young mothers was these considerations is provided in the sections
simply to terminate the demands that were being which follow.
made upon them by their preschool children. This is It occurred to us that if family members could
compounded by the fact that in the home there are not agree as to what is and what is not perceived
many things that can injure or be disrupted by the as aversive, then there would be little hope of pro-
child. As a consequence, a day with a 2-year-old ceeding with the study of coercive processes.
child is a study in vigilance. In her role of monitor, Therefore, one of our first studies investigated the
the mother finds herself issuing frequent stop com- agreement among parents of normal preschool
mands. In a laboratory setting, mothers interact- children in rating the 29 code categories on a nine
ing with 2- and 3-year-old children issue these point scale of “aversiveness-pleasantness” (Jones et
commands at the rate of about one every three al., 1975). The main ratings are summarized in
minutes (Minton, Kagan & Levine, 1971). Even Table 4.1. It can be seen that the categories classi-
higher rates were reported by Forehand et al. fied by the experimenters a priori as aversive were
(1975). It is not surprising to find that mothers re- also perceived as such by mothers. Mothers rated
port feeling maximally stressed when their chil- Destructive child behaviors as the most aversive
dren are 6 years old or under (Campbell, 1975). with a mean rating of 8.5, and Command with a
However, these general findings are only indi- mean of 5.6 as the least. The ordering of the coer-
rectly related to the hypothesis that aversive inter- cive behaviors correlated .86 with comparable rat-

77
 Table 4.2
Covariation of Mother’s Mood and Aversive Events

PPM Correlations of Lubin Score* with:

Subjects Child TAB score Mother TAB score Mean Lubin scores

Pumpkin —.19 — 15 155

Pluto 5 43 0.25

Spring 12) .09 4.30

Tofu 42 .26 Poe ts)

Eclipse 34 .67 3.15

Median WS) 26

*The “Lubin Score” consisted of the frequency of negative adjectives checked.

ings by a sample of mothers of older normal chil- was serially dependent; therefore, no attempt was
dren from Johnson and Bolstad (1973). These made to estimate the significance of these corre-
findings suggest a general consensus among psy- lations.3 In future studies of this kind, it would be
chologists and three samples of parents as to the useful to include coercion rates for all family mem-
rank ordering for aversive child behaviors.” It can bers, as well as a report on Instrumental Dis-
also be seen that the mothers of stealers ranked pleases received from the spouse.
these events in a very similar manner; however,
with four exceptions (CN, HT, NC, CR), they Crises as Stressors
tended to provide ratings that were slightly lower The amount of external stress coming to bear
than ratings by mothers of normals. Their order- upon a family is to some extent fortuitous, e.g.,
ing of the events by aversiveness correlated .60 epidemics, social upheavals, natural disasters, and
with the ratings by mothers of normals in the recessions. The frequency of crises also correlates
OSLC sample, and .69 with the mothers of nor- with the status of the family in the economic hier-
mals in the Johnson and Bolstad (1973) study. archy. For example, writers in the ecological tradi-
The next question concerned the covariation be- tion emphasize the destructive impact of poverty
tween coercive child behavior and the mother’s upon the family and upon individuals. Lewis’ La
mood. To provide a data base, five mother-child Vida (1968) documents the steady attrition of
dyads participated in extended baseline studies spirit and gradual sense of helplessness accompa-
lasting 20 days. On each day the observation data nying prolonged enmeshment in extreme poverty.
collected in the home was used to calculate a Total Systematic interviews by Tonge, James and Hillam
Aversive Behavior (TAB) score for the mother and (1975) documented the overwhelming frequency
a comparable score for the child. During each ses- and intensity of the crises encountered by multi-
sion, the mother also filled out three self-report problem families seeking assistance from commu-
measures. She completed one of the three Lubin nity agencies. In the survey study of a borough in
Mood Scales and the number of negative adjec- London, Brown, Bhrolchain and Harris (1976)
tives checked provided an estimate of her dysphor- found that stressful events were more likely to oc-
ic state for that day. In keeping with the study by cur in the lives of working-class women than mid-
Wessman and Ricks (1966), it was expected there dle-class women. Interestingly enough, the impact
would be sizable shifts in the mood scores from of these crises was decreased if the woman had a
one day to the next. It can be seen from Table 4.2 close friend, husband, or relative with whom cri-
that in four of the five cases there was a covaria- ses could be shared.
tion between mother’s mood and coercive child While suggestive, none of these studies provide
behavior. On days when the mother described her- a direct test of the impact of crises upon the mood
self as being dysphoric, the child was more likely of the mother. To provide data to test such a hy-
to perform coercive behaviors at high rates. Note, pothesis, a family crises checklist was constructed
too, the high covariation between mothers’ moods at OSLC. The crisis checklist is provided in Ap-
and their own aversive behaviors for Pluto and pendix 4.1. The data were collected daily, and
Eclipse. It was assumed that each of the variates sampled events from general categories such as

78
 Table 4.3
Covariations of Mothers’ Mood, Crises and Community Contacts

Correlations of Independent
Mean Insularity Scores Variables with Lubin Scores

Frequency
Crises Contacts Positive Crises Positive
Subject per day perday Minutes Contacts per day Contacts Minutes Contacts Multiple Rt

Pumpkin 4.35 4.35 429.8 2.70 Sy) .02 hil -.11 ies
Pluto 1.60 StS 41.75 1635) 42 .09 -.04 -—.30 .675
Spring 5.50 5.70 343.5 4.05 —.02 —.26 no .02 546

Tofu 2.20 Bris 1222 2.05 34 16 —.23 —.32 .686

Eclipse 5255 3.70 287.0 2.90 sity. —.30 -.11 -.05 .780

+The Multiple R was calculated with Lubin score as criterion. The value includes the contribution of mother and child TAB scores as
independent variables in addition to the four listed in this table.

+The value also includes the contribution of sibling TAB score.

Household, Economic, Health, Employment, Of the five families who participated in the ex-
School, Social Interchange, Legal, Drugs, and tended baseline study, some thought of their chil-
Recreation. dren as being problems and were later treated;
Since its recent inception, the checklist has been others did not. The combination of mixed sample
used with eight normal families. They showed a and minuscule sample size means we are in no po-
range of two to nine relatively minor crises or sition to make simple generalizations about what
“hassles” per week, e.g., bills to pay, arguments covaries with what. Crises do occur in both nor-
with spouse, an illness in the family, the car break- mal and distressed families. Some mothers reported
ing down, or a dinner that was ruined. The mean that the crises covaried with their moods. Howev-
for this sample was 4.75 crises per week. The er, the covariation could reflect the simple fact
checklist was also filled out daily by five mothers that both variables were based upon mothers’ self-
who participated in an extended baseline study of report. We will see in Chapter 12 that mood and
20 sessions. As can be seen in Table 4.3, the five crises covaried with how coercive the mothers
mothers reported a considerably greater incidence were when interacting with their children. But
of crises than did the mothers in the (minuscule) there is no single pattern of covariation that char-
sample of normals. The correlational data showed acterizes all families.
that mood shifts for two of the mothers (Pumpkin Kohn (1973) and others have noted it is not just
and Pluto) covaried more with crises than with the amount of stress or the number of crises that
aversive child behavior. The writer would believe must be considered. It is the individual’s resources
the crises caused the shift in mood; however, the for coping with crises that determine the long-term
alternative hypotheses are equally reasonable. The impact. In his review, he makes the fascinating ob-
mothers filled out the checklists for mood and cri- servation that, at any given level of stress/crisis,
ses at the same time, therefore the correlations persons from lower social classes are more likely
may only describe their efforts to be internally to manifest breakdown symptoms than are mid-
consistent; or perhaps a bad mood in some way dle- or upper-class individuals in similar straits.
generated the increase in crises. However, it can be He goes on to make the point that, with the same
seen that for most of the mothers their moods co- amount of stress, the lower-class person has fewer
varied with some pattern of crises, insularity, and community, financial, and inner coping resources.
coercive interactions with their children. The mul- The Brown et al. (1976) study also implicated the
tiple R values were high enough to suggest that contribution of the spouse, friends, and relatives
mothers’ daily fluctuations in mood may be a sen- as support systems for coping. Several investiga-
sitive indication of what is going on in the family. tors give us reason to believe that families of anti-

a
social children may have reduced community sup-
Contribution of the Composite to Mood
port networks (Tonge et al., 1975; Wahler, 1979).
For some time, multiple-problem families have It is perhaps commonplace to document the dif-
been seen by caseworkers as being relatively iso- ficulties encountered in raising young children. In
lated from the community. The clinical data pro- the centuries prior to the advent of research
vided by Tonge et al. (1975) portray in graphic de- studies in child psychology, these “facts” had
tail the feelings of alienation and mistrust that probably not gone unnoticed. However, it is as-
members of these families have toward school, sumed by the writer that recent cultural changes
government, police, and welfare agencies. These that have broken up the extended family have re-
families tend to have few contacts with friends, duced some of the primary support systems de-
neighbors, or family. The contacts from the com- signed to buttress the role of caretaker. These cul-
munity that do occur are largely aversive in na- tural shifts bring aversive events into greater
ture. These findings for multiple-problem families prominence, particularly as they disrupt the mood
are certainly in accord with our own clinical ex- and self-esteem of those engaged in this role. Giv-
perience. When we attempted to treat families of en the loss of support from relatives, friends, and
stealers, socially aggressive, and multiple-offend- community, then crises could play a more central
ing adolescents, we were confronted with their rel- role in an arena in which they were once balanced
ative lack of support networks. In our experience, by positive support systems such as the extended
these families do not seem to cope with crises. As a family. The aversive events experienced by con-
result, they accumulate problems, and eventually temporary mothers may not necessarily be of
become the concern for a committee of casework- greater frequency or intensity. They are, however,
ers from a half-dozen agencies. These families part of a process in which their presence is felt
seem to be more mobile than other families, a more keenly.
finding also noted in the longitudinal study of ag- The pilot data from the five mother-child pairs
gressive children by Eron et al. (1974). Perhaps suggest that future studies will find many combi-
this increased mobility is a convulsive effort to es- nations of variables, such as coercive child behav-
cape from the mounting crises and largely abrasive ior, crises, and community contacts that are “de-
communications received from neighbors and terminants” for shifts in mothers’ mood. For the
community agencies. present, only an overweening curiosity could lead
Robert Wahler at the University of Tennessee an investigator to ask how much of the variance in
and his colleagues have provided an ingenious mothers’ daily mood shifts can be accounted for
(and simple) means of obtaining data describing by these variables. This task is confounded by
the “insularity” of these families (Wahler, 1979; faulty constructs that read like a litany for a failed
Wahler, Leske, & Rogers, 1977). Daily reports of doctoral dissertation: six variables and an N of
contacts with the community are obtained from only 20, serial dependencies unknown, only five
the mothers. Wahler’s first sample of eight high- cases studied, degree of freedom unknown, and
risk families had an average of 2.6 community true magnitude of correlations unknown. Howevy-
contacts per day; 30% involved friends. This was er, curiosity prevails. The multiple-regression co-
in contrast to an average of 9.5 per day for low- efficient in Table 4.3 employed the Lubin score as
risk families; 58% of these contacts were with the criterion. Thirty to 60% of the variance in the
friends. mood scores covaried with measures of aversive
The OSLC data in Table 4.3 showed three to events and support systems.
four daily contacts with persons from outside the If these correlations are replicated, the writer
family. This could be either a telephone call or would assert that these variables are causally con-
personal contact; it seems that the majority were nected. It appears that high densities of aversive
positive. There was a surprising range in the events and the lack of a support system produces
amount of time involved in these contacts; from mild and perhaps chronic depression in mothers.
less than an hour per day to more than five hours. This should be particularly true for young, iso-
The intraindividual correlation (across days) with lated mothers who lack effective problem-solving
the Lubin scores showed that the fewer the posi- skills (see Chapter 10). In lieu of such studies,
tive contacts, the lower the mother’s mood. How- there are findings-from treatment studies relevant
ever, these correlations were obtained for only to this issue. The test for causal connection will re-
three of the five mothers; they were also of very quire experimental manipulations. It would be ex-
low magnitude. In any case, the Wahler approach pected, for example, that if a sample of mothers
to measuring the important variables is seen as a was trained to use family management procedures
promising beginning. to reduce coercive child behavior, these mothers

80
should report that they feel less depressed. There happy, in contrast to the figure of 60.6% for fath-
are two studies relating to this issue that did not ers of that age whose children had already left
include the necessary control group. However, in home. There were only modest increases related to
both studies the mothers were trained to use social child absence in happiness for mothers in that age
learning procedures to significantly reduce avers- group; however, for mothers in their 50’s, the
ive child behaviors. Both studies showed decreases comparable percentages were 29.4% (children
in mothers’ self-reported depression (Patterson, present) and 41.9% for post parenting. Parents
1980a; Griest et al., 1979). seem to agree that they are happier after the chil-
dren have left home, and least happy during early
The Effect of Stress upon Caretakers
child-rearing years.
Even in a normal family, the caretaker role is Prolonged exposure to aversive events over
analogous to a storm center. It is the pivot point which one has little control might eventually lead
for external and internal stress. If there are many to the evaluation of oneself in negative terms. Se-
children at the preschool level, then the degree of ligman (1975) labels this conditioned helplessness,
stress may be considerable. In addition, if there is which he suggests is accompanied by anxiety and
a disruption in family management skills (see depression. Even in normal families the person oc-
Chapter 10), then external crises may accumulate. cupying the caretaker role must cope with an on-
If this stress persists for long periods of time, it slaught of crises, conflicts with spouse, and avers-
may be thought of as chronic. Selye (1976) sum- ive child interactions. With a lack of skill and/or
marized several decades of programmatic work, an overwhelming density of external stressors, the
making a strong case for a nonspecific physiologi- person occupying the caretaker role may be condi-
cal reaction to chronic stress. Prolonged stress, tioned into a state of helplessness. The self-report
with its accompanying adrenocortical reaction, data reviewed in Chapter 12 present a picture of a
may lead to eventual exhaustion of one or more depressed, angry woman.
physiological systems. However, it remains to be Traditionally, our culture has taken the position
shown that there is a relation between stress accru- that one should ignore these aversive inputs. The
ing to the caretaker role and some manifestations mother is encouraged to: “Turn the other cheek,”
of Selye’s General Adaptation Syndrome (GAS). “Ignore it and the child will grow out of it,” “Ac-
Theorell (1974) did report a set of findings relat- cept it, everybody gets a little depressed.” At the
ing life stress events to physiological changes anal- same time, there is the tacit acceptance of the idea
ogous to the GAS. Twenty-one males were inter- that mothers have the right to be bitchy. Xanthippe,
viewed each week for one to three months. In ad- while an historical figure, always has her modern
dition to having their life change events measured counterparts. Her real-life counterpart’s presence
(Holmes), they also gave a urine sample. The life in a household is accepted as one accepts a visita-
change and catecholamine scores were correlated tion of influenza or any other “natural” phenome-
for each subject. The median correlation was .40.
non.
These data indicate that a week with more life
changes was characterized by an increase in the in- Conclusion
dicators of physiological stress. The studies reviewed thus far suggest that avers-
Russell (1974) obtained data from crises relat- ive events are a ubiquitous feature of family life.
ing to young couples experiencing their first child. However, these events are not innocuous, and
As might be expected, many of the young mothers they should not be ignored. The data presented
perceived the birth process, with its aftermath of here suggest that aversive events may covary with
fatigue, altered physical appearance, and change marital satisfaction and with general happiness.
in mood, as constituting a crisis. Those experienc- At a more molecular level they covary with day-
ing an irritable or colicky infant perceived the cri- by-day changes in mood. These events elicit auto-
sis to be more severe. The 12 studies reviewed by nomic reactions that, depending on the setting,
Rollins and Feldman (1970) consistently under- may also serve as an important source for negative
scored a sharp decline in happiness and/or marital attributions to the intentions of others. These
satisfaction concurrent with the advent of the first changes, in turn, correlate with an increased likeli-
child. The reduction in happiness persisted until hood of aggressive attack. The writer believes that
the children left home, at which time there was an these events are major determinants for the de-
increase in reported happiness. More recent re- pression and unhappiness many mothers describe
sults from national surveys are in accord with as their lot.
these findings. For example, Glenn (1975) found The aversive events found in family interaction
that 45.6% of fathers in their 40’s reported being have a further function: they serve as units in a

81
process, the outcome of which may be even more likely to distort cues in otherwise ambiguous situa-
insidious than a change in mood or decrement in tions. The problem with this formulation is that
self-esteem. Aversive events are the tools-in-trade current measures of self-esteem do not differenti-
for the coercive person who seeks to mold and ate aggressive children from normals (see Chapter
shape the behavior of others. Garden-variety one
aversive events are the basic building blocks for a 2. In a study by Lorber (1978), nine of the
process that may eventually escalate to’ physical aversive codes were also rated by 23 male and 23
assault among family members. The means. by female college students. Their ratings correlated
which this process occurs in families will occupy only .26 (v.s.) with the ratings from the mothers
the bulk of the next six chapters. of preschool children. This suggests that the set to
perceive certain child behaviors as more aversive
than others may reflect a “cultural norm” to only a
very limited degree.
3. It is likely that the data for consecutive days
Footnotes would demonstrate significant serial correlations.
1. Berkowitz (1970) has suggested that low If this is the case, then one is in the position of not
self-esteem plays an important role in determining knowing what the degrees of freedom really are
the attack behaviors of violent adults. This varia- for these correlations (Hoffman, 1967). In addi-
ble may also play a role in the misattributions tion, a positive autocorrelation would mean that
made by aggressive children in the Dodge study, the estimates of magnitude are inflated (Hibbs,
i.e., children with low self-esteem may be more 1974).

Perspective on Chapter 4
by Dr. Robert G. Wahler
The beauty of Patterson’s formulation in this means. But, to present a nuts and bolts schema for
chapter centers on his confirmation that most the concept offers not only a set of guidelines for
aversive events in family life are “innocuous” and system researchers—it also offers potential im-
ubiquitous—yet they are critically important de- provements in the clinical skills of applied work-
terminants of psychopathology. A paradox? Yes, ers. Let me now touch on this behavioristic sys-
if one were to follow an early Watsonian view of tems concept from my perspective.
behaviorism in which stimulus-response associa- When a family therapist examines the basis of a
tions are the fundamental units of analysis. How- clinical referral, the initial “problem” is almost al-
ever, Patterson is clearly one of those Gestalten Be- ways seen in the context of a coercive dyadic rela-
haviorists who seem to be saying that it is the pat- tionship. Two people in the family, usually the
terning or relationship among these S-R events mother and one child, are entrapped in a recipro-
that should be seen as basic units. Thus, “innocu- cally aversive pattern of interchange. The clinical
ous” events become pathology inducing events solution, as parent trainers have documented, in-
when they appear in certain patterns (e.g., a mari- volves some specific means of changing the pat-
tal relationship “feels” o.k. if the ratio of positives terns—a means that is teachable in a didactic sense
to aversives is approximately 30 to 1—when the to both of the entrapped parties. But, there are
ratio drops to 4:1 it feels bad; it seems to be the also related coercive processes operating simulta-
duration of each mother-child coercive exchange neously at other functional levels within this fami-
that may lead to the child’s classification as devi- ly. At the individual level each party can be viewed
ant). as a system of covarying responses, some of them
This chapter provides one of the rare hard data directly instrumental in social relationships and
looks at the family system and the operating or others indirectly related as attentional (or possibly
functional levels within such a system. Certainly attributional) responses. The latter, while not ob-
the system concept is not a novel notion by any viously connected to each party’s coercive input to

82
the dyad, has been shown to exert stimulus con- ment of coercive interchange reveals the presence
trol over such input. Now, the complexity of this of such entrapment at three levels of the system:
attentional or tracking control unit is that its func- mother-child, mother-father, and mother-kinfolk.
tron may be affected not only by the dyadic coer- I believe that a “hierarchical levels” rule of thumb
cion problem; coercive relations beyond the trou- will become evident in judging problem severity.
bled dyad can also add variance to the attentional The greater the number of different dyads en-
processes. Thus, aversive events provided by a trapped within a particular system, the more diffi-
third family member or by members of the com- cult it will be to produce therapeutic change. Since
munity might strongly influence the ways in which these different dyads do demonstrate a functional
each member of the troubled dyad attend to one interdependence, there is probably some sort of
another. For example, a mother’s ability to see additive impact on the coercive stability of each
clearly her coercive entrapment with her son will dyadic operation. Thus, while a parent training
be affected not only by that dyadic problem—co- intervention ought to be successful in changing the
ercive problems with her husband, her own moth- parent-child problem, its therapeutic effects might
er or a helping agency representative will also add be attenuated if the parent is also entrapped with
a certain degree of “cloud cover” to her attentional other people in the family-community system. I
processes. think the treatment issue message is clear. We will
To find coercive processes operating at different develop treatment strategies based on our under-
levels within the family system is of paramount standing of how coercive processes operate within
importance in a diagnostic sense. The severity of the system. Our current strategies are largely
psychopathology is usually viewed in terms of the geared to particular dyadic problems (e.g., parent
likelihood of change: the more unlikely the training, marital contracting). It is reasonable to
change possibilities, the more serious the clinical anticipate the development of new strategies based
problem. Suppose, for example, that an assess- on triadic and even quadratic units.

83
 Chapter 5
Abstract
Coercion theory is focused upon the probabilistic relation between the antecedent behavior of one
family member (Aj) and the reaction (Rj) of some other members of the family. It is proposed that
positive reinforcement increases p(Rj|Aj;) and that this probability value is the proper dependent vari-
able for the measurement of reinforcement effects. Recent studies are reviewed which demonstrate
that reinforcement can have an automatic effect in strengthening the connection between the A; and
the Rj; i.e., the effect occurs without the subject’s “awareness.”
In the homes of normal and distressed families, members provide positive consequences for coer-
cive child behaviors 55%to 72% of the time. There are no significant differences in consequation by
family agents or by sample. Procedures are reviewed which test for the status of positive conse-
quences that serve as reinforcers for coercive behaviors. In addition, some speculations are outlined
as to why children differ in their responsiveness to positive reinforcers.

Glossary of Symbols
antecedent event

aversive antecedent

positive consequence
negative consequence (punishment)
negative reinforcement arrangement
positive reinforcement arrangement
positive reinforcer (this is a C+that has been shown
to strengthen A;—» Rj connections)

response
coercive response
probability of occurrence for a specific response
conditional probability for the occurrence of a specific response given the
occurrence of a specific antecedent behavior
conditional probability that R, will occur given the occurrence of A; at time-one (t;)
conditional probability that Rj will recur at a later point in time (t2) given
a presentation of the same A;

84
 Chapter 5

Positive Reinforcement
for Aggression

There are certain behaviors of family members teractions identifies certain persons and certain be-
which reliably produce aggressive reactions from haviors as weak elicitors for coercive reactions by
antisocial children. The connection between the other members. As used here, the term “elicitor”
aggressive response of the child and the stimulus refers to the probabilistic relation between the be-
which controls its occurrence is maintained by havior of one family member and the reaction of
both positive and negative reinforcement. In this another. It implies neither an invariant nor reflex-
chapter we will examine only the contribution of ive connection. In fact, the relation is so subtle
positive reinforcement to this process. The distinc- that it would be difficult for the untrained observ-
tive feature of positive reinforcement in the con- er to detect its presence; they are subtle counter-
text of familial aggression is that the victim’s reac- points usually obscured by the dramatic ebb and
tion increases the likelihood of future attacks. The flow of the content of family interaction. Who no-
victim is responsible for a dual contribution to this tices amomentary shift in the likelihood of Broth-
process. First, the victim provides a cue which sets er-Yell given Sister-Tease when one is trying to un-
the occasion for an aggressive reaction; and sec- derstand content problems such as: “Why did she
ond, the victim’s reaction functions as a positive tease just then?” or “Why did he yell?” These per-
reinforcement for the aggressive attack. fectly reasonable questions are the traditional con-
As a general case it is assumed that both the cerns for many of us, but they draw our attention
family and the culture provide rich schedules of away from the fact that structural changes are also
positive reinforcement for aggressive behavior. It occurring in these interchanges.
is further assumed that most of the stimuli which What is changing from one interaction to anoth-
control familial aggression are external; i.e., they er is the probability (p) value describing the rela-
are the observable behaviors of other family mem- tion between the antecedent or prior behavior of
bers. A prerequisite for understanding aggression one family member (Aj) and the reaction which
in families is that we understand how some stimuli follows it (Rj). The functional relation can be ex-
control aggressive reactions while others do not. It pressed as p(Rj|Aj). As detailed in Chapter 8, it is
is suggested that the mechanisms of positive rein- thought that these conditional p values are con-
forcement play a crucial role in determining which stantly shifting about some mean value. Increases
specific behaviors of family members will act as in the p value reflect the impact of positive and
controlling stimuli for the performance of aggres- negative reinforcement; decreases result from pun-
sive behavior. A functional analysis of familial in- ishment or extinction arrangements.

85
 One could identify the stimuli controlling the presented here.
behaviors of family members, and use them to The first question is: What is reinforcement? A
make predictions for a future set of interactions. related question is: What do reinforcers have in
That would satisfy the requirements of a perfor- common? Schoenfeld (1978) has most recently
mance theory. However, an understanding of why posed these questions and reviewed the literature
these stimuli control behavior requires the intro- relating to them. As he notes, it was Hull (1943)
duction of concepts from reinforcement theory. who made the major effort to answer these ques-
This question, in turn, is necessary if one is con- tions. From this position, all reinforcers are drive
cerned with matters of intervention and behavior reducing. Unfortunately, this position quickly en-
change. To treat an antisocial child will require countered difficulties. For example, certain drive-
that the functional relations be disrupted between activating events were shown to function as rein-
the controlling stimuli and the child’s aggressive forcers (Berlyne, 1967). While there are other
behavior. This problem in social engineering will more recent explanations, such as the Premack
require family members to learn how to manipu- Principle, each has its own particular flaw. “Simply
late their exchanges of reinforcement and punish- put, it is that amy stimulus can act as a ‘reinforcer.’
ment. There is no special class or group of characteristics
of ‘reinforcement’ which sets it apart from other
About Reinforcement
stimuli, but rather that all stimuli can act so de-
pending upon their intensity, static, and dynamic
The Nature of Reinforcement and patterning.” (Schoenfeld, 1978, p. 142)
Reinforcing Stimuli
Given the state of the art, one can identify a rein-
There seems to be a polarization within psy- forcer only by its effect on behavior. This, of
chology with regard to reinforcement theory. On course, is also Skinner’s position (Skinner, 1948,
the one hand, there is a diminishing number of 1953, 1969).
radical behaviorists who embrace reinforcement When considering the question of how a previ-
theory as a necessary and sufficient concept for the ously neutral event could come to serve as a rein-
explanation of most behaviors. This group expli- forcer for aggression, we are again led back to a
citly rejects all other explanations of human be- simplistic position. As Schoenfeld notes in his re-
havior. On the other hand, there is an expanding view, theorizing in the 1940’s and 1950’s differen-
number of psychologists fervently embracing “ev- tiated between primary and secondary reinforcers.
erything else” (cognitions, perception, motives Presumably, some contingent pairing of the neu-
and genes); they reject reinforcement concepts tral event with a primary reinforcer could alter the
with equal passion. The writer would suggest that stimulus such that it would become a “condi-
it is not necessary to be a radical behaviorist in or- tioned” or “secondary” reinforcer.
der to employ reinforcement concepts. They are
extremely useful as a starting point for studying “The upshot of all this thought and effort regard-
social behavior. However, there are some prob- ing possible distinctions between ‘primary’ and
‘secondary’ reinforcement was a disillusion with
lems associated with the use of reinforcement as
an explanation of behavior. the problem. Where once the literature was
This section is focused upon three questions crowded with studies and discussions of the topic,
which tend to enter into most discussions about there is now a disappointed and exhausted silence.
reinforcement theory: (1) What is reinforcement?; It seems safe to say that the two cannot be distin-
(2) Which stimuli are reinforcers and/or how do guished. When an experimental animal is chosen,
they become reinforcers?; and (3) Are all reinforc- he is taken as he is. A stimulus is a stimulus, and a
ers mediated directly? Some investigators claim ‘reinforcer’ is as a ‘reinforcer’ does.” (Schoenfeld,
that they cannot use the concept of reinforcement 1978, p. 139)
until they have satisfactory answers to these ques- Some investigators may be put off by the stark re-
tions. At present, the empirical base is not suffi- ality of the statement—a reinforcer is as a reinforc-
cient to provide definitive answers to these ques- er does. Worse yet, it makes the tautological nature
tions. As pointed out by Estes (1971), the majority of the reinforcement concept evident. What is a re-
of empirical studies testing reinforcement hypo- inforcer? It is that which strengthens behavior; but
theses lack coherence. In their aggregate, the find- how does one determine whether or not the con-
ings do not define a theory. There is, however, a tingent arrangement “strengthened” a particular
reasonable perspective that has been adopted by response?
many of us who wish to use the ideas from rein- In effect, any stimulus can, under certain condi-
forcement theory. It is this perspective that will be tions, serve as a reinforcer. However, there exists

86
no means for classifying events, in advance, as re- reinforcement and what is performed. Again, Es-
inforcing or nonreinforcing. This appears to be an tes (1971) states the current perspective in a very
insurmountable barrier—can we apply the con- succinct manner (italics by the present author):
cept if we do not really understand how some “For the lower animals, for the very young chil-
events become reinforcers while others remain dren, and to some extent for humans of all ages
neutral? The answer to this question is an empha- who are mentally retarded or subject to severe
tic yes! Concepts relating to genetics, magnetism, neurological or behavior disorders, behavior from
and synapses served very useful functions in scien- moment to moment is largely describable and
tific enterprise long before they could be ex- predictable in terms of response to particular
plained. The concept of reinforcement will proba- stimuli and the rewarding or punishing outcome
bly occupy a similar niche for some time to come; of previous stimulus-response sequences.” (p. 23)
but, in the meantime, it will continue to serve a
useful function. Estes then went on to state that he does not be-
For present purposes we do not need to know lieve this will hold for mature adults! The behav-
why or how something becomes a reinforcer, but ior of the latter is, rather, governed by cognitive
we do need to know how to identify an event as a strategies and anticipated consequences for future
reinforcer (this problem will be considered in due actions. In the same vein, he suggested that man-
course). We can understand the impact of rein- ipulation of reinforcing contingencies would prove
forcers on behavior. That information is available to be of little practical value for persons other than
to us; it forms the basis of reinforcement theory as those in institutional settings.
it stands today. The present writer agrees with the notion that
the power of reinforcing contingencies is most
Learning or Performance? manifest when considering matters relating to per-
Does reward modify learning, or does reward formance. However, I disagree with Estes in my
modify the performance of the Rj given the Aj? belief that reinforcing contingencies serve as a ma-
Again, the present writer draws heavily from Es- jor determinant for many social behaviors of both
tes’ (1971) perspective, “The learning of stimulus- children and adults. The recent successful applica-
response and stimulus-outcome relationships is as- tions of reinforcement concepts to studies of
sumed to proceed simply according to association adults in marital conflict (Jacobson & Margolin,
by contiguity” (p. 25). Bandura (1973) documents 1979), smoking (Lichtenstein, 1971), and obesity
the possibility that much of what a child learns (Stunkard & Rush, 1974) would lead one to be-
about the connection between social events and lieve that this issue has not been completely re-
his or her behavior is based upon observation of solved.
the social environment; i.e., the child sees, and In keeping with the general position taken by
stores, many of the contiguous relations among A; Estes (1971) and Bandura (1973), reinforcement
and R; events. As Bandura notes, the child also during trial and error learning is not the most effi-
stores in memory observations about the relation cient way for a child to learn about driving a car or
between specific behaviors and likely outcomes. about social interaction. Rather, it is more effi-
This kind of learning by contiguity can be based cient to arrange for the child to observe the pro-
on observing social interaction, films, role-play, cess. The child should also observe the outcomes
pictures, or stories. The programmatic work of for various reactions to the same or similar stimu-
Bandura (1973) and his colleagues beautifully li. Now, let us assume that the child has observed
documents the power of this concept as it applies his parent driving the car; in effect, he has learned
to a wide range of children’s social behaviors. As something about the process. The next stage in ac-
noted by Estes (1971) and others, it is not alto- quiring skill as a driver requires that the child be
gether clear as to what, if any, contribution rein- supervised while performing these skills. This sec-
forcing contingencies make to this stage of learn- ond stage in skill acquisition is a polishing and
ing. Reinforcement could augment the saliency of smoothing process. It has been described in detail
the A; and/or R; events. Contingent reward could in B.F. Skinner’s Contingencies of Reinforcement
increase the likelihood that a particular event will (1969). As Skinner points out, driving the car re-
be stored in memory. Learning associations may quires that one repeatedly expose oneself to the
also be more rapid under conditions of reward; natural contingencies produced by this or that re-
however, modern theory gives reinforcement per sponse and then be allowed to correct the S-R con-
se only a secondary role to play in the learning nection accordingly. This takes time; it also takes
process. a sympathetic audience and a safe place. This kind
The next question concerns the relation between of polishing is so demanding that it most typically

87
occurs in our early years and in specially designed that is of concern to us is p(Tease|Whine). When
environments, such as the home or school. If this observed over a two-day period, several thousand
training does not occur, the deficits may be ex- behavioral events (interactions between brother
tremely difficult to make up later. There are liter- and sister) were recorded. Let’s say that among the
ally no social agencies that have the staff available matrix of these interactions, we find that 100 of
for such large investments of time. the sister’s behavioral events were Whine and that
This book describes our effort to analyze family 12 of these were followed by Brother-Tease. So
interaction as an example of performance. All of our baseline conditional value for p(Tease| Whine)
the family members have learned a variety of coer- at time-one is .12. In our example we are going to
cive behaviors, but their performance of these be- assume that Sister-Cry is a reinforcer for the se-
haviors changes from day to day and month to quence Sister-Whine followed by Brother-Tease.
month. Why is this? Some families perform coer- Thus, the three events that we are looking at are
cive behaviors at higher rates than others; and Sister-Whine (A;) > Brother-Tease (Rj) —>Sister-
within families, some members perform these be- Cry (S+). Let’s say that during our baseline study
haviors at higher rates than others. Why is this? In at time-one that Sister-Whine —-»Brother-Tease
the context of a performance theory, reinforce- was followed by occasional outbursts of Sister-
ment concepts are not the main explanation for Cry. However, now on day three (time-two), it
these questions. However, reinforcement concepts happens that every time Brother-Tease occurs, it is
are a necessary part of the picture. followed by Sister-Cry. During day three several
hundred interactions between brother and sister
Aj» Rj connections were recorded. Within the matrix of these interac-
The term reinforcement has been used by asso- tions, Sister-Whine occurred 50 times and was fol-
ciation-based learning theorists, such as Pavlov, lowed by Brother-Tease 20 times. The value of
Thorndike, and Hull, to convey the idea of p(Tease|Whine) at time-two is .40. A comparison
strengthening (Schoenfeld, 1978). As Schoenfeld of this p value with the baseline value of .12 indi-
points out, Pavlov’s use of the term referred to the cates more than a threefold increase in the strength
pairing of the UCS with the CS. What was of the connection between Sister-Whine and
strengthened by these pairings was the increased Brother-Tease. This suggests that Sister-Cry is a
likelihood of the conditioned response being elic- reinforcer for the sequence Sister-Whine —>Broth-
ited by the CS. In instrumental conditioning, as er-Tease.! It should be noted in the context of this
used by Hull and Guthrie, a reward was thought example that Sister-Cry may also serve as a rein-
of as strengthening the connection between a stim- forcer for other sequences of events. Furthermore,
ulus (S) and a response (R). This relationship can there are probably reinforcers other than Sister-
be expressed as p(R|S). Cry which strengthen the connection between
As used here, the functional relationship will be Tease and Whine.
expressed as the likelihood of a particular re- This emphasis on the connection between Aj;
sponse (Rj) given the occurrence of a specific ante- and R; is not in keeping with the current Skinneri-
cedent (Aj) or p(Rj|A;). The term antecedent (A)) is an perspective which stresses the operant as the
more precise than the more traditional term stimu- important functional unit. In that paradigm, the
lus (S). To be identified as an antecedent (A;), the strength of the response is defined by a measure of
event must occur immediately prior to the re- the rate of occurrence or p(R;). Skinner rejected
sponse (Rj). Throughout the discussion, the term the idea that reinforcement strengthed the connec-
R; refers to a specific target event serving as the fo- tion between a stimulus and a response.*
cus in calculating the conditional p value. The problem with the Skinnerian stance is gen-
To understand reinforcement effects we need to erated in part by the fact that in natural settings
have three events in sequence. This sequence can the complexity of social interactions is greater
be represented symbolically as Aj—» Rj» S+. than that found in most laboratory situations. In
The effect of the third event, which is a reinforcer, social interaction most changes in p(Rj) have little
will be to increase the probability that a particular to do with changes in reinforcement and punish-
R; will follow a specific Aj. In other words, it will ment. Rather, itis the case that in social inter-
increase the p(Rj|Aj;). The calculation of this p val- action Rj is evoked by the immediately prior be-
ue is relatively straightforward. For example, let’s havior of the subject or the other person. The pre-
assume that a baseline study shows that there is a sentation of these controlling events is changing all
connection between Sister-Whine and Brother- of the time and the p(R;) changes accordingly.
Tease. In this example we have Sister-Whine as A; If the frequency of R; reflects the impact of con-
and Brother-Tease as Rj. The functional relation trolling stimuli then how is one to test for the im-

88
 pact of a reinforcer? The answer is to use p(Rj|Aj) child will be playing at t,. Responses tend to exert
as the dependent variable. This will sensitively re- their own thematic control over the responses
flect the impact of punishment and reinforcement. which follow. The fact that a particular behavior
When a reinforcing event (S + ) follows an Aj—» R; continues into the next time interval usually has
sequence, the probability that Rj will follow a pre- little to do with the immediate changes in rein-
sentation of the same A; should increase. Here, forcement. Moment-by-moment changes in ongo-
even massive changes in the density of controlling ing behavior are typically reactions to shifts in im-
stimuli do not mask the contribution of reinforce- mediately impinging stimuli. This is stimulus con-
ment to the connection between the A; and the R;j. trol which has its own logic, means of analysis,
Given that a Whine-Tease sequence is followed by and utility (see Chapters 8 and 9). Reinforcement
Cry is there an increase in p(Whine|Tease) on the control must be tested by other means which re-
next trial? The next trial could be in two minutes, quire control for thematic components.
two hours, or two days; this should make very lit- In retrospect, our first efforts to use sequential
tle difference. In addition, what happens in the se- data to identify reinforcement effects for aggres-
quences that follow the Aj—» R;-> S + arrange- sion were misguided. With unerring skill we com-
ment has little bearing on our analysis of rein- mitted all of the errors delineated above. In the
forcement effects per se unless the sequence in- Patterson, Littman, and Bricker (1967) study of
volves our fundamental unit of Aj and R;. nursery school aggression, the immediate recur-
Before leaving this topic it should be noted that rence of an attack by the aggressor served as a
understanding and predicting p(R;) is the task for measure of reinforcement effects. Presumably, a
a performance theory. The determinants for these positive reinforcer provided by the victim in-
variables are considered in detail in Chapters 8 creased the likelihood of a recurrence. By combin-
and 9. Suffice it to say that reinforcement and ing the data for immediate recurrence with data
punishment account for only trivial amounts of from trials that were separated by minutes and/or
variance in measures of p(R;) when analyzed in the hours, the measurement of reinforcement effects
context of social interaction. was inextricably confounded.
As noted by methodologists such as Parton This difficulty with rate as a dependent variable
(1967), measures of rate change used in labora- should have been obvious. But, in the mid-1960’s,
tory studies with human subjects proved to be the problem was not apparent. Literally months
both unreliable and heavily correlated with base- and years of data analysis slipped by before the
line level. There were, in addition, many variables perspective we now have finally emerged. This is a
other than changes in reinforcement schedule small problem, but dozens like it had to be worked
which affected the rate measures. The present through. Most of them will not be detailed in this
writer believes that p(Rj|Aj) will prove to be a fashion. This particular problem is described in
more sensitive measure of response strength. depth because it continues to be a difficulty for
When applied to family interaction, measures of many who wish to study interactional data.
rate change can often underestimate reinforcement Before leaving the concept of A;—» R; connec-
effects. For example, the sequence of child-open- tions, it would be wise to reiterate a point that has
door, followed by mother-thank you, then both been only alluded to. One of the key assumptions
walk to the car, would be tabulated as no rein- regarding coercive family processes is that there
forcement effect. The behavior did not persist into are two reinforcement mechanisms involved—one
the adjacent time interval, therefore, no reinforce- is positive reinforcement and the other is negative
ment effect! That, of course, makes no sense at all. reinforcement. In negative reinforcement, the an-
Child-Whine followed by mother-Talk, then child- tecedent would be an aversive event (A;) supplied
Talk, would be another case in point. Actually, by a family member; the child’s coercive counter-
observations of family or nursery school aggres- attack followed by a termination of the intrusion
sion contain many sequences like this. Again, the would be an example of negative reinforcement
measure p(R;|A;) seems better suited as a depen- (see Chapter 7). The termination could be either a
dent variable. If “thank you” or “mother-Talk” are neutral or a positive reaction by the initiator. In
indeed reinforcers, then the p value describing fu- this instance, what would be strengthened would
ture interactions will show a slight increase. be the connection between the aversive antecedent
What is needed is some means for untangling (A;) and the counterattack. For example, mother-
the obvious thematic control exerted by ongoing Scold —»child-Whine —>mother-Talk would be a
stimulus events from the effects that we wish to at- negative reinforcement arrangement that strength-
tribute to reinforcement effects. For example, if a ens the connection between mother-Scold and
child is playing at t; there is a good chance that the child-Whine. The child’s behavior (Whine) may be

89
controlled by both positive and negative reinforce- al’s information processing time and would quick-
ment. For this reason, it becomes particularly im- ly surpass his or her processing ability. Instead, the
portant to classify A; into positive, aversive and individual relies on his or her own a priori causal
neutral categories. These, in turn, differentially re- theories to acount for the relation between the
late to attack and counterattack response catego- stimuli to which he or she does attend (Nisbett &
ries, as well as to positive and negative reinforce- Wilson, 1977). In Nisbett and Wilson’s review,
ment mechanisms. These matters will be discussed they cite a number of studies in which subjects
in detail in Chapter 7. were unaware of even dramatic changes in their
behavior. In other studies, when asked to explain
Awareness
certain decisions they made, their explanations
Coercion is a social skill that takes practice. had little to do with the variables actually manipu-
Some people are more skilled at it than others. An lated by the experimenter to control their deci-
extremely coercive 10-year-old child, a well-prac- sions. In their attempt to explain their behavior,
ticed master sergeant, and the authoritative pro- the subjects employed the kind of representative
fessor are all masters of pain control. Their level heuristic described by Kahneman and Tversky
of skill comes from thousands of trials and years (1973). Here, the subject first examines the beha-
of exposure to natural contingencies (most of vior and the context. The subject then séarches
which involved “winning”). In the course of this within his or her representative theory for possible
chapter we will show that coercive behaviors that “causes” for that behavior. If the new experience
are performed at high rates are maintained by rich seems to fit the subject’s causal theory then the
schedules of reinforcement. event may be identified as causal when, in fact, it
How aware are the aggressor and/or the victim had no impact upon the subject’s behavior.
of the slow shifts in p (Rj|Aj) that are taking place? The general tenor of these findings is further
Do these shifts take place without the individuals buttressed by the findings reviewed by Slovic,
being aware of them? During social interaction, Fischhoff, and Lichtenstein (1977). Their pro-
the individual can focus only on a limited subset of grammatic research indicated that subjects had lit-
the complex matrix of behavioral events. Of that tle awareness of the weights they attached to cues
subset to which the individual does attend, only a that were being used to make decisions. Slovic et
small proportion of the events find their way into al.’s studies showed serious discrepancies between
long-term memory. The question, then, concerns subjective estimates and the objectively deter-
what it is that the individual attends to, and what mined weights for these cues. In some studies the
it is that goes into short- or long-term memory. If subjects tended to overestimate the importance
the individual did not know that a given event was they attached to minor cues and underestimate the
a reinforcer he or she might attend to a more sa- weighting given major cues. In another series of
lient feature of the interaction. Furthermore, if the studies they showed there was little relation be-
individual is not trained in reinforcement theory, tween a subject’s confidence in his or her rating
he or she might miss the fact that these events oc- and accuracy.
curred contingently, and were followed by The idea that we have little or no accurate
changes in p (R;|Aj) or in p(Rj). awareness of higher order cognitive processes
If one accepts the reality imposed by the OSLC seems well-established, at least as it applies to fair-
coding system, then in one hour of interaction a ly complex situations. However, given an ex-
child would be required to sort through 600 of his tremely “simple” situation, there may be a corre-
or her own behavioral events! In addition, the spondence between the functional relations actual-
child would view at least 600 events generated by ly occurring and the subject’s theory about them
the reactions of other family members. The ques- (Wilson & Nisbett, 1978). An excellent example
tion arises as to just how aware the individual can of such a simplistic situation can be found in the
be of these processes. How accurately might he or verbal conditioning studies that were popular in
she describe the contingencies that are found with- the early 1960's. In these studies, the subject was
in these interactions? The writer assumes that brought into a room with the experimenter, and
most of the “explanations” we give for our own asked to say whatever words came to mind, or
behavior are not isomorphic with the molecular perhaps to make up sentences employing particu-
processes which determine them. Modern devel- lar words. During the baseline period, the experi-
opments in cognitive theory suggest that what we menter interacted very little with the subject. Fol-
do is far more reasonable. To store and actively lowing this, the experimenter began to interact by
search for the functional relations between 1200 saying, “Hmmm,” “Good,” or “Uh-huh” contin-
events would take up a great deal of the individu- gent on a particular response of the subject. Nis-

90
bett et al. (1977) reviewed these early studies and forcer), then “conditioning” occurred. If the sub-
concluded: ject was not aware, there was no conditioning.
These demolition studies were well done; there
“The present analysis makes it clear that there is
was no real argument about either their design or
every reason to expect that subjects in these exper-
interpretation. It had been assumed that reinforce-
iments should be able to accurately report about
ment produced an automatic strengthening be-
cause and effect. (a) The response possibilities al-
tween the S and R without awareness. This as-
lowed the subject are extremely constrained. He is
sumption was no longer tenable. As a result, many
permitted very little latitude in the sorts of behav-
investigators left the area of social reinforcement
ior he may emit. (b) The stimulus situation is even
more fixed and static. In fact, virtually the only to work on other problems.
stimulus that occurs is the experimenter’s ‘uhhuh’ During the next decade, studies concerning au-
or ‘good.’ (c) Finally, the causal connection be- tomatic strengthening were few in number and
tween this critical stimulus or reinforcement and went largely unnoticed. Reinforcement principles
the increased frequency of a particular response did become part of the social learning theory that
class should be a highly plausible one. . . . Devo- was developed in the mid-1960’s. Reinforcement
tees of learning-without-awareness could scarcely remained a crucial variable in the 1970's, but the
have designed a paradigm more likely to result in form in which it was expressed was radically al-
accurate verbal report if they had set out deliber- tered.
ately to do so. There is some evidence, in fact, that “Although the empirical issue is not yet completely
when even relatively minor steps are taken to dis- resolved, there is little evidence that rewards func-
guise the connection between stimulus and re- tion as automatic strengtheners of human con-
sponse, subjects will fail to report such a connec- duct. . . . Behavior is not much affected by its con-
tion.” (pp. 253-254) sequences without awareness of what is being re-
inforced . . . reinforcement—its image has changed
It is a pleasant state of affairs for the iconoclast
when a group considers any variable or arrange- from a ‘mechanical strengthener’ of conduct to an
ment as a necessary and sufficient condition for ‘informative and motivating influence.’” (Bandu-
determining human behavior. If one were to hold ra, 1974, p. 12)
the position that reinforcers must always be pres- Bandura’s statement accurately reflects the cur-
ent in any learning situation, then it is only neces- rent status of reinforcement within social learning
sary to find one exception and the entire logical theory with its emphasis on cognitive variables.
construct is in jeopardy. It requires but one dem- His statement stands as a “strong law of cogni-
onstration of a young chimpanzee learning the use tion.” Subject awareness (as expressed by a verbal-
of a new tool by observing a peer successfully em- ized statement of contingencies) is thought to be
ploying it to produce an intellectual scramble to necessary for reinforcement effects to occur. How-
extend the logic so that it explains the exception. ever, the present writer believes that there are now
Certainly the work of the modeling theorists such a number of studies which show that reinforce-
as Kanfer (1970) and Bandura and Walters (1963) ment effects can be demonstrated under condi-
in the mid-1960’s set limits on the expansive hori- tions in which the subject is not able to verbalize
zons claimed by the reinforcement theorists of that the contingencies. If an awareness of the informa-
time. However, in that most investigators did not tional characteristics of reinforcers is assumed to
take seriously the claim that reinforcement was be a necessary condition, then all that is required
necessary for behavior change, the findings from to refute this is one demonstration that behavior
modeling studies did not require a drastic readjust- changed via reinforcement without the subject be-
ment. ing able to report awareness. The format for these
The assumption at the core of the enterprise studies is simple—provide the subject with salient
concerned the automatic strengthening of the S-R stimuli to which he or she attends, while arranging
bond as a function of reinforcement. The early for contingencies which produce behavioral
studies of verbal conditioning were held up as cen- changes that are not noticed.
terpieces demonstrating this effect in study after Some of the work demonstrating automatic
study. Then the careful laboratory controls intro- strengthening was done prior to the demolition by
duced by Spielberger (1962) and others demolished Spielberger (1962), but these studies had gone un-
that premise. His work and that of other investiga- noticed by him (e.g., Hefferline, Keenan, & Har-
tors published in Eriksen (1962) showed that if the ford, 1959). In a later study by Hefferline, Kee-
subject could verbalize his or her awareness of the nan, and Birch (cited in Hefferline, 1963), 12
connection (between the response and the rein- adults participated in an experiment consisting of

91
 three phases. During the baseline period, they lis- hypothesis. Second, complete data were reported
tened to music while an apparatus recorded very for only one subject. Clearly, the work needed to
small muscle twitches. During the conditioning be replicated. This was provided by Gewirtz and
trial, white noise was introduced—its termination Boyd (1977) working in a different laboratory set-
was made contingent on the occurence of these mi- ting and using mothers as subjects. In this double-
nor muscle movements. An interview revealed that agent design, the mothers had been instructed to
the subjects had no awareness of the contingent use verbal reinforcers to shape their infants’ verbal
arrangement and yet all of the subjects showed a behavior. The infants were in an adjoining room.
conditioning effect. They were then told that some The mother was told that whenever a light came
aspect of their behavior was related to turning off on it meant that her infant had turned its head.
the white noise. In the next phase, the condition- While the mother focused on infant sounds, the
ing trials continued while the subjects searched for experimenter made the information about infant
the arrangement. While they were unsuccessful in head turns contingent on mother smiles. The rate
stating the correct hypothesis, all of the subjects of mothers’ smiling increased without their being
showed increases in muscle twitches while the aware of the change in their behavior or the con-
white noise was present. tingencies that produced it. In the next study, the
This early study has many overtones that are mother was to reinforce head turning. This time
reminiscent of modern biofeedback studies. Even infant sounds were fed back to the mother; the
though the format was of historical interest, the sounds were made contingent (by the experiment-
results could not necessarily be generalized to so- er) on the mother smiling. Again, the reinforce-
cial situations in which the Rj event could, at least ment arrangement had a significant effect on the
in principle, be observed. For this reason, the dou- smiling behavior of the mother. No evidence of
ble-agent studies by Rosenfeld and Baer (1969) awareness could be obtained from the interviews.
and by Gewirtz and Boyd (1977) are more directly It may be the case that more skillful interviewers
relevant to the present discussion. In the first study could demonstrate that the subjects in such ar-
by Rosenfeld and Baer (1969), a college student rangements really are aware. Indeed, this is acom-
was cast in the role of subject; he also served as pelling reason for carrying out double-agent
“double agent.” The experimenter (a graduate stu- studies in other settings. However, as things now
dent) planned to reinforce the subject for “chin- stand, these findings cast serious doubt on the as-
rubs.” However, the student (in collusion with Ro- sumption that an awareness of the reinforcing
senfeld) had a prior agreement to make a chin-rub contingencies is a necessary condition for changes
response contingent on the “experimenter” saying in performance.
“yeah.” This well-designed single-subject experi- The findings from recent developments in cog-
ment began with a baseline measure of the experi- nitive psychology strongly suggest that behavior
menter’s frequency for saying “yeah.” The as- changes occur in complex social interactions with-
sumption was that the target behavior for the ex- out the participants being aware of them. The
perimenter’s experiment would function as a rein- writer believes that most changes in behavior have
forcer if made contingent upon his behavior. Dur- this quality. The changes occur in very small incre-
ing the next phase, chin-rub was made contingent ments and are embedded in a rich matrix of ongo-
upon the experimenter’s “yeah.” The data re- ing stimulation. The shifts are seldom constant
flected a marked increase in rate; during the next and are not necessarily unidirectional. The contin-
baseline the rate fell again. Reinforcing an alterna- gencies producing these changes tend to go unno-
tive response of the experimenter’s had no effect ticed. If there is a systematic arrangement for these
on his rate of saying “yeah.” This was demon- contingencies, then over a period of time these
strated in an ABAB reversal design, in which changes will be directional. Such large changes
“hmmm” was reinforced during the A conditions would lead to labeling or explanations which re-
and “yeah” during the B conditions. A carefully flect the conventional wisdom about such matters.
designed interview probe at the end of the study For most people, the causes and nature of behav-
failed to reveal an awareness on the part of the ex- ior changes are a profound mystery. In fact, most
perimenter that his behavior had been repeatedly parents of antisocial children are convinced that
changed during the experiment. the behavior of their children cannot be changed.
There are, however, several considerations that It seems likely that at the microsocial level many
give one pause in wholeheartedly accepting these changes in behavior do occur, and that most of us
findings. First, the experiments were carried out in are not aware of these changes nor do we have ac-
a setting (University of Kansas) that could not be curate explanations as to their cause. In this sense
thought of as neutral with regard to the awareness we are in agreement with Freud who also believed

92
that most persons were unable to give an accurate modeled by other people. The intricacies of com-
account for behavioral determinants. At the very plex social behaviors, as well as language, aggres-
least, the data call for a wait-and-see attitude re- sion, and other social skills are first observed and,
garding the necessity of cognitive mediation in re- together with the likelihoods of various outcomes,
inforcement. There is obviously much to be stored in memory.
learned about cognitive processes. It is no longer
“In the present context, it was assumed that the
believed that there is a simple relation between
average three-year-old, in our society, has learned
cognition and behavior. For example, recent
all of the 14 noxious behaviors identified by our
studies demonstrate that changes in behavior are
code system. This early acquisition is facilitated by
more likely to determine changes in attitude rather
the ubiquitous presence of coercive models in the
than the more traditional belief that attitudes de-
home, nursery school, and on television. For ex-
termine behavior (Bem, 1967).
ample, observations (Jones et al., 1975), in the
In sum, the relationships between cognition, re-
homes of normal families showed coercive behav-
inforcement, and behavior define a set of complex
iors occurred from a range of .02 to .50 responses
relations that are not yet well understood. For ex-
per minute! Because the code sampled only dyads,
ample, it would be fascinating to study when it is
these figures represent minimal estimates of the
that a victim will reinforce an attack. What attri-
rates with which such aversive stimuli occurred.
butions are given to the victim’s behavior and the
Presumably these events provide rich opportuni-
behavior of the attacker? Our own structural
ties for vicarious learning for young children. Ex-
questioning concerning this kind of relationship
tensive observations from two nursery schools
suggests the participants are poor trackers of what
showed a range of 11 to 40 verbal or nonverbal at-
they do and what is being done to them. They are
tacks per session. . . . In the review of studies eval-
being altered by minutiae that, as isolated events,
uating the content of children’s TV programs,
seem innocuous enough. However, when they are
Friedrich and Stein (1974) cited a study that
repeated a dozen times, they do have an effect on
showed an average of 25.1 aggressive episodes per
performance. Because these events are not consid-
hour for children’s cartoons. For ‘adult’ shows,
ered to be important, they pass unnoticed. If
there was a mean of five violent episodes per hour
brought into awareness, would such knowledge
for shows presented on prime TV time. Most chil-
facilitate or inhibit changes in behavior?
dren surveyed were found to watch three to four
The Learning of Aggression hours of TV per day. . . .” (Patterson, 1976, p.
280)
It is reasonable to believe that some coercive be-
haviors are present at birth, and, in that sense, do While all children in our culture may have
not need to be learned. The infant’s crying is aver- “learned” a variety of coercive behaviors, they dif-
sive. It is elicited by various discomforts experi- fer profoundly in the rate and skill with which
enced by the infant, e.g., extremes of temperature, they perform them. But, as noted earlier in the dis-
or hunger. The ethologist would point out that it cussion, in order to become a complex social skill,
also has survival value; crying is used to alert the what is learned must be subjected to the polishing
caretaker as to when to change diapers and when touch of natural contingencies. One might master
to feed. As time goes on, the infant learns other the art of using a spoon at the table by viewing a
means for coping with hunger or discomfort. It film. A brief period of practice with its concomi-
should be noted that while these particular S-R tant set of contingencies would probably be suffi-
connections are probably innate, they are also def- cient. However, no amount of viewing and model-
initely modifiable. ing would directly produce a concert violinist or a
There are probably other coercive behaviors for race car driver. Similarly, many of the subtle coer-
which the connection between eliciting stimuli and cive skills seen on video tapes of clients in therapy
responses is innate. In his ethological studies of require years of extensive practice and shaping.
primates and children, Eibl-Eibesfeldt (1974) iden- Over a period of time, the to-be-aggressive child
tified temper tantrums as having such a status. He must learn which set of coercive responses will be
found this reaction in a deaf and blind girl. The successful with which family member. For exam-
pattern of facial grimace and posture was very ple, observation data showed both normal and de-
similar to that observed for sighted children and viant boys learn that they may not hit their moth-
young primates. ers, but they can hit siblings (Patterson, 1980a).
Modeling theorists such as Kanfer and Phillips The young boy learns that whining, noncompli-
(1970) and Bandura (1973) put forth the persua- ance, and yelling are appropriate when mothers
sive argument that much of what a child learns is are the targets; teasing, humiliation, and negative

93
commands are useful coercive modes for siblings. its wars and battles. The great powers, including
The coercive child must also learn about set- the United States, have been the most frequent
tings. The reinforcing/punishing contingencies fighters. Wars seem to accompany a country’s
provided during practice sessions will teach the co- march to power; until recent times, wars were re-
ercive child where and when coercive behaviors warding to the winner.
are likely to be successful. For example, most fam- Prior to the nineteenth century and the advent
ilies do not permit coercive interactions in of the modern conscript armies, 30% to 50% of
churches, hospitals and restaurants. Some will those engaged in a battle were likely to be casual-
permit its occurrence when company is present, ties. One wonders what the reinforcer could pos-
others will not. Some parents control its occur- sibly have been for those combat troops. Even a
rence in public places such as supermarkets and cursory examination of history suggests that the
parks; others do not. rewards were well understood and carefully
By the age of 6 years, most normal children spelled out in advance for all participants from
have these lessons well in hand. Their rate is about kings, generals, and lieutenants to foot soldiers.
one aversive event every two to three minutes; For example, Grant’s (1974) analysis of the struc-
most of these are relatively low-intensity events ture of the armies of the Caesars read:
such as disapproval, noncompliance, and teasing.
The patterns of these behaviors are relatively sta- “The imperator called the soldiers my soldiers;
ble over time with regard to overall rate. By this and they were prepared to follow him loyally if he
age the child carefully discriminates between vic- rewarded them sufficiently well. Rewards meant
tims, settings, and responses. In the home, behav- not only their pay and the occasional bonus, but
ior patterns have been shaped by the family mem- above all a suitable provision for their retirement.
bers. People in other settings will teach the child to Without that all the pay in the world would not
make comparable discriminations in those set- have satisfied them.
tings. The normal child’s coerciveness is accepta- “Ex-soldiers enjoyed many privileges, and ex-
emptions. On a papyrus of 31 B.C., the victor of
ble; it /ooks normal in that things are seldom car-
Actium listed some of them. ‘I have decided to de-
ried too far. It is the victims’ reactions (which
function as reinforcement and/or punishment) cree that all veterans be granted exemption from
tribute . . . to grant to them, their parents and chil-
that polish these skills. In effect, the victims teach
the child what is acceptable and what is not. dren, and the wives they have or shall have, ex-
emption of all their property from taxation; and to
Cultural Programming the end that they may be Roman citizens with full-
est legal right, they shall be exempt from the per-
At a time when it is popular to consider aggres-
formance of compulsory public services.’” (Pp.
sion as an instinct (Lorenz, 1966), it is perhaps not
fashionable to consider the possibility that our cul- 79-80).
ture is programmed to reward many kinds of ag- In Tuchman’s (1978) account, this model was
gression. However, the present writer believes our the rule rather than the exception for 14th century
particular culture has a history in which certain English, French, and German armies. For many of
kinds of aggression are highly prized. Our folk the participants the incessant wars represented an
heroes have been the cowboy and the pioneer; escape from the crushing poverty and taxes of that
both are individualists. In our mythology about time. For foot soldiers, knights, and lords alike, it
them, we emphasize their combativeness. Sports was the possibility of dramatic material gain that
such as football, boxing and hockey glorify ag- determined whether an army functioned well or
gressiveness. Shaw’s (1972) illuminating Meat on melted away (as many of them did). What was im-
the Hoof describes the payoffs for professional pressive was the meticulous care with which these
football players who are extremely aggressive. reinforcers were spelled out in advance. For exam-
The reinforcers range from money, cars, and sex ple, when Cortes invaded the Aztec kingdom,
to membership in a supportive group. Winning is land, slaves and gold were the primary reinforcers.
the primary goal. Shaw’s account leaves little
doubt that the dual process of enduring pain and “The gold dollar or sterling value of the total gold
inflicting pain on the opponent is a means to and silver put together, counting Motecucuma’s
achieve that end. treasure and the gold received from the provinces,
At a more macrolevel, war is the ultimate ex- has been calculated at six million three hundred
pression of aggression. In his statistical analysis, thousand dollars. . . . But the distribution could
Wright (1972) noted the correlation between the not possibly be so simple. The deep loyal feelings
status or strength of a nation and the frequency of of the army to the Crown may be seen in the

94
promptness with which the royal fifth is set aside. inferential leap to view the small infantry squad as
Then Cortes claimed the fifth which had been providing the day-by-day reinforcement necessary
promised him in Veracruz. Then, Bernal Diaz to maintain each other’s effective combat perfor-
grumbles, Cortes set aside sums for the expendi- mance. But these contingencies are only distantly
tures he had made towards the expedition, and for related to the processes maintaining aggression in
Diego Valazquez’ expenditures, and for the pro- the home. While reinforcement is thought to be a
curators sent to Castille, and for the men left be- key mechanism, this time it is dispensed by the vic-
hind in Veracruz, and a double share for the tim; the victim’s reaction maintains the behavior
horsemen; and in the end the common soldier of the attacker. If the adults in those settings give
found his share so small that many shouted they their tacit permission, then the more skilled coerc-
would not receive it.” (de Madariaga, 1967, pp. ers will seek out the more compliant victims and
282-283). the coercion process will begin.
The countries that were most involved in pro- There are two general questions to consider.
grams of military expansion also emphasized early The first has to do with the likelihood of positive
training of their youth in organizations espousing consequences for aggression in the classroom and
duty, obedience, competition, and endurance of the home. It seems contrary to common sense to
pain. The external reinforcers for these activities believe that victims provide positive outcomes that
were carefully planned. For many ancient wars the are sufficient to maintain the attacking behavior.
reinforcers were agreed upon prior to the engage- At a simple descriptive level, what is the likelihood
ment. These facts do not prove that the presence of a positive outcome for a particularly coercive
of reinforcers was a contributing cause for wars. response? Is the likelihood of a positive outcome
However, they do suggest that the presence (then stable from one week to another? To understand
and now) of these and other contingencies should the antisocial child it is imperative that we obtain
be considered as playing a role in maintaining so- answers to these questions. A related question
cially condoned aggression such as that found in concerns the impact of these consequences in
football, boxing, and war. War is the exemplar of strenthening the A;—> R; connections. Do these
aggression. It is unfortunate that there are no his- consequences function as reinforcers? How is this
torical studies analyzing the reinforcing contingen- to be demonstrated? This second set of questions
cies supplied for mass killings. will be discussed in a later section of this chapter.
The topic of cultural programming for aggres- Richard Walters suggested several decades ago
sion was introduced to illustrate the feasibility of that the contingencies are likely to be very differ-
viewing this and other macro-problems within the ent for high- and low-amplitude aggression. I am
framework of behaviors perhaps partially deter- not convinced that this is true, but I do believe that
mined by reinforcers. A culture, like a family, may they differ in the processes producing them.
have some unspoken assumptions about the rela- Within family interactions the main function of
tive utility of aggression as it relates to survival. high-amplitude aggression is to produce compli-
Whether or not it is a stated assumption, one ance and/or submission. For the purpose of the
could test for its presence by examining the extent present discussion, suffice it to say that if a sudden
to which competition, aggression, and violence increase in intensity is followed by victim compli-
are presented to its youth as models for the solu- ance then there is an increase in the likelihood that
tion of problems. It is assumed that those cultures the next trial will begin at or above the same inten-
most frequently engaging in war would model sity. The details of this process are presented in
these “solutions” at the highest rates in their Chapter 7. High-amplitude responses have their
stories, drama, and films. Their sports activities beginnings in rather low-amplitude interchanges.
would probably also emphasize body contact and In that context, it becomes imperative to under-
the skillful employment of pain control tech- stand the processes that maintain relatively innoc-
niques. In the values they place upon antisocial be- uous interchanges. These microsocial processes
havior, some families are like those cultures. How- are thought to be necessary as antecedents for
ever, they differ in one fundamental respect; the more violent forms of abuse among family mem-
family permits coercion to occur among its own bers.
members. If a society were to behave in a similar What are the positive outcomes for familial co-
way, anarchy would be the immediate outcome. ercion? For that matter, what are the positive out-
comes for family interaction in general? Descrip-
The Reinforcement of Aggression tive data generated by such questions emphasize
One can identify the external reinforcers the banality of much of familial interaction. At
claimed by the victors of ancient wars. It takes no any particular time there does not seem to be very

95
much going on. As we shall see in later discus- child, or any family member, employ coercive
sions, even in very distressed families only around means to achieve this same end? It should be said
12% of the interactions are coercive. It is also the at the outset that there is not, as yet, an adequate
case that there is a general lack of high-grade posi- answer to this question. However, the author be-
tive consequences such as Approving, Touching, lieves that when the answer is in hand, it will in-
or Kissing. The normative data in Reid (1978, p. clude a reference to some peculiarities in adult-
72) showed Approval was given by parents to child interaction patterns. The essence of it seems
family members only about once every ten min- to be that adults quickly tire of interactions with
utes. Hugs, Touches, and Kisses were forthcom- young children. Technically speaking, adults seem
ing only about once every 200 minutes! The ratio to satiate rather quickly to the reinforcers found in
of positive consequences (Talk, Attend, Physical ongoing interaction with children. The first clue to
Positive, and Approve) to aversive consequences this was to be found in the classic study by the
for normal families was 13.9:1. The findings for Barker group for American and English children
home settings are reminiscent of the conclusions (Barker et al., 1962). They showed that, for Mid-
reached by White (1975), who reviewed 16 obser- western children, 89% of all interaction episodes
vation studies conducted in classroom settings. lasted less than one minute! The comparable fig-
She found that, prior to grade two, teachers’ use of ure for English children was 84%. The term epi-
Disapproval more or less balanced their use of Ap- sode was used to describe a unit of social interac-
proval. However, from that point on they used tion in which some action was completed.
punishment much more often than positive conse- The theme of the brief encounter is reiterated in
quences! White points out that this is also about a study of 30 socially aggressive and 20 normal
the same time that most persons report a loss of children from the OSLC files. The details of the
the sense of joy in being in school. These findings sample and procedures are described in Appendix
suggest that, as a culture, we are not committed to 5.1. For each child the likelihood was calculated
the idea of providing what might be considered that the mother would continue interacting with
high-incentive positive reinforcers for child behav- the child in a prosical manner, given that the child
ior. had been prosocial in the immediately preceding
interaction. The analysis was carried out to the
Why Coerce? eighth interchange in the sequence. The ANOVA
The schedules for Touches, Hugs, and Approv- for repeated measures controlled for age (young/
als are very lean. I think that most family interac- old) and normal or clinical status of the child. The
tions are, instead, maintained bya variety of less analyses showed a significant main effect for tri-
interesting consequences which are in greater sup- als, but nonsignificant main effects for groups or
ply. The consequences would include Attend, age. For this reason the data for all four groups
Talk, and Play. The observation data from the were combined and summarized in Figure 5.1.
homes of 33 normal children by Johnson, Wahl, Given a prior prosocial child response, then the
Martin, and Johansson (1973) showed 17% posi- likelihood was .65 that the mother would con-
tive consequences for prosocial child behavior (for tinue interacting with the child into the adjacent
example, the child Talks and the mother Attends; six-second interval. If she continued through two
the child Plays and a sibling Plays with him). In interchanges, then the conditional p was .35 that
most families, there is an abundance of garden-va- she would continue interacting in the third inter-
riety positive consequences. In disrupted families, val. For both normal and distressed children it is
coercive child behaviors are a fairly reliable means unlikely that interactions with their mothers will
for producing these positive outcomes. extend beyond 18 seconds. I think that it is at in-
In a sense, modern social learning theory is in tervals four, five, etc. that the child becomes in-
agreement with the clinical impressions reported creasingly likely to initiate a coercive response.
by Alfred Adler one-half century ago (Adler, Here the function is to reinstate parental atten-
1929). He believed that most of the so-called dis- tion.
turbed child behavior was exemplary of “attention
Context
getting.” To this we, in turn, add one interesting
bilateral twist. When the family member responds Thus far the formulation about consequences
by Attending or Talking, then the child terminates for coercive behavior has twice introduced as-
the coercive behavior. In so doing, the child rein- sumptions emphasizing the critical role of the an-
forces the victims for supplying the reinforcer. If tecedent in defining the meaning of the coercive
one grants a high likelihood of Talk or Attend giv- event itself. In one instance events having aversive
en prosocial child behavior, then why does the antecedents were set aside as being examples of

96
 Figure 5.1
Mothers’ Responsiveness

ey

¥ oy
oa
-6
) 6

3 AS)
ben
vu
s
-—
°
S 4
a]
)
i}) 3
7)
a
—
of?

R, R, R; R, R,; Rg R, Rg

Consecutive Interactions

negative reinforcement. The preceding section em- quirements for both age and the diagnostic label of
phasized intervals during which people stopped at- socially aggressive. It consisted of children re-
tending as setting the occasion for coercive behav- ferred by parents and/or community agents be-
ior. Are neutral events an important source of the cause of high rates of coercive behaviors directed
antecedents for attacks? If so, then coercive events toward other people (e.g., Fight, Not Mind,
in such a context might be thought of as attention Tease, Temper Tantrums, Whine, Yell). Of those
getting. Attacks that follow positive antecedents referred for such problems, cases were selected for
might be thought of as attention maintaining. Pre- the sample when the observation data collected in
sumably, these functions would be served whether the home showed their rates of coercive behavior
the participants were parents or children (e.g., to exceed .45 responses per minute. The normal
some wives may nag just to get their husbands’ at- sample (N = 20) was obtained by advertising in the
tention). newspaper for families with nonproblem children
To provide the necessary descriptive data, sam- to be observed. The two samples were matched
ples were drawn from a population of problem for father absence, number of siblings, age of the
and non-problem families studied at OSLC since target child, and occupational level. As indicated
1968. (The samples are described in Appendix in Table 5.1, the findings showed the samples to
5.1). Cases were selected that fell into the age be comparable on all variables.
group of 3.0 to 8.5 years old or 8.6 through 13.5 Context was defined in terms of the behavior of
years old. For the clinical sample (N =30), only the other people which occurred as an immediate
those cases were included that met the dual re- antecedent (Aj;) to the coercive target event. The

OF
 Table 5.1
Antecedent Contexts for Coercive Child Behavior

p(Coercive|AntecedentContext)
Nondistressed Socially Aggressive F values for
Younger Older Younger Older
Contexts (N=10) (N=10) Mean (N=15) (N=15) Mean Sample Age Interaction

Aversive <9 1) .26 35 33) .34 pk 1.94 1.18

(9)?
Neutral Siy/ .24 Ail .26 HAs) .28 DAZ .134 0.17
(9)
Positive 50 ie 54 40 38 Bi) 5.70* .14 0.43

*p<.10
Fo<a0'5:
>The numbers in parentheses refer to the size of the sample. The information is given only when the sample differs from the expected
values of 10 for normal subsamples and 15 for deviant subsamples.

Table 5.2
Positive Consequences for Coercive Child Behaviors

Percent of positive outcomes:

Settings Age Provided by Normal samples Deviant samples
1. Institutionsfor delinquents Adolescents Peers & Staff 70%
(Buehler,Patterson, & Furniss,
1966)
2. Nursery school 3-4 Peers 80%
(Patterson,Littman, & Bricker,
1967) N=36
3. Home
(Wahl, 1971) N=33 Parents 47%
4-6 Siblings 28%
(Snyder,1977) N=20 5-10 Family 64.7% 69.5%

comparisons were based on the proportion of co-

ercive episodes involving neutral, positive, and Empirical Findings
aversive antecedents or p (Coercive|A;). As shown The data collected in the home and school have
in Table 5.1, the trends for the clinical sample consistently shown very high rates of positive con-
showed higher proportions of aversive and neutral sequences provided by adults and peers contingent
contexts than was the case for the normal samples. upon coercive child behaviors. For example, in the
These trends were, however, of borderline signifi- study by Walker and Buckley (1973) deviant chil-
cance. The finding with real surprise value in- dren received 77% of one teacher’s attention! A
volved the significantly higher likelihood of atten- similar situation was found for one teacher but not
tion maintaining episodes for the normal sample. for another in the study by Madsen, Becker, and
On the average, 54% of the normal child’s attacks Thomas (1968).? The earlier studies also pointed
occurred in a positive setting, as compared to 39% out that in institutions designed to treat antisocial
for problem children. In fact, across all groups, at- children there were surprisingly rich schedules of
tacks were most likely to occur in a positive con- positive consequences for deviant behavior (Bueh-
text, i.e., the other family member was engaged in ler, Furniss, & Patterson, 1966). The supporting
some prosocial behavior such as talking or playing reactions were typically provided by peers (but on
with the target child. occasion by the staff as well) for verbal statements

98
 Table 5.3
Positive Payoffs for Coercive Child Behaviors: Normal Sample
(from Johnson, Wahl, Martin, & Johanssen, 1973, pp. 44-45)

Number
Percent Percent children
Coercive Rate per positive negative emitting
behavior! minute consequences* = consequences behavior

1. Destructiveness .0013 17.6% 35.2% 6

2. Physical negative .0167 34.5 53.0 2)

3. Command negative .0126 38.2 57.8 16

4. Smart talk .0080 She} 39.0 13

5. Command .1204 Gand 35.8 2

6. High rate behavior .0644 21.6 Gr) 25

7. Negativism -0061 43.1 By? 13

8. Tease .0110 41.7 38.2 13

9. Yell .0410 55.8 21.8 24

1Only behaviors evidenced by four or more subjects were included, e.g., only two showed Temper Tantrums.
*These percentages will not total 100% because neutral reactions were not analyzed here.

Reprinted with permission from “How deviant is the normal child: A behavioral analysis of the preschool child and his family.” In
R.D. Rubin, J.P. Brady, & J.D. Henderson, editors, Advances in Behavior Therapy. Copyright 1973 by Academic Press, Inc.

with antisocial connotations. Gelfand, Gelfand, forcers. If these consequences are reinforcers then
and Dobson (1967) found essentially the same their density is certainly sufficient to maintain be-
pattern for patients in a mental hospital; i.e., staff havior.
and patients tended to react supportively to devi- It would be of some interest to know if some co-
ant behavior. Table 5.2 summarizes the findings ercive responses receive more positive conse-
from several studies. In the nursery school setting quences than others. The study of preschool chil-
coercive responses were reacted to by the victims dren by Johnson, Wahl, Martin, and Johansson
withdrawing from the area, crying, or giving up (1973) provides data that are relevant to this ques-
an object or play setting. As can be seen, peer vic- tion. They used a home observation code that was
tims reacted supportively 80% of the time. a modification of the FICS. They analyzed for
For normal families, the findings represent a each of nine coercive responses both the rate of oc-
range in likelihoods from .47 to .65. Though these currence and the proportion of aversive and posi-
are indeed rich schedules, their existence does not tive consequences for that response.* As described
directly support the hypothesis that the schedules in Table 5.3, child-Command was the most fre-
for positive consequences (C+) are functionally quently occurring coercive behavior. About 62%
related to the target behaviors. The implication is of these behaviors received a positive consequence
that the probability of a positive consequence from family members. There was no covariation
given an attack response or p (C +|R)) is signifi- between response rates and likelihoods of positive
cantly higher than the base-rate p (C+). For ex- consequences. The correlation between base rates
ample, it could be that no matter what the child of occurrence and likelihood of positive outcome
does, he or she will receive positive payoffs that was only .13 (n.s.). The covariation question will
are of this magnitude. In further studies, the base- be considered in more detail in a later section. But
rate values should be included so that results such the Johnson data emphasize the differences among
as those in Table 5.2 can be interpreted. However, coercive responses. Some are more likely to be
at a simple descriptive level these findings are ex- punished, others are more likely to receive positive
tremely interesting. The social environment does consequences.
provide a rich supply of positive consequences for The next question involves potential differences
coercive child behavior. It is not known if they are in the reactions of the various family members.
functionally related, or if they function as rein- Are mothers more or less likely to respond in a

99
 _ Table 5.4
Likelihood Child Coercion Followed by Positive Consequence: p(C + |Rj)

Samples:
Family agent Dependent variable Normal Social Aggressive Abused
All deviant
Father (TAB score) fd 65 .66
Mother fas rife) .70
Brothers 63 .60 oa
Sisters .67 63 66
Hostile
Father .63 61 .64
Mother 69 ed, 74
Brothers Bhs} .67 .67
Sisters .67 .70 .67
Social Aggressive
Father 34 .24 Bos
Mother 34 40 34
Brothers oF asi .76
Sisters .40 .46 2
In this analysis, a positive consequence included neutral plus positive behaviors (i.e., approve, attend, command, comply, indulgence,
laugh, normative, no response, play, physical positive, receive, self-stimulate, talk, touch, work).

positive manner than fathers? Are siblings more Negative) directed toward them. This finding reit-
likely to be supportive than parents? As suggested erates the fine discriminations involved in the per-
by Johnson’s work, family members will respond formance of coercive skills. The skilled antisocial
differently depending on which coercive behaviors child. must learn which people, what coercive be-
are presented to them. To facilitate these compari- havior, where (the setting), and when to practice
sons three different response classes were used as his trade successfully. It also seems certain that a
dependent variables. The first combined the reac- child can make discriminations of a finer order
tions to all 14 coercive responses. The other two than shown here. For example, the antisocial child
variables consisted of functionally defined classes probably knows just when to Tease a brother in
(Patterson, 1977a; Patterson & Cobb, 1973). order to produce a conditional p value for positive
Each class was comprised of responses controlled payoffs which is considerably above the .51 shown
by the same network of antecedent events (see in Table 5.4.
Chapter 8 for details). The comparisons were There is a problem relating to these descriptive
made for three samples drawn from the OSLC data which must be considered. The difficulty lies
data files: 36 Normals, 44 Social Aggressors, and in the fact that 30% to 40% of coercive child be-
19 Abused Children. The age range was from 4 havior comes in bursts. While these matters will be
through 12 years. considered in greater detail both in this chapter
Table 5.4 indicates few differences among fami- and in Chapter 8, suffice it to say here that, for
ly members or between samples. The findings for problem children, one deviant behavior is likely to
the Total Aversive Behavior (TAB) category are di- be followed by another. Antisocial children are
rectly comparable to those obtained by Snyder characterized not only by more of these sequences,
(1977). Comparisons for the Hostile class also but the chains are of longer duration as well (Pat-
showed no significant difference. terson, 1976). If the deviant chains are lengthy, it
In Table 5.4 the fewest positive consequences is quite likely that family members are reacting to
were provided for events in the Social Aggressor the child while the chain is occurring. As will be
class. In both clinical samples siblings were more discussed in Chapter 8, these concomitant events
likely than parents to respond positively to socially are the critical. factor determining the eventual
aggressive behavior (e.g., Teasing or Physical length of the chain and the overall rate of aggres-

100
 Table 5.5
Likelihood of Positive Final Outcome

Normal Socially Aggressive F values for

Younger Older Younger Older
Context N=10 N=10 N=15 N=15 Sample Age Interaction

Positive 369) .47 =p) .40 19S 94S OIG

(9)t (9)
Neutral 20) .68 45 eA 1.74 2.02 LSEZO=
(8) (9) (14)
on EA

+The numbers in parentheses refer to the size of the subsample. These figures are included only when they differ from the expected
values.

sion. In analyzing discrete six-second events, the only a 20% payoff in the neutral context, but a
writer (1976), Johnson, Wahl, Martin, and Jo- whopping 69% in the positive context. There was
hansson (1973), and Snyder (1977) combined the a significant tendency for younger children in both
positive events found in mid-chain with those oc- samples to receive higher payoffs for attention
curring as the outcome. The writer now has rea- maintaining behaviors (i.e., positive contexts). In
son to believe that they serve very different func- the neutral context it was the older normal chil-
tions. If this formulation is correct, then the earlier dren and the younger deviant children who re-
analyses may not provide an accurate description ceived significantly higher payoffs than other age
of the incidence of positive consequences for coer- groups.
cive events (the likelihood of a positive conse-
quence might be very different at mid-chain than
Relation of Consequences to Inter- and
the likelihood as a final outcome).
To investigate this possibility, the same samples
Intraindividual Variance
were used as the samples employed for Table 5.1. One of the primary goals claimed for a coercion
In the present analysis only those events that fol- performance theory is that it will account for sig-
lowed an isolated coercive behavior or those nificant portions of the variance in behavior. In
found at the end of a coercive sequence were tabu- what sense does knowledge of positive conse-
lated. Comparisons were made between normal quences for behavior contribute to this particular
and clinical samples as well as between younger enterprise?
and older target children. As expected, the findings The traditional literature from reinforcement
for final outcomes in Table 5.5 were somewhat theory would lead one to conclude that there will
lower than those obtained for concomitants plus be no correlation between the likelihood of posi-
outcomes. For the normal sample, the mean for tive consequences and individual differences in
the combined data (concomitants plus outcomes) rates of aggression. As Herrnstein (1961, 1974)
was .68. This is in contrast to the mean of .51 for and others have pointed out, there is vot a linear
final outcomes shown in Table 5.5. The compara- monotonic relation between response strength and
ble data for the Social Aggression sample were .65 reinforcement density. The correlation data from
and .40 respectively. One number is no more real field studies are in accord with that position.
or accurate than the other. Rather, they represent Taplin (1974) carried out the first study of the
two different ways of describing the supportive re- relation between coercive child behavior and posi-
actions of family members. It could turn out that tive consequences provided by parents for these
they mean different things, but that is yet to be es- behaviors. He first examined the test-retest stabil-
tablished. ity of p (C + |DevBeh). The likelihoods were based
What stands out among the findings represented on the mean for three sessions from the first week
in Table 5.5 is the intricate relation between final of baseline and the comparable mean for the sec-
outcome, age of the target child, context, and clin- ond week of baseline. The PPM correlations (un-
ical or nonclinical status. Young normals received corrected) for mothers was .48 (df= 23; p<.001),

101
and for fathers .11 (n.s.). For both parents com- for aversive responses. How efficiently does the re-
bined, the correlation was .82 (df=23; p<.02). sponse produce the positive consequence? The
During baseline, the correlations between moth- presence of this structure implies that something is
er’s p (C + |Dev) and child’s rate of coercive behav- being done to control the behavior of the victim as
ior were .10 for week one, and .08 for week two. a reinforcer. In effect, the attacker trains the vic-
The comparable data for fathers were .15 and .15. tim to reinforce her or him for the attack. Chapter
The Taplin analysis did not support the idea of 7 considers this matter in detail as one component
a linear monotonic relation between the strength of the reinforcement trap. For the present, it is suf-
of a child’s coercive behavior and the likelihood of ficient to note that when the victim complies, the
the child being reinforced for it. As noted earlier, aggressor terminates the attack.
this is in keeping with the traditional findings from J.S. Watson (1979) and E. Thomas (personal
learning theory. There is a possibility that Taplin’s communication) make the extremely important
finding was obtained because of the double con- point that in the real world matters do not stop
founding due to combining data across contexts, here. There is a third function that is very impor-
and combining data for concomitants and out- tant. In the laboratory the two p values p (Rj|Aj)
comes. As a check, the data for final outcomes and p (C+|Rj) tell the greater part of the story.
were analyzed for the combined reactions of all However, in social interaction we need a measure
family members to the problem child. The correla- of how often the reinforcer occurs when it is not
tion between the problem child’s baseline TAB contingent upon that particular R;. In the labora-
score and the p(C + |Dev) was .25 (n.s.) for posi- tory, the reinforcer occurs only when the response
tive contexts and .11 (n.s.) for neutral -contexts. occurs. That is, p (prior R;|C +) is always 100%
Incidentally, the families most likely to pay off in (if a positive consequence occurs it is always pre-
one context were also most likely to do so in the ceded by Rj). But in family interaction, Talk,
other. The correlation between the two likeli- Approve, and Comply may follow many child be-
hoods was .47 (df=22; p<.02). The disposition haviors other than coercive acts. In social interac-
to react positively seems moderately stable across tion this important variable is left free to vary and
time and settings. the p (prior Rj|C+) is considerably less than
The prime function assigned to positive conse- 100%. One might think of this value as a measure
quences has been that of strengthening the S-R of the sufficiency with which the response pro-
bond. From this perspective, reinforcers make an duces the reinforcer. As Watson points out, in
indirect contribution to performance. Reinforce- social interaction the values p (prior R;|C +) and p
ment alters the status of the controlling stimuli (C+ [Rj) vary independently. In his laboratory
which, in turn, account for one type of perform- studies of infants, Watson has investigated the rel-
ance variance. It is thought that day-by-day fluctu- ative contribution of each in determining the shap-
ations in coercive behaviors are determined pri- ing of the infants’ behavior. The largest reinforce-
marily by changes in the density of the stimuli con- ment effects were obtained under conditions of
trolling their occurrence. matched indices of contingency. However one
may interpret these specific findings, these seminal
Some Functional Relations studies make a major point that must be con-
At the crux of the interactional stance is the idea sidered in the present context. The impact of a
that each member of a dyad changes the ongoing reinforcer on the Aj—» R; connection may be rela-
behavior of the other (Bandura, 1974; Bell, 1968; tive to the general availability of the reinforcer as
Bell & Harper, 1977; Gottman, 1979). This has expressed in Watson’s sufficiency index. This
some interesting implications for reinforcement point will be considered in more detail in a later
theory which have been largely ignored. section. What a reinforcer means, as expressed by
There are three functional relations of interest. its impact on the A;—» R; connection, is partly a
The first is the familiar p(Rj|Aj). The Aj is the pri- function of the matrix in which it occurs.
or behavior of the victim that elicited the attack.
Some Further Considerations
The conditional p value describes how efficacious
that A; is in producing the Rj. Here, the victim To understand the structure of social interac-
trains the attacker by providing positive reinforce- tion it is useful to think of reinforcement as one
ment for the attack. determinant for many of the functional relations.
The second relation describes the likelihood of a The review of the empirical findings emphasized
positive consequence (C + ) given an attack (Rj) or that the social environment is characterized by a
p(C+|R;). One might think of this as a statement rich supply of positive consequences for coercive
about the richness of the reinforcement schedule child behaviors. However, is it the case that these

102
positive consequences are reinforcers? What is the out the thematic effect of the subject’s own ongo-
evidence that the C +’s identified in field studies ing behavior. Chapter 8 documents the fact that
function as reinforcers? These issues are discussed one of a subject’s aggressive events will tend to fol-
in the section which follows. In addition, some low another, i.e., aggression comes in “chunks.”
brief speculations are presented concerning the When analyzing for shifts in the attacker’s behav-
question of why people respond to social rein-
forcers, and why there are individual differences in
this responsiveness.
Table 5.6
Tests for Status as a Reinforcer Shifts in Attack and/or Victim
How is one to determine that a C + identified in as a Function of Victim Reaction
a field setting functions as a reinforcer? This sec- (from Patterson, Littman, & Bricker,1967, p. 22)
tion examines three approaches found in the cur-
rent literature. They are ordered from the least to
Frequency
the most powerful as tests.
Subject assertive events Chi square
Changes in A;—» R; Connections. The first at-
tempt to employ an A;—» Rj format was a rather SC 106 iS eye a
crude affair because we did not fully appreciate 8C 86 Da
the complexity of the problems involved. For this 11C 56 10365
analysis, the nine most aggressive nursery school 14C 401 6:97e=
children were selected as targets (Patterson, Litt- 1A 335 14.68***
man & Bricker, 1967). At a minimum, each pro- 12P 56 DSi:
duced two or more aggressive responses per ses- 13P Zi; NOMS
sion. The positive consequences provided by the 14P 340 7180s
victim included: gives up toy, withdraws, does not 15P 49 0.04
respond, cries, and defensive posture (e.g., covers p00
up head). If these outcomes functioned as reinforc- ***p<.001
ers, they should have been associated with the in-
creased likelihoods that for his or her next attack
the aggressor would select the same victim (Aj)
and employ the same category of attack (Rj). At- ior, the 1967 study combined immediately follow-
tacks included the following: physical with object, ing events with those occurring several days later.
verbal, invades territory. Punishment would be We know now that only the latter could be
associated with a change in the A; and/or a change thought of as being a test of reinforcement effects,
in Rj. Either the attack would shift its form, or a i.e., changes in the p of R2 given A).°
new victim would be selected. Punishment in- One-half decade later the problem had become
cluded: tell the teacher, hit back, and recover ob- clearer. A proper test of reinforcement effects
ject. As found in Table 5.6, the shifts were in ac- using sequential data requires that the thematic ef-
cord with these predictions. The chi square analy- fects found in ongoing behavior be parceled out.
sis showed a significant effect for six of the nine This requires that one identify an A; and R; fol-
children. lowed by a C + that is suspected to be a reinforcer.
The analysis was clumsy in several important Then, at some later time, when the same agent
respects. First, the behavior change reflected the performs the same Aj, what is the likelihood that it
effects of both rewards and punishment. How- will be followed by the same R;? If the conse-
ever, in that the study was designed to demon- quence functioned as a reinforcer, then the A; >
strate that these two variables controlled aggres- R; connection should be strengthened. There are
sive behavior, that was not a serious problem. To many problems associated with doing such an
provide a fit to the Aj—» R; format, it would have analysis, e.g., what is the appropriate time inter-
required that p (Rj|Aj) be calculated separately for val? Is the stimulus the presence of the agent, the
each of the aggressor’s victims. That simple recal- behavior of the agent, or both?
culation could easily be carried out. The reason Thus far, we have carried out only one pilot
this was not done is that there was a more serious analysis of this problem (Patterson & Cobb,
confound. In fact, it is of such magnitude as to call 1971). Those data were from the first 24 families
into question any conclusion about whether these referred to OSLC. We were searching for an Aj >
data reflected reinforcement effects at all. R; connection at t; that was followed during the
Thematic Effects. The problem lies in parceling same five minutes by another presentation of the

103
same Aj. Given that constraint, then the question founds. There could be variables embedded in so-
of how to control for thematic effect becomes very cial interaction, which could contribute to a rein-
difficult. How long must one wait to control for forcement-like effect, but have nothing to do per
“chunking?” We do not know the answer to that se with the phenomenon. Ultimately the test of
question. It is, however, the kind of question for causal connections must include an experimental
which G. Sackett’s sequential lag analyses seem manipulation.
ideally suited to provide an answer (Sackett, Experimental Manipulations. Motivational the-
1977). Patterson and Cobb arbitrarily imposed an ories served as the primary focus for much of the
interruption period consisting of at least three six- early work on aggression, e.g., the frustration-
second units following interchanges in which a aggression hypothesis by Dollard, Doob, Miller,
former attacker who had hit a victim interacted Mowrer and Sears (1939). The emphasis on moti-
with some family member other than the former vational variables was so overwhelming that only
victim. a few studies were carried out testing the effect of
There were 69 episodes in which the same agent positive reinforcers on aggressive behavior. For
reappeared, but emitted a behavior other than the this reason, the classic study by Miller (1948) was
A, associated with the reinforced hit. The p(Hit,| of particular importance. He showed that with-
Same Agent) at t, was .061. Similarly, if the same drawal of shock could serve as a negative rein-
A, behavior reappeared, but was performed by a forcer for aggression, and demonstrated the po-
different agent, the p(Hit,|Same Behavior) at t, tential importance of reinforcement as a determi-
was .064. One could think of each of the p values nant. It took more than a decade to return to this
as a kind of base rate. line of thought. Then, the study by Reynolds, Ca-
In addition, those episodes were analyzed in tania, and Skinner (1963) showed that pigeon at-
which the former victim who had been hit reap- tacks could be shaped by food as a reinforcer. Ul-
peared or the same behavior appeared, but with a rich, Johnston, Richardson, and Wolff (1963,
new agent. The likelihood for Hit was .002. The cited by Ulrich et al., 1973) used water to shape
data showed only ten instances in which the same fighting behavior in water-deprived rats.
agent and agent behavior recurred in the same In the early 1960’s, a generation of young chil-
five-minute segment; the p (R2|A2) was .200. dren participated in studies in which they were re-
These meager findings suggest that agent pres- inforced and/or punished for striking rubber
ence and agent behavior may combine in a unique Bobo dolls. These studies (reviewed in Patterson
way to form the A,;. The data available suggest & Cobb, 1971) established the fact that social and
that the positive consequences provided by the vic- nonsocial reinforcers increased the rate of hitting
tims served as positive reinforcers for hitting. Be- Bobo dolls. Programmatic studies at Stanford pro-
cause of the limitations in the number of rein- vided a careful analysis of the variables such as
forced events, the analysis can only be thought of sex, age of children, and sex of reinforcing agent.
as illustrative. However, the results provide such a These studies were summarized in Bandura
close fit to expectations that they emphasize the (1973):
potential of this mode of testing for status as a re- There were some methodological problems in-
inforcer. Large amounts of data are required un- herent in these early studies but they did not seem
derlining, in turn, the need to collect the data in overwhelming. For example, an extended baseline
larger time blocks than the five-minute segments study by R. Jones at the University of Oregon
used in the 1971 study. It would also be impera- found that a laboratory measure of the rate of hit-
tive to carry out empirical studies demonstrating ting was an unstable measure. In his study the rate
the relation of time following the episode and the increased in the absence of reinforcement! This led
calculation of p (R2|R1). How long does it take to to our decision to use amplitude as well as fre-
overcome the thematic effect of R; as an elicitor quency as dependent variables. It also led to the
for R2; is it six seconds, 60 seconds, or five min- construction of a high-tech Bobo doll (amplitude
utes? This is crucial because it determines how one and frequency were automatically recorded) who
forms the base-rate values against which the p squatted in automated splendor in his own throne
(R2|A2) is compared. room. The studies which followed demonstrated
One would also be considerably reassured if the satisfactory reliability and validity for the new de-
investigator went on to submit C + events surviv- pendent variables. In the last study in the series,
ing such analyses to experimental manipulation. A positive social reinforcement from adults signifi-
number of such demonstrations would be needed cantly increased the amplitude of the child’s blows
to convince a reasonable critic that the conditional to the Bobo doll. The magnitude of the increase in
p format did not contain some unsuspected con- amplitude correlated .66 with peer sociometric

104
 Table 5.7
Mother as Reinforcer for Child Coercion

Phases:
Baseline, Experimental Baseline,
Day p(attack) p(reinforced) p(attack) p(reinforced) p(attack) p(reinforced)
First oS 50 81 .88 .70 70
Second .00 — 49 58 74 Hil

ratings of aggressiveness in school. A second study son, Littman, & Bricker, 1967), the rates at which
replicated this general relation between laboratory victim-dispensed reinforcers were supplied seemed
conditioning for aggression and school behavior. astonishingly high.
Similar generalizability from laboratory measures It took another ten years to learn how to design
had been demonstrated by Walters and Brown experimental tests for the hypothesis that victims
(1963) and Lovaas (1961b). dispensed the primary reinforcers for children’s
At that time, it seemed we had some leverage on aggression. The studies that have been designed to
the problem. Aggression-like behaviors could be test for the effect of negative reinforcement will be
shaped in the laboratory. Antisocial children were summarized in Chapter 7. The general format for
generally more “conditionable” than were nonag- all these studies has been a single-subject ABA re-
gressive children. We began thinking about how versal design. All of the sessions take place in the
this rather consistent set of laboratory findings home; the reinforcing agents have been the
might apply to children’s aggression in the real mothers. Training mothers to use reinforcement
world. The general idea was that other persons has proven to be a valuable prelude to training
(e.g., spectators or parents) might give positive re- them to manage their contumacious preschool
inforcers for successful aggression. In that vein, children. In these studies the mother participates
the parent interview study by Bandura and Wal- in prior role-playing sessions in order to facilitate
ters (1959) reported that parents of aggressive her delivering the right reinforcers at the right
boys supported the child’s aggression toward per- time. During the more recent studies, the mothers
sons outside the family. It seemed reasonable that not only received several pretrial training sessions,
in certain machismo-oriented subcultures the par- but in addition, cuing devices were used during the
ents, as well as peers, might provide explicit ap- experiments. Typically, the design also included a
proval for aggressive behavior directed toward replication of the ABA procedures on a second and
outsiders. Peer-dispensed social reinforcers for ag- third day.
gression seemed like another possibility as a For example, in the case of the preschool child,
source for the reinforcement of aggressive behav- Eclipse, the mother was observed in the home to
ior. provide positive consequences for coercive behav-
At about this time, a university colleague (K. iors. A younger sibling also provided a rich sched-
Polk) called the author and offered to provide a ule. As can be seen in Table 5.7, on the first day
small amount of funding to support a project re- the mother reinforced (Talk, Attend) half of the
lated to antisocial children. The actual amount target child’s coercive interactions with her. At
was too small to finance an experiment (which at that time about 25% of the child’s interactions
that time was our preferred approach). However, with her were coercive. During the experimental
it was just enough to pay two observers to go to phase the mother was cued to respond positively
one nursery school for two or three weeks. The re- when a coercive behavior occurred. On the first
sults from that pilot study revealed to us what, in day she was able to reinforce 88%, and on the sec-
retrospect, was probably intuitively obvious to ond day 58%. The likelihood of the child’s attacks
everyone but a Ph.D. psychologist. Positive conse- (given interaction with her) increased on both
quences were seldom delivered by onlookers, days. Note that during the second baseline, even
whether peers or adults. The reinforcers were sup- though the mother was no longer cued, she con-
plied by the victim! As shown by the results from tinued to reinforce the child’s coercive attacks.
the NIMH supported study that followed (Patter- The case for parent and/or sibling attention as a

105
reinforcer for some coercive behaviors is not prov- Most of the tests for victim pain as a variable
en by a single case study. The case for sibling at- have been made in laboratory analogue situations
tention and/or compliance as a reinforcer: for co- in which the Buss type shock machine was em-
ercion has not been tested. The study of Eclipse ployed. The study by Hartmann (1969) was a pro-
and his mother is an example of how such studies totype for much of the later work. Antisocial adol-
might be done. The thought is that they should be escents served as subjects in an experiment in
done in the home, and with as little disruption of which they were to deliver shocks to another sub-
the natural setting as possible. ject engaged in learning a task. In this situation
those subjects with longer histories of antisocial
Pain as the Reinforcer behavior tended to be more punitive. Hearing a re-
The majority of the outcomes for coercive child action from the victim that signified pain was cor-
behavior consists of positive behaviors such as related with an increased likelihood of aggression.
Talk or Attend. A study of the impact of these out- That finding is in keeping with Berkowitz’s
comes upon low-amplitude aversive events will (1973a, 1973b) speculations about victim pain re-
presumably contribute to our understanding of actions as a reinforcer. However, it is not clear in
why antisocial children are rejected by peers and such a procedure whether pain reactions served to
family alike. As noted earlier, understanding how elicit attacks or reinforce them.
this process develops may also clarify the antece- In reviewing the extensive literature on this sub-
dents for some high-amplitude familial violence. ject, Rule and Nesdale (1976) cited a number of
However, it is difficult to believe that Talk and At- studies showing that pain cues from a victim were
tend will prove to be the primary reinforcers for associated with a reduction in anger. The study by
high-amplitude attacks. More likely, the rein- Baron (1971a, 1971b) demonstrated that pain
forcer will prove to be compliance and/or submis- cues reduced delivery of shocks by nonangered
sion. In that context, it may also be the case that people, but increased the delivery of shocks by an-
the pain reaction of the victim may function as a gered people. These findings support the general
positive reinforcer for both high- and low-ampli- distinction between angry and nonangry aggres-
tude coercion. Unfortunately, the limitations of sion made by Bandura (1973) and by Hartup
the current OSLC code systems were such that (1974). It seems reasonable to suppose that the re-
measures of pain reactions (other than Cry) were inforcer for angry aggression would be a pain re-
not feasible. The new code (MOSAIC) should action from the victim.
remedy this situation. In the meantime, a discus- Swart and Berkowitz (1976) provided a power-
sion of the positive reinforcers for aggression ful test for the effect of pain cues as a reinforcer.
would seem to be incomplete if it did not take into They reasoned that if pain cues functioned as a re-
account the possibility of pain cues functioning as inforcer, then pairing a neutral stimulus with the
a powerful reinforcer for some aggression in some pain cue should produce a conditional reinforcer.
settings. In subsequent trials, this should function as a rein-
Buss (1961) differentiated outcomes for aggres- forcer. In their study, a light was paired with the
sive behavior into extrinsic reinforcers and those victim’s pain reaction. In later trials with another
reinforcers supplied by the victim. The latter in- victim, the conditioned stimulus was associated
cluded pain reactions by the victim as well as sub- with more intense attacks. In that there was no
mission and compliance reactions. Observers of group controlling for noncontingent employment
children’s play in six cultures coded five percent of of the pain cue or for the conditioned stimulus it-
their aggression as designed to hurt the victim self, one may well ask whether the conditioned
(Lambert, 1974). Toch (1969), in his classic Vio- stimulus served to arouse or reinforce. In either
lent Men, emphasized the fact that both violent case, it seems clear that pain cues were a powerful
prison inmates and violent police officers reported determinant for attacks.
that they enjoyed inflicting pain on their victims. There are a number of studies demonstrating
B. F. Skinner (1969) also emphasized that a reac- that the impact of pain cues on attacks is a joint
tion by the victim that indicated damage had oc- function of several other variables. Baron (1971a,
curred was a likely reinforcer for aggression. Inci- 1971b) and Buss (1966a), cited by Perry & Perry,
dentally, in the same volume he makes a general 1974) found victims’ pain reactions produced an
case for genetic determinants of differential re- inhibition of aggressive attacks (shorter duration
sponsiveness to certain reinforcers. When applied and less intensity) rather than an escalation. John
to aggression, the argument could be generalized Knutson (personal communication) described a
to suggest a genetically determined differential re- study carried out at his laboratory that also failed
sponsiveness to the victim’s pain reaction. to find the relation. Perry and Perry (1974) take

106
 the position that the pain reinforcement effect
_ would occur only if the subject were angry, and Changes in Responsiveness
then perhaps only if the subject were also antiso- Prosocial behaviors such as Talk and Attend are
cial. They tested for these possibilities in a well- considered to be primary reinforcers for a wide
controlled large-scale study. First, a peer socio- range of prosocial and coercive behaviors. How-
metric was used to identify 128 aggressive and ever, from early on, our clinical experience with
nonaggressive elementary-school-aged boys. The families of antisocial children suggested that these
2X2X2 factorial design also controlled for angry children were nei particularly responsive to posi-
and nonangry, and high-and low-pain victim feed- tive reinforcers. The empirical studies reviewed in
back conditions. The main effects were significant Chapter 11 support this assumption. It was also
for both the aggressive and the anger arousal con- the case that following treatment many of these
ditions. The antisocial and the angry (borderline antisocial children seemed to become more re-
significance) conditions were associated with high- sponsive to social reinforcers. When taken togeth-
er scores on the Buss aggression machine proce- er, these findings and our clinical impressions
dure. However, the effects of feedback about vic- point to the fact that at some level, responsiveness
tim pain inhibited the reactions of aggressors and to social reinforcers is not a static trait. There are
nonaggressors alike. It should be kept in mind that not only individual differences, but intraindividual
in this study the children labeled as aggressive shifts over time. This last section is a highly specu-
were normal children in the public schools. They lative account of how this might come about.
were not, as in the case of Hartmann’s (1969) There is an interesting sense in which the per-
sample, institutionalized delinquents. In this spective of the ethologists such as Hinde (1974)
writers opinion, this is a crucial variable. As melds with that of operant psychology (Skinner,
noted earlier for the Radke-Yarrow and Waxler 1969). Both emphasize the function of evolution-
(1979) study, the covariations of measures of ag- ary selection in producing species that are differen-
gression were reversed in sign depending on tially responsive to certain reinforcers. In that con-
whether the children were in the normal or ex- text, it is reasonable to speculate that, among
treme ranges for aggression. other things, humans may be uniquely responsive
Perry and Bussey (1977) repeated this general to reinforcers found in the behavior of other per-
design with a second sample of normal children. sons. For example, even very young infants can be
Again, feedback from the victim inhibited attack conditioned by the contingent presentation of a
behaviors in both aggressive and nonaggressive face. It is also the case that even at that age there is
children. The inhibition effect was, however, sig- a marked variability in responsiveness.
nificant only for the nonaggressive sample. In this The seminal work of J.S. Watson (1979) ex-
study, they added a new component which in- tends this formulation. He introduced a concept
volved the question of the relation between the which may explain why responsiveness to positive
victim’s pain reaction and the self-evaluation of reinforcers varies over time. He proposed that
the attacker. For the nonaggressive sample the cor- even infants directly perceive the contingent rela-
relation between the average pain-cue feedback re- tion between their behavior and its impact on the
ceived and the amount of self-reward at the end of environment. His study of eight-week-old infants
the test was —.65 (p< .001). For the more aggres- demonstrated that they were sensitive to very
sive sample, the comparable correlation was —.10 small changes in contingencies. The behaviors
(ness): controlled by contingent presentations were ac-
As things stand, the notion of victim pain reac- companied by smiling, while noncontingent pre-
tion as a reinforcer for aggression remains plausi- sentations were not. Watson posits a relation be-
ble but unproven. Adequate controls for arousal tween attraction on the one hand and contingent
effects have not been used. The literature would reactions on the other. We are attracted to a part-
suggest that anger, arousal and prior status as (ex- ner because he or she responds contingently (and
tremely) antisocial may also be necessary condi- nonaversively to our behavior. Presumably the
tions. These hypotheses should be tested for fami- same could apply to pinball machines, trout fish-
ly interactions. For example, does victim Cry serve ing and to our affection for our mothers as well.
as a reinforcer, i.e., strengthen the p (R2|Az)?° Is Watson adds one further postulate of funda-
Cry most likely to follow high-amplitude attacks? mental importance. If the event is presented non-
Is Cry a reinforcer that contributes to the escalat- contingently, it does not shape behavior, but non-
ing cycles of high-amplitude interchanges in wife- contingent presentations do alter the responsive-
and child-abusive families? ness of the individual to those presentations made
contingently. To understand responsiveness, two

107
conditional values are required. It is, in fact, their
ratio that determines behavior control. The first is Acknowledgments
the by now familiar p (C +|Rj),and the second is The writer gratefully acknowledges the efforts
some estimate of how likely C + is to occur in situ- of L. Lippsitt and R. Loeber in correcting some of
ations not involving Rj. If the likeliheed is very the misconceptions in earlier drafts of this manu-
high that C + will occur in the absense of R;, then script. I am also grateful to Beverly Fagot for her
p (C+|R;) will have less effect. Watson’s own careful editing of the final draft of the chapter. Her
work is concerned with the problem of estimating efforts resulted in several major changes in the or-
the ratio that gives maximum control. Those is- ganization of the material which did much to en-
sues have yet to be resolved, but the general con- hance its readability.
cept has important implications for many areas of
socialization research.
From this viewpoint, the impact of a contingen-
Footnotes
cy is a relative matter. This, of course, is a general
concept having a long history in experimental psy- 1. It should be noted that the binomial Z has
chology. For example, the Crespi-type studies of found some favor as a statistic for making such
negative contrast effects still find consistent empir- comparisons (Sackett, 1977). However, as noted
ical support (Black, 1968). Similarly, the finding by Gottman (1979), the Z statistic approximates a
from Rotter-type studies emphasize the general normal distribution only when N is larger than 25.
importance of the past history of reinforcement in If the base-rate value is close to zero or one, then
assessing the impact of current contingencies (Bar- even larger values are required. This point consti-
on, 1966). The Watson formulation takes this rel- tutes a major difficulty in that many social behav-
ativity stance, and puts it into the immediate pres- iors of interest occur at a base rate so low as to vi-
ent. In a given setting, it is some ratio of the con- olate the assumptions underlying the use of the Z
tingent to noncontingent characteristic of a rein- statistic.
forcer which determines its impact. This is remi- 2. As discussed by Schoenfeld (1978), this re-
niscent of Estes’ (1971) focus on the informational jection more clearly delineated the differences be-
values of reinforcers as differentiated from their tween the theories of Skinner and Pavlov.
hedonic value. 3. It would be important to have a series of
The position is directly related to the systematic studies to determine just how widespread the phe-
studies by R. Cairns’ students reviewed in Chapter nomenon really is. The writer is aware of two oth-
11. Those studies demonstrated that noncontin- er studies with similar findings: Hall, Lund, and
gent presentations of social reinforcers were fol- Jackson (1968), and Hotchkiss (1966). In their ag-
lowed by reduced responsiveness when the same gregate, these studies sample only a very small
events were later presented contingently. Cairns number of teachers. However, it cannot be true
emphasized that the noncontingent presentations that coercive behaviors work that effectively in all
reduced the informational value of the events. classrooms.
Is it true that parents of antisocial children are 4. Each investigator tends to define categories
about as likely to reinforce one response as anoth- of positive consequences in a slightly different
er? It is our impression that this is the case, but as way. For example, Johnson, Wahl, Martin and Jo-
yet there are no analyses that are appropriate for hansson (1973) included Approve, Attend, Com-
this question. As a hunch, I believe that it is this ply, Indulge, Laugh, Physical Positive, Receive,
noncontingent aspect of the parents’ behavior Talk, Touch, and Nonverbal.
which produces the relatively nonresponsive child. 5. As shown in Patterson (1977b, 1980b), for
I also think that one of the key features of the family interactions, the effect of an aversive conse-
OSLC treatment is training parents to be contin- quence is to increase the likelihood of an immedi-
gent (to reinforce contingently and to punish con- ate continuation of the attack. In the nursery
tingently). The conventional wisdom gives the school data this would have been counted as a
adage “be consistent.” To this we would add be “miss,” i.e., the effect of a negative consequence
contingent. A well-intentioned parent who is often did not lead to a change in victim or attack re-
noncontingently warm and affectionate may cre- sponse, but rather a continuation. Thus, the 1967
ate a situation in which she or he will produce a study underestimated the effect of punishment.
problem. C+ reactions are associated with the termination
of ongoing attacks. As a result, in the 1967 study,
the inclusion of the short-term effects for C + also
led to an underestimate of the true relationship

108
(the analysis was too conservative). The problem large extent on bursts or chains in which one child
is that we do not know if these assumptions apply attacked the other, sometimes in a sequence last-
to school interactions. Therefore, the most sensi- ing for several moments. Thus, some predictions
ble position is to be extremely cautious in accept- were based on the adjacent attack in the chain,
ing the 1967 findings as offering direct support for and some were based on an attack during the next
the hypothesis. day. The latter may reflect reinforcement effects,
6. Two field studies included Cry as a conse- but the former more likely reflect thematic arousal
quence for aggressive behavior. In both studies, and/or suppression effects. Because the 1967
however, the analyses were partially confounded study summed across both kinds of predictions,
at least as they relate to the issue of Cry as a rein- the data are simply not interpretable. At that time
forcer for attacks. In the first study, Patterson, the writer simply did not understand the differ-
Littman, and Bricker (1967) studied nursery ences between accelerators and reinforcers.
school attacks. Cry was one of three outcomes The next analysis by Patterson and Cobb (1971)
grouped a priori as an example of a positive conse- of Cry as a consequence for Hit was based on fam-
quence. The analyses suggested that this class of ily interaction data. The focus for the analysis was
outcomes was associated with an increase in the only on its function as an accelerator or decelera-
probability of the same type of attack being di- tor, and not as a reinforcer. The base rate for a
rected to the same victim. Such an increase was second Hit given the first Hit for family members
demonstrated, but the specific contribution of Cry was .24. If Cry occurred as a consequence for the
was lost in this analysis. Unfortunately, the results first Hit, then the conditional p was .22. It seems
for even the general class of consequences cannot Cry did not function as an accelerator for ongoing
be viewed as demonstrating their status as a rein- hitting. No test was made for Cry as a reinforcing
forcer. The 1967 analysis was based to a rather event.

109
 Chapter 6
Abstract
Effective parental punishment is thought to weaken the connection between a stimulus and the
child’s coercive response. Most parental punishment consists of threats to punish, scolding, and so
on. To be effective, these threats must occasionally be paired with back-up punishment. Parental
threats which are not backed up are categorized as “nattering.” The effect of parental nattering is to
increase the likelihood that the child’s aversive behavior will continue.
This chapter reviews survey studies on the incidence of various kinds of parental punishment. Ob-
servation data are presented comparing normal and deviant samples for the likelihood of various par-
ental punishments for various coercive responses. The samples are also compared for the effective-
ness of parental punishment in suppressing coercive behavior. Generally, the impact is to increase the
likelihood of the immediate recurrence of the deviant behavior. This “facilitation effect” relates dif-
ferentially to context, age, agent, and kind of coercive behavior. Various explanations for this phe-
nomenon are presented: mixed schedules of reward and punishment, timing of the punishment, inad-
equate pairings of CS and UCS, lack of clearly stated rules, and heritability for hyporesponsiveness to
aversive events.

110
 Chapter 6

Punishment for
Aggression

If I were allowed to select only one concept to serving two very different functions. In some con-
use in training parents of antisocial children, | texts aversive consequences indicate parental irri-
would teach them how to punish more effectively. tation. It is a kind of irritable aggression, a reac-
It is the key to understanding familial aggression. tion to the child’s annoying behavior. Sometimes
The study of punishment is very complex; we now these reactions indiciate that more intense punish-
know that we have only begun to understand this ment is forthcoming. They function as condi-
aspect of the coercion process. This is peculiar be- tioned stimuli (CS) for threats or warnings of more
cause it was the study of punishment that led to intensive punishment (UCS). The threats must pe-
the development of much of what we now call riodically be “backed up” or they lose their power
coercion theory. to control behavior. Typically the back-up events
The same aversive consequences that effectively would be loss of privilege, time out from rein-
altered family members’ prosocial behavior were forcement, work, or, for the unskilled parent,
relatively ineffective in suppressing Hit (Patterson physical assault (spanking). If threats and scolding
& Cobb, 1971).! How could an event serve as a are seldom followed through, then in their role as
punishment for one response but not for another? CS they will have only weak control over antiso-
It was this question that eventually led us to the cial aggression.
study of family structure and the irritability cycle For each family there are some aversive conse-
(Chapter 8). This question, however, still stands quences which all members understand to have
as a kind of Gordian knot. After a decade of work, special status. They do not indicate that a more in-
we understand this question much better. We also tensive punishment (UCS) is forthcoming. Their
continue to find new ways of studying it. This function is to make the other person feel bad but
dedicated effort was engendered by a second sur- they do not necessarily mean that he or she must
prise that we encountered very early in our investi- “stop.” Our culture defines the meanings of cer-
gation. We found that when parents tried to pun- tain aversive words, gestures, and facial expres-
ish the coercive behavior of problem children, the sions. Events such as insults or humiliations are
immediate effect was to make things worse (Pat- experienced as aversive for this reason rather than
terson, 1976, 1977a)! Again a surprise, and, because of their prior association with more pow-
again, the question of how can this be? This chap- erful punishment. When these aversives are used
ter addresses these two questions. as consequences, they elicit a counterattack rather
We have come to view parental punishment as than suppressing coercive behaviors. I assume that

isa
in distressed families there are many more catego- up punishments used by most parents. Depriva-
ries of aversive consequences that function in this tion of privileges often results in an extended ver-
way than there are those which signal, “Stop or bal confrontation which is aversive to all parties.
else!” It also requires the parents to give up their own
As a general case, the writer assumes that parents pleasures in order to monitor the child for several
of antisocial children seldom follow through on days. Physical assault, in contrast, is brief and sat-
their threats or scolding (Patterson, 1976). The isfying. After all the effort involved in deprivation
child is aware of this and responds to most aver- of privileges, the child will eventually do it again.
sive consequences with counterattacks. From his To the parent of the antisocial child, this means
standpoint, insults, humiliations, and scolding are the punishment did not really work. They do not
all aversives; few of them signal that punishment understand the concept of “training” which re-
will follow. They are responded to not as signals quires the parents to react consistently during
for him to “stop” but rather as irritants and elicit- many trials using punishment repeatedly.
ors of counterattack. They are left, then, with nattering as their only
My longstanding colleague, John Reid, has means of suppressing unwanted behavior. In a
coined the word “nattering” to describe this pro- sense they punish, or at least are aversive, more
cess (he claims the term was created by Mavis E. often than most parents. They also beat the child
Hetherington). The parent experiences some as- more often; but they are less likely to confront
pect of the child’s behavior as unpleasant and him. In contrast, normal parents are more likely
wishes it to stop. Nattering is an expression of either to ignore an event or to confront it and ef-
parental displeasure. It signifies irritation with no fectively stop the behavior. They are less likely to
intention of following through. We believe that natter and less likely to beat the child. From this
parents of antisocial children employ nattering at viewpoint, we would have to assume that natter-
higher rates than parents of normal children. In ing and beating covary.
this way they avoid the major conflict which
would accompany their efforts to confront the Some Background Considerations
child and try to make him stop what he is doing. When viewed historically within experimental
They do not really try to stop coercive behavior; psychology, the concept of punishment is charac-
they only ineptly meddle in it. The effect of their terized by strange twists and turns. First of all,
nattering is to produce extensions of the behaviors punishment describes an arrangement which in-
which elicit their displeasure. cludes a painful stimulus. However, very little is
In most families the parents use certain cues that understood about the perception of pain. There is,
are well understood. They indicate that the parent as yet, no physiological basis for believing that a
will now back up his or her threats (e.g., they pain center exists in the brain (Barber, 1959). The
glare, or use key words with accompanying voice second peculiar characteristic of punishment lies
inflections). In the heat of the moment, the antiso- in the efficiency with which it controls human be-
cial child often misses these cues and may be badly havior. Arrangements involving positive reinforc-
beaten as a result. One-third of our antisocial chil- ers often require hundreds of trials to produce mi-
dren were abused (Reid, Taplin, & Lorber, 1981). nor changes in behavior. In contrast, there are
By definition, a beaten child means that the par- studies which demonstrate that the use of punish-
ents are unskilled; but it could also mean that the ment arrangements results in the total suppression
child is unskilled. He should track his parents of certain human and animal behaviors in a single
more carefully. I think that even child-abusive par- trial (Church, 1963)! The classic avoidance condi-
ents send these signals, but they are missed by the tioning studies by Solomon and Wynne (1954)
child. At this point, our problem as investigators is showed a failure of extinction for avoidance re-
that we have no ready means of differentiating sponses. After two or three pairings of the CS and
nattering from a punishment confrontation. We UCS, the presentation of the CS elicited up to 650
need some new assessment device for this task. avoidance responses! This “irreversibility” had
Beatings and spankings work. They effectively also been noted in earlier studies by Pavlovian in-
stop annoying child behavior for a short time. For vestigators.
several hours or even days following this, scolding At a cultural level, punishment has always
and threats may have an increased effect. Even played a central role in the control of human be-
though the parents may express extreme guilt havior. The more important the goal, the shorter
about the assault, this one back-up punishment is the amount of time available, then the more likely
all that they have. They have neither the skills nor it is that punishment will play a central role. When
the discipline required to effectively use the back- faced with a national emergency requiring the

Liz
 training of millions of young men for a wide range havior functions as a positive reinforcement for
of new roles, the armed forces resorted to estab- other family members. When this is pointed out to
* lished punishment procedures. The trainees were them, they are not inclined to believe it. The
isolated from the usual distractions and were sub- changes in behavior brought about by reinforce-
jected to massive amounts of information and ment are simply too slow for them to understand
modeling for the new skills. The performance of the relevance of the concept. When they do at-
these skills was maintained largely by punishment. tempt to use positive social reinforcement, they
If one did not follow time schedules, did not dress find it a difficult and embarrassing task to make
in the appropriate uniform, did not follow mili- their reactions contingent on certain child behay-
tary protocol when addressing an officer, or gave a iors and not others. In the majority of cases, ques-
perfunctory performance on the drill field, then tions to parents concerning how to get effective
punishment was almost certain to follow. The performance result in discussions about punish-
punishments employed were often highly innova- ment. However, at the same time, the general
tive and multidimensional, including components question of why a child does what he or she does
of physical pain, fatigue, boredom, shame, and on remains a profound mystery. The most frightening
occasion, were extended to include one’s peer child of all is one who is cunning enough to con-
group as well. There was no question about the re- vince his or her parents that he or she does not re-
lationship between threats and back-up punish- spond to parental punishment.
ments. Most parents rely heavily upon information-giv-
Punishment is an ancient idea that has been well ing to control behavior. In practice, this means
understood and deftly used by most persons re- giving a “lecture”: the assumption is that knowing
sponsible for administrating large social units. It is is doing.For some parents the lecture is a primary
an integral part of our public school system with mechanism for behavior change; nagging, threats,
its focus on producing a few winners. As a by- and punishment serve as occasional back-up con-
product, it also produces an enormous number of tingencies. Studies carried out in the classroom
“losers.”? In the classroom the ratio of critical to showed rule-giving without back-up contingencies
supportive comments is overwhelmingly in favor does not change behavior (Becker, Madsen, Ar-
of the former. The process of training graduate nold, & Thomas, 1967; Greenwood, Hops, Del-
students to “do” a dissertation is so punishing that quardri, & Guild, 1974). This same finding would
90% of Ph.D.’s never repeat the process of doing probably also hold true in the home. It comes
research. down to the fact that most parents must use some
It seems that a heavy reliance on punishment kind of punishment. The question is which pun-
characterizes many families as well. While as a ishments work and which do not?
culture we extol the family as a concept synony-
mous with love, survey studies suggest that this is The Practice of Punishment
not the experience of many. Straus and his co-
workers have found that for many people, mar- The Stark and McEvoy (1970) survey found
riage can be characterized as a “license to hit.” that even middle-class parents found it necessary
They found that in the family, hate is more likely to use physical force in dealing with recalcitrant
than love (Steinmetz & Straus, 1974). As a socie- children. They found that 84% of American par-
ents had spanked their children. The earlier study
ty, we emphasize parental warmth and laissez-
by Sears, Maccoby, and Levin (1957) found a
faire attitudes as crucial in child rearing; but at the
comparable figure of 98%. All of these parents
same time we condone physical assault as a neces-
were presumably advocates of the cultural adage,
sary mechanism for socializing the child (Stark &
spare the rod and spoil the child.
McEvoy, 1970).
Gelles (1972) intensively interviewed a small
Making an aversive event contingent upon an
sample of families identified as violent, together
undesirable behavior is our culture’s most efficient
with a sample from the community at large. In this
means for altering performance. Punishment
study, he found a general attitude toward physi-
works. Its effectiveness is experienced by one and
cal punishment which he characterized as “normal
all. In fact, most members of our culture are un-
violence,” i.e., it is normal and thought to be
aware of any other mechanism for bringing about
necessary to achieve goals that are valued by par-
behavior change. Most parents dismiss the recent
ents and society (italics added by the present au-
emphasis on positive reinforcement as being rela-
tively unimportant. “It’s okay for pigeons and thor):
rats; but it won’t work on people.” Most parents “Parents use the slap, the spanking, and the
are unaware of the fact that some of their own be- strap to teach their children not to do things, to

113
pay attention, and to control behavior. Force The ferula was a whip ingeniously designed to
often is used as a resource when the parent cannot raise blisters as well as inflict pain. The blisters
think of anything else that would be as effective. served as a reminder. Stone describes these early
“Mrs. (56): ‘Well, if I put him on the toilet and practices as follows:
he won’t do it and then I leave him there for an
“This disciplinary practice can be followed in
hour and then I take him off, and then ten minutes
great detail in that of the young heir and King
later he’s done it in his pants, I mean that upsets
Louis IV of France. . . . The child was first
me. What do Ido? I spank him and let him know
whipped at the age of two, and the punishment
it’s wrong then the time after that for a couple
continued after he became King at the age of nine.
days he’s all right. And then he'll do it again. I
He was whipped on the buttocks with a birch or
think he avenges me. I don’t know what it is. I
switch, administered first by his nurse, the time
think it’s psychological—he’s out to get me.’
being immediately after he woke up. . . . The
“There is one pattern that emerges from the dis-
whippings increased in frequency when he was
cussion of violence used to teach and control.
three . . . as he grew older, his nurse could not
Each parent who employs violence in this manner
control him, and the child was held down by sol-
believes that force is necessary and cannot be
diers while she beat him. At the age of ten he still
avoided. Another important rationale is simply
had nightmares of being whipped. . . . The threats
that striking a child in these situations is not con-
to whip him only stopped at the age of thirteen,
sidered violent: it is normative... .
not long before his marriage.” (Stone, 1977, p.
“Mrs. (75): ‘If he spits his food at me I slap his
169)
leg. No time to learn like the present—if he is old
enough to do it, he’s old enough to learn not to do In our culture, what is shocking is the frequency
it.” Mrs. (75) was speaking about her six-month- with which punishment practices escalate to high-
old baby.” (Gelles, 1972, pp. 65-67) intensity interchanges. For example, Gelles (1972)
found that 20% of the wives and 32% of the hus-
In the Gelles (1972) study, 23% of the families bands in a combined sample actually engaged in
suspected to be violent reported hitting their chil- slapping, hitting (with an open hand), scratching,
dren at least daily. One of the 38 families not
known to be violent also reported punishing their
child at this rate. A surprising number of families Table 6.1
also reported hitting their children at least once a Percent of Mothers and Fathers Who
week or once a month; the figures were 21% and Practiced Violence on their Children
42.5% respectively. The review of studies by
(from Gelles, 1972, p. 57)
Blumberg (1974) showed this phenomenon is not
restricted to the American culture. In one study of
700 average English families, 62% of the babies Father Mother
had been “smacked” by the time they were one Violent Act (N = 78) (N=78)
year old. Spank on bottom 60 92
Physical punishment is a ubiquitous facet of Spank using object 19 28
parenting in our culture. This is, perhaps, not so Slap on body 13 AM
surprising when one evaluates the general attitude Slap in face 5 14
of males in our culture toward violence. The ex- Slam or push into wall 3
tensive survey of 1,374 males by Blumenthal, Punch 3 1
Kohn, Andrews, and Head (1972) reported that Hit with hard object 1 1
violence was viewed by the majority as a useful Choke 0 1
tool to produce changes that were desirable or to
control changes that were undesirable. Fifty per-
Reprinted with permission from The Violent Home, by R.]J.
cent thought that shooting was almost always a
Gelles, ©1972 (revised, 1974) by Sage Publications, Beverly
good way of handling campus disturbances. In Hills/London.
keeping with this attitude, the survey by Stark and
McEvoy (1970) found that about one-fifth of
adults approved of slapping one’s spouse. It or grabbing behavior with each other. Table 6.1
seems, too, that physical punishment of children is summarizes the dolorous findings from the same
a part of our historical heritage. Stone (1977) doc- study regarding the intensity of child punishment
uments the ubiquitous use of the ferula in the employed by parents. One child in seven was
homes and schools of sixteenth century Europe. slapped in the face, one in five was slapped on the

114
 body, and one in four was spanked using an ob- ous kinds of punishment employed by parents for
ject. Mothers were more likely than fathers to be coercive child behaviors. In both the clinical and
’ involved in these high-amplitude attacks. the normal samples, Command and Disapproval
Survey studies are important in that they under- were the most likely and Humiliate the least likely
score the attitudes held by parents toward the to occur. The rho between the rankings from the
physical punishment of children. However, the two samples for parental punishment was .83
parents’ verbal reports may not provide accurate (df=9; p<.01).
estimates for some of these events. The classic There is probably also a consensus as to which
study by Goodenough (1931) provides data ger- coercive child behaviors should be punished and
mane to this point. In that study, mothers re- which should not. The observation study by John-
corded in their daily diaries the punitive measures son and Bolstad (1973) employed a derivative of
that they employed in dealing with their preschool the FICS to collect data in the homes for a sample
children. Before the study began, they had also of 33 families of normal 3- and 4-year-old chil-
filled out a questionnaire asking for information dren. The likelihood of parental punishment was
regarding the punishments that they typically em- calculated for each of 16 coercive child behaviors.
ployed. The daily report data did not correspond The findings are summarized in Table 6.3. As can
with the information they had given on the ques-
tionnaire. However, it may not matter too much
that our estimates of incidence are somewhat un-
reliable. It is, perhaps, sufficient to know that Table 6.3
there are a great number of children subjected to Proportion of Negative Consequences
the daily or weekly indignity of being beaten by an
for Coercive Child Behavior:
enraged parent. There are, perhaps, an equal
number of parents feeling both guilty and fright- A Normal Sample
ened by their participation in these rituals. (from Johnson & Bolstad, 1973, p. 53)
The writer believes that, in our culture, there
are shared norms about the kinds of parental pun-
Likelihood of
ishment that should be practiced most and least;
Child Behavior Parental Punishment
e.g., it is acceptable to scold frequently, if neces-
sary, but hitting should be used sparingly. This as- CommandNegative sows
sumption is based on the fact that when families PhysicalNegative 293
are observed in their homes, their punishment AversiveCommand 43
practices are surprisingly consistent across sam- SmartTalk oo)
ples. Table 6.2 summarizes the likelihoods of vari- Tease 38
Negativism 38
Command 36
Destructive 3)
Table 6.2
Disapproval 24
Likelihood of Various Kinds of Parental Yell “2p
Punishment for Coercive Child Behaviors Ignore vA
Noncompliance 18
Whine 13
Consequence Distressed Nondistressed Cry 10
Command 30 .38 DemandAttention .08
CommandNegative 14 10 High Rate .06
Disapproval 5S) fehl
Humiliate 01 01
Ignore 05 05S be seen there, if a child engaged in Smart Talk,
Noncomply 0S .02
there was a likelihood of .39 that the parents
Negativism .02 .03
would punish. If the child Hit a family member,
PhysicalNegative .06 0S then the likelihood was .53. These mores about
Tease 00 .03 what is to be punished probably also take into ac-
Yell .02 .02
count the sex and age of the child. For example,
100% 100% Goodenough (1931) found a decrease in mothers’
use of physical punishment for older children.

115.
 Bronfenbrenner (1958) noted the-trend for the was only temporary (italics added by the present
middle class, and later the working class, to use author):
more permissive or laissez-faire methods of child
“Tt is true that there is a temporary suppression
rearing. The later review of survey studies by Er-
of responses, but all responses originally in the re-
langer (1974) suggested cultural changes in this
serve eventually emerge without further positive
century in the attitude toward parental punish-
reinforcement. Such an effect is, by definition,
ment. He suggests a curvilinear relation in which
emotional. It is an effect upon the relation be-
less parental punishment occurred both early and
tween the reserve and the rate, not upon the re-
late in this century.
serve itself. In this experiment there is no evidence
Formulations about Punishment whatsoever for a process of negative conditioning
directly the opposite of positive conditioning. The
More than six decades of experimental psychol-
behavior of the rat, on the other hand, is quite in
ogy have produced findings from laboratory
accord with the assumption that the slap estab-
studies which make the cultural predilection for
lishes an emotional state of such a sort that any be-
the use of punishment more understandable
havior associated with feeding is temporarily sup-
(Campbell & Church, 1969). I assume that pun-
pressed and that eventually the lever itself and in-
ishment is frequently employed in child training
cipient movements of pressing the lever become
because it works so well. There have been a num-
conditioned stimuli capable of evoking the same
ber of attempts to explain why punishment is so
state.” (Skinner, 1978, p. 107)
efficient as a behavior control mechanism. The
earlier reviews by Church (1963), Solomon and Skinner further developed the implication of
Wynne (1954), and Solomon (1964) support this this position in his Walden Two (1948). Here he
position. They also capture the variety of perspec- emphasized the long-term effects of positive rein-
tives taken in viewing some of the paradoxical ef- forcement and the high cost of employing punish-
fects associated with this phenomenon. Some the- ment. Punishment was presented as an impractical
orists emphasize the role of emotional states, such means for producing behavior change. In addi-
as fear, which are elicited by the aversive event. tion, it was claimed that punishment produced
This state can also be elicited by previously neutral side effects, such as neurosis, fear, and shame.
stimuli which have been contiguously associated While Solomon (1964) and Church (1963)
with the aversive event itself. From this stance, the pointed out at that time that the research findings
connection between a stimulus and the escape/ were inconclusive, their statements were ignored.
avoidance behaviors which it elicits is reinforced Skinner’s position coalesced into what Solomon
by the reduction of the elicited fear. Other theo- labeled The Legend, that punishment is ineffective
rists, such as Guthrie (1978), denied the relevance (italics added by the present author):
of elicited emotional states and focused instead
“Later on, avoidance-training experiments in
upon the competing skeletal responses elicited by
the 1940’s and 1950’s added impressive data on
the aversive event or by its conditioned stimulus.
the long-lasting behavioral control exerted by
In effect, punishment suppresses ongoing behavior
noxious stimuli (Solomon & Brush, 1956). In
because it elicits incompatible reactions (i.e., the
spite of this empirical development, many writers
competing response changes the situation).
of books in the field of learning now devote but a
These and related theoretical issues have yet to
few lines to the problems of punishment. . . . Most
be resolved. There was, however, a second set of
contemporary introductory psychology texts de-
issues raised in the earlier reviews by Church
vote but a paragraph or two to punishment as a
(1963) and Solomon (1964) that has been partially
scientific problem. . . . Perhaps one reason for the
clarified. Estes (1944) and Thorndike (1932), as
usual textbook relegation of the topic of punish-
cited by Rachlin and Herrnstein (1969), took the
ment to the fringe of experimental psychology is
position that punishment did not really alter be-
the widespread belief that punishment is unimpor-
havior but only temporarily suppressed it. This
tant because it does not really weaken habits; that
would mean that punishment would not change
it pragmatically-is a poor controller of behavior;
the probability for R, when, on future occasions,
that it is extremely cruel and unnecessary; and that
the same antecedent (Aj) was presented (i.e., p
it is a technique leading to neurosis and worse.”
(R,|Aj) does not change. From that position, pun-
(Solomon, 1964, pp. 248-249)
ishment temporarily suppressed R, by increasing
the likelihood of R2. It did not, however, alter the The studies by Church (1969) and by Rachlin
basic connection between Aj and Rj. Skinner also and Herrnstein (1969) address several aspects of
adopted the position that the effect of punishment this issue. Both showed that the effect of contin-

116
 Figure 6.1
Short-Term and Long-Term Effects of Punishment
Short-term effects Long-term effects
Antecedentevent (Aj)—»R,—* C, - Increase p(R2) Increase p(A;—» other than R,)
(Escape) (Avoidance)
R2 Decrease p(A;—» R,)

gent shock was significantly greater than noncon- The other effect is a more long-range reduction
tingent shock. This would suggest that the sup- in the connection between the S and R. This long-
pression of ongoing behavior could not be ac- range effect was what Rachlin and Herrnstein
counted for by mere distractions due to emotional (1969) referred to as the negative law of effect. In
states or the disrupting effect of competing re- the context of social interactions, punishment is
sponses elicited by the aversive stimulus. Studies assumed to alter the connection between a social
from both laboratories showed that the effect of behavior and its antecedent. In Figure 6.1, the
contingent aversive stimuli was more than could phrase long-term effect refers to the p value which
be accounted for by a pseudo-instrumental condi- describes the connection between the Aj and Rj at
tioning (noncontingent). Rachlin and Herrnstein some future time. At the time when A; is presented
(1969) go on to argue for Thorndike’s original for- again, the impact of the punishment will be re-
mulation. They present a case for the idea that flected in the reduced p value describing the likeli-
positive reinforcement and punishment lead to op- hood of R;. The long-term effect of punishment is
posite effects on behavior. The former strengthens a reduction in the connection between the re-
the S-R connection, and the latter weakens it. sponse and its controlling stimulus.
This writer would suggest that pending a wide As noted by Church (1969) and others, one
acceptance for a negative law of effect, it is a con- does not have to assume that punishment wipes
venient metaphor for examining several aspects of out a connection to believe that the connection can
social interactions. Figure 6.1 illustrates that pun- be weakened by punishment. Extinction (nonrein-
ishment of one family member by another can be forcement) also weakens these connections. Per-
thought of as having two simultaneous effects. haps only extinction reduces these p values to
There is the immediate effect in that the aversive zero. The net effect is that these p values are con-
event suppresses the response upon which it was stantly shifting as a result of punishment, rein-
made contingent. In social interaction terms, this forcement, and extinction contingencies. These p
means there is a reduced likelihood that the re- values define a dynamic interactional structure.
sponse will continue as the next event in the se-
quence. Under these circumstances, some re- Praise and Disapproval in
sponse other than R; is likely to occur. As noted in Discrimination Training
the previous chapters, it is assumed that aversive One of the lines of research which undermined
events often elicit autonomic reactions; but in the what Solomon called The Legend of punishment
present context, this state does not necessarily ineffectiveness were the studies on discrimination
play a part. The reduction in p value for the imme- learning. In these studies, the individual was re-
diate recurrence of the punished response does not quired to learn to respond in one way to Stimulus
require arousal as an explanation. The child has X and in some other way to Stimulus Y. The
previously learned a variety of alternative re- studies of particular interest are those in which the
sponses which he can make to a stimulus. In Fig- individual was rewarded for the correct response
ure 6.1 these alternatives are symbolized by R2. and punished for the incorrect response. The
They represent coping responses of a kind. In the study by Warden and Aylesworth (1927) em-
laboratory studies they are the escape responses. ployed such a design and found that rewarded rats
The immediate effect of punishment is to increase learned twice as fast as did a punishment-only
the likelihood of one of these alternatives. Their group. However, reward for a correct response
occurrence marks a deflection of an ongoing be- and punishment for incorrect responses produced
havior. Of course, this could also be called re- learning to criteria nine times as quickly as a re-
sponse suppression. ward-only condition. Meyer and Seidman (1961)

117
obtained similar results using two age groups of
children as subjects. Both preschool- and elemen- Some Parametric Studies
tary-school-aged children were more responsive to of Punishment
“wrong” than to “right.” However, one suspects Laboratory studies have shown that the effec-
that the outcome of such comparisons must de- tiveness of a CS for controlling behavior was de-
pend upon the kind of reward or the kind of pun- termined by: (a) the frequency of the CS-UCS pair-
ishment employed. For example, Terrell and Ken- ings, (b) the intensity of the UCS, and (c) the laten-
nedy (1957) found learning to be equal under con- cy between the CS and the UCS (Church, 1963;
ditions for praise and disapproval; but candy was Solomon, 1964). From this perspective, the moth-
more effective than either of the social conse- er who threatens (CS) but seldom follows through
quences. (UCS) will be less effective in using her threats and
There are two studies which compare the effects warnings to control deviant child behavior. This
of social consequences provided by parents in a arrangement produces the parent who natters.
discrimination-like procedure. A mobile labora- They express their annoyance by threats and
tory was taken to the home and provided a means scolding, which serve only to elicit counterattacks.
for making parent reinforcement contingent upon It may be that these natterers simply do not know
the child’s making a correct discrimination. In one what to use as a nonviolent back-up (USC). They
study, parents were cued to use a variety of posi- periodically respond with physical assault; but the
tive social reinforcers, resulting in an average of a favorite back-up for parents of antisocial children
7% shift in preference (Patterson, Littman, & seems to be “grounding the child,” e.g., for a
Hinsey, 1964). In a second study using the same month. When this occurs, the parent and the child
apparatus, a second sample of parents were cued know that the parent does not really mean it.
to use such terms as “no,” and “wrong.” During Within two days of being grounded (staying at
reinforcement the mean shift in preference from home) the child can slip away; and when he or she
the baseline condition was .155; the punishment does, the parents simply natter. The infrequent as-
condition was about twice as effective as the posi- saults and groundings produce some brief respite,
tive reinforcement. These findings were supported but things quickly return to “normal.” As a result,
by Casey (1967), who used the same apparatus the parents report that, in the long run, nothing
but male adults served as reinforcing agents. One works.
group of children was reinforced and the other
Whether they confront their children or natter
was punished. Punishment changed children’s
at them, the parents of out-of-control children
preferences significantly more than did positive re-
inforcement. Kelly and Stephens (1964) found a also seem to wait too long before intervening. Co-
ercive interchanges often occur over long periods
comparable effect, using the marble-drop proce-
of time. In that context, a parent saying “Don’t do
dure with preschool boys. It should be noted in
that” means very little. If, instead, the parents had
passing that punishment was not always found to
intervened when the conflict first began, they
be more effective. Brackbill ard O’Hara (1958)
might have had some chance of success.
found that withdrawal of a reinforcer (candy) was
less effective than giving the candy contingent on The treatment procedure for parents of antiso-
the correct response. cial children involves, among other things, teach-
These findings suggest that one reason our cul- ing the parents to consistently pair their stop-com-
ture condones the use of parental punishment is mands with a five-minute time out (TO) (Patter-
that it tends to be a very effective means for chang- son, Reid, Jones, & Conger, 1975). Toobert’s
ing behavior. If the parent is maximizing short- (1980) extensive review of the research literature
term gains, a technique producing an immediate on time out showed that a simple statement of the
impact is likely to be employed. In a situation in- house rule, followed by four minutes of time out
volving a contumacious child and an impatient was effective for most (nonautistic) children ages 3
parent, punishment is likely to occur. If that pun- through 12 years. As she points out, this nonphys-
ishment suppresses the noxious child behavior, ical form of punishment is replete with complexi-
then the parent is likely to use it again. Negative ties such that parents of problem children should
reinforcement strengthens the connection between receive instruction and supervision. For older chil-
child noxious behavior and parental punitiveness. dren, work chores and loss of privileges serve an
Just as there are reinforcers supplied by the victims analogous function as a back-up UCS. In either
for attack behaviors, there are also reinforcers case, the parent is given back-up consequences
which strengthen the use of punishment by par- which substitute for the role served by physical vi-
ents. olence and grounding as UCS.

118
 Figure 6.2
Mean Rates of Deviant Behavior for the Sequence 1, 15, and 30 minutes of Time Out
(from White, Nielsen, & Johnson, 1972, p. 116)

.500

1.400
ae
ae)
=]
i“)
I~]
cP)
rv
is]
5
‘Bie 1400
to
uv
a
i?2)
I
2
>
=
© .200
fe)
-—
[=|
8
>
uv
v2]
iB)
Ob
S .100
9
<

.000

BL, 1 min. time out BL, 15 min. time out BL; 30 min. time out BL,
- 61.35% — §3.27% —40.29%

Reprinted with permission from “Time out duration and the suppression of deviant behavior in children,” Journal of Applied Behav-
ior Analysis, 5, 111-120. Copyright 1972 by the Society for the Experimental Analysis of Behavior, Inc.

and running away. The ABABABA design per-

Time Out mitted a test for the effect of varying time intervals
Most time out studies have used control groups and for order effects as well. It was, in fact, their
or reversal designs to provide a clear demonstra- analyses of the order effects that provided the
tion of the control of behavior. Some of the studies most intriguing outcome of the study.
have been carried out in natural settings. The The data for Group A are presented in Figure
study by White, Nielsen, and Johnson (1972) is a 6.2. Following an extended baseline, the subjects
good example. Three groups of retarded, institu- received one minute of time out contingent upon
tionalized subjects were randomly assigned to any of the deviant responses listed above. After
three different sequences of time out intervals. In three weeks of this phase, the contingencies were
these sequences, time intervals of one, 15, and 30 removed for a two-week baseline period. Each of
minutes were counterbalanced. The dependent the remaining experimental periods lasted for
variable consisted of a measure of deviant behav- three weeks and were followed by baseline periods
ior such as aggression, self-destruction, tantrums, of two weeks. The numbers at the base of the

119
 Figure 6.3
Mean Rates of Deviant Behavior for the Sequence 30, 1, and 15 minutes of Time Out
(fromWhite,Nielsen,& Johnson,1972,p. 116)

.500

is)
2 — .400
re)
|
i?2)
ie)
vu
Qa,
bw
=}
°
= «300
bo
vu
Qa
is}
a}
>
fas
<=
uv
<8 200
=
ay
>
uv
sS
Vv
iY0)
cs
oe e100
>
<

.000

BL, 30 min. time out BL, 1 min. time out BL, 15 min. time out BL,
—40.51% + 30.94% — 31.25%

Reprinted with permission from “Time out duration and the suppression of deviant behavior in children,” Journal of Applied Behav-
ior Analysis, 5, 111-120. Copyright 1972 by the Society for the Experimental Analysis of Behavior, Inc.

graph refer to the drop in deviancy rate compared outcome for this manipulation. As reported there,
to the preceding baseline. It can be seen there that the first time out interval of 30 minutes produced
the introduction of even a one-minute time out a dramatic reduction in rates of deviant behavior.
punishment was dramatically effective for most of During the next two weeks (Baseline) there was a
the children. In fact, there was no increase in con- gradual increase in deviancy. When a time out pe-
trol when longer intervals were added in later riod of one minute was introduced, it had little im-
manipulations. It should be noted at Baseline,, pact; in fact, the acceleration in deviancy contin-
when time out was no longer in use, the behavior ued at the same rate as before. Introducing a time
again came under reinforcement control and out period of 15 minutes reinstated control.
quickly returned to its former level. The effective use of nonviolent punishment,
It should be noted that if one began with 30 such as TO, has an additional important effect. Its
minutes of time out, and then later attempted to use can have a vicarious modeling effect for the
reduce the interval to one minute, there was an im- other children in that setting. For example, in a
mediate loss of control. Figure 6.3 summarizes the study of classroom intervention, various reinforc-

120
ing and punishing contingencies were employed to in that study to control for the fact that much of
bring the problem children under control (Patter- the sibling aversive behavior was elicited by be-
son, Cobb, & Ray, 1972). Pre- and post-observa- haviors of the target child. In addition, there was
tion data for peers in the classroom indicated that no control for the fact that many of the parents ap-
there were dramatic reductions in deviancy levels plied the same programs directly to the siblings.
in four of the five classrooms. Wilson, Robertson, The reactive and vicarious effects cannot be separ-
Herlong, and Haynes (1979) cite four additional ated from this treatment study; but it seems plausi-
studies which also demonstrated vicarious effects ble to assume that family members were influ-
in untreated peers. In their own study, Wilson et enced by watching punishment being used effec-
al. (1979) collected observation data in a nursery tively on one of the members. Similarly, if a moth-
school setting for an aggressive child and his 13 er loses control of one child, she may very well
peers. The data demonstrated the expected impact lose control of the others.
of time out in reducing the aggressive behavior of A careful analysis of the treatment procedures
the target child. During Baseline;, on the average, for institutionalized, chronic schizophrenic adults
the peers were observed to be aggressive in about revealed that assaultive behavior was one of the
4% of the intervals. When the target child received most difficult episodes to handle for all treatment
time out, the average for peers fell to 1.27%, then strategies employed (Paul & Lentz, 1978). The
rose to 3.17% during Baseline,. The mean for only effective treatment for this behavior was time
peers during the second time out phase was out. It effectively controlled the aggressive behav-
1.97%. The findings suggest that, in some re- ior and helped make it possible to eventually re-
spects, the peers were being affected by the manip- turn these people to the community.
ulations designed for the target child. It was hy-
Parental Warmth
pothesized that the changes in peer aggression
might reflect the fact that the target child initiated The traditional child development literature has
fewer attacks during the experimental phase than strongly emphasized the nature of the relationship
in the baseline phase. A second hypothesis con- between parent and child as a primary determi-
cerned differential changes in the teachers’ reac- nant for the effectiveness of parental punishment
tions to peer aggression during the three phases. (Bandura & Walters, 1963; Sears, Maccoby, &
However, their analyses of attacks by the target Levin, 1957). Control theorists, such as Hirschi
child and teacher reactions offered no support for (1969), represent a modern variation of this posi-
either assumption. The authors concluded that the tion; it is suggested that the attachment or bond of
use of time out in the classroom can have a vicari- the child to the parent determines the impact of
ous impact on the peer group; when it is being parental efforts to control the child.
used, peers can see that aggression is likely to be The influential paper by Sears, Whiting, Now-
punished. This conclusion is in keeping with the lis, and Sears (1953) provided one of the first data
studies by Bandura (1973) which indicated that bases for these speculations. Their interview data
children’s aggressive behavior was significantly al- suggested the nurturant parent was more effective
tered by vicarious reinforcement and punishment. in using punishment. Presumably, punishment by
One of the tenets of coercion theory is that it is such a parent would be associated implicitly or ex-
the adult who determines how much aggression plicitly with the loss of love. In his review of the
will occur in a setting. The performance level is re- punishment literature, Parke (1970) notes that this
lated to the willingness of the adult to use an effec- position assumes that parental punishment has
tive punishment. For example, in the home, if sib- two components: (a) the aversive event itself (e.g.,
lings see that one of their peers “gets away with a spanking, or threat of loss of privileges), and (b)
it,” this serves as further incentive for their own an aversive state analogous to anxiety relating to a
coercive behavior. Similarly, if they observe that a possible loss of love.
fellow sibling is now being punished with alarm-
ing consistency and effectiveness, it may lead to an “In fact, a certain degree of positive interaction
inhibition of their own deviant behavior. Indirect and affection between a parent and a child is nec-
support for this hypothesis is to be found in the essary if social punishment is to be an effective
correlation of .74 (df=26; p<.001) between means of producing response inhibition. This ar-
baseline TAB scores for siblings and problem chil- gument rests on the assumption that withdrawal
dren (Arnold, Levine, & Patterson, 1975). The of affection is an effective component of all forms
pre- and post-treatment comparison also showed of social punishment. . . . On this basis, it is as-
significant reductions in sibling coerciveness fol- sumed that a nurturant punishing agent arouses a
lowing treatment. However, no attempt was made greater degree of anxiety than a neutral agent and

121
that consequently the former agent is more effec- tions, the room was stocked with attractive toys,
tive in producing response inhibition in the and the adult provided encouragement, support,
child.” (Parke, 1970, p. 88) ; and approval. The other group had two sessions
Casey (1967) tested the effect of a brief period with a relatively distant adult and unattractive
of warm versus distant interaction in the control toys. The children experiencing the positive inter-
of adult-dispensed punishment. Neither condition actions later showed significantly greater respon-
had a significant effect on the children’s respon- siveness in a punishment situation. This general
siveness to punishment or reward. The primary effect was then replicated in the study by Parke
data in support of this position came from the (1970). He used a 2x2x2x2 factorial design con-
large-scale survey by Sears, Maccoby, and Levin trolling for timing and intensity of the punishment
(1957). Mothers who were rated as warm and and levels of nurturance and cognitive structure.
who claimed to use frequent physical punishment The nurturance manipulation was similar to that
said that spankings were effective. Those mothers for the Parke and Walters (1967) study. There
described as cold or hostile, who also used fre- were two levels of cognitive structuring. At one
quent physical punishment, said that spankings phase of the experiment, children in all groups
were not effective. Schuck (1974) carried out a were told that they should not touch or play with
path analysis of the data for both the Sears, Mac- some of the toys; and if they picked an incorrect
coby, & Levin (1957) sample and the replication toy, the buzzer would sound. The low-cognitive-
study by Radke-Yarrow et al. (1968). The structure group heard only the buzzer. Those in
latter sample consisted of highly educated the high-cognitive-structure group received an
and/or professional families. His analysis showed elaborate explanation as to why the restrictions
that for the Sears study, mothers’ warmth did play were placed on their behavior. In addition, each
a significant role in child aggression. The effect time the buzzer sounded, the experimenter said,
was, however, indirect, being mediated by both “No, that one might get broken.” Following this
the physical punishment and punishment of ag- phase, all subjects were left alone with the toys
gression variables. Mothers lacking in warmth that they had been punished for choosing.
tended to use more physical punishment and to The number of deviations served as dependent
punish aggression more often. However, his path variables. The findings summarized in Figure 6.4
analyses for the replication study reported the re- report high-nurturance subjects deviated signifi-
verse relation. Mothers lacking in warmth tended cantly less than did low-nurturance subjects; but
to use less physical punishment and to punish ag- this held only under conditions of low-cognitive
gression less often. However, neither of these path structure; i.e., the condition obtained in the earlier
coefficients for warmth were significant. Inciden- study. Under conditions of high-cognitive struc-
tally, Schuck (1974) found two variables directly ture, there was no significant difference between
relating to child aggression for both of these mid- the nurturance conditions. The reader should
dle-class samples. In both instances, parental per- note, however, that under the high-cognitive-
missiveness caused child aggression. Secondly, structure condition the trend was the reverse of
those parents permitting aggressiveness tended to what might be suggested by the warmth-attach-
not punish it as severely. ment theorists: high-nurturance was associated
One of the problems with using parent-report with more deviations.
data to describe both the child and the parent is When viewed from the perspective of a tradi-
that the correlations obtained may reflect the im- tional development theorist, there is a thread of
plicit theory of the parent as much as the true state consistency here. Across the studies there were
of affairs. This potential confound was controlled low-level negative correlations between parental
in the large-scale study by Eron, Walder, Hues- warmth and aggressive child behavior. Low nur-
mann, and Lefkowitz (1974). Ratings from parent turance and high deviations in child behavior were
interviews on parental nurturance for boys in both also found to correlate in two experimental
the third and thirteenth grades correlated —.16 studies. The findings from the correlational
and —.15 with peer nominations for aggressive- studies also showed the expected (low-order) posi-
ness. Parental punishment correlated .18 and .13 tive correlations between the amount of parental
respectively.? punishment and aggressive behavior. The attach-
Parke and Walters (1967) experimentally ment theorist might well claim that these results
manipulated the warmth of the relationship to de- support his or her position: low-nurturance par-
termine its impact on punishment effectiveness. ents were less effective as punishing agents; and
The manipulation involved two 10-minute ses- they, therefore, punished more although their
sions in a free-play situation. In the warmth situa- child continued to be more aggressive. There is,

122. -
 Figure 6.4
The Relationship of Cognitive Structure and Agent-Nurturance
to Punishment Effectiveness
(from Parke, 1970, p. 96)

Cognitive structure
Agent-nurturance relationship
16

14 Low nurturance

iz)
=|
2 12
a}
>
~ 10
hens
fo)
5 8 High nurturance
2
E
S 6
l=}
g 4
=

High Low
cognitive structure cognitive structure

Reprinted with permission from “The role of punishment in the socialization process.” In R.A. Hoppe et al., editors, Early Experi-
ences and the Process of Socialization. Copyright 1970 by Academic Press, Inc.

however, an equally plausible alternative explana-

tion which fits these same data. The parents of an-
Mixed Schedules, Timing, Rules,
tisocial children tended to be less attached to their and Contingencies
children; they were also less skilled in family man- All parents effectively punish some child behav-
agement (see Chapter 10). They tended to natter iors (e.g., the toddler ambling toward the busy
rather than to stop the deviant child behavior. The street). Even parents of antisocial children do not
effect of nattering was to increase the performance permit all forms of out-of-control behavior. Most
of coercive child behavior. Less skillful parents of them are quite selective. They effectively punish
punished more often but with less effect. It is not some deviant behaviors and stop their occurence;
that one thing causes the other; it is, rather, that while for other kinds of deviant behavior they sim-
the effect is interactive. The traditional finding of ply express displeasure by nattering. Selectivity
antisocial children receiving more punishment is implies that the parent must cue the child as to
caused by this interplay of effects. These same un- when he or she means stop and when he or she
skilled parents tend to be less attached to their simply means, “I don’t like that.” These cues may
children. It is not a lack of warmth, per se, lower- be idiosyncratic to each parent. For some it is a
ing the impact of parental punishment; it is that glare, for others it is a facial grimace or a rising
unskilled and unattached parents do not back up voice inflection. Our current inability to differenti-
their threats. ate one cue from the other represents a major im-

123
 Figure 6.5
Mean Number of Punches in the Post-Training Period as Related to the Type of Training
(fromDeur& Parke,1970,p. 407)

60

Ces) Continuous
punishment
50 Peinetion

i?2)
=
= 40
=
(=)
a.
>S 30
to
vo
=
apn a)
Z

10

100% Reward 50% Reward 50% Reward

50% Nonreward 50% Punishment

Training conditions

Reprinted with permission from “Effects of inconsistent punishment on aggression in children,” Developmental Psychology, 2,
403-411. Copyright 1970 by the American Psychological Association.

passe in the development of a coercion theory. as a cue for reward, or as an arouser.” (Solomon,
Parents use mixed schedules of reinforcement 1964, p. 241)
and punishment. The actual schedule varies from
In his review, Martin (1963) pointed out that as
one coercive response to another, reflecting paren-
a general case, mixed schedules of reward and
tal sensitivity to certain types of coercive child be-
punishment produced greater resistance to extinc-
haviors. The formulation by Solomon (1964) sug-
tion. Deur and Parke (1970) tested this hypothesis
gests that behaviors maintained by mixed sched-
for its relevance to aggression using children as
ules might be particularly resistant to punishment
subjects. Subjects were assigned to three. training
control. Ordinarily, the resistance of a response to
experiences: continuous reinforcement (marbles),
extinction can be decreased if punishment is intro-
partial reward and nonreward, partial reward and
duced. However, there is an arrangement in which
punishment (loud noise). After they received train-
the effect of punishment may be reversed.
ing, the groups were then further divided into
“If the subject is habituated to receiving shock either punishment or extinction conditions. The
together with positive reinforcement during re- results are summarized in Figure 6.5. While pun-
ward training, the relationship can be reversed, ishment slowed down the acquisition of aggres-
and punishment during extinction can actually in- sion, it was associated with significantly greater
crease resistance to extinction (Holz & Azrin, resistance to extinction. The mixed schedule con-
1961). Evidently, punishment, so employed, can dition was also the most resistant to continuous
functionally operate as a secondary reinforcer, or punishment.

124
 Figure 6.6
The Relationship of Punishment Intensity and Timing
(from Parke, 1970, p. 93)

deviations
of
number
Mean Late punishment

PRI
tara Site arly punishment
Treat

High Low
intensity intensity

Reprinted with permission from “The role of punishment in the socialization process.” In R.A. Hoppe et al., editors, Early Experi-
ences and the Process of Socialization. Copyright 1970 by Academic Press, Inc.

Ross Parke has intensively analyzed the parame- child’s cognitions about the punishment. He as-
ters relating to punishment effectiveness for chil- sumed, along with Aronfreed, that giving the child
dren (Parke, 1970; Parke & Levy, 1972; Parke, a rule would enhance the inhibiting effect of pun-
Deur, & Sawin, 1970; Parke & Walters, 1967). ishment. He first tested the possibility that restat-
His work on timing is most germane to the ques- ing the rule would overcome some of the liabilities
tion of parental effectiveness. Parke (1970) as- thought to accrue to delays in punishment; e.g., “I
sumes that punishment for events that occur ear- told you at the store that if you whined there you
lier in a sequence more effectively suppresses be- would have to go to time out when we got home.
havior than punishment for those events that oc- Go to time out now.” The study cited by Parke
cur later in a sequence of deviant acts. He cites and Walters (1967) directly addressed this issue.
several studies in support of this hypothesis. Fig- In that study, punishment following either a video-
ure 6.6 summarizes the outcome for one of his tape replay of the deviation or a verbal restate-
own studies in which he manipulated the intensity ment of the rule was as effective as punishing the
of the punishment (90 db versus 65 db tone) and original deviation itself. This effect was replicated
early versus late timing. The response being pun- in a single-subject design for a 10-year-old child
ished was the child’s deviations from the adult’s re- (Verna, 1977). The rule statement at the time of
quest to not touch certain toys. As shown there, delayed punishment (four hours) was equally as
late punishment became particularly ineffective effective as immediate punishment. Parke (1970)
under conditions of low intensity. While this inter- went on to demonstrate that rule statements inter-
action of timing and intensity is not always found, acted significantly with almost all of the other par-
the significantly greater impact of high-intensity ameters he manipulated (nurturance, timing, and
punishment seems to be a consistent finding. intensity).
In a series of studies, Parke also manipulated the Parke’s (1975) update of his earlier reviews fur-

125
 Table 6.4
Aversive Reactions by Family Members to Target Child’s Coercive Behavior

Mother Father Male Sibling Female Sibling Grand

Sample Mean N_ SD Mean N Mean N_ SD Mean N_ SD Means

Normal 18 36uneee 18 25 34 236 OF 27 732 25

Social
Aggressor 290 «F490 LS cole 24 FAV 2323 384 23ie <8 34

Abused LI RE RNS: 33 13 ath Oi 34) 15° 28 32

“There were fewer than 10 subjects.

ther substantiated the significant contribution of nection. Presumably the more intense the aversive
intensity and timing; and again, rule statements event, the more it would weaken the connection.
were shown to increase the effectiveness of punish- The studies reviewed in Chapter 4 showed that
ment. In fact, conditions of rule statements were parents of normal children and parents of antiso-
so powerful that they overcame the usual effects of cial children agreed in their perceptions about
both timing (early and late equally effective) and what is and what is not aversive about children’s
intensity (high and low equal). Peer endorsements social behavior. The next question is whether
further enhanced the impact of rule statements. If there is a correlation between what the parents
the child was allowed to choose the means of rule perceive as aversive and how they react in the
enforcement, there was significantly greater resis- home. The Johnson and Bolstad (1973) study
tance to deviation from the rules. found a correlation of .73 between parent ratings
These studies are of particular interest because of deviancy for code categories and the likelihood
they relate directly to our clinical experience in that the parents would be observed to react aver-
working with families of antisocial children. In sively to these behaviors. Also, the more deviant
these homes there are not only fewer house rules the child behavior was perceived to be, the more
but those that do exist are not clearly stated. For likely the parent was to react in a punitive manner.
example, ambiguities as to who should clear the Do family members differ in the likelihood of
table after supper may result in endless bickering. providing aversive consequences for deviant child
Because the rule about when to return in the eve- behavior? Do these reactions vary as a function of
ning is not clearly stated the child can talk his way membership in a clinical or nonclinical sample?
out of being punished. Parental commands are Data relating to these questions were analyzed
given more often in these families, but they also from three different samples. The samples con-
tend to be vaguely stated. Many of these com- sisted of 36 Normals, 43 Social Aggressors, and
mands seem to be more an affirmation of parental 19 Abused children. For each subject the reactions
displeasure than a statement of what they want the of parents to coercive child behaviors were com-
child to do. Given these kinds of ambiguities it is bined across events occurring in initial, middle,
perhaps not surprising to find that their punish- and final sections of coercive chains. Combining
ment is less effective. data across agents, the mean likelihood of paren-
tal aversive consequences for deviant behavior was
Field Studies of Parental Punishment
.250 for the normal sample.* The differences in
The first problem in carrying out field studies of likelihood between mothers and fathers or be-
punishment was to identify the aversive events. tween male and female siblings were relatively
Which aspects of social interaction could weaken small. However, the siblings were significantly
A,—® Rj connections in a fashion analogous to more likely than their parents to react aversively to
electric shock or loud buzzer sounds? Although coercive behaviors presented by the problem child.
never clearly stated, our implicit assumption was Johnson and Lobitz (1974a) also found this to be
that the aversive events found in social interaction the case in their study of normal preschoolers and
would have a similar effect. This would mean that their families. Siblings from the Social Aggressor
even the low-intensity aversive events found in so- sample were more likely than parents to respond
cial interaction would weaken the Aj—» Rj con- punitively. The trends were consistent for all com-

126
parisons. While a one-way ANOVA for repeated
measures would be the analysis of choice, there
Table 6.5
were only nine families for which data were avail-
able for all four family agents. For this reason, Parental Reactions to Out-of-Control
agent-by-agent comparisons were made; two of Child Behavior (“Outbursts”)
the six possible comparisons were significant.
(adapted from Goodenough, 1931, pp. 197-201)
Male siblings were more punitive than mothers
(t=2.63; p< .01 two-tailed). Female siblings were
also more punitive than were mothers (t=2.67; Parental Reaction Boys Girls
p<.01 two-tailed). Ridicule 0.5% 1.2%
Family members of socially aggressive children Deprivation privileges 2.3% 0.3%
were generally more likely than their normal coun- Ignore 29.6% 38.9%
terparts to be punitive. The differences between Deprivation food 1.0% 0.5%
fathers from the normal and the clinical samples Put child chair, bed, room 10.7% 3.4%
were of borderline significance (t=1.62; p<.06 Scold 8.8% 4.5%
one-tailed), as were the differences between moth- Threat 6.4% 3.1%
ers (t=2.23; p<.05 one-tailed). These findings Spank, slap 7.7% 3.4%
plus the comparable nonsignificant trends for the Physical force 8.7% 2.4%
siblings were in keeping with the findings re-
ported in the traditional child development litera-
ture (Feshbach, 1970). Most of those studies have
shown that antisocial children receive more pun- (.06). Note that the punishments emphasized by
ishment than Normals (Eron et al., 1971; Sears, social learning approaches (deprivation of privi-
Maccoby, & Levin, 1957). Typically, these find- leges, time out) were used in less than 10% of the
ings were based on parents’ reports of how fre- conflicts involving boys.
quently they punished their children. The findings The data already presented in Table 6.2 sum-
presented in Table 6.4 are in agreement in that marize the comparable data from OSLC samples.
parents of antisocial boys reacted aversively to de- As noted earlier, parental reactions to coercive
viant behavior more frequently than parents of child behavior for distressed and non-distressed
normal boys. The present data add one compo- samples were very similar. Parents in both samples
nent to the traditional findings by demonstrating were most likely to use (CM) Command, e.g.,
that parents of antisocial boys were also more like- “Stop teasing your sister.” They were also very
ly to react aversively given that the child was being likely to use (DI) Disapproval or (CN) Command
coercive. Negative. Presumably, parents in the two samples
The pattern of findings for the Abused-Child differ in that parents in distressed families were
sample was quite different. Here, parents and sib- less likely to follow through on their threats. Un-
lings were equally punitive in their reactions to the fortunately, this most crucial assumption has yet
problem child. In fact, these parents were more to be tested.°
punitive than were parents from any other sample. The cross-cultural observation study by Lam-
Mothers of abused children were significantly bert (1974) showed an interesting shift in punitive
more punitive than mothers of Normals (t= 3.79; reactions to aggression by peers as a function of
p<.001 one-tailed). Also, fathers of abused chil- age. When children aggressed against 3- and 4-
dren were significantly more likely than fathers of year-olds, they were punished by their peers only
normal children to be punitive (t=2.22; p<.05 36% of the time. By age four, the likelihood of be-
one-tailed). The differences between parents of ing punished for aggression exceeded the likeli-
abused and socially aggressive children were not hood of a positive outcome. Aggression directed
significant. toward adolescents was punished 67% of the
Goodenough’s (1931) analysis of parental dia- time.
ries showed that mothers of preschool children
often cope with conflicts by resorting to bribery, The Impact(s) of Aversive Consequences
reasoning, surrendering to child demands, coax- Do these aversive reactions function as punish-
ing, or diversionary tactics. As illustrated in Table ment? For an event to function as punishment it
6.5, mothers were generally more likely to punish must first be identified as aversive and then be pre-
boys’ conflict behavior (.42) than girls’ (.18). In sented contingent upon the occurrence of a target
addition, they were more likely to use spanking response. If these contingent arrangements are fol-
and physical force for boys (.16) than for girls lowed by decreases in the likelihood of the target

127
behavior R; given the occurrence of its controlling coercive behavior? This paradox confronted us in
antecedent Aj, then an event may be said to be the early 1970s. It was eventually resolved by re-
“punishment.” The weakening of the Aj—»R; moving the short-term impact of coercive events
connection describes the long-term effect of a pun- from the punishment concept and considering it,
ishing event. That same punishing event.also has instead, as an example of the eliciting effect of co-
an immediate short-term effect. In fact, in the con- ercive events upon ongoing behavior. Given that
text of social interaction in a natural setting, these aversive events elicit aversive reactions, then,
aversive consequences have such a profound im- when used as a consequence, the effect will be to
pact upon ongoing interaction that our efforts to “decelerate” ongoing prosocial behaviors and “ac-
understand punishment have been limited almost celerate” ongoing coercive behaviors.
entirely to the study of these short-term effects.
We satisfied ourselves that an effective punish-
ment (time out) significantly reduced the likeli-
hood of a child’s Noncomply given mother-Com- Aversive Events as Accelerating
mand; i.e., the Aj—» R; connection was reduced. Consequences
The data for these two single case studies are de-
scribed in Chapter 8. However, the long-term im- In the first study of the eliciting effect of aversive
pact of the 14 aversive consequences has yet to be consequences, Patterson and Cobb (1971) anal-
determined. The study would require both a cor- yzed 56,632 interactions involving Talk from
relational and an experimental format analogous members of 24 families. If a family member was
to what was described in Chapter S for positive re- talking, the likelihood was .519 that she or he
inforcement. Each of the 14 consequences would would continue talking into the next six-second
be considered separately. Given that it was made time frame. The data indicated that each of the 14
contingent upon a specific A; and target response noxious behaviors was effective in deflecting (sup-
Rj, was this accompanied by a reduced likelihood pressing) this ongoing prosocial behavior. Some
for the Rj when the A; was next presented? aversive events, such as Command or Hit, were
As an alternative, a single-subject ABA reversal very effective; the conditional p of Talk, given
design would seem to be the experimental manipu- Talk, followed by Command or Hit was .014 and
lation of choice. Here, the mother could be cued .OS9 respectively. More recently, Snyder (1977)
during the B phase to use one of the 14 aversive replicated this decelerating effect for a sample of
events contingent upon a specific Aj and target Rj. normal and distressed families.
The punishment should be accompanied by a re- The data from the Patterson and Cobb (1971)
duced likelihood of Rj given A; during the experi- study of 24 disturbed families generated a total of
mental phase. This effect should persist during the 615 Hits, Pushes, and Shoves. These target behav-
second baseline. When a parent plans to confront iors were performed by parents, siblings, and
and stop a deviant child behavior, which punish- problem children alike. In these clinical families,
ment does he or she use? How effective is it? In the 14 aversive consequences did not suppress Hit.
that we have not carried out these studies, our un- Quite the reverse, aversive consequences were as-
derstanding of parental punishment is very lim- sociated with increased likelihoods for continu-
ited. ance of the attack. The more aversive the conse-
An aversive consequence is very likely to elicit quence, the greater its effect as an accelerator for
an aversive reaction from the other person. If the Hit. A comparable sequential analysis of problem
other person is doing something prosocial, such as children in an institutional setting by Kopfstein
Talk, and you provide an aversive reaction, such (1972) replicated the. finding and extended it to
as Humiliate, then its eliciting function will very the full range of coercive target responses. Gar-
likely suppress ongoing Talk behavior. When we den-variety punishments facilitated the recurrence
first encountered this phenomenon, we thought it of coercive child behaviors. More recently, this ef-
was an example of punishment functioning as a fect was also replicated for families of normal and
temporary suppressor of behavior (Patterson & antisocial children by Snyder (1977). Snyder ob-
Cobb, 1971). This was the view of punishment in tained ratings of aversiveness for each of the code
vogue in the late 1960’s. Imagine our surprise categories separately for mothers, fathers, and the
when we discovered in the same analysis that these children. For the normal families, the correlations
aversive consequences had an accelerating effect between a family member’s perception of the aver-
when made contingent upon coercive behavior! siveness of an event correlated very highly with the
How could the same aversive event suppress (i.e. event’s observed impact in accelerating the indi-
punish) prosocial behavior and accelerate ongoing vidual’s ongoing coercive behavior.

128
 Table 6.6
The Impact of Aversive Consequences Provided by Parents for Coercive Child Behaviors
(from Patterson, 1976)

The likelihood of immediate recurrence for:

Hostile! Social Aggression? Total Aversive Behavior

Parental Normal Problem Normal Problem Normal Problem
Consequences Sample Sample Sample Sample Sample Sample
Aversive 25 41 S12 2X) 36 50
AllNonaversive
Consequences? Bo 26 30 te) sey 34
1Hostile is comprised of DI, NE, HU, WH, and IG.
Social Aggression is defined by PN and TE.
’This includes nonaversive reactions by parents and siblings. Nonaversive would include events classified as neutral and prosocial.

Furman and Masters (1978) were also con- responses which formed each class also shared a
cerned with the validity of a priori classifications common set of aversive consequences which func-
of events as aversive or positive. They assumed tioned as “accelerators” (Patterson, 1977a). These
that positive events should produce more positive data will be discussed in detail in Chapter 8.
reactions (e.g., Laugh, Talk) from the recipient. The first comparison study for differential reac-
Similarly, events identified a priori as aversive tivity to aversive consequences examined the dif-
should be followed by more negative reactions ferential effect of parent and sibling punishments
(e.g., Disapprove, Yell) and fewer neutral and for coercive behavior of the problem child (Patter-
positive reactions. Preschool children’s interac- son, 1976). The comparison included samples
tions in classroom and gym settings served as the from nonproblem families and from families re-
data base. These events were negatively correlated ferred to OSLC for treatment. The dependent var-
with the likelihood for both neutral and positive iables were three classes of coercive child behav-
reactions. Their analysis showed a convergent cor- ior: Hostile, Social Aggressive, and a composite
relation of .58 between the data for a priori cate- score (TAB) comprised of all 14 coercive re-
gorizing for aversive consequences and the data sponses. For both samples, the findings consistent-
for the recipient’s negative reactions. In other ly showed that punishment by siblings was ineffec-
words, the investigators’ a priori notions about tive in suppressing coercive behavior. The results
aversive consequences were in agreement with the for parental punishments are summarized in Table
data from the children’s actual reactions to these 6.6. In that table, the corrected base rates de-
events. The investigators also carried out a dis- scribed the residual effect when the impact of aver-
criminant analysis showing the heterotrait-mono- sive consequences was removed from the calcula-
method correlations to be less than the conver- tion of the base rate. If one looks only at these cor-
gent, monotrait-heteromethod correlation. These rected values, then about two out of three times,
data demonstrated the convergent and discrimi- boys in both samples tend to perform a single six-
nant validities for their measures of aversive con- second coercive act and then stop (given no paren-
sequences. tal punishment). In families of normal children,
Once this accelerating effect was identified, we the impact of parental aversive consequences
began to study differences between age groups, (C —) varies as a function of the class of coercive
and normal and clinical samples concerning inter- responses. For the total composite score, (TAB),
actions between various family agents. It was also there is a slight increase in recurrence (.36) as a
at this time that we identified two response function of Parental C—. When similar conse-
classes, each of which was controlled by a com- quences are applied to the child’s socially aggres-
mon network of antecedent events (Patterson & sive behavior, the effect is to suppress the ongoing
Cobb, 1973). When we studied these classes (So- behavior. The conditional p of .12 for the recur-
cial Aggressive and Hostile) further, we found that rence of socially aggressive behavior showed a re-

129
duction when compared to the corrected base-rate by John Reid, the real deviant is a person whose
value of .30. This is of particular interest in that behavior cannot be controlled by the reactions of
the responses which comprise this class were con- other people. If such a person cannot be controlled
trolled by antecedents and consequences provided even by punishment, then there is cause for alarm.
largely by siblings (Patterson, 1977a). Parents of These considerations led to the decision to carry
normal children seem to be effective in suppressing out yet further methodological studies on the in-
sibling conflicts. However, note that they were dex, p(R2|R:—» C-—). This index represents the
only moderately successful in suppressing the probability that a coercive response will recur giv-
coercive events in the Hostile class. Interestingly, en that the first presentation of the aversive re-
this class was controlled by antecedents and conse- sponse was followed by a negative consequence.
quences that were mainly provided by mothers Up to this point, estimates of the effect had been
and older sisters. based upon samples of events rather than subjects.
The aversive reactions by parents of antisocial Secondly, it would be of some interest to compare
children did not effectively suppress ongoing be- various clinical samples and to differentiate
haviors for any of the response classes. For overall among the reactions of various agents. For this
coercive child behavior there was roughly a 50% purpose, OSLC files were used to form samples
increase in the likelihood of occurrence as com- comprised of 36 Normals, 43 Social Aggressors,
pared to the corrected base-rate value of .34. In ef- and 19 Abused children. The differential effective-
fect, their attempts to punish coercive child behav- ness of aversive consequences for the ongoing co-
ior made things worse. The same study also ercive behavior of the problem child was exam-
showed that these parents punished more often ined for mothers, fathers, and siblings. The data
but less effectively. No wonder some of these par- summed across the effects of punishment at all
ents described their children as being possessed by junctures in the target child’s coercive chains. In
an evil spirit which required the expertise of an ex- Appendix 6.1 the corrected base rates specified the
orcist rather than a psychologist to remedy the sit- likelihood of the target child’s Stop given that the
uation. The problem child was out of control both target child was interacting with various family
in his referral symptoms and in his failure to re- members, and given that they reacted to his coer-
spond to this basic mechanism of socialization. cive behavior in a neutral and/or positive manner.
Our confidence in these findings was consider- It can be seen there that socially aggressive boys
ably enhanced by the comparison study of prob- were generally more likely to continue their coer-
lem and nonproblem families by Snyder (1977). cive behavior than were boys in normal families.
Using a derivative of the OSLC code, he first rep- They were also more likely to continue their coer-
licated the finding from our 1971 study that par- cive behavior when interacting with their mothers
ental aversives decelerated prosocial behaviors. than with their sisters.
He then examined the impact of these parental The data from the samples are summarized in
aversive consequences for overall child coercive Table 6.7. The mean likelihood of the child’s Stop
behaviors. He found a corrected base-rate value of given an aversive reaction to the prior coercive be-
.28 for immediate recurrence of coercive child be- havior was calculated separately for each of four
haviors in his normal sample. This is very close to family agents and for the normal and two clinical
the comparable value of .32 found in Table 6.6. In samples. The mean, standard deviations, and N’s
Snyder’s study, given parental punishment, the are presented for each of the cells. The t-tests for
conditional p for recurrence was .32. The results correlated means compared the conditional p
for his clinical sample were in keeping with those values to the corrected base-rate values (Appendix
presented in Table 6.6, but the acceleration effect 6.1) for that cell. A negative value indicated that
was not as dramatic. In his clinical sample, the the conditional p value was smaller than the cor-
base rate for recurrence was .27. When punished, rected base-rate value; i.e., the child was less likely
the deviant child was significantly more likely to to Stop when he was “punished.”
continue, as shown by: the conditional p value of The effect of aversive consequences upon coer-
.36. Note, however, that the acceleration effect is cive child behavior varied as a function of sample
only minimally greater for the clinical sample than and the family member with whom the child inter-
for the normal sample. acted. In keeping with the earlier findings (Table
A simple index of the child’s reaction to aversive 6.6), the acceleration effect was more likely to be
consequences which are made contingent upon his significant for socially aggressive boys than for
coercive behavior has some interesting implica- boys from normal families. As shown by the nega-
tions. For example, it seems like a good operation- tive “t” values, for normal boys the effect of nat-
al definition of “deviancy” in general. As suggested tering by family members was generally to reduce

130
 Table 6.7
The Likelihood of Problem Child Stop Given Family Members’ Aversive Consequences
p(Stop|Prior Response Punished)

Mother Father Male Sibling FemaleSibling

Sample Mean N_ SD Mean N_ SD Mean N SD Mean N_ SD

Normal .66 29) 4 ay? AS 50 pd e3h7f .60 21 .40

pa —1.14 —1.57 See NES)

Social Aggressor 163, 44. 36 59 24 38 Si opwks6 160.0) Batis

ay = 0.55 = 1:80" ~2.39** = Wop

Abused 47 LSia= 33) 118 1330 66 Nee eS)

ie —3.06** —0.09 — 1.64

*p<.05

**p<.01
#5 <001
*All comparisons are two-tailed tests.

the likelihood that he would Stop. Nattering pro- There are more parameters which need to be
duced a small increase in the likelihood of an im- considered. What is the effect of age of the target
mediate recurrence of a coercive behavior. This ef- child given that the parent tries to control his coer-
fect was significant only when the target child was cive behavior? Second, what is the effect of con-
interacting with male siblings. This suggests the text upon the likelihood of acceleration? These
necessity for studying sibling interaction in partic- questions are considered for mothers as agents in
ular when trying to understand coercion in fami- Chapter 12. The analyses showed that the nonsig-
lies. nificant acceleration produced by mothers in Ta-
This sibling theme is reiterated and even more ble 6.7 most likely involved younger Social Ag-
strongly emphasized in the Social Aggressor sam- gressors, where the mothers “started” the process
ple. Here, interactions with both male and female with an aversive antecedent. For mothers of Nor-
siblings are implicated. In both cases their aversive mals, the borderline acceleration effect in Table
consequences produced significant increases in the 6.6 was most likely produced by mothers interact-
likelihood of an immediate continuance of the ing with older normal siblings. Further compara-
problem child’s deviant behavior. Note, too, the tive studies are badly needed, particularly those
fact that it was the nattering of the father, not the which involve siblings. As shown in Table 6.7, it is
mother, which significantly accelerated ongoing siblings who play the central role for producing
coerciveness for this sample. acceleration effects for Social Aggressors and for
The two clinical samples (Social Aggressors and Normals.
Child Abusive) seem fundamentally different, a We are beginning to see measures of this effect
fact also noted by Reid, Taplin, and Lorber as central to evaluating treatment effectiveness.
(1981). Notice first that in the Abused-Child sam- The central concern of parents who bring their
ple aversive consequences by fathers did not pro- children to OSLC is not that they are out of con-
duce an acceleration effect! Only nattering by the trol; it is that they cannot control their children.
mother had a significant impact. The situation for The measures of punishment acceleration (p[R2|R;
the Abused-Child sample seems analogous to the —» C —}) indicate the magnitude of this disruption
Sadomasochistic Arabesque families described in in control. As a single index, the p combines two
Chapter 13. In such families one parent has very kinds of information that relate to the loss of con-
tight punitive control over the child and the other trol. Its magnitude reflects, in part, the child’s gen-
has no control at all. This seems exactly the state eral disposition to extend coercive chains and his
of affairs here—the father has control, but no one disposition to extend these chains when punished.
else does. Given that the OSLC treatment program empha-

131
 Table 6.8
Likelihood of Punishment Acceleration by Dyads and Samples
p(Go|Prior Deviant Behavior Punished)

Mean Likelihood for Correlation with

Samples of Child TAB score
Social
Dyads Normals Aggressors Stealers F Values PPM N

p(Child Accelerate|Mother Punish) els) .42 34 11.64*** ATES 111

p(Mother Accelerate|Child Punish) ree) 41 30 2.45 BT 111
p(Child Accelerate|Father Punish .04 41 .24 USE Fes A6nN 68
p(Father Accelerate|Child Punish) ve) 50 31 3.11 a2, 68
p(Child Accelerate|Sibling Punish) .24 41 38 See Sse 103
p(Sibling Accelerate|Child Punish) ney 43 .44 1.74 16 103

*p<.05
**p<.01
***p <.001

sizes the use of nonviolent back-up punishments, scores for 33 referrals were correlated .34 (df= 31;
this should lead to a situation where parent Stop- p<.06) with the acceleration index (Patterson,
Commands are more effective in controlling devi- 1979c). This sample was comprised of both high-
ant child behavior. A comparison of pre- and post- rate Social Aggressors and lower rate problem
treatment measures showed significant decreases children (Stealers). As a replication, three samples
in the p values for the acceleration index (Patter- from OSLC files were examined for 38 Normals,
son, 1976). At termination of treatment, parental 37 Social Aggressors, and 38 Stealers (see Appen-
attempts to control coercive child behavior were dix 6.2 for details of study). The punishment ac-
as effective as they were for Normals. This effect celeration index was calculated separately for each
was also replicated by Szykula (1979), who used dyad involving the target child. As shown in Table
the same coding system and treatment procedures 6.8, the children in the clinical samples were sig-
at another site. nificantly more likely than Normals to accelerate
The results from the intervention studies are in when punished.
keeping with our general formulations about pa- The correlational data in Table 6.8 show that
rental nattering. An increase in the pairing of the child’s disposition to accelerate when punished
threats or Stop-Commands with back-up punish- correlates significantly with his TAB score. The
ment should give the parents better control. Sup- likelihood of accelerating when parents punished
port for the hypothesis would be even more con- accounted for twice as much variance as the com-
vincing if experimental data could be brought for- parable likelihood for siblings. The reader should
ward. Does noncontingent use of punishment keep in mind the fact that this acceleration index is
(nattering without back-up) produce an increment a composite which partially reflects the child’s
in the p value for chain extension? Until these general disposition to persist in his coerciveness.
studies are done, we cannot be sure of exactly This would tend to insure a correlation with his
what this index means. overall performance.
The parameters which account for variance in
individual differences in coercion performance lev- Summary
el will be developed in Chapters 8, 11, and 12. Field studies indicate that parents of antisocial
The acceleration index is thought to be an impor- children are more likely than parents of Normals
tant covariate, particularly in accounting for vari- to provide aversive consequences for coercive
ance in the lower ranges of performance. For a child behaviors. These consequences have a short-
heterogeneous sample of antisocial children, the term effect upon ongoing social interaction that is
index shows a modest correlation with perfor- produced by the aversive events which elicit aver-
mance level. In an earlier study the baseline TAB sive interactions. These same consequences are

132
 also thought to produce long-term changes in ents of antisocial children noncontingent in their
‘A;> R; connections; but, as yet, there are no field use of aversives? From the Watson point of view, it
studies which have investigated this crucial aspect could be the ratio of contingent to noncontingent
of punishment. Our assumption is that for the punishment that is at issue here. How often does
antisocial child, there are fewer consequences that the parent “punish” a prosocial behavior? The
would weaken these connections, and that the data in Chapters 8 and 12 show that this happens
magnitude of the reduction would be /ess. In that more often than you might think, even in normal
there are no systematic studies of this phenome- families. The unprovoked aversive event was sig-
non, it remains a purely clinical impression. nificantly more likely to be provided by mothers of
All of the studies to date have focused upon the antisocial children than by mothers of Normals.
short-term impact of aversive consequences as
they accelerate ongoing coercive behavior. The an-
tisocial child is about twice as likely as a normal
child to extend his coercive behavior. The analyses Differences in Responsiveness
showed that aversive consequences significantly to Punishment
accelerated these chains when the socially aggres-
sive child was “punished” by female or male sib- The focus of the OSLC treatment program is
lings and by fathers. For Child Abusive samples, a upon training parents to use nonviolent punish-
significant acceleration effect was produced by ment in a consistent manner. If parents can learn
maternal punishment. For Normals, only aversive to do this, they will be successful in changing their
consequences provided by male siblings produced problem child; if they can’t, the child is likely to
a significant effect. remain a problem child. This position has two im-
In families of Social Aggressors and, to a lesser plications for the etiology of antisocial behavior;
degree, families of Normals, acceleration follow- both are equally tenable. One is that antisocial be-
ing punishment seems related to sibling interac- havior occurs because the parents can’t (or won’t)
tions. It seems likely that in these families they confront the child in a consistent manner when he
play a crucial role in determining performance lev- is deviant. Personally, I favor this hypothesis. The
els for the target child. Perhaps one can consider a alternate hypothesis is not necessarily antithetical.
high acceleration index as a kind of dominance The assumption is that antisocial children are by
measure. The problem child is more likely to re- temperament /ess responsive to punishment than
spond to threats, scolding, and warnings with a normal children. There are laboratory studies
continuation of his antisocial behavior. In effect, which suggest that this indeed may be the case.
he is about as powerful as his siblings or his par- There is some evidence for the hypothesis that
ents. I think that very high indices for chain exten- antisocial children may be less responsive than
sion and for acceleration when punished means Normals to mild (social) punishment on neutral
that the child is in control most of the time. laboratory analogue tasks (Kuenstler, 1970; John-
A proper understanding of the role of punish- ston, 1976; Orzech, 1962, cited by Hedlund,
ment in coercion processes requires a new genera- 1971). However, the design for many of these
tion of studies such as those by Ross Parke. These studies is similar to the design employed by the
studies should compare antisocial and normal verbal conditioning studies in the 1960’s. The dif-
children for their responsiveness to such nonsocial ficulties with these procedures were detailed in
punishment as time out, point loss, and depriva- Chapter 5. The modern counterparts of these la-
tion of privileges. We also need to compare these boratory studies are found in the attentional defi-
samples for their reactions to low key aversive cit studies reviewed in Chapter 11. These studies
events (e.g., statements such as “wrong,” and showed that the attentional deficits of antisocial
“no”). The question here is how generalizable is children were quickly remedied when money was
this nonresponsiveness? Do any of the 14 aversive used as a reinforcer for accurate performance.
consequences produce long-term reductions in the What at first seemed to be an attentional problem
A;—» R; connections? Is the reduction the same turned out to be a motivational problem. I wonder
for antisocial and for normal children? if the studies which showed a lack of responsive-
The nattering hypothesis has yet to be directly ness to punishment might demonstrate little more
tested. Does pairing of parental Stop-Commands than a lack of involvement in the task.
with time out increase the decelerating impact of It is possible that antisocial children are less re-
future commands? Finally, there is an intriguing sponsive to aversive events. However, before that
idea based upon J. Watson’s (1979) formulation conclusion is tenable we need a series of more con-
that merits investigation. To what extent are par- vincing studies than those published to date.

133
 what intensity. The vehicles for the training are
Variables Relating to Dispositions the familiar tools of modeling, reinforcement, and
within the Child punishment.
The child’s responsiveness to social reinforcers As viewed by Hamburg and vanLawick-Goodall
and to punishment is not a static disposition. Re- (1974) and Hinde (1974), the development of ag-
sponsiveness to these consequences is thought to gressive behavior would involve movement
be related to states that are, themselves, readily through a progression of such experiences. The
manipulable (Patterson, 1969). For example, the unskilled individual may be particularly respon-
studies reviewed by Gewirtz (1967) and Stevenson sive to certain stimuli eliciting, in turn, partially
(1965) suggested that responsiveness to social re- formed aggressive responses. These skills are
inforcers was altered by prior conditions of satia- shaped by experience. There is no single path to
tion and deprivation. Other variables relating to aggression; each individual finds his or her own
responsiveness to positive reinforcers were re- way. The learning process may be facilitated,
viewed in Chapter 5. Similar considerations are however, by the individual’s genetic history, by se-
thought to apply to responsiveness to punishment. lective responsiveness to certain kinds of eliciting
It may be that involvement in a coercive system stimuli and to certain kinds of reinforcers, as well
produces, as a side effect, members that are rela- as by the presence of some partially formed but
tively less responsive to each other’s coercive be- unlearned aggressive responses (such as temper
haviors. Quinsey (1970) suggests that being in a tantrums). Regarding both animals and humans,
coercive system alters one’s general adaptation lev- this general formulation is still tentative. The re-
el so that the threshold for what is perceived as ex- cent text by R. Cairns (1979b) gives a first-rate
tremely aversive is altered. This would suggest overview of this body of literature. Hamburg
that siblings and parents from antisocial families (1974) goes on to say, “We certainly do not believe
would be less responsive than normal siblings and that there is anything like definitive evidence on
parents to garden-variety aversive consequences. these matters at the present time. Rather, we are
This may be in keeping with the findings discussed trying to construct plausible models that would
in this chapter, which indicated that Social Aggres- stimulate and guide inquiry . . .” (p. 62).
sors reacted significantly more than Normals to For animal species, including primates, the re-
aversive consequences as elicitors rather than as search findings clearly support a genetic differen-
suppressors of antisocial behavior. tiation for aggressiveness. As developed by Hinde
(1974) and by Cairns (1979b), the model is clearly
Heritability and Responsiveness an interactional one, i.e., genetic dispositions
to Aversive Events modified by experience. As we shall see, the find-
The literature suggests that species differ in ings for genetic and/or constitutional variables
the (innate) disposition to learn aggression. For which are related to individual differences among
the animal lacking social experience, aggressive humans are not so clear-cut. Nevertheless, the
behavior may be expressed in an incomplete form. data for humans are sufficient to be taken serious-
Skill in aggression requires additional learning. ly. The current attitude seems to be “wait and
The form in which aggressive behavior is ex- see,” with some cautious speculation about which
pressed may also change as a function of age. For variables might possibly be involved. For ex-
example, its earliest manifestation in primates ample, variables, such as hyporesponsiveness to
may be temper tantrums, which do not have to be aversive stimuli, differences in activity level, tes-
learned. D. Hebb (1972, cited by Hamburg & tosterone, or irritability could all covary with co-
vanLawick-Goodall, 1974) writes: ercive performance. All of them could also have
genetic and/or constitutional components.
“Neither a human nor a chimpanzee baby needs to
One cannot properly evaluate the importance of
learn how to have a temper tantrum . . . the baby
genetic contributions to the performance of ag-
does not have to practice it (nor to see how others
gressive behavior without considering the matrix
do it) in order to produce, on first try, a first-class
of social interactions within which it occurs. This
sample. It is therefore ‘unlearned.’ But it is not in- is also the position of behavioral geneticists such
dependent of learning.” (p. 63)
as McClearn and DeFries (1973). They have ex-
Eibl-Eibesfeldt (1974) reported that a deaf child’s tensively reviewed the literature demonstrating the
pattern of temper tantrum was indistinguishable well known fact that some strains of mice and
from those of normal children or young primates. dogs are more aggressive than others. An example
Experience will teach the young child when to use of these studies is provided by the programmatic
the temper tantrum, where, with whom, and at work of Lagerspetz (1964). In this study she mea-

134
 sured aggression in male mice in successive genera- more homogeneous than for DZ twins.
‘ tions selectively bred for high and low aggressive- Mednick and Hutchings (1977) proposed a
ness. The findings provide compelling evidence for cross-fostering design to obviate these difficulties.
some type of genetic determinants for aggressive- As part of their studies in Denmark, they obtained
ness in mice. The precise mechanisms involved are the files for 1,145 male adoptions, including infor-
not clear; but the search for them constitutes the mation about criminal behavior of the adoptee,
focus for Lagerspetz’s current work. This pro- the biological father, and the foster father. There
grammatic work is also very much in the interac- were 143 adoptees with criminal records for
tional mode. Among other things, she has found whom complete data were available for both bio-
that being defeated reduced aggressiveness for logical and foster fathers. This sample was
both aggressive and nonaggressive strains (Lager- matched for age and social class of the father with
spetz, 1980). a sample of noncriminal adoptees. In both samples
In the same vein, the fostering studies reviewed there was little or no contact between the adoptee
by Denenberg (1973) demonstrated the interac- and the biological father. Given that neither father
tion between genetic endowment and environ- was criminal, then 10.5% of the adoptees were
ment. Kuo (1930, 1938, cited by Denenberg, criminal; given a noncriminal biological father
1973) reared kittens and rats together and found and a criminal adoptive father, the figure for the
that this experience inhibited later rat-killing be- adoptee was 11.5%; if the foster father was not
havior. Denenberg’s (1973) own elegant cross-fos- criminal and the biological father was, then the
tering studies demonstrated that rearing mice with rate was 21.4%; if both fathers were criminal, the
rat mothers reduced the rats’ later aggression to- rate was 36.2%. These findings suggest a genetic
ward mice. Thus, while other studies demonstrate base for criminality. Mednick’s programmatic
that rats’ mice-killing behavior was unlearned, work continues and has been expanded to include
these studies demonstrated that the same behavior speculations as to how fathers transmit such pro-
was alterable. For these species, then, aggressive clivities to their sons.
behavior was the outcome of the interaction be- There are several variables identified in the re-
tween genetic contributions and social experience. search literature which may be genetically (or con-
How generalizable are these findings to pri- stitutionally) determined that also relate to the
mates and/or humans? Studies of monozygotic learning of aggressive behavior. For example,
(MZ) and dizygotic (DZ) twins are relevant to this Lagerspetz (1980) and McClearn and DeFries
question. Presumably, MZ and DZ twins have the (1973) reported that aggressive strains of mice
same social environment but vary in their genetic were more active. The next question is, in what
similarities. In the Scarr (1966) studies of parents’ sense might activity level determine aggressive be-
ratings of their children, she found an intraclass havior? Vigorous activity was the dependent vari-
correlation of .35 for MZ twins on aggression. able identified by Cairns (1972) in his studies of
The comparable correlation for DZ twins was the aggressiveness of mice reared in isolation.
—.08. In their review of the literature on studies When placed together isolated mice tended to re-
of twins, Mednick and Hutchings (1977) note spond in a hyperstimulated fashion to overtures
concordance rates for criminality of 60% for MZ from other mice. This, in turn, often produced a
twins and 30% for DZ twins. As the authors vigorous counterreaction from the isolated sub-
pointed out, there were probably serious sampling ject. These reciprocal interchanges quickly esca-
errors that characterized most of these early lated to fighting.
studies. For example, in many studies the propor- Willerman (1973) presented data for twins
tion of MZ twins was higher than expected, sug- showing an intraclass correlation of .71 for hyper-
gesting biased samples. However, the recent study activity in MZ twins. He also cited several other
by Christiansen (1974) overcame many of these studies supporting these findings. Bell’s (1968)
difficulties. He sampled all twins born in Copen- careful review also provides a number of studies
hagen between 1884 and 1910. Of 3,586 twin supporting the notion of heritability of impulsive-
pairs, the concordance rates for crimes were 36% ness and assertiveness in children. What is missing
for MZ and 13% for DZ twins. While these find- here are data which demonstrate that more active
ings suggest a genetic component for criminality, children are also more likely to learn aggressive
Christiansen (1974) and Mednick and Hutchings behavior. The observation study of nursery school
(1977) note that the key assumption underlying aggression by Patterson, Littman, and Bricker
these studies may be open to question. As they (1967) gave partial support to this hypothesis; but
point out, there is some reason to expect that the the interpretation varied depending on how one
shared social experience for MZ twins may be defined activity level. Children interacting at high

135
rates tended to be more aggressive. The correla- to son. They also speculate as to how this could
tions of activity level with aggression were positive relate to the faulty development of self-control
(but nonsignificant) for measures of distance trav- mechanisms. Their current studies include a longi-
eled and for ratings by observers of intensity of tudinal design providing for electrodermal recov-
motor behavior. However, observer ratings for vo- ery rates for parents and children. The long-term
calization rate and intensity correlated .66 (p< follow-up data will demonstrate if this measure
.05) with observed aggression. predicts later criminality.
There is yet another plausible set of variables It seems reasonable to suppose that small indi-
stemming from Lykken’s (1957) well known stud- vidual differences in activity rates and/or respon-
ies demonstrating that psychopaths were less ef- siveness to aversive stimuli might interact with in-
fective in learning escape responses to aversive ept child management techniques to produce an
events. The variable implicated was the psycho- antisocial child. The empirical basis for specifying
paths’ hyporesponsiveness to the aversive stimuli. these relationships is far from satisfactory and, at
The position has been reaffirmed by Hare (1968). best, is only suggestive. However, the data which
He showed that psychopaths had lower resting are available indicate that a simplistic environ-
levels for skin conductance and less responsivity to mentalist position is probably inadequate. One
mildly aversive auditory tones. They also reacted cannot conclude which of these variables is more
slower than Normals on two measures of auto- important based on present data. However, my
nomic activity: cardiac deceleration and digital own subjective bias is that inept parenting will be
vasoconstriction. the key variable and that the problem child’s hypo-
Mednick and Christiansen’s (1977) edited vol- responsiveness can be altered by improvements in
ume brings into focus the literature investigating the parents’ child management skills.
the hyporeactivity of adult and juvenile criminals
to aversive stimuli. They present two key hypothe- Some Implications
ses. One suggests that hyporesponsiveness to pun-
Punishment is Treatment
ishment impedes avoidance learning. The second
hypothesis is that a slow autonomic recovery con- The findings from studies analyzing punishment
tributes to this impedance. The studies relating to have direct relevance to the treatment of families
these hypotheses were reviewed by Siddle (1977). with antisocial children. Parents of normal chil-
Most of these studies involved selected subgroups dren tend to ignore most coercive child behavior;
of adult criminals, primary and secondary psycho- as a\result, the episodes tend to be of short dura-
paths. His review suggests that these out-of-con- tion. When they want to, these parents are able to
trol adults cannot be considered to be generally use punishment to stop or suppress these behav-
underaroused because, at best, the findings from iors. Parents of antisocial children ignore less and
at-rest conditions were ambiguous. However, dur- natter more. However, unless they use extreme
ing aversive stimulation or the threat of aversive measures, they cannot stop coercive behaviors.
stimulation most of the studies cited by Siddle sup- Furthermore, their nattering contributes directly
ported the hypothesis of hypoarousal. The psy- to extended coercive episodes. During the last five
chopaths showed both smaller skin conductance years of treating samples of extremely difficult
response to the signal and the shock, as well as a chronic delinquent and abused children, we have
slower recovery. Presumably, a lowered respon- become convinced that training parents to use a
siveness and slower recovery from punishment nonviolent form of punishment is a necessary
would lead to slower learning of avoidance re- component of successful intervention. That posi-
sponses. Siddle cites the well known studies by tion is certainly not in keeping with our original
Lykken (1957) and Schmauk (1970) in support of stance (Patterson, 1965a; Patterson, Littman, &
this hypothesis. Bricker, 1967). At that time, it was thought that
Mednick and Hutchings (1977) reviewed a ser- positive reinforcement of prosocial responses that
ies of studies demonstrating that electrodermal re- competed with the deviant responses would serve
covery rates were significantly slower for criminal as the treatment base. However, clinical experi-
and delinquent samples than for noncriminal pop- ence quickly underscored the fact that this was not
ulations. One of the studies cited involved a sam- to be the case. The studies reviewed here and in
ple of twins for which these measures were availa- Chapter 11 showed that extinction, per se, was
ble. The heritability coefficient for that sample not successful even when combined with rein-
was .83. The authors make the point that a slow forcement of competing responses. As one of the
electrodermal recovery rate is a physiological original “true believers,” I found it extremely diffi-
mechanism that might be transmitted from father cult to accept the idea that extinction would not

136
 work. If it were possible to stop all positive and As shown in this chapter, the parents of socially
negative reinforcement for coercive behaviors then aggressive children did punish more often; at least
" it seems that these responses would eventually be they were more likely to provide an aversive con-
extinguished. In principle, extinction should sequence in reaction to sibling and problem child
work. But I think that it is simply not possible for aggressive behavior. Others have noted this corre-
parents to control all of their own reactions plus lation and have put forth several interpretations of
the reactions of siblings to the skillful intrusions of what it means. The most widely accepted position
the problem child. At a practical level, in social sit- is the modeling-frustration hypothesis, which sug-
uations the reinforcement schedules can be altered gests that parental punishment produces an emo-
but never reduced to zero. Therefore, the prag- tional reaction and provides a model of aggressive
matics of family interactions are such that punish- behavior for the child. This theme was succinctly
ment is a necessary, but not sufficient, component stated by Bandura (1973):
for effective child management (Patterson, Reid,
“In exercising punitive control, prohibitive
Jones, & Conger, 1975).
agents model aggressive styles of behavior not un-
A beautifully controlled study by Walker, Hops, like those they wish to discourage in others. Recip-
and Greenwood (in preparation) addresses this is-
ients may, on later occasions, adopt similar ag-
sue. They applied a multiple baseline ABABABde- gressive solutions in coping with the problems
sign to a small token-culture classroom. After a confronting them. . . . Although the direction of
baseline period and a prior manipulation, the dis- causal relationships cannot be unequivocally es-
ruptive children were placed on a schedule in tablished from correlational data, it is clear from
which they received tokens for positive, achieve- controlled studies that aggressive modeling breeds
ment-oriented responses. The observation data aggression.” (p. 226)
showed the expected steady increase in rate for
task-oriented behaviors. After a slight reduction, Other writers sweep aside Bandura’s cautions
the disruptive behaviors returned to their baseline regarding the correlational nature of the findings.
level. Following another baseline phase, the teach- For example, Welsh (1976) leaps to the inference
er punished disruptive behavior by taking away of causality: “A number of field studies indicate
points or, occasionally, by using time out proce- that severe parental punishment is definitely a pre-
dures. Task-oriented behaviors were again rein- cursor of aggression in humans” (p. 17). He goes
forced by giving points. Over sessions, the disrup- on to document the frequency with which extreme
tive behaviors were dramatically reduced, and the physical punishment (using a belt, a board, an ex-
prosocial behaviors remained at a high level. The tension cord, or a fist) are reported by samples of
study demonstrated that for extremely antisocial male delinquents. He labels this causal relation his
children achievement and disruptive behaviors are belt theory of juvenile delinquency. In effect, he
functionally independent of each other. Second, states that parental punishment causes aggressive
the findings indicated that in order to control devi- behavior, and that extreme punishment causes ex-
ant behavior, it was necessary to employ an effec- treme acting out. '
tive punishment. Other writers, such as Berkowitz (1973a), take
a more moderate stance in interpreting these cor-
Punishment Causes Aggression? relational findings. He first examines the possibili-
The position taken in this volume is that the ty that other variables may mediate parental effec-
control of antisocial behavior requires the contin- tiveness as punishing agents, e.g., nurturance,
gent use of some kind of punishment. This posi- consistency, and rule statements. While accepting
tion is contrary to a well-established finding in de- the importance of these variables, he emphasizes
velopmental psychology. Studies that investigated the possibility that the kind of punishment used by
parental reports about their punitive practices parents of aggressive children may be ineffective.
consistently showed a positive relation with anti- He goes on to provide a case for the necessity of
social child behavior (Feshbach, 1970). Parents of punishing aggressive child behaviors, but in the
problem children report that they use punishment context of a warm, loving parent, who uses rea-
more frequently than parents of normal children. soning or explanations in conjunction with nonvi-
Their punitive practices are also more likely to be olent punishment, such as time out.
extreme. These covariations also stand up in some The present writer is in total accord with Berko-
of the more recent studies which control for the witz’s position. His statement perfectly reflects our
possible confounds inherent in using parent re- conclusions from the last decade’s intervention
ports for both the dependent and the independent studies with families of aggressive children. Time
variables (Eron et al., 1971, 1974). out and analogous consequences (such as work de-

137
tails or loss of privileges) are certainly aversive.
Relatively speaking, they are very effective; how-
Conclusion
ever, they are not violent. When point loss, time What then has been learned? For one thing, the
out, and loss of privileges are effectively applied, concept of punishment seems to be at the core of
the correlation found in traditional development antisocial child behavior. The parents of these
studies will be reversed. After a few weeks the like- children cannot or will not stop these behaviors
lihood of p (Time Out|Coercive) should correlate from occurring. To understand more, we must
negatively with observed TAB score; i.e., the fre- now shift our focus somewhat. We need to know
quently punished child is less aggressive. It is as- more about signal cues, about back-up punish-
sumed the traditional finding of a positive correla- ment, and about hyporesponsiveness to aversive
tion between parent punishment and child aggres- stimuli.
sion is a process outcome, not the cause, as sug- It also seems the case that we may have to train
gested by Welsh (1976). parents and teachers in our culture to accept the
Again, more studies are needed. What do the idea that punishment is not a bad thing for a child.
parents of the antisocial child do when they want In moving away from the excesses of physical vio-
to stop coercive behavior? How do they signal to lence previously directed against children, we have
the child that they really mean it this time? When created a dilemma for those who must devote their
they natter, do they want to stop the behavior, or lives to training children. The old adage, “It takes
do they only want to “bitch” a lot? Our present as- two to fight,” is not true. One well-trained coercer
sessment devices cannot differentiate effective can disrupt an entire classroom, an office, or a
punishment from nattering, so these concepts are, home. In dealing with a problem child it helps to
for the moment, not directly testable. Obviously have rules clearly stated and rational discussion,
some new measures are needed. Perhaps some var- but sooner or later someone has to confront him
iant of the Parent Daily Report (Chapter 3) would and teach him that coercion does not work. If the
give us daily records of parents’ back-up punish- teacher or the parent has been taught that all pun-
ment. Is it true that parents of normal children are ishment is bad then they will not employ the one
more likely to ignore and to use back-up punish- mechanism that we have found necessary for the
ment, and less likely to natter? control of antisocial behavior.
Given that the child coerces and the father nat-
ters, there is no change in the likelihood of future
problems. Nattering is analogous to a whine or a
Footnotes
whimper, an indication that things are getting out 1. It should be noted that the study cited here
of control. It seems to be a danger signal of some measured only the short-term suppressing impact
kind. It is a declaration that the parent has taken a of aversive consequences. It remains necessary to
nontraining stance. I think that the normal parent use these events in a comparison of their long-term
intervenes to reduce long-term misery and to teach effects in altering p(Rj|A;) when applied to proso-
the child. The parent of the antisocial child does cial and to coercive family behaviors.
not seem to make this commitment. 2. The attribution studies by Lepper (1980)
There is another facet of this process that would demonstrate in a convincing manner that even
be fascinating to study. It has to do with the ap- those children who successfully compete in that
parent skill with which some parents (and chil- system experience it as aversive. Prizes won in
dren) employ guilt induction as punishment. On competition are not experienced as rewards.
one occasion, for example, I was present when an These effects have been replicated in a number of
adolescent brought home three D’s as term grades. laboratories. They serve as a major indictment of
In the following discussion he skillfully turned it modern educational practice, with its emphasis on
around so that his mother was to blame for his competition and grades, which compare the child
bad grades! He first pointed out that she worked to all other children. The alternative to the compe-
full time and this meant she could not supervise tition model is, of course, to provide meaningful
him enough. He then skillfully punished her for positive external reinforcement contingent on the
even bringing up the topic of his poor grades. One child’s moving-from his or her own baseline level.
has the feeling this technique is used often, espe- Why not design schools where feedback is tailored
cially in some subcultures. It also seems to be terri- to the individual? Why not use external reinforc-
bly effective in the hands of the skilled practition- ers, such as praise, instead of relying upon the fear
er; but its subtlety defies our existing code system. of not competing successfully?
3. Unfortunately, there is a confound here.
Given that parents of normal and antisocial chil-

138
 dren were equally /ikely to punish deviant behav- was events rather than subjects; some high-rate
ior, then it would follow that the latter should re- boys contributed many events to the analysis,
. port more frequent attempts to punish. These chil- others contributed only a few. For the 1976 study,
dren are performing deviant behaviors at nearly the mean likelihood of parental punishment for
twice the rate so their parents should report al- the distressed sample was .62, and for the nondis-
most twice as much punishment. As seen in Table tressed .40. A comparison to the findings in Table
6.4, the parents of Social Aggressors not only pun- 6.4, based on samples of subjects, shows marked
ish more often, they are also more likely to punish. discrepancies. Evidently, some individuals in the
If one uses this likelihood index and correlates it 1976 sample did contribute extreme scores. I
with the child’s level of aggression (TAB score), think the analysis of events rather than subjects
the relation is just the reverse of the traditional gives an overestimate of the likelihood of parental
finding. The correlation between p (Parental punishment.
Command Negative|Child Deviant) and the 5. The OSLC code makes no provision for re-
child’s TAB score was —.10. This was based upon cording critical information on the use of time out,
a sample of 30 OSLC cases. In either case, the par- withdrawal of privileges, and so on. After the code
ent frequency and/or likelihood of punishment ac- was developed, John B. Reid added the Observer
count for very little of the variance. Impression Checklist. The three pages of ratings
4. An earlier analysis by Patterson (1976) used contain items relating to such events. The check-
a heterogeneous sample of out-of-control boys list is filled out after each session. These data are
and a normal sample similar to that employed in presently available but have not yet been analyzed.
Table 6.4. In the earlier analysis, the population

139
 Chapter 7
Abstract
Negative reinforcement (NR) is an arrangement in which an aversive event is followed by a re-
sponse which, in turn, has a neutral or positive outcome. Such arrangements apply to about one-
fourth of the problem child’s coercive behavior. These NR arrangements serve in several different
roles. First, it is a means by which one may cope with the aversive intrusions of other family mem-
bers. NR arrangements are also involved in the process by which the aggressor trains his victim to
reinforce him. The third contribution is its key role in escalating interchanges to high-amplitude vio-
lence.
Data are presented for normal and clinical samples, describing the likelihood of aversive antece-
dents and the proportion of episodes in which the counterattack effectively terminates the aversive
antecedent.
Experiments are described which show that the termination of an aversive event is followed by a
strengthening of the A;—» Rj connection. Victim compliance is also shown to reliably produce a ces-
sation of the counterattack.
The literature describing physical violence in families is reviewed. It is assumed that this is the end
product of an escalation process. Increases in amplitude are more likely to occur during extended co-
ercive chains. If an increase in amplitude produces victim compliance, it increases the likelihood of
higher amplitude interchanges during future trials. Given both members of a dyad have relatively
equal power, escalation on the part of one member will lead to synchronous increases by the other.
This process continues until one person submits.

140
 Negative Reinforcement
and Escalation
Coercive behaviors are maintained by both pos- sor may seem counterintuitive. If the mother gives
itive and negative reinforcement. Data to be re- in, does the whining child really stop within sec-
viewed in a later section show that about one- onds? Data relating to the victim-training hypoth-
fourth of the time the child’s coercive response is esis will be presented in a later section of this chap-
followed by the removal of an aversive intrusion ter. Suffice it to say here that the data support
by another family member. If the counterattack is these ideas.
successful, the intruder withdraws and the connec- Let’s assume that a young sibling has observed
tion between the antecedent (intrusion) and the re- and thus learned about coercive responses. On oc-
sponse (counter) is strengthened. This arrange- casion he serves as a victim for these attacks.
ment is negative reinforcement (NR).! One of the Given that the attacks are not overwhelming, then
striking features of negative reinforcement is that he or she will sometimes counterattack. If the
it can produce massive changes in antecedent-re- counterattack meets with some success, then the
sponse connections with very few trials. lamb may become a lion. The child may also in-
NR plays a crucial role in coercive interchanges. crease the likelihood that he or she will perform
NR is not only responsible for the maintenance of these same attack behaviors. Again, NR is thought
many coercive events, it is also the means by to play the crucial role. The formulation was
which the aggressor trains a victim to comply or based on the results of the study of nursery school
submit. In this context, the person initiating a co- conflicts (Patterson, Littman, & Bricker, 1967).
ercive event is the attacker. The person who ends a In that study, 21 children were identified, each of
coercive exchange with a positive or neutral reac- whom had displayed two or less coercive behav-
tion is the victim. Within coercion theory it is the iors during the first five nursery school sessions.
victim who trains the aggressor when, how, and During the following sessions, 12 children in this
where to attack; the aggressor also trains the vic- group were observed to interact at high rates with
tim how and when to reinforce the attack. This their peers and were victimized an average of 70
Machiavellian twist is achieved by a NR arrange- times. When they counterattacked, they were rein-
ment. If the victim complies, then the aggressor forced, on the average, 69% of the time by a ter-
immediately terminates his attack. In this manner mination of the attack. These NR arrangements
submission is reinforced by the withdrawal of the seemed, in turn, to be accompanied by an increase
aversive antecedent. This (hypothesized) sudden in the rate at which the children initiated their own
termination of hostilities on the part of the aggres- attacks. Children who were seldom victimized, or

141
who were not successful in their counterattacks, currences of Aj, the NR arrangement also pro-
showed little or no change in the rate at which duces immediate changes in behavior that are not
they later initiated coercive interchanges. Presum- related to the probabilistic connection between A;
ably, a similar process takes place in the home as and R;. Not only does NR produce a duality of ef-
well. fects (long-term and short-term), NR also pro-
A further contribution of NR concerns the key duces greater changes in the A;—»R; connection
assumption that physical violence among family with fewer trials. Given the presentation of highly
members is the outcome of a process whose com- aversive stimuli, a few training trials may produce
ponents consist of low-amplitude aversive events. escape or avoidance responses which are extreme-
The mechanism serving this process is negative re- ly resistant to change (Maier, 1961; Solomon &
inforcement. Wynne, 1954). Taken together, these characteris-
In each context the NR arrangement makes an tics make NR a powerful mechanism for behavior
important contribution to our understanding of change. In view of the potential it has for contrib-
social interaction. The arrangement is a variation uting to our understanding of family interaction, it
on a theme of pain control. In the matrix of ongo- is unfortunate that it has been ignored for so long.
ing social interaction, it has some things in com- Before addressing these issues, there is one fur-
mon with punishment. There are a substantial ther problem that must be discussed. When the
number of experimental findings which relate to format is applied to ongoing social interaction,
NR arrangements. The laboratory studiés of this there are some subtle changes in NR arrange-
phenomenon are straightforward. Typically, an ments. Because the aversive events are embedded
aversive stimulus is presented and is terminated within a sequence, the interpretation becomes
only when the animal performs the desired re- more complex than for NR analogues found in the
sponse. Contingent termination of pain has been laboratory. This increase in complexity occurs at
shown to be a powerful method for teaching a va- several different points. In laboratory studies, the
riety of responses. In these experimental designs, typical NR sequence begins by presenting an aver-
learning varies as a function of the intensity of the sive event continuously until the animal makes the
painful stimulus and the number of training trials escape response which has been preselected by the
(Hineline, 1977). The precursor for studies of this experimenter. In another type of design, the aver-
kind was clearly the escape and avoidance learning sive event is programmed to occur, but the animal
paradigm, with its long tradition in the history of can avoid this possibility by pressing a lever, de-
learning theories. The recent extensions of this laying the occurrence of the aversive event for a
work within the operant paradigm have explored certain period of time. Social interaction, of
the function of schedules of negative reinforce- course, typically does not have such an urgent
ment and their contribution to concurrent and quality to it, but there are some examples which
multiple-chain components (Hineline, 1977). This come close. One can think of a mother continuing
new work has established a solid empirical base to glare until the child closes the door, a harried
for the concept. spouse continuing to nag until her husband fixes
the faucet, a sibling crying until the mother comes
in and forces the older sibling to return a toy, or
Some Peculiar Properties of the mother using guilt induction until her son gets
NR Arrangements straight “A’s.” The most analogous situation in
There are a number of ways in which NR differs our data would be one in which a person pre-
from positive reinforcement (PR) in its impact sented a series of two or three aversive events until
upon behavior. For example, NR is associated compliance was obtained.
with stimulus reduction (withdrawal of A;), while There is, however, a fundamental difference be-
positive reinforcement can be thought of as stim- tween the NR arrangements found within the ma-
ulus producing. While NR arrangements are asso- trix of social interaction and the NR arrangements
ciated with significant reductions in physiological found in the laboratory. In social interaction, all
arousal, the studies reviewed by Fowles (1980) the components in the NR arrangement are con-
showed that positive incentives were related to in- nected only in a probabilistic sense. The mother
creases in arousal. does not always issue commands until the child
Both PR and NR arrangements have a long- complies; when she issues acommand (Aj), there is
term impact. The contingent arrangement a certain likelihood that the child will comply. If
strengthens the connection between the antecedent the child complies, there is a certain likelihood
and response; i.e., p(Rj|Aj) increases. However, in that the mother’s next response will be neutral or
addition to this long-term impact upon future oc- positive. How does this collection of fluctuating

142
 Figure 7.1
The Negative Reinforcement and Punishment Family

Event, Event, Event, Event, Event,

Subject Ry R,
Other person A, A, A; =
ODEs a
Punishment Sequence
beesiad: SE eee
NR Sequence

Effects of Punishment Effects of NR =

1 Weakens A,» R, 1 Strengthens A,—» R,
2 Fortuitous strengthening of A,» R, 2 Fortuitous weakening of A,» R,

conditional p values constitute a NR arrange- ping segments (A, through R2) of the same se-
ment? I think that when the social-interaction se- quences. Those who investigate punishments are
quence is: aversive —> response —->nonaversive, generally interested in the effect of A, in weaken-
then it is appropriate to think of it as a NR ar- ing the probabilistic connection between A; and
rangement. The utility of this convention lies in R,. However, if the analysis of the punishment ep-
the prediction that the Aj—» R; connection was isode were continued, it would include the NR seg-
strengthened. ment. In most punishment studies there is little in-
There is another sense in which NR arrange- terest in what happens next. When an animal sub-
ments become more dynamic when embedded ject is shocked for R,, it does something else (for
within the context of social interaction. In the rel- example, crouch down on the opposite side of the
atively fixed laboratory environment, the aversive cage); this is R,. Note this response, R2, is fol-
stimulus is either present or absent. This binary lowed by a termination of the shock. Some theo-
condition is in contrast to the situation found in rists, such as Guthrie, have explained punishment
social interaction. Here the escape response usual- effects in this way. The competing response, R2,
ly changes only part of the stimulus. For example, replaces the original response. As Church and Get-
following a successful counterattack, the family ty (1972) point out, an aversive event must always
member may still be present, even though he or terminate at some point. The immediately prior
she has stopped behaving in an aversive manner. response will then be reinforced. Given the occur-
The subject does not leave the situation; he rence of an aversive event then, by definition, NR
changes it. His behavior has altered the social-in- will occur for some response.
teractional setting; the matrix of events now mean When studying NR, the focus shifts from
something different than they did prior to his re- strengthening A, and R, to strengthening the
sponse. probabilistic connection between A, and R,. No-
When NR is embedded within social interac- tice, however, that punishment and NR have an
tions, it becomes obvious that some of the seg- effect on events beyond the segment which is typi-
ments which define it are also related to punish- cally at the center of the experimenter’s attention.
ment arrangements. Figure 7.1 describes an inter- What seems to be a simple NR process that
change between two people. The subject reacts strengthens A,—R, also weakens the A,» R,
with a sequence of two responses (R;’s) to what the connection. What seems to be a simple punish-
other person is doing (Aj). The other person ment arrangement that weakens A;—®R, also
changes the valence of his or her behavior from strengthens A, —>R,. This mutuality of effects de-
neutral A, to aversive A and then to either a posi- fines punishment and NR arrangements as mem-
tive or a neutral A;. As is shown in Figure 7.1, bers of the same family.
punishment and NR arrangements share overlap-

143
 / Table 7.1
Changes in Latency and Duration Following Reinforcement
(from Devine, 1971)

Group Reinforced Mean Duration (seconds) Mean Latency (seconds)

for: Pre Post F Pre Post F

Prosocial Behaviors 29.9 64.2 6.122 46.0 15.4 51815

Coercive Behaviors 1229 40.0 21579 60.3 16.6 DS

*p< .05
**y< 01

response strength. Devine assumed that reinforce-

The Efficiency of NR ment would increase the duration of a response
The first attempts to produce an experimental and decrease the latency of its occurence given the
test of NR revealed one particularly surprising presentation of the antecedent stimulus. These
finding. NR arrangements produced very rapid data indicated that as few as four trials were suffi-
changes in behavior. Devine (1971) created a lab- cient to produce significant changes in the rein-
oratory situation in which the mother initiated an forced responses. Negative reinforcement seemed
aversive behavior; the aversive behavior was then equally effective in changing prosocial and coer-
terminated contingent upon specific child behav- cive responses.
iors. The arrangement produced significant In discussing these results, Devine made the very
changes in the behavior of the child in only four important observation that when NR was em-
trials. This effect is in stark contrast to earlier bedded within the matrix of social interaction, it
OSLC studies of aggression in which up to 120 so- may have a double payoff. A response may be as-
cial reinforcers were required to produce even sociated both with the removal of the aversive an-
minimally reliable increases in aggressive behav- tecedent and with the positive consequence which
iors directed toward Bobo dolls (Patterson & follows (such as Attend, Talk, or Laugh). His
Cobb, 1971). In our treatment programs, positive point is well taken. Laboratory studies are needed
social reinforcers plus nonsocial reinforcers (such that test the impact of neutral and positive out-
as points and money) produced changes only in comes upon NR arrangements. Is NR more effi-
prosocial behavior and these changes were pro- cient than positive reinforcement in changing be-
duced over a period of weeks or months. havior? It seems to be, but as yet there have been
In Devine’s (1971) study, 22 mother-child pairs no systematic comparisons. Nevertheless, if this is
participated in a laboratory task. After a baseline the case, it may relate to the rapid shifts in perfor-
period, the mother was instructed to work on a mance level which describe some children.
task and ignore the child. Previous studies had
shown this to be highly aversive for preschool chil- The Reinforcement Trap
dren (Atkinson, 1971). For one group, the moth- This section is concerned with the multiplicity
er’s attention was made contingent upon the of effects that can be brought about by. introduc-
child’s prosocial behavior; e.g., if the child played, ing a single aversive event into an interaction se-
the mother immediately attended to the child. For quence. One of the most fascinating features of so-
the second group, the mother’s attention was cial interaction is the possibility of simultaneous
made contingent upon the child’s performance of a short- and long-term outcomes for a single event.
coercive response. Each trial was followed, in From the situation in Figure 7.2, one may obtain
turn, by baseline and free-play periods. There five different outcomes from a single exchange at
were a series of four training trials alternated with time-one (t,). In the example, the mother and child
baseline phases. Table 7.1 summarizes the data. both behave in such a way as to maximize short-
The two groups were compared on their baseline term gains. The child behaves in such a way as to
and post-test trials with respect to two measures of maximize the likelihood that Mother will termin-

144
 Figure 7.2
The Reinforcement Trap

Situation at t, Situation at t,

Mother If he Whines
Mother Mother less likely to Mother more
Scolds Talks Scold likely to Talk

Child Child Child more

Whines stops likely to Whine
Whining

Day 1 Day 2

ate her scolding. The mother behaves in such a Talk produced a termination of child-Whine. This
way as to maximize the likelihood that the child is in keeping with findings from the laboratory
will terminate his whining. What makes this inter- analogue study by Dengerink, Schnedler, and
esting is the fact that the short-term effects seem, Covey (1977). In a Buss-type shock situation,
to the participants at least, to be unrelated to the when the other member terminated attacks, sub-
long-term effects. For example, in the short run, jects quickly reduced their level of aggression.
the mother got “rid of” child-Whine but in doing Within family interaction does the coercer termi-
so, increased the likelihood that child-Whine will nate the attack within seconds of victim compli-
occur in the future! This lack of agreement be- ance? In a study of three mother-child pairs, Woo
tween short-term and long-term effects makes it (1978) found that during baseline the mean dura-
extremely difficult for the participants to under- tion of child-Whine was 28 seconds. Given an
stand what has happened. It is, in other words, an aversive stimulus followed by child-Whine and
ideal situation for behavior change without aware- mother-Comply, the child stopped whining in an
ness (see Chapter 5). It is this characteristic that average of 4.6 seconds!
leads the writer to describe this arrangement as a In the example above, the child reinforced (NR)
“reinforcement trap.” the mother’s shift from nagging to talking by ter-
The situation depicted in Figure 7.2 has a his- minating his whining. In the short term, both
tory. There have been many previous trials in members reduced the number of aversive events
which the mother nagged the child about his impinging upon them. It should be noted, how-
messy room. At t,, the scene opens with mother- ever, that the child’s room remained a mess. Be-
Scold; as previously rehearsed, the child begins to cause the mother gave in, the child won the battle.
whine. Normally this would induce the mother to The mother (who is the victim) lost because she
continue scolding (Patterson, 1977a), but for the opted for an immediate reduction in pain, not
sake of illustration, let’s assume that child-Whine knowing, of course, it was at the expense of an in-
effectively terminates mother-Scold. She now talks creased likelihood that messy rooms will occur in
to the child in a soothing manner. The first out- the future.
come of interest is that child-Whine produced I believe that this general scenario is repeated
mother-Talk. The second outcome is that mother- many times for parents of the Social Aggressor.

145
While the parents may repeat their demands sever- lowing frustration by a colleague. Those male sub-
al times, they eventually give in. At time-two (t2) jects who were permitted to aggress against their
the situation is repeated. However, there is techni- frustrator showed a more rapid decline in systolic
cally a slight reduction in the likelihood of the blood pressure than a control group that was not
mother nagging. This third outcome is the result given an opportunity to counteraggress. Bandura
of the child punishing the mother for nagging on (1973) cites three additional studies demonstrat-
the previous trial. Incidentally, most antisocial ing a similar effect. However, as he notes, this
children have trained their parents in this manner. arousal-reduction effect is not invariant. For ex-
As a group, these children are given very few ample, it was not found for female aggressors in
chores to do. Fourth, the child is more likely to the Hokanson and Edelman study, nor was the ef-
whine because it worked so well earlier. Fifth, the fect always reflected in measures of heart rate or
mother is more likely to give in if the child whines dystolic blood pressure.
because of the conditioning she received on the A variety of responses may terminate an attack.
preceding trial; i.e., he terminated his whining as For example, a friendly overture in response to a
soon as she submitted. The mother attends only to hostile initiation may serve this function. Such a
the short-term impact of her behavior upon the response, if successful, would also be accom-
child. In so doing she contributes to her own mis- panied by reduction in autonomic level.;The Ho-
ery. On future occasions, the child is even more kanson et al. (1968) study demonstrated that both
likely to behave in a coercive fashion. friendliness and counterattacks were accompanied
by rapid reductions in arousal if they were fol-
Arousal Reduction and Rigidity lowed by termination of the attacks. In this regard,
This section reviews two additional characteris- the methodological study by Szpiler and Epstein
tics of NR that may, or may not be interrelated. (1976) used shock as the aversive stimulus and
First, it seems that when a response is reinforced several measures of heart rate and electrodermal
by the withdrawal of an aversive event, it may also response were the dependent variables. All of the
be accompanied by a reduction in arousal. This re- arousal measures consistently reflected the impact
lation has some very interesting implications for of shock. However, only the measure of nonspe-
the clinical processes associated with aggression. cific skin-conductance responses reflect the arous-
The second characteristic is that some escape- al-reduction effect for the group of subjects who
avoidance responses become particularly resistant were provided with a nonaggressive means of
to extinction. This may relate, in turn, to some avoiding the shock.
problems frequently encountered in the treatment What is needed is a study that tests for a reduc-
of antisocial children. tion in arousal following successful counteraggres-
The studies reviewed in Chapter 4 showed that sion and demonstrates the strengthening effect of
a wide range of aversive stimuli produced auto- NR as it relates to future attacks. The study by
nomic arousal. These studies investigated the ef- Green, Stonner, and Shope (1975) found a con-
fect of shock and frustration on highly generalized nection between arousal reduction and the
measures of arousal such as blood pressure and strengthening of aggression. In that study, 90
skin conductance. Interestingly, one study showed males were either attacked or treated in a neutral
that antisocial people showed greater arousal (Ho- manner by a confederate. Then in the maze-learn-
kanson, 1961). It should be noted that recent ad- ing trials which followed, one group shocked the
vances in psychophysiology are suggesting the confederate (counteraggressed), a second group
possibility of three activating systems, one of which waited, and a third group observed the experimen-
reflects reactions to aversive stimulation, and an- ter attack the confederate. The counteraggressive
other of which reflects active coping responses to group showed the greatest reduction in dystolic
aversive stimuli (Fowles, 1980). According to this blood pressure during these second trials. Vicari-
new formulation, electrodermal activation may be ous counteraggression, or not responding, re-
a more sensitive measure of the arousal that is pro- sulted in less arousal reduction. All three groups
duced by an aversive stimulus. then participated in a third set of trials in which all
It is hypothesized that responses which produce of the subjects could shock a confederate who was
a termination of an aversive stimulus will be asso- working on coding problems. The group that had
ciated with arousal reduction. The case for this been the counteraggressors during the previous tri-
arousal-reduction hypothesis is only marginally als employed shocks of greater intensity on this
convincing, but sufficient to merit our attention. new task. Interestingly, they also expressed a
The study by Hokanson and Edelman (1966) greater dislike for the confederates in their ratings
found the expected elevation in blood pressure fol- of them. A greater reduction in arousal was

146
 associated with an increased likelihood for high- reasonable arrangement. Given that a substantial
intensity future attacks. share of coercive behavior maintenance is due to
Incidentally, this finding is contrary to some of NR arrangements, then only punishment would
’ the prevailing interpretations of catharsis theory alter the connection between the aversive antece-
which suggest that draining off anger is associated dent and the response. For child management pur-
with a reduced likelihood for future aggression. poses, extinction of coercive behavior is not a use-
One should accept the concept of arousal reduc- ful concept.
tion with some caution. However, these studies In the traditional studies of escape and avoid-
have important implications for the concept of ca- ance learning, a number of writers have noted the
tharsis. They should give pause to anyone enter- extreme rigidity of behaviors acquired under these
taining the notion that catharsis is a necessary arrangements (Maier, 1961; Seligman, 1975; Sol-
and/or sufficient condition for the treatment of omon & Wynne, 1954). In those studies, the ani-
antisocial children. The arousal-reduction data mals continued to display stereotypic escape re-
support the idea that letting the child have a tem- sponses long after the original conditioning trials
per tantrum will make him feel better. However, (Solomon & Wynne, 1954). Hineline (1977) re-
the side effects are just the opposite of what one viewed the extinction literature relating to shock-
wants. If the child “won” by using a temper tan- delay studies of NR. Leaving the animal in the set-
trum, then he will not only feel better but he will ting with the shock turned off eventually produced
be more likely to do it again. The catharsis theo- extinction for the lever presses which previously
rists accurately portray the fact that the child feels delayed shock onset. However, extinction effects
better after the tantrum, but these theorists ignore were not obtained for those subjects for whom
the reinforcement trap. shock was made noncontingent. These groups
In passing, there is another aspect of the arous- continued lever pressing for thousands of re-
al-reduction studies which is relevant to the ca- sponses when the response no longer delayed
tharsis concept. Catharsis theory suggests that shock onset. Animals trained in situations where
substitute activities, such as viewing aggression they could not avoid the pain also displayed stere-
that is directed against others, “drains off anger,” otypic responses persisting for many trials (Maier,
thereby reducing aggression. In his review, Ban- 1961). These studies reiterate the theme that NR
dura (1973) noted that arousal reduction was not arrangements may result in a rapid acquisition of
likely to occur if the counter-response was dis- coercive behaviors that are highly resistant to ex-
placed toward a target other than the tormentor; tinction. For treatment purposes, there are two
nor did it occur if the counterattack was expressed options. One is to reduce the number of aversive
as a fantasy. His review suggests that arousal re- events presented by family members. The other is
duction is maximized when the counterattack is to effectively punish each counterattack. The
immediate and directed toward the tormentor. OSLC treatment program emphasizes both com-
In Chapter 11, studies are reviewed which dem- ponents (Patterson, Reid, Jones & Conger, 1975).
onstrate that extinction is an ineffective method In extending these earlier studies, Seligman
for altering coercive child behaviors. In these un- (1975) noted that when the animals (from the es-
controlled treatment studies, it is difficult to deter- cape and avoidance studies) were later placed in
mine why nonreinforcement of aggressive behav- solvable escape situations, they showed them-
ior by parents is not accompanied by a reduction selves to be slow learners. The series of studies by
in rates of aggression. One reasonable hypothesis Nelson and Knutson (1978), as well as their re-
is that a behavior that is maintained by mixed view of the literature, showed that the metaphor,
schedules of NR and positive reinforcement may “learned helplessness,” has only limited generaliza-
be more resistant to extinction. As pointed out by bility when applied to aggression. They found that
J. Knutson, there is an extensive literature which reduced responsiveness was specific to the sex and
suggests that nonreinforcement is itself aversive. strain of the animal and whether or not the animal
Parent nonreinforcement may, therefore, induce had been housed in isolation or in a community
further coercive child behaviors. Extinction for cage. While the status of learned helplessness as a
NR would be equally complex. The parent would scientific concept may be in doubt, it is an appeal-
have to present an aversive such as scolding and ing metaphor. Clinical experience brings to mind
continue scolding until the child complied. Few the picture of some parents of aggressive children
parents could be so persistent in their presentation who long ago learned that nothing they could do
of an aversive. What this would really mean is that would work in their attempts to deal with the
the child’s noncompliance is being punished. In a problem child. Their expressed feelings of help-
social context, extinction for NR is simply not a lessness and stereotypical responses to the prob-

147
 Table 7.2
Aversive Antecedents Given that the Subject Performed a Coercive Response

Clinical Sample (N=33) Nonproblem Sample (N = 27)

Category N* p(Ai[Ri) N* p(Ai|R;)

CommandNegative 17 S37) 8 44
Cry 8 .34 — —
Dependency a sil 5 .20
Destructiveness 12 ae —— —
Disapprove 33 $8! 26 IPB)
High Rate 15 .34 9 gp
Humiliate 19 PHY 7 -00
Ignore iW .63 8 45
Negativeness 25 23 17 .16
Noncomply 32 gs) 25 16
PhysicalNegative 24 So) 14 .40
Tease Dil ashi 17 Vi
Whine 23 a5| 2 23)
Yell 19 36 8 ye!
Mean nee? 25)
* refers to the number of subjects performing the response. It was this N that was used to calculate p(A; > Rj).

lem child seem to fit the Seligman model. two different samples provided the data base. One
It is a fact that children’s aggressive behaviors sample consisted of 33 boys referred to OSLC as
are extremely difficult to change. The studies de- out of control; the other was a group of 27 non-
tailed in Chapter 5 suggest the difficulty of apply- problem boys matched for age, occupation of fa-
ing extinction procedures to alter coercive behavy- ther, family size, and father presence. Whether the
iors. It is an open question as to why this is the coercive behavior occurred within a chain or in
case. However, I believe that part of the difficulty isolation, the computer identified the antecedent
lies in the fact that NR arrangements are an im- event that was supplied by other family members.
portant source of maintenance for coercive behav- These A; were classified as aversive or nonaver-
ior. Given an aversive antecedent, you can comply sive. This calculation for the likelihood of an aver-
with it, punish it, or try to pretend it doesn’t exist. sive A; antecedent was carried out separately for
The field studies which demonstrated the failure each coercive response. The data are summarized
of extinction arrangements manipulated only posi- in Table 7.2. The two samples differed in several
tive reinforcers. Would the outcomes have been important respects. There were proportionately
different if they had focused only upon the child more children in the clinical sample who used De-
coercive response previously shown to be main- structive, Yell, Whine, Humiliate, Command
tained only by positive reinforcement? I don’t Negative, and Cry. Given that there was a coercive
think so; as noted earlier, extinction arrangements response, there was a slightly higher likelihood of
may evoke further coercion. What are the charac- an aversive antecedent for the clinical sample (.32)
teristics of responses maintained by mixed sched- than for the normal sample (.25). In effect, a siza-
ules of PR and NR? These are key questions for ble proportion of coercive child behaviors could
the next round of studies. be thought of as counterattacks, i.e., they were re-
actions to noxious intrusions. In effect, for both
Descriptive Data normal and clinical samples the stage is frequently
set for the possibility of aNR arrangement. Clear-
The Likelihood of an Aversive ly, then, much of children’s aggression is not at-
Antecedent tack but counterattack behavior.
First, what proportion of coercive child behav- There were also dramatic differences among the
iors have aversive antecedents? Data collected for responses. For both samples the coercive re-

148
 Table 7.3
Likelihood of Aversive Antecedents for Isolated or Initial Coercive Events

p(A;|Coercive Sequence) for F Values for

Sample Younger Older Mean Sample Age Interaction
Normals 33 HL9 .26
(10) (9)t Mepipt 1.94 1.18
SocialAggressors 55)9) 133 34
(15) (15)
t+Numbersin parenthesesreferto samplesize.

Table 7.4
Likelihood of “Start-Up” by Family Agents

Mean likelihood of p( —| +) by sample

Social
Dyads Aggressors Stealers Normals F Values

- +
p(Mother|Child) 0S .04 .02 ROO meas
~ +
p(Child| Mother) .04 .03 .01 835°
N 37 36 Sif
- +
p(Father|Child) .04 .02 01 Sie
- +
p(Child|Father) .02 .02 .00 1.58
N i 26 26
- +
p(Sibling|Child) .03 .03 01 Sey
- +
p(Child|Sibling) 02 .02 01 1.67
N 36 35 33
*p~<.05
DSA Vil
p00

sponses of Physical Negative, Ignore, Command comparisons were based upon data from 30 fami-
Negative, and Tease were most likely to have aver- lies of 30 socially aggressive boys and 20 families
sive antecedents. of 20 normal boys. In the former, the chains from
It is conceivable that the OSLC figures for aver- younger (3.1 to 8.5 years old) and older (8.5 to
sive antecedents were distorted by the six-second 13.5 years old) coercive boys provided the data
chunking procedure built into the code.? For this base. The data are very similar to the findings
reason, a second analysis was conducted employ- based on six-second units. Note that the differ-
ing a “natural” unit. Here the entire uninterrupted ences between samples were of only borderline sig-
sequence of coercive behaviors was tabulated as a nificance.
chain. Only the antecedents for the chain were an- Now the question should be turned around. We
alyzed (see Appendix 5.1 for details). Table 7.3 know that roughly one-third of the child’s coercive
summarizes the likelihood for aversive antecedents behavior was evoked by an aversive antecedent.
for these chains—p(Aj|Coercive Sequence). The However, what is the likelihood that an A; will be

149
 Table 7.5
Aversive Antecedents for Counterattacks in Distressed Families

p(ChildinitiateR;\A;)
Hit Tease Disapproval Whine

Aversive
Antecedent p(Ai) (.010)t (.011) (.061) (.011)

Command .085 .000 .000 .028 .011

Disapprove 035) .000 _ .200* 10325
Hit .007 276" — .141 —
Humiliate .003 — _ ran —
Tease .006 .060* nlaoe .209 “=

Whine .002 — nl60* -160* —_

Yell .004 -— .077* _ —

Talkt .367 .060 .007 .074 019

+The figures in parentheses are base-rate values for p(R;).

“In the Patterson and Cobb (1973) report, where conditional p values are listed, these A; were identified as controlling stimuli for that
target response. The term, controlling, means that the Aj was accompanied bya significant increase in the likelihood of the target re-
sponse.
+Although Talk is not considered an aversive antecedent, it has been included for comparison purposes.

presented to him? To answer this question, data mean that NR mechanisms play an important role
were examined for samples of Normals, Stealers, in maintaining coercive child behavior.
and Social Aggressors (see Appendix 6.2 for a de-
The Likelihood of Counterattack
scription). The data in Table 7.4 were tabulated
separately for each dyad. Given that the target As Bandura (1973) noted, the optimum condi-
child’s antecedent had been prosocial (+), what tion for eliciting an aggressive reaction is simply to
was the likelihood that a family member would do something aversive to the other person. For ex-
crossover and launch an “unprovoked attack?” The ample, interactions in a residential treatment set-
figure in row one column one showed that in the ting showed very high likelihoods of counterattack
Social Aggressor sample 5% of the mother’s inter- (Raush, 1965). His study of disturbed and nondis-
action with the problem child fit into this category. turbed boys showed that for both samples the like-
The data for the clinical sample showed that lihood of an unfriendly act following an unfriend-
each family agent was significantly more likely ly initiation by the other person was about .80.
than their normal counterparts to launch such at- Aversive behaviors elicit aversive reactions.
tacks. Mothers were more likely than fathers or The likelihood of attack and counterattack is
siblings to engage in crossover attacks. Note, too, determined by a complex set of variables, includ-
that the problem children were invariably less like- ing setting, age, and sex of the other person. For
ly than the parents to initiate a conflict. When example, in their review, Maccoby and Jacklin
compared to Normals, problem children were two (1974) noted boys were more likely than girls to
to three times more likely to cross over. This is in retaliate (counterattack).
keeping with the earlier findings of Raush (1965), Several investigators have found an interesting
who showed that this variable differentiated interaction between the type of provocation and
hyperaggressive from normal boys observed in a the type of counterattack. In a nursery school set-
residential setting. He found that crossover was ting, Hartup (1974) found that if the target child
five times more likely to occur in the interactions was insulting or derogatory, a large proportion of
of the clinical sample than in the normal sample. the younger child’s counterattacks were likely to
It seems, then, that a child from clinical samples be Hit. With the same provocation, only 22% of
must learn to cope with noxious intrusions by oth- the older child’s counterattacks were Hit; 78%
er family members. If his coping behaviors suc- were reactions in kind.
cessfully terminate these intrusions, this would To study this among family members, Patterson

150
 Table 7.6
Likelihood of Counterattack by Sample and Agents

Mean likelihood of counterattack

Social

Dyads Normals Aggressors Stealers F Values

p(Mother|Child) 15 25 .16 4.12*

p(Child|Mother) .09 oH) .16 Z2ORS25

p(Father|Child) 08 14 .10 233

p(Child|Father) 08 .18 ails; 4.67*

p(Sibling|Child) 19 ell .28 2.84

p(Child|Sibling) 10 sf lo) SESaae
*p<.05
**p<.01
“p< .001

and Cobb (1973) analyzed the antecedents for the were taken from all points within coercive chains.
initiations of a coercive response. Events were Given that the other person was coercive, what
classed as initiations if the child had previously was the likelihood that the target family member
been engaged in 18 seconds or more of prosocial would counterattack? As shown in Table 7.6, the
behaviors. For example, given all prosocial and likelihood of counterattack varied as a function of
deviant initiations, the base rate for the initiation sample and dyad. In normal families the figure
of Hit was .010 and for Disapproval was .061. ranged from .08 to .19, the range in the clinical
The first column of figures in Table 7.5, headed by samples was .10 to .28. In all comparisons the
p(A;), lists the base-rate values for the antecedents problem children in clinical samples were signifi-
for the four most frequent target responses.* For cantly more likely to counterattack than were nor-
example, for all initiations the most likely A; mal children.
would be Talk. Only the functional relations
thought to be significant were listed in Table 7.5. Utility of NR
Each coercive response was controlled by slight- The key assumption is that coercive child be-
ly different networks of aversive antecedents. Giv- havior may serve the important function of ter-
en that a family member Disapproved, the prob- minating aversive intrusions by other family mem-
lem child’s most likely counterattack was to recip- bers. Buss (1966a) states that the major reinforcer
rocate in kind. The conditional probability for for attack-instigated aggression is termination of
Disapprove, given Disapprove, was .200. If a fam- the attack. Do counterattacks by problem children
ily member Teased, the most likely aversive re- serve this function? Just how effective are these be-
sponse was to Disapprove (.209) or Tease (.149). haviors?
Hitting functioned as a means of coping with fam- The first analysis of this problem was based on
ily members’ Tease and Hit. If a family member six-second units of data from clinical and nonclini-
Hits, the most likely counter was for the problem cal samples. In what proportion of these sequences
child to respond in kind (.296). In the 1973 analy- was the counterattack followed by a favorable
sis of Normals, the comparable value was .200. In outcome (i.e., the interchange was terminated by a
distressed and nondistressed samples alike, Hit positive and/or neutral outcome)? The findings
evokes Hit. are summarized in Table 7.7. The first column
The likelihood of counterattack was calculated identifies the number of subjects in each sample
separately for each dyad involving a target child who actually engaged in one or more counterat-
for samples of Normals, Stealers, and Social Ag- tacks. The second column gives the mean propor-
gressors (see Appendix 6.2 for details). The events tion of successful outcomes for each response. On

151
 Table 7.7
Favorable Outcomes (Utility) for Counterattacks in Two Samples

Distressed Sample Nondistressed Sample

Number of Number of
Child’s subjectswith Mean subjectswith Mean
Counterattacks Ai> R; episodes p(C* |A;> R;)t A,> Rj episodes p(C* |A;> R;)

CommandNegative 17 30 8 38
Cry 8 30 —_ oa
Dependency i} 36 .00
Destructiveness 12 a29 0 .00
Disapprove 33 a5 26 48
High Rate 15 ao 9 .00
Humiliate 19 .16 .00
Ignore 17 .68 8 AS
Negativeness 25 29 17 .28
Noncomply 32 43 25 31
PhysicalNegative 24 .43 14 wZS
Tease 27 35 bz 41
Whine 23 41 12 my)
Yell 19 537. 8 36
tThe likelihood of a positive and/or neutral outcome for a counterattack.

Table 7.8
Successful Final Outcomes for Counterattacks

Normals Social Aggressors F values for:

Agent(s) Younger Older Younger Older Group Age Interaction

Likelihood Successful

All family members

% Ge 40 .62 42 Sy 1671 0.95 2.48*
YorGe 23 222 py) 31 0.31 0.28 0.59
Total (C+ +C°) .63 84 .64 .68

Likelihood Negative Reinforcement

All family members 16 alle) Pal aoe Teh Tes 0.08 0.30

*p<.05
**p<,001

the average, the counterattacks were about 30% this? Why do we allow some kinds of counterat-
effective in producing successful outcomes. The tacks to work but not others? Do these utilities
general trend was for a slightly greater likelihood vary as a function of age and sex of persons in the
of success for counterattacks when employed in dyad? These are questions for the next round of
distressed families. studies.
As shown in Table 7.7, the major differences in Again, there was a possibility that the six-sec-
utility were among the coercive responses. Disap- ond chunking used in the code distorted the find-
proval functioned effectively about half the time, ings. The utility score could have been con-
while Humiliate was relatively ineffective. Why is founded because each score summed across out-

152
comes for the different components of a chain four NR pairings significantly strengthened the be-
such that outcomes early in the sequence are havior on which it was made contingent. The ef-
added to those at midpoint, and so on. The next fects were significant for both prosocial and devi-
analysis focused only on the final outcomes for ant behaviors. But to provide a closer fit to this
chains. The data were based on an analysis of 30 discussion of NR it would be necessary to intro-
socially aggressive children and target boys from duce two refinements in the experimental design
20 normal families (see Appendix 5.1). The analy- used by Devine (1971). First, it was an analogue
sis also compared the likelihoods for younger and study carried out in an artificial setting; can these
older subjects as they interacted with other family results be generalized to the home? The second
members. The findings are summarized in Table problem involves the means by which response
7.8. The final outcomes for counterattack chains strength was measured. While the latency and dur-
were successful about 40% of the time. This figure ation measures were germane to the general issue,
did not vary significantly as a function of sample they are not directly relevant to this discussion. As
or age. stated here, a test of the NR hypothesis requires a
The next question required that two different direct measure of the change in the probability of
pieces of information be combined into a single in- R; given Aj;. For the present, nothing is known
dex. One piece describes the proportion of coer- about the relation between duration, latency, and
cive responses having an aversive antecedent. The p(R;|A;) as measures of response strength.
second describes the proportion of this subset John Knutson and his colleagues at the Univer-
which had a successful outcome. The index, Neg- sity of Iowa have studied pain-elicited aggression
ative Reinforcement, combines these two vari- with animal subjects. The dependent variable that
ables. As shown in Table 7.8, there was a signifi- they employed met the present requirements, and
cantly larger proportion of coercive behavior their design provided a powerful test of the rela-
maintained by NR arrangements in the distressed tion of NR to aggression. The victim was re-
samples than in the nondistressed samples. For strained. If the experimental subject was shocked,
normal children, about 15% of their coercive re- but the victim was not, there was a very low likeli-
sponses could be said to be maintained by NR hood of an attack. Typically, the attacks were
schedules. For the clinical sample, the comparable most likely to occur when both the experimental
figure was about 21%. subject and the victim received inescapable
The descriptive data support the assumption shocks. During five consecutive trials, both ani-
that NR arrangements do occur in family interac- mals received a shock; if an attack on the re-
tion. They are particularly likely to occur in a clin- strained victim occurred, then the shock was ter-
ical sample. As a general rule, they are surprisingly minated; i.e., NR occurred. Then, in a subsequent
effective in terminating the noxious intrusions of trial, anew victim was introduced; this victim was
other family members. Asa result, it seems plausi- allowed to move freely about the cage. Given a
ble that NR may, indeed, play an important role shock, the probability of attack was significantly
in maintaining coercive behavior. The next ques- higher for the NR-trained group. In a second
tion is, do these NR arrangements strengthen an- study, the training sessions involved a freely mov-
tecedent —->response connections? ing victim, and the test condition involved a re-
strained victim. The effects were replicated. Nega-
Experimental Manipulation tive reinforcement arrangements strengthened at-
of NR in Situ tack behavior.
The fact that aversive events and NR arrange- The next question is whether such NR arrange-
ments occur in family interactions does not mean ments serve a similar function in family interac-
that they actually control behavior. Only experi- tion. Woo (1978) carried out the first such study.
mental manipulations can demonstrate that an After several days of baseline observation data
event or an arrangement actually exerts significant were collected in the home, the mother and child
control. This section reviews the evidence for the participated in two days of experimental manipu-
contention that altering NR contingencies will al- lations. During these sessions the mother received
ter the probabilistic connection between the aver- telephone calls (from an experimenter). When a
sive antecedent (Aj) and the coercive response (R)) call came, she told the child: “I’m going to be busy
or p(Rj\A;). It is assumed that NR arrangements on the phone now; please don’t bother me, okay?”
will increase this likelihood (this is analogous to She then talked on the phone while an observer
the earlier discussion of positive reinforcement). collected data on the likelihood and duration of
The results from the study by Devine (1971) the child’s whining.* On each day the pair partici-
summarized in Table 7.1 showed that as few as pated in an ABA reversal design. During the first

153
 ‘Table:7.9
Experimental Tests of NR for Mother-Child Interactions
p(Rj|Ai)
Subjects Arrangement BL, Manipulation BL,
Woo(1978,p. 19)
Chris p(Child Whine|Mother on Telephone) 05 33 48
Marty p(Child Whine|Mother on Telephone) .08 Va .28

Poy p(Child Whine|Mother on Telephone) PP, oh) 225)

Mean bz 32 34

OSLC Replication

Autumn p(Child Complain|Mother Restrain) 538 .67 oF

Pumpkin p(Child Provoke|Mother Command) “its 30 .04

baseline the mother received two calis, during ference was 3.72, significant at p< .01; (df=16).
which time she talked for up to 15 minutes. If the The mean duration for his Complain was 6.98
child whined during this time, she waited until the seconds. On each of two days an ABA design was
child had stopped whining for 12 seconds and followed; each condition lasted 20 minutes. Fol-
then hung up the phone. If calls that were received lowing condition A, a period of normal interac-
during the experimental phase led to whining, tion, the mother was instructed to use the proce-
then the mother was to hang up the telephone im- dures she had previously practiced with the experi-
mediately. Over the two days, each child received menter. She was cued to restrain Autumn; when
about 12 NR pairings during the experimental he complained, she was to behave in a prosocial
phase. In the Woo study, each of the three subjects manner. There were approximately 10 of these
showed a dramatic increase over baseline levels in pairings each day. The data in Table 7.9 summar-
the probability of Whine given the telephone call ize the findings. As shown there, the predicted re-
(see Table 7.9). Note also that the conditional p inforcement effect was obtained (on both days).
value for the post-test did mot return to Baseline, During the experimental condition, given that the
levels. mother restrained, Autumn was increasingly more
The Woo study provided data that was relevant likely to complain. The mean p value of .67 for
to the study of NR arrangements in family interac- the two days during which the manipulations took
tion. However, the experimental procedure intro- place was larger than the mean baseline value of
duced an aversive event that was atypical for par- .33. The binomial Z comparing these values was
ent-child interaction. The next studies were de- 4.24 (df=35; p< .001). The hypothesis that (neg-
signed to use events that are commonly found in ative) reinforcement arrangements may have an
mother-child interaction. They also studied coer- immediate effect on functional relationships be-
cive child behaviors other than whining. Note, tween responses and controlling events was sup-
however, that all of these studies share in common ported by these findings. As was the case in the
the “confound” noted by Devine (1971). Within Woo (1978) study, the effects persisted into the
the matrix of social interaction, a NR arrange- (20-minute) second baseline period. This differ-
ment is very likely to be accompanied by a positive ence between the first and second baseline periods
reinforcer. f was also significant, as shown by the binomial Z
Autumn was a 5-year-old boy from a middle- value of 1.89 (df=13; p<.10).
class family (see Appendix 5.1) The baseline data The data for Pumpkin describe a similar experi-
showed that when Autumn was restrained by his ment for a second mother-child pair. The ABA de-
mother, he was likely to complain. The condition- sign was again followed on each of two consecu-
al probability of Complain, given Restraint, was tive days. When Mother gave a Command, and
.33. The corrected base rate for Autumn-Com- Pumpkin behaved provocatively, the mother either
plain was .15. The binomial Z testing for the dif- reacted positively or ignored it. The experimental

154
phase was associated with increases in p (Pump- known and enters directly into the correlation.
kin-Provoke|Mother-Command). Does Disapproval get more reinforcement than
This brief series of studies showed that NR ar- does Hit? The assumption is that if this is so, it
rangements did increase the likelihood of coercive will occur more often. If NR is, in fact, the ar-
child behavior.. Even though positive reinforce- rangement determining the strength of the re-
ment components may also have been involved, I sponse, then there should be a significant correla-
think that NR arrangements were responsible for tion between p(R;) and p(NR) among the coercive
the bulk of the experimental effect. This assump- responses being compared.
tion requires further testing. The observations To study the variance associated with differ-
made in homes showed that these arrangements ences in response rates, it would be useful to have
occur with sufficient frequency to warrant being some prior knowledge about which aggressive re-
taken seriously as a major mechanism relating to sponses are primarily controlled by NR and which
deviant child behavior. are not. For example, if all 14 coercive responses
are controlled by mixed schedules of NR and PR
Variance Accounted for by NR arrangements, then neither arrangement by itself
It was not expected that NR components would would correlate with p(R;). The analysis by Patter-
contribute directly to the problem of accounting son and Dawes (1975) employed a replicated
for the variance associated with individual differ- Guttman Scalogram analysis for observation code
ences among subjects. For reasons discussed in variables to identify a transitive progression for
Chapter 5, it was not expected that there would be the following coercive categories: Disapproval,
a linear relation between p(NR) and the strength Noncomply, Negativism, Tease, Physical Nega-
of the coercive response. The analysis in Patterson tive, Command Negative, and Humiliate. The
(1979b) showed that, indeed, there was no correla- writers suggested that the underlying dimension
tion between p(NR) and TAB scores. While NR shared in common by these seven variables might
increases p(R;\ Aj), this has little to do directly with be that all were controlled primarily by NR ar-
overall performance level. As shown in Chapters 8 rangements. A. Harris and J. Reid (in preparation)
and 11, the performance level for deviant behavior replicated this progression for the classroom. Pat-
is determined primarily by the disposition of other terson (1979b) then showed that the response hi-
family members to provide stimuli which start up erarchies for these seven coercion variables covar-
conflict and stimuli which accelerate conflict once ied with the richness of their NR schedules. The
it starts. In effect, individual differences in perfor- correlation for one sample was .59 (n.s.) and for
mance are determined by variations in the density the second, .93 (df=5; p<.05). Differences in
of the stimuli which control that behavior. Day- rates among these seven coercive responses varied
by-day fluctuations in performance are also as a function of their usefulness in terminating
thought to covary with shifts in the density of con- aversive intrusions. Those working most effective-
trolling stimuli. It is NR (and PR) arrangements ly (e.g., Disapproval) tended to occur most often.
which give these controlling stimuli their power;
Escalation
however, NR contributes only indirectly to indi-
vidual differences in performance. One of the key assumptions in coercion theory
In social interaction the main direct contribu- is that the analyses of processes comprised of in-
tion of NR arrangements is to account for the vari- nocuous, garden-variety aversive events will lead
ance associated with differences among aggressive to an understanding of physical violence among
responses. The reader might well ask why NR family members. Wife and child beatings are
should account for variance relating to differences thought to be the outcome of processes that have
among responses but not for variance relating to been set in motion for some time. The progression
differences among subjects or sessions. The gener- is thought to move in a transitive fashion from
al formulation is in keeping with the earlier state- low-amplitude to high-amplitude aggression (e.g.,
ment of this matter (see Chapter 5). To under- from threaten, yell, throw objects, push, hit, hit
stand what implications the probability of rein- with fist, to hit with an object). Each increment in
forcement has for a given response, it is necessary intensity has been reinforced by the submissive re-
to compare it to the schedules provided for com- action of the victim to the prior increase in ampli-
peting responses. The base-rate values for re- tude. These increases in amplitude were thought
sponses sum across subjects, thus surmounting the to be most likely to occur during extended coer-
individual differences problem. In the case of a set cive interchanges. A recent analysis by Loeber
of responses, the information for the probability (1980) showed that hitting was, indeed, signifi-
of reinforcement for each relative to the other is cantly more likely to occur in coercive chains of

155
longer duration. Surrender or compliance by the How is it they end up attacking their own chil-
victim constitutes a NR, increasing the likelihood dren? How do middle-class, nonaggressive cou-
of future high-amplitude attacks. One further as- ples come to use physical violence? These and re-
sumption is that escalation in families is most like- lated questions serve as the focus for the following
ly to occur when the participating members are discussion.
equally likely to employ pain-control techniques,
Violence in the Home
i.e., they are of equal power. Escalation is ex-
tremely difficult to study in field settings. The In our culture violence has been given a central
problem lies in the difficulty of defining the dimen- role as a problem-solving technique. However, it
sion of intensity as it relates to social behavior.*® is generally thought that violence is not a tech-
One could take a static-trait position and as- nique that is used in the home except by a few ex-
sume that angry people marry angry people, and tremely aggressive persons (Forster, 1966). The
angry mothers produce aggressive children; i.e., age of innocence concerning conditions in the
these traits existed prior to the formation of the American home is rapidly coming to an end. The
family. From this position, beatings are simply the extensive survey studies carried out by M. Straus
logical outcome of a Markov process describing and his colleagues suggest that violence in the
interactions among people with a prior disposition home is as likely to occur as love (Steinmetz &
to be aggressive. This might well be true. How- Straus, 1974; Straus, Gelles, & Steinmetz, 1979).
ever, there are published findings describing physi- As pointed out by Steinmetz and Straus (1974),
cal aggression as a concomitant for divorce pro- “Most parents feel not only that it is okay to hit.
ceedings among otherwise nonaggressive people. In the case of children, it is more than just okay.
For example, Wallerstein and Kelly (1980) found Most American parents see it as an obligation
that in the process of separating during divorce, . and this carries over to the relationship be-
about one-half of the middle-class couples studied tween husbands and wives” (p. 3). They also
engaged in physical aggression. These people cite a study by Stark and McEvoy (1970) which
tended not to be aggressive either before or follow- showed that about one parent in four takes the ex-
ing the separation. I believe that this may be an ex- plicit view that there are circumstances in which it
ample of a process in which relatively nonaggres- is acceptable for one spouse to hit another. Straus
sive people can become physically aggressive. The goes on to review other studies which lend support
increases in rates of coercive behavior and in ex- to his claim that a marriage license is a hitting li-
tended coercive chains would place them at in- cense. According to his studies, one couple in ten
creasing risk for escalation. Given that both claim that hitting is acommon experience. For ex-
spouses have a disposition to be aggressive, then ample, he cites the in-depth interview study of 80
the escalation process would undoubtedly be ac- families by Gelles (1973). About 60% of the fami-
celerated. lies described at least one instance in which physi-
What is novel about this formulation is that cal force was used between spouses. Margolin
people who were initially nonaggressive can find (1979) cites studies asserting that 16% of Ameri-
themselves trapped by the escalation process. For can couples engage in physical violence at least
example, I have worked with several teachers who once a year; for 10% of the couples the inter-
were overwhelmed with guilt because they had changes were extremely violent. It is, therefore,
lost their tempers and struck children in the class- not surprising to find that among lower class ap-
room. Many of the parents referring their antiso- plicants for divorce 40% mention physical abuse
cial children to OSLC are gentle, nonaggressive (Levinger, 1966). The comparable figure among
people. They perceive both themselves and their middle-class participants was 22%. Marital vio-
child as being under the influence of forces beyond lence is not selective; it occurs in all classes.
their control. Some parents express extreme guilt It is acommonly shared assumption among in-
because of the beatings they have given their chil- vestigators in this area that fighting between
dren. A study of children referred to OSLC spouses often gets out of control. Occasionally
showed that about one-third of the parents of anti- beatings escalate to homicides. In keeping with
social boys have physically abused their children this assumption, there are a dramatic number of
(Reid et al., 1979). Was this the outcome of a homicides among family members. Field and Field
gradual progression in which discipline was in- (1973) showed that during 1965, killings within
creasingly supplemented by physical assault? By the family represented 31% of all murders. In one-
definition, mothers who beat their children are ag- half of these instances one spouse was the victim
gressive, but many of them will tell you that they and the other spouse was the murderer. The vio-
have never physically assaulted another person. lent character of spouse fighting is well known to

156
the police, whose task it is to intervene in these do- intensity. These synchronous increases in intensity
mestic quarrels. Twenty-two percent of police have been well documented in the laboratory stud-
deaths and 40% of their wounds result from at- ies reviewed by Dengerink (1976). Over a period
tempts to intercede (Singer, 1971). of weeks or months, the level at which coercive in-
Familial violence is, of course, not limited to al- terchanges begin increases in intensity, and/or it
tercations between spouses. Helfer and Kempe very quickly moves to a higher level during ensu-
(1968) brought the battered child to national at- ing interchanges. In a given episode resulting in
tention. In the national survey involving 1,000 hitting, the dyad may move quickly through sever-
families by Gelles (1977), 58% of parents re- al increases. If one member is a well-trained ag-
ported slapping or spanking their child during the gressor, he may move so quickly through a se-
past year. Thirteen percent reported hitting the quence that the victim is hardly aware that the
child with an object; 3% attacked the child by confrontation has begun.
kicking, biting, or pushing. It might be interjected Probably the best source of descriptive material
here that many of these battered children are for such interchanges is to be found in the clinical
themselves extremely coercive. Several observa- study by H. Toch, Violent Men (1969). He inter-
tion studies of abused children revealed them to viewed policemen who had a history of frequent
have significantly higher rates of coercive behav- violent assaults and prison inmates who were in-
iors than normal children (Burgess, 1978; Reid, carcerated because of their violent behavior. The
Taplin, & Lorber, 1981). These findings suggest policemen described a process that began when
the possibility that the child may contribute to the police officer perceived the suspect as having a
family processes in which he or she is beaten. “negative attitude.” The policeman then reacted to
Straus et al. (1979) makes the point that violence this with a blunt demand for immediate compli-
is an accepted part of family life in this culture. ance. From the perspective of the suspect, the offi-
Eighty to ninety percent of parents said that some- cer’s tone of voice and physical gestures may be in-
times physical violence involving the child may be terpreted as a threat, leading the suspect, in turn,
necessary. Most parents do not allow siblings to to overt noncompliance. Thus, the assault-prone
fight it out and interfere regularly in sibling quar- officer may produce noncompliance in an other-
rels. However, some of the parents referred to wise compliant suspect. The suspect’s noncompli-
OSLC report that they sometimes allow a good ance is a real stimulus, to which the officer may
fight to run its course so that the children can “get justifiably respond by using physical force, e.g.,
it out of their systems.” In one survey study, mod- push the suspect over the hood of the car. This fur-
ern practitioners of the catharsis theory found that ther exacerbates the situation, leading to a scuffle,
62% of high school seniors had struck a sibling clubbing, or shooting. Similar progressions char-
during the past year (Straus et al., 1979). Thirty- acterized both officers and inmates with records of
five percent had also hit someone outside the fami- assault.
ly. It is interesting to note that the official records In either case, the individual in our culture has
indicate that most violent crimes against persons learned that when things get out of control, a rap-
are committed by older teenagers. The follow-up id escalation in violence may provide a solution to
study by Wolfgang (1977) showed that violent de- the situation. In this manner, over time and en-
linquents were those with a prior history of many counters, NR shapes the individual to become in-
police offenses. The analyses in Chapter 11 sug- creasingly violent. It may also be the case that an
gest that these delinquents are likely to have been increasing spectrum of cues become signals for
children who received little parental monitoring threatening situations. Both Toch (1969) and Ber-
and ineffective punishment for antisocial behav- kowitz (1978) emphasize that for violent people it
ior. I think that much of society’s problem with vi- is their perception of a situation as being threaten-
olence is due to disruptions in the processes of ing that is most often the antecedent for an as-
family management. sault.
Berkowitz (1978) interviewed 65 white males
Escalation Process
with long histories of violence. The great majority
As used here, the term escalation refers to (83%) had prior convictions for crimes; two-
changes in intensity over time. When the coercive thirds of these convictions were for violent behav-
interchanges between members of a dyad move ior. The attacks were most likely to occur on
from lower to greater intensity, the increments are weekends, at night, and 85% were related to
likely to be small. First, one person increases the drinking. The incident was most likely to start
intensity of his or her attack, and the other eventu- with an argument or a perceived insult by another
ally reciprocates by responding at an equally high person. When provoked, the goal for most in-

157
 Figure 7.3
Two Components of Escalation

Prior history of
NR for high
intensity attack

External
p(High-intensity attack)
event

Attribution of
hostile intention
to other person

Low self-esteem

mates was to hurt the other person. This is in or 24 hours, i.e., a variant of Parent Daily Report.
keeping with the findings from laboratory studies During the probe, the informant is asked for infor-
of victim pain reactions as a positive reinforcer for mation about setting, antecedents, and outcomes.
the angry, extremely antisocial person (see Chap- This approach could easily be expanded to include
ter). information about anger and attribution.
In both of these clinical studies, the first stage Straus (1973) has gone on to provide what he
was generated by an insult or an argument. If this believes is a quantifiable model for describing es-
is the first step in a sequence, then it may be possi- calating family processes. He particularly empha-
ble to disrupt the ensuing steps by defusing the sit- sizes positive-feedback loops supporting the up-
uation. That possibility was investigated by Sykes ward spiral in violence. He also emphasizes vio-
and Brent (1978). In a situation where an officer’s lence as a systems by-product. In one analysis,
request was met by the suspect’s noncompliance, families were classified according to the power of
the officers were trained to simply repeat the re- the wife and the husband. This score was based on
quest in a civil manner. Their field study showed information about who made decisions such as
that this simple expedient increased the likelihood which car to buy, how to spend family income,
of compliance and reduced violence. where to vacation, which house to buy, how the
The clinical studies by Berkowitz and Toch sug- children were disciplined and whether or not the
gest that at some point early in the interchange, wife worked. In families where the wife was more
the violent person is more likely to attribute a ma- powerful than the husband, the mean violence di-
levolent intent to the behavior of the other person. rected by wives to husbands was highest. The
This attribution is accompanied by autonomic wives’ violence score was lowest in families where
arousal and an otherwise simple argument be- the husbands’ power was highest. Husbands, on
comes a potential battleground. These people all the other hand, tended to be more violent to their
believed that they were provoked. As shown in wives under either extreme of power differential;
Figure 7.3, the variables of negative attribution i.e., either the wife or husband was much more
and anger seem to be likely candidates as determi- powerful than the spouse.
nants or concomitants of assaults. There is, then,
Some Formulations about Escalation
a need to investigate these two variables empirical-
ly. But how is it one can obtain measures of anger The foregoing clinical studies and our own ex-
and attribution in situ? My colleague, J.B. Reid, is perience with over 200 treated cases emphasizes
exploring the possibility of using participants’ re- two interrelated components for the escalation
ports of critical incidents occurring during the pri- process. It is assumed that the first increase in in-

158
 tensity comes about during an extended aversive tions which are frequently coercive place the dyad
interchange. This is the first component. I think at greater risk for NR based increases in ampli-
_ these extended chains may also be accompanied tude. The question is how does this relation be-
by increasing anger and hostile attribution. The tween frequency and intensity come about?
second component concerns the relation between The beaten wife or child has probably been a
NR and intensity changes. Given a successful out- participant in prior escalating interchanges. Both
come for an increase in amplitude, NR increases the wife and child have probably increased the am-
the likelihood of a high-intensity response with fu- plitude of their attacks and counterattacks prior to
ture presentations of that same stimulus. the first beating. The first beating was one addi-
In a normal family, if the parent escalates and tional step in a process which had been going on
spanks the child, the child is likely to submit. Nei- for some time.’ If both members of a dyad are dis-
ther the parent nor the child are likely to escalate posed to react in an irritable fashion (see Chapter
during this trial or during the one which follows. 8), then the dyad is at greater risk for escalation.
However, if the parent of the socially aggressive Given a system where escalation is an accepted
child tries to spank, he or she is likely to be met means for problem solving, then physical violence
with anything but submission. The child may re- is a predictable by-product. Incidentally, our re-
spond in a synchronous fashion by hitting back. cent work in treating child abusive families con-
The reaction may be a temper tantrum, loud yell- vinces me that there may be several progressions
ing, or whatever, but it is very likely that the child leading to violence in families, in addition to the
will respond in some high-amplitude coercive escalation model. There seem to be some families
fashion. The formulation by Straus (1973) about in which the child and wife use very little coercion
power differentials can also be applied to this situ- but still are beaten by the father.
ation. One can certainly think of the socially ag- It is also assumed that once a high-amplitude re-
gressive child as being “powerful.” I think the sponse is well established in the repertoire, it may
powerful person is more likely to counteraggress; be evoked under an increasing range of situations;
he or she is also more likely to provide a synchron- i.e., the individual is likely to “explode” at school,
ous increase in amplitude. A pilot analysis by J. as well as at home. I would expect that some of the
Reid of TAB scores for mother and child sup- violent inmates and policemen studied by Toch
ported this hypothesis. In a related study, Reid and Berkowitz also beat their wives and children.
demonstrated that in Normal and Stealer samples Why do some individuals develop a general dispo-
80% to 90% of the mothers tended to have higher sition to quickly escalate to violence in many situa-
TAB scores than did their target children! Howev- tions while others are violent in only one or two
er, a larger proportion of dyads in the Social Ag- settings? The studies reviewed in Chapter 4 estab-
gressor sample showed the child to be higher than lished the correlation between anger, arousal, neg-
the mothers. This asymmetry, noted by Reid, may ative attribution, and the likelihood of attack.
be extremely important because a more recent What is now required are studies demonstrating
study by Loeber and Reid (in preparation) showed that these variables relate to high-amplitude at-
that this variable related to the child’s later becom- tacks and the disposition to generalize across
ing an assaultive adolescent. In that retrospective many settings.
study, those families studied at OSLC whose chil- The Toch-Berkowitz formulation also stresses
dren later showed up in police files as assaultive, the contribution of low self-esteem to this process.
Stealers, or “clean” were examined for variables From their viewpoint, it is the person with low
which might differentiate among them. The pat- self-esteem that is most likely to be caught up in
tern for the 12-year-old who was later identified as the anger /negative-attribution cycle. This seems
assaultive was to have the highest TAB score like a sensible idea that is difficult, but not impos-
among family members. Relative to other family sible, to test. It is included here because it fits what
members, these children occupy a powerful role. many of the OSLC families have described.
A recent study by Reid, Patterson, and Loeber While speculating about ideas that have (as yet)
(1981) showed that families in which the members little empirical support, there is one further note
perform coercive behavior at high rates (TAB that may be of use. Is it possible that a short fuse,
scores) are more at risk for hitting. The hypothesis or anger related arousal itself, are subject to rein-
was that the frequency of coercion covaries with forcement? Attack reactions that are accompanied
intensity. The analysis showed correlations in the by high levels of arousal are probably more likely
range of .4 to .7 between TAB and Hitting scores. to be successful. The recent work in biofeedback
The correlations were consistent across family provides an interesting method for testing this pos-
members and samples. This suggests that interac- sibility. If this formulation is true, then the arousal

159
 Figure 7.4
Escalations in Intensity During Extended Chains
(adapted from Patterson, 1980, pp. 31-33)

Younger sibs

—=-=-=-@ Older sibs

Rating
Aversiveness
Mean Mothers

Events in Sequence

itself may have been shaped by the same contin- drugged, the attacks were not likely to occur in
gencies which shape the assaultive responses. An- PEA procedures.
ger may then be a concomitant rather than a deter- These studies are of great interest because they
minant for assault. Are low boiling points and as- relate to the irritability hypothesis described in
saults shaped by similar NR contingencies? Chapter 8. As shown there, it is the general dispo-
sition of both members to react in an irritable
Trained Fighters Escalate More Quickly fashion that determines the child’s performance
The early researchers who studied animal ag- level. I think that this disposition also relates to es-
gression in laboratory settings had commented on calation. Cairns and Nakelski (1971) carried out a
how rapidly trained fighting mice escalated in in- classic series of studies which relate to this issue.
tensity (Lagerspetz, 1964). Ulrich, Johnston, The first study established that isolated mice were
Richardson, and Wolff (1963) also made this ob- more vigorous in their explorations when later in-
servation in their studies of pain-elicited aggres- troduced to cage mates in their home cage. The
sion (PEA). If both the aggressor and the victim vigor of their greeting was reciprocated by equally
had been previously trained, then the fighting intense reactions for the normally-reared mice.
quickly escalated to high-intensity attacks. The re- These vigorous interchanges were associated with
action of the victim may also contribute to the es- frequent escalations to physical violence. Cairns
calation process. The reviews by Knutson (1973) reasoned that the likelihood of such escalation
and by Cairns and Scholz (1973) showed that the from vigorous greeting to fighting would be great-
intensity of victim reaction to the attack correlated est if both members had previously been isolated,
with the likelihood of escalation to yet higher in- less if one member had been isolated, and least of
tensities during future attacks. At the other ex- all if neither had been. For the three groups, the
treme, when the victim was nonreactive or data showed the incidence of fighting to be .33,

160
 Table 7.10
Escalation in Intensity of Coercive Sequences as a Function of Prior Pay Off
Likelihood that reaction of mother to prior chain extension was:

Base-ratep Mother Mother Mother

third event “backed off’ “backed off” remained Mother

longer than from aversive from high to the same increased Reinforcement
Subject secondevent Nonaversive to nonaversivelower aversive aversive aversive effect

Summer 49 (.43)t foul BIL7/ .05 .03 24 +

Autumn 30 (.24) 74 Lye .00 .04 .00 50
+Corrected base rate.
+Probability that the child’s coercive Event 3 was longer than the child’s Event 2, given that Mother “backed off.”

.18, and .00 respectively. It may be that the iso-

lated mice ignore cues from the other mice which Experimental Tests of NR and
would normally dampen an escalation process. Escalation
They may also fail to provide such cues. I’m not The key assumption is that NR arrangements
sure that “vigor” is directly analogous to “irritabil- can produce across-trial increases in the intensity
ity,” but the Cairns studies describe a process of attack behavior. Thus far, there are only two
which quickly gets out of control that bears more single-subject studies which directly test this hy-
than a metaphorical relation to escalation in fami- pothesis. For these studies, it was assumed that a
lies. NR arrangement would increase the likelihood
In the context of family interaction, Patterson that the child’s next coercive event in the sequence
(1980a) hypothesized that the problem child was a would be of higher intensity or of longer dura-
more highly trained aggressor than the mother. If tion.® The duration of the coercive response
this is the case, one would expect problem chil- served as the major dependent variable (increases
dren to escalate the intensity of their coercive be- in rated intensity for an event were a less frequent
haviors more quickly than mothers. Fifty-nine ex- form of escalation in intensity).
tended coercive chains consisting of four or more Four to six hours of baseline data were scored
events in sequence served as the data base (Patter- for two preschool children, Autumn and Summer.
son, 1980a). Mothers were involved in two-thirds The protocols were scored twice. First, intensity
of these chains. The aversiveness for each com- ratings were assigned to each coercive event in the
ponent was rated in each extended chain. The interactions of these preschoolers with their moth-
mean ratings for younger and older siblings and ers. The intensity ratings were based on the mean
mothers as they tried to cope with the problem ratings of the code categories provided by mothers
child are shown in Figure 7.4. of normal children (Jones, Reid, & Patterson,
The initial reaction of the problem child (not 1975). Then, each coercive chain for the child was
shown in Figure 7.4) was to immediately escalate scored for duration and/or intensity changes by
to almost maximum intensity (between 7 and 8) six-second units. If the child escalated in intensity
and remain there. The writer assumes such rapid during the second coercive event in a sequence,
escalation is characteristic of well-trained aggres- what was the impact on the mother? The data in
sors in general and problem children in particular. Table 7.10 describe the outcomes for the various
The rapid escalation of the younger and older sib- stages of the interchange. First it is necessary to
lings is due to the relatively sophisticated training know what the base-rate likelihood would be that
to which they have been subjected. The ANOVA the third coercive event would be longer than the
for repeated measures showed an F value of 11.17 second. In general, how likely were the children to
(p<.001) for groups and an F value of 5.13 escalate in intensity? The p values were .43 for
(p< .003) for trials. The question that is yet to be Summer and .24 for Autumn. The mother’s most
answered is why did the mothers escalate so slow- likely reaction to such increases was to remain
ly? nonaversive (.51 for Summer’s mother and .74 for
Autumn’s).

161
 Given that the mother had been abrasive and quickly escalated the amplitude of their attacks on
the child increased the duration of his counterat- the doll. One can turn this idea around as Cairns
tack, how likely was it that the mother would had done with mice and immobilize the other
back off? If the mother backed off, the child, in ef- member of a dyad in a situation which might oth-
fect, had “won.” If negative reinforcement func- erwise elicit attacks. Given a nonreactive victim,
tions as it should, then winning should be associ- there was not only no escalation, there were fewer
ated with an increased likelihood that ensuing attacks. This same effect was obtained in the stud-
events could be of greater intensity. This next ies by J. Knutson described earlier.
event is number three in the sequence. As can be What are the other variables which might pre-
seen in the last column in Table 7.10, for Autumn program one member of a dyad in such a way as to
the conditional probability was .50; this com- alter the likelihood of escalation? Medication for
pared favorably to the corrected base rate of .24. the mother or child might produce a decrease in
At a microsocial level, the effect of the NR ar- the risk of escalation, but it is, at best, a short-
rangement was to increase the likelihood that the term solution. In the long run, the family would be
next trial would be of greater intensity. Escalation better served by procedures which would perma-
not only “worked” on a short-term basis, but, in nently change the system. For example, the obser-
addition, it altered the p values for ensuing events. vation study by Minton, Kagan, and |Levine
Note, however, the long-term impact of “escala- (1971) showed that mothers were most likely to
tion —>win” was not strongly reflected in the data escalate the amplitude of their punishment, given
for Summer. The conditional p value was only that the first punishment was ineffective. This
.55, as compared to the corrected base rate of .43. makes sense; if the first punishment doesn’t work,
Two single case studies do not constitute a con- they try harder! This suggests that when the child
firmation of the hypothesis, especially when one engages in punishment acceleration, the parent
of them (Summer) gave only marginal support. As may escalate in amplitude. In fact, a study of
they stand, the analyses illustrate the fact that abused children showed that they were more likely
family interaction data can be used to test such hy- to accelerate when punished than were other anti-
potheses. A heavy reliance upon the use of dura- social children or normals. It also suggests that
tion as a dependent measure for intensity change training mothers to use a punishment (such as time
does not seem satisfactory until studies demon- out) that works the first time will reduce the likeli-
strate that duration changes are equivalent to in- hood of escalation.
tensity changes. The new OSLC coding system
provides measures of changes in intensity. This Implications
improved technology should make it possible to When the victim complies, the attack is termi-
construct a more conclusive test of the NR escala- nated. The price for submission is that there is an
tion hypothesis. increase in the likelihood of future attacks. If,
What are some of the parameters which increase however, the victim counterattacks, then the inter-
the likelihood of escalation? The best source of change may escalate in intensity. What is one to
data and ideas on this topic are to be found in the do? The data from laboratory studies suggest if
programmatic work by Cairns and his colleagues one person /owers the rate of attacks, then the oth-
(Cairns, 1979a,b). One of the parameters which er person is likely to follow suit (Dengerink,
they have identified was discussed earlier. They 1976). Lindskold (1978) reviews the evidence sug-
found that mice being reared in isolation were gesting that in game-theory settings, conciliatory
more vigorous in their explorations of new cage efforts and declaration of cooperative intent re-
mates and also reacted more vigorously to their duces conflict. However, findings reviewed in this
explorations. He also cites the study by Hall chapter suggest that while backing down or sub-
(1973) which demonstrated that it is possible to mission on the part of one person reduces conflict
manipulate the likelihood that one member of the over the short term, it also increases the likelihood
dyad will use high-amplitude aggression. This, in of future attacks. If backing down is followed by
turn, alters the likelihood for the other member. an increase in amplitude, then the future may well
Dyad interchanges can be manipulated by prepro- bring more high-amplitude attacks. A permissive,
gramming the behavior of one member—a simple accepting stance by the victim does not work; nei-
idea, but a powerful means for exploring the esca- ther does nattering or half-hearted counterattacks.
lation hypothesis. In the Hall (1973) study, one If the coercive process is well under way, there
child was exposed to a film portraying high-ampli- are several things that must be done. First, the sit-
tude aggression. When he later joined another uations producing the conflict should be changed.
child to play with an inflated Bobo doll, the pair If the exchanges are between equals, as perhaps in

162
 marital conflicts, then the differences must be ne- about the wisdom of encouraging couples to fight
gotiated, as detailed in the social learning ap- as a means of changing relationships. I think that
proaches to marital conflict, e.g., Jacobson and we should teach families how to control conflicts
“Margolin (1979). In the case of parents and chil- and at the same time teach them humane ways of
dren, much of the conflict may be about ambigui- changing each other’s behavior.
ties in the definitions of what is and what is not ac- At an even more speculative level, I would like
ceptable behavior. For them, it may be necessary to consider the role of NR as it relates to guilt in-
to begin by clearly spelling out the rules. However, duction, and socialization. There is one pattern
given an antisocial child, it is more the case that he which I have observed in middle- and upper-mid-
or she does not accept the fact that the parent has dle-class homes. They use a category of aversives
the right to control the child’s behavior. If the not listed in the OSLC code—guilt. These guilt in-
child dominates the situation, then NR arrange- ductions are swiftly done and are so subtle that of-
ments are given full play. ten they can only be tracked by the family members
Second, the parents must learn to stay out of co- themselves. For example, an implicit assumption
ercive interchanges, to ignore them as much as is that if Family Member A performed more per-
possible. When they do get involved, they must fectly, then Family Member B could relinquish re-
use (nonviolent) punishment, and they must win sponsibility for making Family Member A feel
each time they do so. As outlined in Chapter 6, the guilty. When this is accomplished with a smile or a
punishment “should” be nonviolent in nature, joke, the visitor is hardly aware that it has taken
e.g., time out, work details, or loss of privileges. place. Aversives of this kind must be extremely dif-
The point is that high-rate, intense aggression is ficult for a child to learn to cope with. I think that
not amenable to warm, accepting treatment. As they may relate to cerain neurotic patterns. In any
the studies in Chapter 11 demonstrate, coercive case, guilt induction would be a fascinating vari-
child behavior does not change as a result of ex- ant of NR to study. What does the child learn to
tinction arrangements (withdrawing social atten- do in order to terminate that kind of discomfort?
tion), nor is it reduced by successful efforts to One of the things which has always intrigued
shape prosocial behaviors. Extremely antisocial me is the apparent “sameness” of people after they
behavior must be punished to be changed. The al- pass through adolescence. Over long periods of
ternative is to live in a family system in which NR time their clothing changes, hair color changes,
plays a dominant role in determining the quality of and they have new job titles. However, their social
interaction among family members. behaviors remain surprisingly constant. The long-
The implication is that for families of antisocial term follow-up studies of adults seem to corrobo-
children fighting is bad. For them, fighting is a rate this impression. I think that as adults we only
process which often gets out of control. It becomes change in reaction to massive intrusions, usually
a system run by NR arrangements rather than by crises of some kind. However, even here some per-
the participants. This position on fighting is in sons react to crises by escape and avoidance rather
marked contrast to the current emphasis on “let- than by learning new skills. By running away these
ting it all hang out.” Prestigious publications, such people fail to learn new skills and new perspec-
as Bach and Wyden (1968) emphasize fighting as a tives. NR strengthens the avoidance behavior of
necessity for a couple’s survival; e.g., couples who those who run away. It can also strengthen the de-
fight together are couples who stay together, pro- velopment of new coping skills for those who stay
vided that they know how to fight properly. It may to confront the situation. It is our clinical impres-
be that there are some deeply inhibited people who sion that many of the families of antisocial chil-
cannot say that they need some facet of their rela- dren use escape-avoidance techniques: “If it’s un-
tionship changed. Perhaps fighting is a means for pleasant, stay away from it; pretend it doesn’t ex-
such a person to change a situation in which he or ist, and it will go away.” The alternative is to solve
she is miserable. However, even there, one won- the problem and come up with a means of reduc-
ders at how often fighting leads to physical vio- ing the likelihood of future crises of the same kind.
lence. Out of 47 million married couples, 3.3 mil- But problem solving and looking toward the fu-
lion wives and .25 million husbands experience se- ture are perspectives which seem difficult for these
vere beatings from their spouses (Margolin, distressed families to assimilate. Why is this?
1979). These are appalling statistics; it means that I believe that, in general, the role of NR (and
married couples constitute a population at risk for punishment) has been underestimated in most
violence. Straus (1978) and his colleagues present modern social learning theories. For example, it is
equally impressive statistics for the occurrence of possible that children’s prosocial behaviors are ac-
violence in families. He, too, shares a concern quired as much as a means of avoiding/escaping

163
parental scolding than as a result of positive rein- functional relations. In Table 7.2, some events
forcement for success. Field studies of the acquisi- tabulated (such as Aj; occurring during an ex-
tion of prosocial behaviors may very well demon- tended coercive chain) would be thought of as an-
strate that NR plays an equal role in socialization tecedents for the next component in a coercive
with positive reinforcement. My own impression chain. It is not known if an Aj occurring prior to
is that adults provide rich sources of modelling the first coercive event in a sequence can be
and positive reinforcement for infants and pre- thought of as being similar to an A; that is an ante-
school children but that, by school age, positive cedent for an event occurring midway in the chain.
reinforcement schedules become very lean. That is One way of clarifying this problem would be to
not to say they are insufficient; that remains to be analyze only those Aj occurring prior to the first
determined. Until these much needed studies be- coercive event in a sequence. Data were prepared
gin, I think we should stay open to the possibility in such a format for the Patterson and Cobb
that, for the socialization of older children, pain- (1973) study. These data were tabulated to gener-
control techniques play a role equal in importance ate the probability of A;|Coercive Initiation. The
to positive reinforcement. probability value for the distressed sample was .38
and for the nondistressed sample was .28 (Patter-
son, 1976). Those values were based on popula-
Acknowledgment tions of events; the findings summarized in Table
The writer gratefully acknowledges the careful 7.2 were based on samples of subjects.
critique of this chapter by John Knutson. His in- 3. Note the conditional probability for coun-
terattack is p(R;\ Aj). This, in turn, is functionally
cisive comments led to numerous changes in the
earlier version. As it now stands, there remain related to the likelihood of an aversive Aj, given
that a coercive response has occurred—p(A;)R)).
some areas in which John would accuse me of be-
ing “soft”; but this was as close as I could come to To derive the latter from the former, all that is re-
his hard-headed, Iowa position.
quired is knowledge of p(A;) and p(R;). For exam-
ple, regarding the data in Table 7.5, application of
Bayes’ theorem would show .20 of the antecedents
for the child’s Hit were someone else’s Hit (Phil-
lips, 1973). In effect, knowledge of his particular
antecedent accounts for a good deal of what it is
Footnotes
we need to understand about the problem child’s
1. Hineline (1977) discusses some of the rea- hitting. (See Chapters 8 and 9 for a more extensive
sons for choosing this term rather than using one discussion. )
drawn from the more familiar escape/avoidance 4. During baseline phases, if the child did
paradigm. The Skinnerian psychologists have a whine, the mother was instructed to wait until he
general tendency to define terms in such a way as stopped for 12 seconds and then hang up. In so
to clearly differentiate themselves from investigat- doing, Woo avoided the danger of strengthening
ors working within alternative frameworks. In the Whine but exposed himself to the alternate prob-
long run, these distinctions may not be useful. For lem of strengthening a competing behavior. In ret-
example, the perspective taken by Hineline and rospect, it would have been better to commit the
others is to focus on the impact of NR on the re- error leading to the most conservative interpreta-
sponse as an operant. The dependent variable is tion of the data; e.g., have the mother hang up the
the rate of responding. This de-emphasis on the phone if the child whined at baseline.
S-R aspect of avoidance learning differentiates the 5. The nature of these studies was explained to
operant from the Pavlovian approach. As stated the mother, including the fact that she could be
earlier, the present writer believes that, in the long asked to do something that would for a short time
run, we must understand the details of the process make her child more obnoxious. At the comple-
by which stimuli acquire control over responses. tion of the studies, a careful debriefing was carried
For this purpose, p(R;|A;) seems to be a more ap- out, including a discussion of the data itself if the
propriate dependent variable. parents were interested.
2. In the first study, the analyses were based on Many parents’volunteered for these studies, less
six-second units; i.e., the code system sampled an for the considerable amounts of money involved
event and a reaction to the event every six seconds. than for the possibility of receiving treatment. Five
The effect of this decision was to split some con- of the six families in the extended baseline series
tinuous events, such as Cry, into six-second epi- went on to receive treatment at OSLC. Interesting-
sodes. This decision undoubtedly distorts some ly enough, the therapists often remarked upon

164
 how quickly these families responded to interven- has been described by Megargee (1967) in his
tion. There was a good deal of speculation about studies of “overcontrolled aggression.” He believes
whether being a subject in these sample experi- that for some violent people the antecedent is not a
‘ments would be an effective preparatory proce- gradual escalation in amplitude; rather, it is a
dure; i.e., it teaches the parent that her behavior process of rumination and isolation. This pro-
has an immediate and powerful impact. duces a disruption of inhibitions against killing,
6. There are two options to the investigator i1.e., a cognitive rather than dyadic escalation
choosing to work with intensity changes. On the process.
one hand, he can ask victims or participants to 8. In the present discussion, an intensity in-
rate the intensity of these events. This is the sim- crease referred to either the amplitude of the event
pler approach and the one adopted in these pre- itself (volume of yelling, intensity of the blow, or
liminary analyses (see Chapter 4). A second and increase in general aversiveness), or it consisted of
more difficult approach is training the observers to extending the duration of the event; e.g., 30 sec-
scale events for their intensity. J. Knutson (person- onds of crying is considered a more intense reac-
al communication) carried out pilot studies in tion than 10 seconds of crying. A response fol-
1979 demonstrating the feasibility of this ap- lowed by the termination of a long chain should be
proach. The new code being developed at OSLC strengthened more (due to NR) than a response
will also attempt to measure intensity. followed by the termination of shorter chains.
7. There must, of course, be several paths by Similarly, termination of a high-intensity Aj;
which an individual moves to high-amplitude ag- should be a more effective NR arrangement than
gression. One of the more interesting alternatives termination of a low intensity Aj.

Perspective on Chapter 7
by Dr. John F. Knutson
In analyses of both human and nonhuman be- tions the conditions for negative reinforcement ex-
havior, a pattern of increasingly intense interac- ist for both members of the dyad. Since escalated
tions characterizing extended aversive exchanges and nonescalated aggressive exchanges do not
is commonly reported. Described as escalation by continue indefinitely, both members of the coer-
several authors, it seems crucial for any theory of cive dyad are exposed to negative reinforcement
aggressive behavior to provide an explanation for regardless of whether escalation occurs. In nonre-
this pattern. The present chapter develops the the- ciprocated coercive exchanges, orily one member
sis that negative reinforcement provides a suitable of a dyad would be negatively reinforced (the vic-
account of the escalation that occurs during ex- tim) and the other member might be positively re-
tended aversive dyadic exchanges, and as the inforced. Given an arrangement in which some
chapter documents, there are many reasons that negative reinforcement is always occurring in co-
make a negative reinforcement analysis of escala- ercive exchanges, to invoke negative reinforce-
tion very appealing. However, from my perspec- ment as a causal factor in escalation it is necessary
tive, negative reinforcement provides an incom- to show that there is some differential effect of
plete analysis of escalation. negative reinforcement on some persons (escala-
One of my primary concerns regarding the via- tors) or on some coercive responses. The assess-
bility of a negative reinforcement analysis of esca- ment of different utility functions for the various
lated coercive exchanges is that whenever a recip- counterattacks described in this chapter is consis-
rocated coercive exchange occurs, negative rein- tent with the possibility that some counterattacks
forcement has to occur. That is, when one mem- (reciprocal responses) have greater utility than
ber of a dyad provides an aversive event to the oth- others, and that they could be differentially rein-
er member of a dyad, and the second member re- forceable and differentially affected in escalated
ciprocates with an aversive, as soon as the ex- exchanges. However, one of the complexities that
change terminates or shifts to nonaversive interac- concerns me is the implicit assumption in the utili-

165
ty analysis that social stimuli are relatively stable to be negatively reinforced by shifts to nonaversive
or transsituationally consistent with respect to interactions or the termination of exchanges.
their positive or negative valence. Since basic la- Work in my laboratory with nonhuman subjects
boratory research with nonhuman subjects docu- suggests that the density of noxious stimulation in
ments that stimuli are not transsituationally effec- a finite period preceding an aggressive sequence is
tive as reinforcers or punishers, it is my position a factor that relates to evoking an escalated re-
that social stimuli are also not likely to be transsit- sponse. Thus, an analysis of the length of a dyadic
uationally equally reinforcing or equally aversive. exchange (time or number of behaviors) and some
Thus, it seems that the relationship between nega- weighted mean or weighted moving average of the
tive reinforcement and the escalation hypothesis aversiveness of preceding antecedents might be in-
will need further research regarding the utility cluded within a negative reinforcement model of
functions of particular reciprocated attacks given escalation.
specific antecedents. Obviously, such analyses will One of my primary concerns about the negative
require enormous amounts of data to assess se- reinforcement analysis of escalation is the position
quential patterns in altered conditional probabili- that negative reinforcement of coercive behaviors
ties. I believe such data will have both theoretical contributes to intraindividual differences among
and practical significance as intervention strategies responses to stimuli but is not a major factor in in-
and further research on aggression are considered. dividual differences in the likelihood of respond-
My second concern regarding the present nega- ing with escalated coercive behaviors. I agree that
tive reinforcement analysis of escalation is that negative reinforcement can contribute to intraindi-
there is the implication that it can account for the vidual differences in responses to stimuli; howev-
development of escalated patterns when it is actu- er, the contribution of negative reinforcement to
ally a model that accounts for the maintenance of individual differences in coercive patterns is also
escalated behavior. To provide a complete analysis important. Research in my laboratory (again with
of escalation, the negative reinforcement model nonhuman subjects) has demonstrated that nega-
should be expanded to consider factors that cause tive reinforcement can result in differences among
the emergence of escalated exchanges that are then groups in aggressive responding under standard-
negatively reinforced. One area of research that ized aggression test settings. It seems to me that
might be conducted is an assessment of the num- the analysis presented in this chapter could be ex-
ber of behavioral units (reciprocated dyadic ex- panded to include negative reinforcement of esca-
changes) that must occur before a dyad displays an lation contributing to differences between deviant
escalated pattern. I also believe that there needs to and nondeviant groups. Since Patterson and his
be some consideration of individual differences in colleagues have demonstrated that the ambient
reactivity to social exchanges. That is, it seems level of noxious exchanges is greater in socially ag-
plausible that the number of presentations of so- gressive families, the probability that negative re-
cial aversives required to push a deviant child over inforcement will occur is greater and the possibili-
the threshold for engaging in more intense ex- ty of reciprocating effects of negative reinforce-
changes would be less than the number of aver- ment leading to an increase in group differences
sives required for a nondeviant child. While nega- seems obvious.
tive reinforcement can account for an escalated In summary, I would like to see the negative re-
pattern on succeeding interactions, it suggests a inforcement analysis include a greater considera-
trial and error or random process account of esca- tion of the development of individual differences
lation on initial interactions. I would also like to as well as an expansion of the analysis to consider
see a consideration of the possibility that some in- the role of the antecedent events contributing to
dividuals are more reactive to the negative aspects the escalated pattern that is then negatively rein-
of social exchanges, and therefore are more likely forced.

166
 RE ET Xo piehein ied
*
va i te ‘

Aroniegine: Use Pree

tT iaiot
i NigaeD
algae
-/choheheohbe
tie
PaO ¥ a
THHATE TewsoytT

aaa ce) WWr

id oan « i

al

4 UL Ate Ae

ateie rev :
ee

Dtuat yin Tix; «

Aer ee tat ayety
RRROLA
wipires j

a bthatiel
 Chapter 8
Abstract
As described here, interactional psychology consists of two components: structure and process.
Structure is defined by the functional relations between events adjacent in time. One form of struc-
ture, the intrasubject, is defined by the functional relation between a subject’s own behavior at one
point in time and his subsequent behavior. Intrasubject structure describes patterns or series of the in-
dividual’s reactions that tend to repeat themselves. The second form of structure is intersubject. Here
the functional relation is between one person’s behavior and the other person’s reaction to it.
Process refers to changes in structure. It is assumed that there are mechanisms within interaction
that can alter either inter- or intrasubject structure. Amolecular focus upon structure and process de-
fines what is meant by a microsocial analysis. It is assumed that such an analysis provides an alterna-
tive means for understanding some forms of deviant behavior.
It is also assumed that some aspects of microsocial process are altered by events that impinge from
outside the dyad. In effect, some microsocial variables may be altered by macrosocial variables, such
as crises or stress.
Findings from observation data and from experimental studies are reviewed as they relate to these
issues.

168
 Chapter 8

A Microsocial Analysis of
Structure and Process

“ee
. . . physics does not endeavor to explain nature. to build an understanding of something as com-
In fact, the great success of physics is due to a re- plex as family interaction. However, at this stage
striction of its objectives: it only endeavors to ex- in the development of social-interactional psychol-
plain the regularities in the behavior of objects. ogy, it seems wise to begin with simple two-event
This renunciation of the broader aim, and the or three-event patterns. How can variance in per-
specification of the domain for which an explana- formance levels be accounted for by measures of
tion can be sought, now appear to us an obvious these simple patterns? Once we know this, we can
necessity. In fact, the specification of the explaina- proceed to more complex patterns and determine
ble may have been the greatest discovery of phys- empirically how much additional information is
ics so far.” (Wigner, 1964, p. 995, cited by Fiske, needed. Another consideration is the fact that the
1974, p. 10). more complex the pattern, the larger the data set
required to establish it.
If one closely examines the minutiae describing Functional relations are not immutable; they
the interchange between two persons, it becomes change over time. If data are collected repeatedly
apparent that certain sequences of events tend to in the same setting, then it is possible to describe
be repeated. For present purposes, the focus will these changes precisely. This makes it possible to
be upon events that are immediately adjacent in know when a change has occurred and to begin
time. A tabulation of event sequences will identify the search for the variables that produced the
some events that are reliably associated with other change. In this sense, a microsocial analysis is also
events. In the discussions that follow I will use the concerned with process.
terms functional relations, covariation, and pat- There are two major components of interaction-
tern to describe this probabilistic relation between al structure. One is the intrasubject and the other
one event and another. These patterns define what the intersubject component. Intrasubject refers to
I mean by structure in social interaction. The as- the disposition of the individual to continue what
sumption underlying this approach to studying he is doing. Given that he is talking, there is a bet-
family interaction is that an empirical analysis of ter-than-chance likelihood that this same behavior
these structures will contribute to our understand- will reappear in the sequences that follow. As we
ing of aggression. shall see, this is a significant feature for members
The analysis of only two or three adjacent of families of antisocial children. Once they react
events might seem too limited a base from which in an irritable fashion, they are likely to continue.

169
The intersubject component refers to the fact that persons. The general stance had been described in
some aspects of the individual’s behavior are cor- hortatory volumes by Kantor (1959) and others.
related with the behavior of the other member of Furthermore, investigators in ethology were al-
the dyad. In this type of dyadic exchange the other ready doing these kinds of studies (Hinde, 1974).
person consistently reacts with B whenever the They had laid down an impressive empirical foun-
subject does A (i.e., the pattern of B given A, (BA) dation for this assumption, at least as it applied to
recurs). Again, there are certain intersubject com- primates, birds, and fish (Tinbergen, 1951). For
ponents that are of particular interest. For exam- example, Altmann (1965) observed the social inter-
ple, in his study of normal and aggressive preado- actions of rhesus monkeys. His code sampled the
lescents, Raush (1965) concluded that the variable entire repertoire of their social interactions. The
that differentiated them was the disposition to uncertainty of prediction, given just the informa-
start a conflict. Given that the other child was be- tion for the number of code categories in the sys-
having in a prosocial manner, the aggressive child tem, was 6.91 bits. The term bits is a logarithmic
was more likely than his normal counterpart to expression of uncertainty which, in turn, relates to
start a conflict. the number of alternatives and their likelihoods of
Normally the probability values describing a occurrence. As shown in Figure 8.1, given data for
functional relation fluctuate about a mean value. the base-rate values for the categories (approxima-
The likelihood of B|A may increase in one session tion 1), there was a reduction in uncertainty of
and decrease in another. However, for many func- 2.09 bits. Further information about immediate
tional relations, the mean value based upon three antecedent events produced a comparable reduc-
or four sessions in one week will correlate well tion in uncertainty. Information about two im-
with a mean value based upon three or four ses- mediately prior antecedents produced an addition-
sions during the following week; i.e., the mean is al reduction of about the same magnitude. The
stable. I think that many of the day-by-day fluctu- more information available, the greater the ability
ations about this mean are determined. The ques- to predict. In fact, the increase in predictability
tion is, what causes the functional relation be- seems linear as one moves from just information
tween event A and event B to shift from one day to about base rates to first and second antecedents.
another? Under what conditions will the mean We were further heartened by Raush’s (1965)
value itself shift from a zero slope and begin to re- work. He collected data for disturbed boys and
flect long-term increases or decreases? showed that the immediately prior social behavior
As a general case there are two sets of variables accounted for 30% of the variance in the perfor-
that I believe are associated with stable changes in mance of ongoing social behavior.
structure. One set has to do with stressors that im- How many prior events must be included for a
pinge upon the individual from outside the family. proper understanding of coercive behavior? Kar-
Chronic family crises, for example, increase the powitz (1972) showed that roughly half of the so-
likelihood of irritable reactions by the mother (see cial events of significance in controlling the child’s
Chapters 4 and 10). The second (and more likely) behavior were to be found in the immediately pre-
source of change is to be found in long-term shifts ceding event. The other half were to be found in
in outcomes for interchanges with the other per- patterns of the second and third antecedent events.
son. Each cumulative step back along interactional se-
quences requires large increases in data. At some
Predicting Behavioral Events point one simply must decide that the increment in
The entire set of assumptions underlying the prediction from adding information about prior
analysis of structure and process rests upon the behaviors of the dyad is not worth the cost. What
idea that behavior is governed by immediately im- happens when we begin with information about
pinging events. If this is true, then careful observa- target subject base-rate values and then add infor-
tions of external events should allow us to under- mation about the prior behavior of the other mem-
stand and predict some types of behavioral events. ber of the dyad (intersubject component) plus the
Our training in motivational and cognitive psy- prior behavior of the target subject (intrasubject
chology led us to approach this assumption with a component)? The Karpowitz (1972) and Raush
good deal of trepidation. We thought initially that (1965) studies ‘reassure us that much is to be
it would be true for some limited behaviors of very gained by considering the immediately prior be-
young children. In retrospect, it should not have havior of the other person; but what are the incre-
been so difficult for us to give credence to the idea ments if we proceed further back through the ante-
that much of ongoing social behavior is controlled cedents?
or determined by the observable behavior of other Loeber (in preparation) studied ten mother-

170
 Figure 8.1
Relation Between the Order of Approximation
and the Conditional Uncertainty of any Behavioral Event

(from Altmann, 1965, p. 510)

7
2
at
.-
rep
eel. >.
bo]
-—
i]
Bu A
<
3)
ane
f=}
2
2-—
<
°
01

0
2 3 4 5

Order of Approximation

Reprinted with permission from “Sociobiology of the rhesus monkey. II. Stochastics of social communication,” Journal of Theoretical
Biology, 1965, 8. Copyright by Academic Press Inc. (London) Ltd.

child dyads plus siblings to determine the answer behavior. Do these prior events “control” ongoing
to that question. Five of the families were normal behavior, or are these events merely correlated
and five were referred because of an antisocial with events that follow? How do we demonstrate
child. The interactions among members of fami- which prior behaviors are important and which
lies in the clinical sample were consistently less are not? These questions will be considered in the
predictable at all junctures than were the interac- sections that follow.
tions of members of the normal families. Summing
across the two samples, the behavior of the target
The Search for Structure
children was, on the average, 15% predictable if The search begins with the investigation identi-
based only upon knowledge of base-rate values for fying a target event. For convenience, these events
code categories. The comparable figure from Alt- are labeled R;’s. The first target event in a sequence
mann’s (1965) study of rhesus monkeys was 30%. is R,. The next target event by the same subject is
Information about the immediately prior antece- labeled R2 and so on. Events that precede target
dent gave an average predictability of .19, an aver- events are antecedents (A;). Those events that fol-
age of .23 for two antecedents, and .28 for three low a target response are labeled consequences
antecedents. The findings for mother-child inter- (C;). The first event in a sequence occurs at time f,
action are similar to those from the Altmann the second at t+1, the third at t+2, and so on.
study. The antecedent immediately prior to a target event
These studies by Raush (1965), Altmann (1965), occurs at t—1; t—2 describes the antecedent be-
Karpowitz (1972), and Loeber (in preparation) fore that, and so on.
emphasize the importance of using immediately Any given event in the dyad sequence serves a
prior antecedents as a predictive base for ongoing dual role. Its label depends upon which member of

171
a dyad is the “target subject.” When he or she is actions with his mother. Of these, 15 involved the
identified and an entry point is arbitrarily selected, child’s whining. This gives a base-rate value of
then the first event, R,, is assumed to be a reaction .023. Is there anything that the mother does that is
to the prior behavior of the other person. Howev- associated with an altered probability for child-
er, if we suddenly shift the focus to the other mem- Whine? Does scolding increase the likelihood?
ber of the dyad, then this same target event would Will a hug be associated with a decrease or an in-
serve as a Cj for the prior Rj of the other person. crease? The next step is to tabulate all of the moth-
In our own analyses the target events have al- er behaviors that immediately preceded the child’s
most always been the 14 categories from the FICS 15 Whines. In tabulating these antecedents, any
that define coercive behaviors. The subject has one of the 29 code categories might occur. By way
usually been a problem child. During the first dec- of illustration, assume that in the mother’s 640 re-
ade, our work focused primarily upon the func- actions to the child she yelled 18 times. Four of her
tional relations that were defined by the event that Yells (YE) were followed by the child’s Whine
immediately preceded or followed the child’s coer- (WH) behavior. The p(Child Whine|Mother Yell)
cive target response. These studies revealed two is 4/18 or .222. The next question concerns the
interwoven themes that describe (at a behavioral significance of this tenfold increase in predictabil-
level) what is meant by social interaction. One of ity of child-Whine. The binomial-Z value for this
these themes is the intersubject component, and comparison was 5.57. It is reasonable to conclude
the other is the intrasubject component. that mother-Yell and child-Whine are functionally
related.
The Intersubject Component The binomial-Z statistic provides a useful
for Coercion means for comparing base-rate and conditional p
values given the qualifiers noted by writers such as
“., . like other forms of social conduct, perfor-
Gottman (1979) and Gottman and Bakeman
mance of injurious actions is extensively regulated
(1979). They recommend not presenting the confi-
by environmental cues (Bandura & Walters,
dence values for the statistic because it might lend
1959). A theory of aggression must therefore ex-
an aura of unwarranted precision to the discus-
plain not only how aggressive patterns are devel-
sion. Gardner, Mitchell, and Hartmann (1979)
oped, but also how it is that some are elicitors of
note that there are two kinds of dependence tested
aggression while others are not. . . .” (Bandura,
in the 2 x 2 chi-square table employed by most of
1973, p. 115)
us. One is the serial dependency for within-subject
The study of intersubject components consti- behavior; the other is the functional relation be-
tutes a search for those behaviors of the other per- tween the behavior of one person and another. As
son that control the occurrence of aggressive be- they point out, if the level of within-subject (across
haviors. The first question is, how do you identify time) serial dependency exceeds a correlation of
these controlling events in ongoing social interac- .50, it seriously biases a test of independence.
tion? The second question is, how do you prove Even lower levels of serial dependency produce
that they control attacks? The identification of a large increases in Type I errors. As they note, this
functional relation (or correlation) is the first step problem applies both to chi square and to the use
in analyzing structure. Proof of a causal connec- of binomial Z. The reader is advised to consult the
tion is the second, necessary step. paper by Gardner et al. (1979) for a careful review
To find a functional relation, one begins with of the problems in this type of analysis.
the data for the base-rate value for the target At this point, however, there is a problem to
event. For example, let’s give Whine the high base consider. Even a simple event, such as Whine, re-
rate of .023. This means that the child spends flects both inter- and intrasubject control. The
2.3% of his interactions whining. This base-rate likelihood of child-Whine may be related to the
value is important because it serves as a reference mother’s behavior at t —1; it may also be related to
point for comparing the conditional p values for what the child was doing at t- 2. While this point
this or that functional relation. To be of interest, has been made by Thomas and Martin (1976), it
the conditional value must be significantly differ- has, unfortunately, been largely ignored. As they
ent from the base-rate value. According to Sackett point out, one can control for this duality by using
(1976), if the base-rate value falls below 1%, then either a multiple-regression or a conditional-prob-
one cannot effectively employ the binomial Z sta- ability format. The paper by Martin, Maccoby,
tistical analyses in making formal comparisons be- Baron, and Jacklin (1980) compared the two tech-
tween the conditional and the base-rate values. niques in their analyses of infant and mother inter-
Let’s assume that the child engaged in 640 inter- actions and found the results to be comparable.

172
When I encountered this problem a decade ago, I boys in the clinical sample and 4.7% for boys in
was not aware of these more elegant solutions. the normal sample. Only those conditional values
Examination of the interaction protocols sug- that were appreciably different from the base rates
gested that when the child was coercively engaged, are listed; the decision rules for identifying these
there was an increased likelihood at t+1 for the controlling antecedents are presented in Patterson
occurrence of all 14 coercive behaviors. To control and Cobb (1973) and Patterson (1977a).
for this “thematic effect,” I selected only those tar- The interpretation of these values is straightfor-
get events in which the target event was the first ward. For example, in the clinical sample, given
event in a coercive sequence, or a single, isolated the target child’s Disapprove, then the likelihood
coercive event, i.e., only those Whines not em- of the other family member’s Whine in the prior
bedded in a prior coercive matrix. The procedure 6-second interval was .160. This conditional p
followed for the study by Patterson and Cobb value was several times larger than the base rate of
(1973) was to take only those target events pre- .061 for Disapproval. This suggests that Whine
ceded by 18 seconds of interaction in which the and Disapprove were functionally related, i.e.,
target child’s prior behaviors were prosocial. The correlated. Note, too, that a similar relation holds
analysis was based upon the pooled observation for the normal sample. The reader should note
data from 32 families of antisocial children re- that the p values were calculated from populations
ferred to OSLC and data from 26 matched normal of events; some families probably contributed dis-
families. These were essentially working-class proportionately to these relations.
families; about 30% were father-absent homes. There are several features about Table 8.1 that
Each of the families was observed for a minimum are worth noting. First, it is apparent that some
of six to 12 sessions. Target events were initia- variables were much better defined, or accounted
tions. An initiation of a coercive behavior was de- for, by an Aj controlling stimulus than were oth-
fined as an isolated or initial event in a coercive se- ers. The low-rate events, such as Destructive,
quence; by definition, an initiation of a prosocial High-Rate, and Humiliate, remain poorly under-
behavior meant that it was preceded by 18 seconds stood; Dependency, Cry, and Command Negative
in which neither member of a dyad had interacted occurred at such low rates that they were not ana-
with the other.! Only those events in which the lyzable. It was possible to provide a clear delinea-
other member served as agent for both the antece- tion only for the higher rate events. That, of
dent and the consequence were included. There course, is a sensible outcome; it should be easier to
were 10,626 initiations (prosocial and coercive); identify significant functional relations for target
of these, 2,041 were coercive (.192). The compa- events that occur at high rates.
rable value for a sample of nondistressed boys was Each target response had one or more antece-
142. dent events that seemed correlated with its occur-
The analysis was carried out in order to exam- rence. This set of empirical relations for a target
ine intersubject components that would be rela- response defines what is meant by the term net-
tively free of thematic effects from the child’s on- work. Notice that the sizes of the networks vary
going coercive behavior. Given that the child initi- considerably even among target responses of com-
ates a Whine, what is the likelihood that it was parable base rates. For example, Tease has about
preceded by mother-Disapproval, etc.? As Don twice as many correlated A;’s as Yell. This var-
Hartmann (personal communication) points out, iation in the size of networks could represent
a more balanced view would include the results of omissions built into the original code. We may not
a comparable analysis based upon the excluded have been sampling the appropriate behaviors, at
data from coercive chains. least not the ones that relate to Yell. On the other
The functional relations for ten coercive behav- hand, some responses may be more predictable
iors are summarized in Table 8.1. The first col- than others. For example, some responses may be
umn lists 24 categories that might have served as less correlated with immediately impinging envi-
antecedents for the target child’s coercive behav- ronmental events.
iors. The conditional p values in the next two col- The second feature of the data presented in Ta-
umns describe the likelihood that a variable served ble 8.1 has to do with the kind of antecedents that
as an Aj for the total body of initiations; e.g., for most powerfully “elicit” the target response. For
the deviant boys, only .004 of all their initiations six out of the ten target events, the most powerful
were preceded by Approval; for Normals the value elicitor was an antecedent of the same code cate-
was .010. The values that head each column are gory as the target response. For the clinical sam-
the base rates for initiations. For example, Disap- ple, for example, the likelihood of Physical Nega-
proval made up only 6.1% of all initiations for tive as an antecedent, given Physical Negative,

17S
 Table 8.1
Antecedents Relating to Changes in the Likelihood of Initiating Coercive Responses

Given an initiation of R;, then the likelihood of A; was

Physical
p(A;) for Destructive Disapproval High Rate Humiliate Ignore Negativism Negative Tease Whine Yell

Initiations .002+ .061 .047 .008 .002 .004 .002 -007 .002 .026 .0#9 .010 .003 .011 .007 .011 .012 .011 .007
Dev. Nor. Dev. Nor. Dev. Nor. Dev. Nor. Dev. Nor. Dev. Nor. Dev. Nor. Dev. Nor. Dev. Nor. Dev. Nor. Dev. Nor.
Antecedents 18t 0 646 159 88 6 38 5 72 7 278" “64, 106" “11 117 25 120 | 39°" Tie gree

Approve .004 .010 -046

Attend .146 .138 -025 .028 .004 -003 -006 006

Command 085.104 028 .020 001 -001 001

Command Neg. .010 .007 010

Comply 006.006 132.158 053 074 044

Cry 002 .002 .087

Dependency 001
Disapproval 055.055 .200 .217 017 041 .044 022 .032 .044

Destructive .001 .000 .125

High Rate .004 .000 .614
Humiliate .004 .002 .270 -162 -167

Ignore -004 001 lll

Laugh -009 .019 -115 .042 042 .048 032

Noncomply .006 .004 -167

Negativism .007 .006 158 .026 218 .316 .051 051

Normative 117.063 .008 .014 .005 005 002

Play W4 ERS 048 021 009 006 .003 .015

Physical Negative .007 .004 -141 .200 -133 .296 .200 -070

Physical Positive .005 .005 -080

Talk .367 389 -074 .040 .003 .003 .005 012 .037 .024 .006 .001 .008 .002 .019

Tease .006 .004 .209 045 -060 149) 7.357% -060

Receive -003 002 059.200 .088

Whine 002 .003 -160 .222 111 .160

Yell 004 001 .077 .077 .128

+Values in the row are base rates for :mitiations of target event by target child.

+Number of initiations.

was .30; Tease as an antecedent, given Tease, was parisons could be made.
.15; and Yell as an antecedent, given Yell, was In passing, it should be noted that there are
.13. While all of these conditional probabilities far some difficulties in locating stimuli that are associ-
exceed their respective base-rate values, it is clear ated with reduced likelihood of occurrence. One
that when we speak of “connections” or “bonds,” simply looks for p(Rj|Aj) values that are apprecia-
the reference is probabilistic. There is nothing bly /ower than the base rate for the target event,
even in the comparatively large value of .30 that i.e., CN as an A; for Disapprove. However, given
implies the mechanistic, wired-in connections usu- a Clinical event that has a base rate of .010 or less,
ally ascribed to theories about stimulus-response then even when the conditional p is .000, there is
bonds. the difficulty that most statistical comparisons are
The third characteristic of these data that stands not appropriate. This approach is not an effective
out is the similarity in patterns of functionally re- means for studying controlling stimuli when the
lated stimuli for the two samples. Generally speak- target events are low-frequency events. For this
ing, a stimulus that controls a response for the reason, very few of our publications have dealt
normal sample will also control the same response with the problem of antecedent events associated
for the clinical sample. There are some interesting with a decreased likelihood of occurrence for the
exceptions to this but enough overlap to lead one target event.
to suspect that the basic structure for the two sam-
ples may be very similar. In future investigations it The Utility of Functional Relations
would be wise to calculate these values separately What do networks tell us about target events?
for individuals so that appropriate statistical com- This section outlines a series of steps for determin-

174
 Figure 8.2
The Utility of Functional Relations

Antecedent Target Response

Disapproval
-060
(.08)
Whine
.250 (.02)
Hit
(.005)

The figures in parentheses are base-rate values for the antecedents and the target response.
The figures not in parentheses are the conditional probabilities describing the functional relation p(R;|A)).

ing how much a single functional relation, or an There are a number of ways in which this can be
entire network, tells us about a target event. For done. The joint contribution of the base rate for
present purposes, I am equating the terms useful- the antecedent and its impact upon the target
ness or utility with information about target event could be expressed in terms of reduction in
events. The more complete the information about uncertainty (Attneave, 1959); or, as Suomi (1979)
which antecedents are significantly related to suggests, it could be described by the ratio:
child-Hit, the more useful it is to me as a therapist.
The two key pieces of information are the magni- p(RjAi).
tude of the probability value for the functional re- p(R;)
lation and the frequency with which the antece-
dent occurs. The magnitude of this ratio is an index of one’s
The information about networks of controlling ability to predict R; given a knowledge of the func-
antecedents is a useful beginning to our under- tional relation Rj|Aj. But, as can be seen, it leaves
standing of children’s aggressive behavior, but it is out information about how frequently the antece-
conceivable that such a network, or a single func- dent occurs. The approach by Patterson and Cobb
tional relation, might be of little predictive value. (1973) provides a means for including p(Aj) as fol-
By way of illustration, assume that both Disap- lows:
proval and Hit are significant controlling antece-
dents for Whine. Given Disapproval as an antece- = power
dent, then the likelihood of Whine was about once
in 16 trials. Given Hit as an antecedent, then the Taking the data in Figure 8.2, each conditional p
odds of a Whine were one in four. Both of these value is multiplied by the base rate for the antece-
antecedents serve significant functions as controll- dent. Then dividing this figure by the p(Whine)
ing stimuli for Whine. Both sets of functional rela- value of .02 gives the estimate of the “power” of
tions could also be of interest from the standpoint that antecedent. Twenty-four percent of the occur-
of theory construction. rences of Whine were related to Disapproval; only
If we raise the question of which antecedent has about 6% were related to Hit as the antecedent.
the greater impact, then clearly Hit has the higher Actually, neither one of these variables meets the
conditional p value. However, note that in the il- arbitrary criterion of > 40% used by Patterson
lustration the base-rate value for Hit is extremely and Cobb (1973) when constructing functionally
low. Given that our task is to understand or to defined response classes. Don Hartmann (personal
manage Whine, then this particular antecedent communication) suggests that this p value could
adds little to what it is that we must know. Clear- also be expressed in correlational terms. This
ly, from the standpoint of management or predic- would describe power in terms of variance ac-
tability, it is necessary to take into account the counted for. If these findings regarding child-
base rate for the controlling antecedent stimulus. Whine were available to the therapist, it would

175
 Table 8.2
Power Indices for Coercive Responses
(adapted from Patterson & Cobb, 1973, p. 180)

Problem Family Nonproblem Family

CoerciveResponse p(R) PowerIndexfor R; p(Rj) PowerIndexfor R;
Command
Negative .002 .472 if
Disapproval .061 .780 .047 .798
Destructiveness .002 .053 1F
High Rate .008 .850 t
Humiliate .004 .734 1
Ignore .007 -956 ih
Noncomply¢ .038 997 .038 1999
Negativism .026 .743 .019 796
PhysicalNegative .010 .673 .003 -631
Tease 011 584 .007 96
Whine 011 1959) .012 256
Yell .O11 .326 .007 -091
tOnly Rj with 10 or more events included.
+By definition (in the code), the A; must be Command.

suggest that he could most easily reduce child- Whine as a target event, and the analysis shows
Whine by training family members to cease their that the antecedents account for only 30% of the
use of Disapproval. In so doing, child-Whine occurrences of Whine. There are several steps one
would be reduced by about one-fourth. might take. First the problem may be that the code
In the above example, information about only is defective. Perhaps the more powerful antece-
these two variables would provide a poor basis for dents for Whine were not sampled. Given this situ-
designing an experiment in which one proposes to ation, I would begin by carefully observing some
manipulate Whine. There are simply too many high-rate Whiners to see what I had missed when
possible alternative antecedents that might exert constructing the first code. Second, the antece-
more powerful control than either Disapprove or dents for Whine may not be consistent across fam-
Hit. The equation that we developed to describe ilies; therefore, a series of single case studies might
power is simply a restatement of Bayes’ theorem. result in greater precision. Third, given a sample
Given information from both antecedents, the of dedicated, high-rate Whiners, I would system-
p(Rj| Aj) is only .30. This, in turn, means that atically sample settings for each child. I assume,
there is much about Whine that is left unexplained then, there would be some settings associated with
by these analyses. We shall return in a later section higher p(Whine) than others. By concentrating
to the implication of such findings for the design upon those subsets, the base-rate values for both
of experiments. the target response and the antecedents would
It should be noted in passing that a lower base- likely be more manageable (higher). In this fash-
rate value for the target event covaries with a low- ion, one uses whatever data are at hand to design a
er p value for power. The Patterson and Cobb more powerful approach to the question.
(1973) study showed a correlation of .49 (p <.10) The data in Table 8.2 were based upon the
between the frequency with which a coercion vari- study of initiations from samples of clinical and
able occurred and the power index. This makes normal families (Patterson & Cobb, 1973). For
sense in that only significant functional relations both samples the controlling antecedents ac-
enter into calculations for the power index. It is counted for 70% to 80% of the occurrence of Dis-
simply more difficult to find significant relations approval and Negativism. In both samples the net-
for low base-rate events. works account for 50% to 60% of the occurrences
Suppose that it is important that one understand of Tease and Physical Negative. Presumably, if the

176
 networks were extended backwards to include the coercive behavior. For example, when questioned,
child’s own prior behavior (as in the Loeber [in most mothers “explained” coercive child behavior
preparation] study), we might account for more as being caused by the child’s being in a bad mood,
variance. However, even a reliance upon immedi- being tired, or being upset. The mothers’ explana-
ately prior antecedents tells us much of what we tions reflected the current theories, i.e., deviant
need to know about ten of the 14 responses. While behaviors were caused by underlying motives,
the information from the two samples is generally drives, or affective states. Certainly it is reason-
similar, note the interesting discrepancy in the case able to think of fatigue or depressed mood as a
of Whine as a target event. Here the controlling partial elicitor for coercive reactions. However, as
variables sampled by the code provided a very outlined in Chapter 4, it is not a requirement that
high power index for the clinical sample but much all elicitors be accompanied by arousal. Some con-
less for Normals. trolling events (for example, prosocial ones) seem
It should be clear that social interaction is an in- to control behavior without any noticeable change
credibly flexible structure. The p values that de- in autonomic state. This being the case, we rea-
scribe the functional relations between adjacent soned that perhaps some coercive reactions could
events are of low magnitude, and they shift over be under fairly tight stimulus control. It seemed
time. Even the key concept, “base rate,” is not a that a single subject ABAB reversal design might
static quantity. It may vary across time and as a best demonstrate this control. If the controlling
function of which settings are sampled.” It may events were presented during the manipulation at
also be the case that the boundaries that define the B,, there should be a corresponding increase in the
antecedent event may be changed. For example, target behavior, as compared to the rate during
other events whose meaning is similar to the orig- either the first or second baseline periods. Then a
inally conditioned antecedent may elicit the same presentation of the controlling event at B, should
response. Studies by Berkowitz (1973a) and Staats replicate the effect. If the outcome provided a tight
and Staats (1958) suggest that this might be the fit to our experimental expectations, it would
case. For example, stimuli that are contiguously demonstrate the causal status of the antecedent
paired with the eliciting event may temporarily event. It would, however, leave open questions
produce the “same” reaction (Staats, Staats, & concerning its concomitants; it would not necessa-
Crawford, 1962). rily rule out autonomic arousal.
If not reinforced, the antecedent —-» response The next problem was to decide what and how
connection may eventually weaken and be re- to “manipulate.” To learn how to carry out such
placed by still others, that, in turn, may prove to studies, we began in the laboratory where some
be transitory examples of controlling events. In modicum of control was possible. It was hoped
any case, change is the rule. This requires a certain that what we learned there could be used to design
amount of flexibility in how one defines an event experiments that could then be carried out in field
and flexibility in how one thinks about connec- settings.
tions between events. The structure of human in- A number of pilot tests led us to believe that the
teraction is in many ways analogous to that of the phenomenon, “mother-on-the-phone,” might be a
Eskimo umiak. This ocean-going craft is beautiful powerful event that controlled many coercive be-
in design and function. It shifts and adjusts itself haviors of preschool children. The first study used
as it moves over waves but, at the same time, re- an ABAB design in which the mother freely inter-
tains its essential design and purpose. acted with the child during both baseline (A)
phases (Atkinson, 1971). During the experimental
Causal Status
(B) phases, the mother was totally involved in a
Sequential data collected in the field represent task that required making visual discriminations
only the first stage in constructing a performance while listening to a telephone, and she was placed
theory about social interaction. Significant corre- behind a barricade so the child could see her but
lations do not prove causality. To test for causal not touch her. In retrospect, the procedure was a
relations experimental manipulations are re- good example of overkill; the controlling event
quired. Since most of our work in the early 1970’s was much more powerful than it needed to be.
was focused upon identifying which antecedents However, at that time we simply did not know
were significant and which were not, we decided what would be necessary in order to achieve ex-
to study the causal implications of the significant perimental control.
functional relations first. However, our initial Each of the 14 coercive code categories were
findings did not seem in keeping with the conven- analyzed separately to determine the magnitude of
tional wisdom about what determined children’s behavior changes across the four conditions. As

PF
 Table 8.3
Experimental Manipulation of Coercive Child Behaviors
(adapted from Atkinson, 1971, p. 14)

Coercive Mean Rate per Minute

Child Behaviors Baseline1 Experimental1 Baseline 2 Experimental 2 F Values

Command 0.014 0.029 0.007 0.021 Tez

Cry 0.003 0.211 0.030 0.183 TS

Destructiveness 0.014 0.125 0.002 0.069 1953

High rate noise 0.004 0.022 0.001 0.017 1.03

Negativism 0.000 0.011 0.009 0.028 ihe

Whine 0.007 0.049 0.038 0.043 3.227

Yell 0.003 0.074 0.001 0.074 Sia bills

Mean .006 .074 013 062

*p<.05
2 p=001

shown in Table 8.3, the ANOVA for repeated each phase lasted for ten minutes. During the ex-
measures produced significant F values for four of perimental period, the previously identified signi-
the comparisons (Cry, Whine, Yell, and Destruc- ficant antecedent was repeatedly presented. For
tive). The mean rates for the target responses were the ten preschoolers, the mean likelihood of
significantly higher during the manipulation than Whine during the three phases was .007, .317,
during the baseline periods. The data from this and .043 respectively. The differences among the
sample offered firm support for the idea that some means were highly significant (Patterson, 1977b).
coercive child behaviors were under tight stimulus It is of some interest to note that the likelihood of
control. Using a similar procedure in the course of child-Whine, given the mother’s antecedent, was
his work on negative reinforcement, Devine quite stable; the correlation for the conditional p
(1971) replicated this general effect. values across sessions, separated by several days,
The finding that some mother behaviors reliably was .67 (p<.05).
produced child-Yell, child-Whine, child-Cry, and The three studies showed that the general idea
child-Destructive was not very surprising. Howev- of testing for stimulus control was feasible. It was
er, the fact that removing the controlling stimulus possible to observe mother-child interaction in the
produced abrupt cessation of the target behaviors laboratory and identify an event that controlled
did surprise both the experimenters and the moth- child-Whine. When the controlling event was pre-
ers. These results encouraged us to go on and de- sented in the second session, it increased the likeli-
sign studies that would be closer to what could be hood of child-Whine. When the mother stopped
seen taking place in the home. presenting it, the child stopped whining. Howev-
The search began for an event that could be er, the studies left some key questions unan-
manipulated in the matrix of ongoing social inter- swered.
action. We decided to focus upon a target response First, could one simply go into a home, observe
that occurred often when normal preschool chil- for three or four hours, and identify stimuli that
dren and their mothers visited the OSLC labora- controlled behavior? We were not quite ready to
tory. In that situation, the child’s Whine was more believe that such a level of precision would be pos-
to be expected than his “hello.” The setting for the sible for data from field settings. We knew that
next several studies was still the laboratory. The field data produced significant correlations, but
mother-child pair was observed during semistruc- would these functional relations stand up under
tured interaction, and the data were analyzed to experimental manipulations? It had already re-
identify the behavior of the mother that reliably quired two years of devoted effort by a team of as-
preceded child-Whine. The dyad returned several sistants to come this far. However, for all that ef-
days later to participate in an ABA procedure; fort, we really had only studied one target re-

178
 sponse. In the last study (which most closely ap- ly with her son. During the five-minute B phase,
proximated field conditions) the majority of the the mother was told to ignore every initiation
preschoolers’ Whines were controlled by the same made by the child to communicate with her. If,
‘antecedent (mother-Tease). It was not at all clear during that time, his whining/crying got out of
that we would be successful in situ, where the lati- hand, she was instructed to interact briefly with
tude open to the mother and child seemed practi- him and then return to the mother-Ignore condi-
cally infinite. tion. Following this, the baseline condition was re-
It was with some trepidation that the first field introduced (A,). Identical ABA procedures were
study was designed by the writer, Kate Whalen followed on the second day. Figure 8.3 summar-
and Hal Dengerink. The experimental design now izes the findings testing the status of the control-
ling stimulus identified during the baseline ses-
sions. The data offered strong support for the hy-
pothesis that mother-Ignore was an event that reli-
Figure 8.3 ably controlled the child’s whining behavior.
Experimental Manipulation There was at least a twofold increase in p(Whine)
when this controlling event was introduced. When
of Child-Whine
removed, the behavior returned to its baseline lev-
(from Patterson, 1979b, p. 151)) el on Day 1 and partially returned to its baseline
level on Day 2. These replicated findings gave
modest support to the idea that for this child a
functional analysis of family interactions was in-
deed an effective means for identifying a stimulus
that reliably controlled his whining behavior.
As we pursued this problem further, we soon
produced two or three studies that either failed or,
at best, produced inconclusive outcomes.? The
problem lay in our confusing the concept of signi-
ficant functional relation (p[R;|Aj]) with the term

p(Whine) “control.” As noted earlier, it is possible to have a

significant correlation between an antecedent and
an Rj, but it may provide little information about
the target response. The antecedent may occur in-
frequently, and/or it may be only one of many
“controlling” antecedents. That being the case, it
would be difficult to design a study that could con-
trol for these unknown variables. During an ex-
Baseline, Manipulation Baseline,
perimental phase in a laboratory study, these other
Experimental Phases antecedent events are left free to vary. This, in
turn, means that in some studies the data for rate
of target responses will not demonstrate tight ex-
perimental control.
included two stages: first, to observe in the home One can think of a significant functional rela-
and identify an antecedent that controlled a target- tion as denoting “control” in a statistical sense, but
child behavior; then to design an ABA experiment the prediction of experimental or treatment ma-
that would demonstrate that the antecedent really nipulations requires that we specify how much
controlled the behavior. Then we wanted to repli- control. In earlier analyses of functionally defined
cate the experiment on the following day. Several response classes, we used the concept of a power
baseline sessions in the home of a 3-year-old child index to decide which functional relations be-
showed that mother-Ignore reliably elicited child- longed to a class of responses. It was decided to
Whine. For two baseline sessions, the conditional apply this concept to the problem of predicting the
probability values were .83 and .43; the base rates outcomes of experimental manipulations. Again,
for Whine on those two days were .23 and .07 re- the information from the base rate for Rj, the base
spectively. Following the baseline study, an ABA rate for Aj, and p(Rj|Aj) are used to calculate the
reversal design was employed on two consecutive power index. If the antecedent variable accounted
days. During each of the five-minute baseline peri- for more than 40% of the occurrence of the target
ods, the mother was instructed to interact normal- event, then it seemed reasonable to bet that an ex-

179.
perimental manipulation would be successful. A this amount of data was an absolute minimum for
more conservative approach would be to identify estimating p(R;). The trade-off for increased sta-
the two or three antecedents that control most of bility was that both mother and experimenter
the occurrences for Rj, and then design an experi- were under greater stress. It is this stress that sets
ment that controls for all but one antecedent. an upper limit on the amount of time that can be
Three stages were involved in the design of each used in field-study manipulations. Less than 20
study. First, we observed in the home and identi- minutes per phase means a less stable estimate of
fied the functional relations for several high-rate the experimental effect. However, more than 20
coercive child behaviors. Then we calculated the minutes of manipulation is difficult to tolerate.
power scores for each and selected the relation in The second increase in precision involved more
which the antecedent accounted for more than careful training of the mother prior to the study.
40% of the occurrences of Rj. Then, to establish One of the experimenters played the role of the
the reliability of the data, we replicated the manip- child and, when cued, the mother began to argue.
ulation for one or two sessions. We also began to Figure 8.4 summarizes the results of the two
role-play the experimental procedures with the days during which an ABA design was followed.
mother prior to actually running the study. For ex-
ample, if she was to give a command when cued,
then an experimenter would role-play several Figure 8.4
practice runs with her while another staff member
Experimental Manipulation of
removed the child to another room. As these stud-
ies progressed and the procedures became increas- Child-Argue
ingly intricate, the number of training trials were
increased accordingly.
The studies also involved an improved version
of the FICS. Marion Forgatch took over the com-
plex task of coordinating the observers and data
tabulators (we had no funds for computers). She
organized the data in such a way that we could re-
trieve the conditional p values from field data in a
day or two, and thus design the experiment for the
of
Child-Argue
Frequency
following session. This reorganization was neces-
sary for the new series of studies. The first study in
this new series took place in the home of a mother
and her preschool child. Several days of observa-
»
tion identified a functional relation that was of
particular interest because it described a target re-
sponse (Argue) that had not previously been stud-
ied. The base rate for Argue was .031. Given
mother-Argue as an antecedent, then the condi- A, B A,
tional p for child-Argue was .531. That seemed Baseline Manipulation Baseline
encouraging. The next question was about power.
Did mother-Argue exert enough control to serve Experimental Phases
as a proper base for an experiment? The base rate
for mother-Argue was .024; the compound proba- The figures in parentheses refer to the frequency of
bility (.024 x .531) was .013. The power for this Mother-Argue.
antecedent was obtained by dividing this figure by
the base rate for child-Argue (.031); 42% of the
target events were accounted for by this single The data demonstrate the controlling status of
controlling event. Our best guess, then, would be mother-Argue. When the mother increased the
that an experiment employing this controlling rate with which-she presented these elicitors dur-
event should be successful. ing the B phases, there were commensurate in-
Several further improvements were made in the creases in child-Argue. Mother-Argue determines
experimental procedures and design. First, each of child-Argue; so the prediction was confirmed. But
the three phases (ABA) in a session was increased note the failure of the target event to return alto-
to 20 minutes. This was more in agreement with gether to the first baseline (A,) levels. This reflects
our stability estimates (Chapter 3) that showed the problem of controlling the behavior of the

180
 mother. As shown in Figure 8.4, during A, the tecedent networks controlling a behavior. As a pi-
mother initiated many more Argues than she had lot study, two 6-year-old boys were observed in
in A;. This happened on both days. Controlling the home and in the classroom. In both settings
the controller is obviously a complicated affair.* antecedents were identified that correlated with
The next study is included as an illustration of the boys’ most frequently occurring coercive re-
what happens when the impact of antecedents sponses. There was very little overlap in the stim-
upon R; was significant but not powerful. This uli for the two settings in the events controlling the
could come about because the antecedent occurs same responses. The Fisher exact chi square was
infrequently and/or because the magnitude of used to test for the overlap in networks controlling
p(Rj|Aj) is low. Very low power values mean that coercive behaviors. The chi square (p<.02)
there are variables other than those being manipu- showed a significant association. Given a control-
lated that control the target event. During any ling stimulus for a target event in one setting, then
phase of the study, increases in rate of the target there was an excellent chance that the same ante-
event may occur as the result of these “uncon- cedent would not control the same target event in
trolled” controlling events. Summer was a moder- the other setting. Obviously, with an N of only
ately well-behaved preschool child with a likeli- two cases, one cannot generalize from these find-
hood of .069 of ignoring her mother. This was ings, but they do suggest the importance of carry-
even more likely to occur following mother-Com- ing out microsocial analyses in different settings.
mand (conditional p of .208). The binomial-Z val- In the same study there was one antecedent and
ue was 2.69; so the functional relation is note- target event connection that was built into the
worthy; however, the power for mother-Com- code, so it was found in both settings (child-Non-
mand was a very low 15.9%. This meant that comply|adult-Command). It was decided to alter
there were many events other than mother-Com- the A;—»R; bond in the home and determine
mand that controlled Summer-Ignore. Each of whether the effects generalized to the school. To
four sessions involved an ABA design. On each alter the connection found in the home, both
day during Ai, the mother made zero Com- mothers were trained to use time out in the home
mands. During the B phase she was cued to make whenever the children noncomplied following
20 Commands. During A, she was cued to make mother-Command. Following the manipulation,
five Commands. Increasing mother-Commands both boys showed a significant increase in p(Com-
produced the expected increase in Summer-Ignore. ply|mother-Command). The effects did not gener-
However, even though mother reduced her com- alize to teacher-Command and child-Noncomply
mands during A,, Summer continued to ignore. in the school. For Ben, the conditional baseline p
Variables other than those we manipulated were in school was .20; the post-manipulation p value
obviously controlling the target event. A low pow- was .20. For Bobby, the comparable values were
er score means a sloppy experiment. .30 and .23 respectively. Altering Aj» R; connec-
These studies are an illustration of what re- tions in one setting may not generalize to the same
mains to be done. They are not sufficient for test- stimulus and response in a new setting. I think this
ing the hypothesis that statistically defined func- is especially the case when the stimulus is provided
tional relations denote a causal status for the ante- by different agents; this is in keeping with an earli-
cedent event. The studies completed thus far do er analysis in which we found that agent and agent
support this assumption. In passing, it should be behavior had very different implications for con-
noted that the mothers who participated in these trol of behavior (Patterson & Cobb, 1971).
studies learned much from their experience. To In his 1968 publication, Mischel made a com-
them, it came as a surprise to find how much their pelling case for the necessity of studying the im-
behavior controlled the reactions of their children. pact of settings upon behavior. The pilot studies
Those who later went into parent training made outlined here illustrate how such studies might
very rapid progress in learning child management proceed. For example, the studies reviewed in
skills. Chapter 2 showed that response hierarchies for
the performance of aggressive behaviors are high-
An Analysis of Settings ly predictable from one setting to another. How-
The next hypothesis tested was that the net- ever, it also seems that the specific antecedents
works controlling a particular coercive behavior (and perhaps consequences) that control them
vary from one setting to another. Each setting re- vary from one setting to another. I think future
quires its own discriminations as to when, who, studies will show that the networks that control
what, and how one can maximize payoffs for co- the responses will vary from one setting to anoth-
ercion. Settings are specific with regard to the an- er. What are the implications of such an idea? One

181
implication is concerned with the problem of gen- works of significant antecedents, and that these
eralization of treatment. If one alters the antece- networks generalize across families.
dent —>response structure in one setting, then it Antecedent behaviors provided by family mem-
would have little or no effect on the antecedent > bers were examined to determine which, if any,
response structures in the other. This is certainly covaried with the initiation of each of the 14 coer-
in accord with reviews of the clinical literature on cive responses (see Table 8.1). These functional
treatment outcomes. Hill Walker’s classic studies relations were, in turn, examined to determine
on generalization from token-culture classroom to whether the same networks controlled several co-
regular classroom are a good example. His studies ercive responses. In effect, this second step would
showed little generalization unless steps were tak- identify a functionally defined class of coercive re-
en to involve the teachers and peers in the regular sponses. Patterson (1977a) reanalyzed these con-
classroom (Walker, Hops, & Johnson, 1975). trolling antecedent networks, described in the Pat-
These studies and our own clinical experience are terson and Cobb (1973) study, utilizing a more
in agreement—to obtain “generalization” for rigorous set of decision rules that slightly altered
treatment, the microsocial structure must be al- the constituents of the functional relations. The
tered in both settings. Even if the same stimulus- revised classes are presented in Figures 8.5 and
response bond is found in two settings, the bond is 8.6.
provided by different agents. The stimulus com- The first one was labeled Social Aggression. As
pound (agent + agent behavior) is different from shown in Figure 8.5, there are five controlling
one setting to another. events that share in common the fact that each is
These speculations are based upon a very small associated with an increased likelihood for the ini-
set of analyses. Again, the studies presented were tiation of Tease and Physical Negative. For exam-
intended to be more illustrative of a microsocial ple, given Noncomply as an antecedent, the condi-
approach than confirmatory. Are all the Aj» R; tional p was .088 that the problem child would
connections this setting specific? We don’t know. tease and .059 that he would hit (Physical Nega-
Under what conditions will a child generalize ex- tive). Hitting and teasing are functionally related
periences across settings? It must be the case that in that both are controlled by similar networks of
such generalizations do occur, but what are the antecedents.
parameters that facilitate or inhibit this most im- Do target responses that “share” networks of
portant process? If the controlling networks differ, significant antecedents also share networks of
how can it be that the response hierarchies are sta- functionally related reactions (consequences)? To
ble across settings? These are questions for the answer this question, the data were analyzed from
next generation of studies. a sample of antisocial children (Patterson, 1977a).
The analysis showed that the Social Aggressor re-
Functionally Defined Classes of Rjs sponse class primarily involved siblings. The find-
At a microsocial level it is possible that each ings for consequences, summarized in Figure 8.5,
child has a unique network of functional relations. were from the reactions of male siblings to coercive
Significant functional relations that are found for behaviors of problem children. There were three
one child could be specific to that child alone. consequences shared by the target events that de-
What is understood about one child may not gen- fined this response class. These were Physical Neg-
eralize to other children. If that is so, then interac- ative, Tease, and Yell. In each instance the p(C;|Rj)
tional psychology would stand immobilized in a value was greater than the p(C;). The latter are the
small cloud of disparate facts. The writer believes figures in parentheses. It may be that these shared
that this is not the case. Instead, it is assumed that networks occur only if the target subject is a prob-
structural units exist that generalize across sub- lem child. To test the generalizability of these find-
jects; these units are sufficient in their complexity ings, the siblings were used as target subjects, and
to challenge the most avid seeker. The implicit as- the reactions of the problem child were examined.
sumption is that members of our culture share cer- Sibling-Tease and Physical Negative were signifi-
tain beliefs about how-to react to this or that stim- cantly correlated with the following reactions
ulus. If this is true, then similar functional rela- from the problem child: Physical Negative, Tease,
tions will be found across families. For example, Yell, Disapproval, and Whine.
most adults do not expect perfect obedience from The target events that comprise the response
their children, so there is, then, a functional rela- class, Social Aggression, are functionally related
tion between mother-Command and child-Non- to their antecedents. The responses that make up
comply that holds across families. It also assumed this class share in common a network of five ante-
that certain target responses share similar net- cedent stimuli. They also share in common a set of

182
 Figure 8.5
Functionally Defined Response Class Social Aggression
(adapted from Patterson, 1977a)

Functionally
Controlling A;’s Target R? Related C;’s
WH (.160) arget
a R;’s -209
IG (.111) TE (.011) NR
CR (.087)
YE (.077)

Accelerating C;’s
io)
cs 2.Ws “Os
O
PN’ .450
——$___

PN TE YE
(.011) (.003) (.005)

LA .474
ee
ae) 9S 9
sf SY

.040 LA

The figures in parentheses are base-rate values. Key

The figures not in parentheses are conditional probabilities. CO Command
CR Cry
IG Ignore
LA Laugh
NC Noncomply
NE_ Negativism
NR_ No Response
PN PhysicalNegative
TE Tease
WH Whine
YE Yell

three consequences. These consequences are func- upon future behavior. It may function as a rein-
tionally related to these target events; the re- forcer or a punisher and have a long-term effect in
sponses reliably produce the consequences. altering the connection between the antecedent
Given that a consequence is functionally related and target event. The C; may also have an immedi-
to a target response, it may have a dual impact ate impact in terms of accelerating or decelerating

183
 Figure 8.6
The Functionally Defined Response Class Hostile
(adapted from Patterson, 1977a)

Functionally
Controlling A;’s Target R;’s Related Cs

HUel
TE(.209)
Accelerating C;
vS DI .344
So SSS

-041 Female Mothers

DI Siblings DI (.042)
DI(.027)
NR_ .020
°O.
2, xANY
Decelerating C;

The figures in parentheses are base-rate values. Key

The figures not in parentheses are conditional probabilities. DI __Disapproval
HU Humiliate
NE_ Negativism
NR_ No Response
cE: ) ease
WH Whine

the likelihood of an immediate recurrence for the sponses for a class labeled Hostile. The antecedent
target response. The data used to test this hypothe- controlling the initiation of each of the three target
sis involved the reactions of all family members responses was Disapproval. The family members
(Patterson, 1977a). Given a socially aggressive re- most involved in providing these antecedents were
sponse as a target event and a Physical Negative as the mother and female siblings (Patterson, 1977a).
a consequence, the likelihood of a recurrence of the Hostile behaviors are the problem child’s means of
socially aggressive response was .450. This value coping with verbal assaults by female members of
was compared to the base rate for recurrence of so- the family. Note, too, that the counterattacks to
cially aggressive responses and was found to be sig- these intrusions were also verbal.
nificant. As consequences, both Physical Negative As shown in Figure 8.6, his hostile reactions
and Laugh serve to accelerate the ongoing occur- were a reliable means for producing yet further
rence of members of this response class. Disapproval reactions from both mothers and fe-
Figure 8.6 summarizes the findings for the sec- male siblings. This second analysis was based
ond functionally defined response class. The data upon the interactions of 37 socially aggressive
were obtained from the study on initiation by the problem children. The figures in parentheses give
problem child by Patterson and Cobb (1973). The the base-rate values for these consequences sepa-
search for shared networks produced three re- rately for female siblings and for mothers. As can

184
 be seen there, the problem child’s hostile reactions tion. During this phase, whenever Pumpkin pro-
were associated with increased likelihoods of Dis- voked, the mother was to terminate the aversive
_ approval as a consequence for both mothers and condition and talk nicely to Pumpkin. As ex-
female siblings. pected, the negative reinforcement arrangement
The third stage of the analysis searched for con- strengthened the Aj—» R; bond over the baseline
sequences that would decelerate or accelerate the levels on each of the three days. Strengthening
likelihood of recurrence of the child’s hostile be- mother-Command —*child-Provoke did not alter
havior. In that study (Patterson, 1977a), the base the connection between mother-Command—»>
rate for the comparison was the likelihood for im- child-Ignore. The mean conditional p value for the
mediate recurrence of hostile behavior based upon other A;—» R; connection remained at its baseline
interactions with all family members. The results value. Even though both connections share the
of the analyses are also shown in Figure 8.6. The same antecedent event, they seem to be autono-
consequence, Disapproval (by family members), mous. More studies of this kind must be carried
produces an accelerating impact, as shown by the out before we decide this crucial issue. At this
p value of .344 for immediate recurrence of Hos- juncture, we should think of responses as being
tile. If, however, family members acted as if no- functionally equivalent but also autonomous.
thing had transpired (No Response), the effect was Manipulations of classes. There is an interesting
to decelerate Hostile. sense in which one can manipulate members of a
What is the utility of generating functionally de- functionally defined class. If one increases the den-
fined classes of this kind? I think it is useful to sity with which functionally significant antecedent
know that male siblings are more likely to produce events are presented, then this should increase the
socially aggressive reactions from the problem frequency for all target responses within that class.
child, while mothers and female siblings are more However, the impact of this manipulation will not
likely to produce hostile reactions. As noted in be the same for all members of the response class.
Chapter 5, parental “nattering” differentially af- It will vary as a function of how much control is
fects the two functionally defined classes. Hartup exerted over each target response by the antece-
(1979) introduced the felicitous phrase, “function- dent event. The power index should be of some as-
al equivalence,” to describe functionally defined sistance in predicting which members will be af-
response classes. In Figure 8.6 the findings show fected by manipulating the antecedent and which
that child reactions, such as Whine, Negativism, will not. The one study reported in Patterson
and Disapproval, have very different topographies (1979b) identified five coercive child behaviors as
but produce similar reactions from the female members of a functionally defined response class.
adults against whom they are directed. They are All members were controlled by mother-Com-
also functionally equivalent in the sense that each mand. During the ABA manipulation, mother
is elicited by Disapproval. As we shall see in Chap- commands were increased during the B phase. It
ter 11, these responses also tend to covary when was predicted that during the experimental phase
subjected to an across-subject correlational study. there would be increases in frequency for each of
The findings raise an interesting question about the five target responses beyond what was ob-
what a functionally defined response class means. tained during the two baseline periods. The ma-
For example, if the connection between an antece- nipulation showed the predicted effect for one re-
dent and one target response were altered would it sponse for which the power score was .51 and for
generalize to all other members of that functional- a second for which the power score was only .27.
ly defined class? Two responses for which the power score was less
Thus far, one study has been completed that re- than .10 showed no significant effects for the ma-
lates to this issue. In this study, negative reinforce- nipulation. One response for which the power
ment was used to strengthen the Aj—» R; connec- score was only .04 showed an increase during the
tion for one member of a functionally defined class manipulation and then a return to baseline.
but not for the other. What was of interest was the This manipulation has been replicated several
alteration in the nonmanipulated Aj—» R; connec- times; it has some interesting implications for
tion. The details of the procedures are presented in treatment. Given a baseline study for a clinical
Appendix 8.1. The analysis of baseline data case, a functional analysis would likely show that
showed that Pumpkin’s Ignore and Provoke were certain mother behaviors reliably control some
controlled by the same antecedent, mother-Com- class of functionally related coercive child behav-
mand. On three consecutive days the mother was iors. By having her carefully track and reduce the
cued to issue four commands during each base- frequency with which she presents the controlling
line period and 16 during the experimental condi- event, there would be a predictable drop in all

185
child responses with power scores over .40. This inous implications. A telephone call a year later in-
would not only give the mother some-small respite dicated that the child had been institutionalized.
but also teach her about the impact of her behav- In summary, then, some aspects of the behavior
ior upon that of the child. (We haven’t tried this of family members show orderly relations among
yet). events adjacent in interactional sequences. The be-
havior of one person elicits, in a probabilistic
Variance Accounted For
sense, certain reactions from other members of the
The stated function of a coercion performance family. There are certain networks of antecedent
theory is to account for variance in observed be- events that reliably control several responses.
havior. Identifying the controlling events provided These functionally defined response classes occur
by the behavior of the other person is a step to- across families. The experimental studies carried
ward accomplishing this goal. As shown in this out thus far suggest that antecedent events that are
chapter and in the more general case by Loeber (in correlated with target events may also be thought
preparation), the conditional p for (Rj|Aj) com- of as “controlling events”; i.e., increases or de-
pared to the p(R;) describes the increment in pre- creases in the frequency with which the antecedent
dictability for both intersubject and intrasubject events are presented produce correspomdime
components; p(Rj|A;) defines a small part of the changes in the target events.
events that determine extended coercive chains.
However, by itself, information about the density Intrasubject Component
of controlling antecedents would account for only In social interaction sequences our immediate
very small portions of the variance for the child’s reactions reflect, in part, an extension of what we
overall level of coercive performance. As we shall were doing a moment ago. In that sense, the indi-
see in the next section, it is the variables that deter- vidual’s ongoing social behavior has its own intra-
mine extended chains that also account for the ma- subject structure. Most of us have a sense about
jor portion of performance variance. There are this; at a phenomenological level there is an
some interpersonal components that do contribute awareness of continuity in our own ongoing be-
to the likelihood of chain extension. These compo- havior. While our next reaction may surprise oth-
nents will be discussed later in this chapter. The ers, it is seldom a surprise to us. This sense of or-
variance accounted for in the child’s performance derliness or predictability about our own behavior
level will be presented in Chapters 11 and 12. agrees with the findings from a large number of
One would expect that effective treatment studies by such investigators as Sackett (1977),
would result in several kinds of changes in the Bartlett (1951), Raush (1965), and Gottman
stimuli that control the behavior of the problem (1979). The findings also emphasize the fact that
child. The connections between antecedents and much of what we call social behavior comes in
the child’s coercive target behaviors should be- “chunks.” Once an event begins, it is likely to con-
come weaker. Abrasive antecedents presented by tinue.
family members should decrease in number. The This, in turn, leads us to reconsider in what
network of antecedents controlling prosocial be- form we should express our dependent variables
haviors may increase. In one study, 94.7 hours of for microsocial analyses. As pointed out by Har-
therapist time produced only modest changes in an tup (1979), frequency and rate measures neces-
extremely difficult family; the comparison of base- sarily distort this most important characteristic of
line to intervention showed a reduction of 50% in social behavior. Behavioral events must be de-
overall deviancy for the problem child (Patterson, scribed by both their duration and frequency. The
1973). A functional analysis showed that of the 35 dependent variable for future interactional studies
antecedents controlling the child’s 14 coercive at OSLC will almost certainly be duration meas-
Rj’s, only a few continued to do so at termination. ures.
Eighty-five percent of the controlling stimuli This perspective has several implications. For
showed some reduction in their power scores at one thing, the concept of duration introduces the
termination. While many of these reductions in necessity of considering boundaries for an event.
power score were minor, the trends were in keep- These were also questions considered in the 1960's
ing with the general notion that effective treatment by R. Barker and his colleagues (Barker, 1963).
alters the structure of family interaction. When the When does an event begin, and when does it end?
family moved to a job in a different city, the thera- When the contingency is embedded in the continu-
py was far from complete. The residual high rates ous flow of social interaction, what is the unit of
of deviant behavior and the fact that at termina- behavior being reinforced? What is strengthened;
tion nine new controlling events appeared had om- is it the event that preceded the reinforcer? Or is it

186
 the structure in which the target response is em-
bedded? Must structure be taken into account Serial Dependencies
when defining the boundaries? There are two techniques in general use that
This problem of units, boundaries, and relevant identify intraindividual serial dependencies, i.e.,
contingencies has always been with us, even in the the relation between the target person’s behavior
more tightly controlled laboratory setting. I be- at t and at t+1. One procedure is correlational
came intimately aware of the relation between re- and the other is conditional probability analyses
inforcement effects and response structure while (Martin et al., 1980; Gottman, 1979; Sackett,
studying children’s preference in a two-choice 1977). Both provide essentially the same informa-
preference task involving a continuous series of tion. Both can be lagged for the subject’s own im-
trials (Patterson & Hinsey, 1964). Following one mediately prior behavior at t—1 or prior behavior
study, the event recorder tapes from the perfor- at t—2,etc. For the moment, it seems that the one
mance of 24 subjects during baseline and “condi- that is used depends upon which computer pro-
tioning” were festooned across the walls. By walk- grams are available and whose paper most recent-
ing and looking, it became apparent that during ly influenced the investigator.
baseline the response sequence for many subjects Regardless of which statistic is employed, there
was highly structured. They simply alternated are certain questions about the interconnection of
ABABA, etc. Was the reinforcer contingent upon the individual’s own behavior that must be an-
the child’s preference for A, or the pattern ABAB swered. Unfortunately, these necessary studies
in which it was embedded? Most of the children have hardly begun. For example, it would be use-
whose baseline preference scores for A and B were ful to know the correlations between adjacent seg-
close to 50/50 splits employed the alternation ments (Lag 1) for prosocial and for coercive be-
strategy. The proportion of responses involved in haviors separately for each family member. It
such alternation patterns was highly stable would also be useful to have some estimate for the
(r=.88), as shown in comparing the p values for duration of such thematic effects. Given that the
the first and second 100 trials in an extended base- problem child has engaged in a coercive behavior,
line study. Only 49% of these “alternation” chil- how many subsequent behaviors are characterized
dren showed a shift in preference during condi- by an increased probability for yet further coercive
tioning. This was in contrast to the 80% for those behaviors? Sackett (1977) has analyzed such se-
with infrequent alternation and 63% for those quences for primates, as has Gottman (1979) for
with moderate alternation. married couples, but no comparable studies for
Other investigators working with more complex families have been made. Are the coercive child’s
manipulations involving the operant shaping of own behaviors interconnected if we examine adja-
speech noted similar problems. Salzinger, Portnoy, cent ten- or 20-minute time blocks? Are the behav-
Zlotogura, and Keisner (1963) found that when ior frequencies correlated if based upon observa-
attempting to reinforce pronouns, they often pro- tions made over a series of days? To understand
duced side effects, such as changes in rate of how the problem child’s behavior is structured, we
speech, or in certain subgroups of words. must first have the answers to these questions.
Thomas and Martin (1976), Thomas and Ma- In our own analyses, we began by asking wheth-
lone (1979), and Martin, Maccoby, Baron, and er some problem-child behaviors were more likely
Jacklin (1980) raise one further implication. They to persist (have longer durations) than others. The
note that a covariation between Aj and R; reflects analysis focused upon the likelihood of a second
two sets of determinants. One is the eliciting status response given the first (p[R,+1|R;]). Patterson
of the controlling antecedent event; the other is the (1976) analyzed the data for a mixed sample of 27
subject’s own prior behavior. For example, the antisocial and 27 matched nonproblem boys. The
target subject’s first response may elicit two out- six to ten baseline sessions for each child were
comes. One is the reaction of the other person, combined to form a population of response
which increases the likelihood of R,; but R; may events. The p values for the continuance or recur-
also directly elicit R,. For example, one study of rence of each of the 14 coercive responses are giv-
18-month-old infants interacting with their moth- en in Table 8.4. Given the first Physical Negative,
ers showed that both members were more influ- there was a very good chance that behavior would
enced by their own prior behavior than by the an- continue to the next six seconds for both the clini-
tecedent reaction of the other person (Martin et cal (.23) and nonclinical samples (.17). Notice
al., 1980). This paper underscores the importance that for Cry the probability for continuance was
of the intrasubject component as a determinant for very high for both samples. Nine of the 14 coer-
ongoing behavior. cive responses showed greater likelihoods of con-

187
 badly skew the resulting p values. The next analy-
sis was based upon the samples of 30 socially ag-
Table 8.4 gressive boys and 20 matched normal boys de-
Likelihood of Recurrence/Persistence scribed in Appendix 5.1. The analyses were car-
of a Coercive Response ried out separately by age and by the quality of the
antecedent. Each prior behavior of the other person
(adapted from Patterson, 1976, p. 272) was categorized as positive, neutral, or aversive.
The data in Table 8.5 summarize two different
Nonproblem Antisocial ways of describing coercive sequences. The first
Boys, N=27 Boys, N=27 was simply counting the number of events in each
Code Category —p(Rr|Rr-1) p(Re|Re-1) coercive sequence. This was tabulated separately
for each child; i.e., the sample was children, not
CommandNegative .000 .144 events. As shown there, the coercive sequences for
Cry .658 594 socially aggressive boys were significantly longer
Disapproval .114 .203 (mean 1.50 events) than were those for nonprob-
Dependency PRIN 225 lem boys (mean 1.16 events).
Destructive .182 .610 Another way of expressing this is to calculate
High Rate 281 .492 the conditional p of a second coercive event (or a
Humiliate .000 .203 repetition) given a first event. As expected, the
Ignore 231 .100 boys in the clinical sample were significantly more
Noncomply 158 .187 likely than Normals to continue their coercive
Negativism Sali .284 chains. Across contexts (antecedents) the values
PhysicalNegative .167 eS) were .28 and .13 respectively.’ The likelihood of
Tease MES bys) chain extension was significantly greater for the
Whine 379 293) clinical sample when the antecedents were either
Yell 254 .206 positive or aversive. It can also be seen for both
SocialAggression samples that chain extension was most likely to
(PN 6, TE) TA. 263 occur when the antecedent was positive. This find-
Hostility (DI, NE, ing was counterintuitive. One would have thought
HU, IG, WH) .270 334 chain extensions were more likely to occur during
Total Deviant
the attack and counterattack responses that char-
Behavior acterize episodes with aversive antecedents.
(all 14 R;’s) .302 .388
These lag, conditional, or correlational analyses
hardly scratch the surface of possible response de-
pendencies. What about the patterns involving
tinuance for the clinical than for the nonproblem more than a single category, e.g., episodes such as
sample. Yell —»Whine —»Hit? Thus far at OSLC we have
The data reviewed earlier showed that for the considered only those possibilities that involve ho-
clinical sample there is a higher likelihood of a co- mogeneous sequences. When one considers the
ercive behavior being initiated (.19 versus .14 for permutations and combinations generated by 29
Normals). The data in Table 8.4 showed that alternatives, the prospect is overwhelming. Fur-
there was a greater likelihood of continuance for thermore, the intrasubject thematic effect does not
the clinical than for the normal sample (.39 versus necessarily have to appear as a patterned se-
.30 for Normals). In examining more recent data quence. Given a coercive event, there may be an
relating to the likelihood of continuance for Nor- increased likelihood for further coercive events at
mals, I was surprised to see p values of 0.0 for any one or all of the next dozen subject reactions.
about half of the sample! These children always Sackett’s (1977) sequential analysis is ameans for
emitted a single coercive behavior and then identifying such an effect (Gottman, 1979; Gott-
stopped. There was only one child in the Social man & Bakeman, 1979). The procedure requires
Aggressor sample for whom the p value was 0.0. that one begin with a criterion behavior, such as a
The 1976 study was based upon a mixed sample coercive behavior, and then search at each lag that
of antisocial boys, some referred because of steal- follows for a recurrence of coercive behaviors.
ing and others for social aggression. There was, in The likelihood of a coercive event is calculated for
addition, the already noted complication that re- each lag; Sackett carries this out to 16 lags in his
sults from sampling events rather than subjects; analysis of primate interactions. I think this is po-
i.e., the contribution of a few individuals may tentially a powerful tool for the analysis of family

188
 Table 8.5
Comparisons for Duration of Coercive Sequences

Normals Social Aggressors F values for

Younger Older Younger Older

Antecedent (N=10) (N=10) (N=15) (N=15) Samples Age Interaction
Mean Number of Events in Sequence
General ilGa
|B) 1.16 (9) 1.54 1.46 BOE rs 0.31 0.60
Mean Likelihood Event Extension p(R,|R; —1)
General sale ailis’ 30 yp) DRTye 0.85 0.60
Aversive 06 (9)t 5i5.((74) 53 30 1168 ~ 0.61 1.04
Neutral .35 (8) .16 (9) .22 (14) 14 1.24 B98 0.64
Positive 14 (9) .14 (9) soy 5)0) 1346 0.07 0.06
“p<.05
eae SAOM
***p<.001
+Numbers in parentheses refer to analyses that are based on a subgroup of the entire sample.

interactions. For example, is the duration of these structure would be a giant probability tree. Given
ripple-like effects greater for some family members massive amounts of data, we could empirically
than for others? Are the likelihoods for the recur- trace out the multiple paths or outcomes that are
rence of a coercive behavior greater at Lag 6 or possible from a single initiating event. As an ex-
Lag 16 for members of clinical than for members ample of this, Figure 8.7 follows the course for a
of nonclinical samples? If so, it means the possibil- single target event and the consequences that con-
ity for extended conflicts and/or the restarting of trolled its recurrence. A prior analysis for a single
conflicts is higher for members of the clinical sam- subject identified a functionally defined class
ples. (Hostile) of behaviors, all controlled by the same
network of antecedents (Patterson, 1974c). The
base rate for hostile initiations was .120. The data
Inter- and Intrasubject Components from the 50 observation sessions also showed a
Together common set of consequences that increased the
For ease of discussion, the inter- and intrasub- likelihood for recurrence of events in this class.
ject components of structure have been viewed Given the target event, then the likelihood that
separately. In actuality, the labels, inter- and intra- the parents would supply an accelerating conse-
subject components, are a kind of convenient fic- quence was .285. This type of consequence had
tion. The separate labels imply an independence been identified as the intersubject component most
that does not exist. For example, given an A; that likely to contribute to the child’s continuing to
controls R2, the subject’s prior behavior, R,, is perform the target event. As can been seen in the
also functionally related to R;. It is my assumption figure, the occurrence of the accelerating conse-
that the subject’s R; and the other person’s A; are quence at this juncture was correlated with a dra-
also functionally related. Each juncture in the se- matic increase in the likelihood of a recurrence for
quence is functionally related to what follows. the target event (.519). Notice, however, that the
The relative contribution of one or the other must intrasubject component is very much in evidence
vary as a function of type of target response and (.481). This means that regardless of parent reac-
who the reactive agent is. As yet there are few tions, the child is likely to perform two target
comparative studies, although the few that have events in sequence. It cannot be said, then, that at
been done note the larger contribution of the sub- this juncture the intersubject component contrib-
ject’s own prior behavior (Martin et al., 1980; utes very much.
Thomas & Malone, 1979). Now there are shifts in structure that emphasize
The most convenient metaphor for the overall the nonstationary and dependent status of the con-

189
 Figure 8.7
Inter- and Intrasubject Components
(adapted from Patterson, 1974c, p. 908)

Target Response
(.120)

Intersubject Component Accelerating Consequence

(.285)

ae ie ae en
Intrasubject Component Nontarget Event Target Event Recur
(.481)
(.519)
eae Gy i
Intersubject Component Accelerating Consequence OtherConsequence
(.408) (.592)

Intrasubject Component Target Event Recur Target Event Recur

(.471) (.407)

Numbers in parentheses refer to the likelihood of occurrence.

sequences for this family; i.e., the events are not coercive component is an indication of this.
independent of trials and sequences (Coombs, The simplest assumption that one can make
Dawes, & Tversky, 1970). First, given two target about interaction sequences is that they can be cat-
events in sequence, the parents become increasing- egorized by a limited number of states (e.g., posi-
ly likely to provide the reaction (.408) that is most tive, negative, and neutral), and that the state at
likely to accelerate the child’s coercive behavior! any given time is a function of the immediately pri-
The more extended the child’s chain, the more or state (Coombs et al., 1970). These are the as-
likely he was to receive the accelerating conse- sumptions underlying the Markov chain. Does
quence. However, recent studies of such sequences family interaction meet the requirements for a
by E. Thomas (in preparation) suggest that these Markov chain? I don’t think so, but we are just be-
increases may be expected because of probability ginning to examine this question. The findings
summation and do not necessarily demonstrate from the probability tree data suggest that some
that longer chains have some unexpected function- sequences are associated with marked increases in
al utility. the likelihood of coercive child behavior. Could
Note in the original study the intrasubject com- not such a sequence be considered an “event”? I
ponent remains in evidence. Given that the parents have a sense that this very important dialogue is
reacted by providing first an accelerating conse- just beginning. For example, Raush (1965) also
quence and then a nonaccelerating consequence in noted that for antisocial children the likelihood of
sequence, the likelihood of a third target event re- unfriendly acts did not seem to follow a simple
mained a very high .407. This is in contrast to the Markov process. What is required are studies that
likelihood of .471 when the intersubject compo- formally compare predictions from Markov analy-
nents had been supportive of his prior two target ses to the empirical findings.
events. This, incidentally, was a very deviant Patterson and Moore (1979) studied a mother-
child. The extreme persistence of his intrasubject child dyad that was illustrative of the functional

190
 Figure 8.8
Inter- and Intrasubject Components for Tina and Mother
(from Patterson & Moore, 1979, p. 92)

given punish- given punish-

Mother prosocial prosocial prosocial
ment, then ment, then

Mother
p(prosocial,/prosocial,)= .18
p(Aversive,/Aversive,)= .52
p(Aversive;/Aversive,)= .30
p(Aversive;/Aversive,)
( = .57

Patterson & Moore, Interactive patterns as units of behavior. In S$.Suomi, M. Lamb, and G. Stephenson, editors, Social Interaction
Analysis: Methodological Issues (Madison: The University of Wisconsin Press; © 1979 by the Board of Regents of the University of
Wisconsin System). Reprinted with permission.

relation between the inter- and intrasubject com- Complain, then the likelihood of a second Com-
ponents. The study was based upon 60 sessions in plain was .44. The intersubject component (moth-
the home of 6-year-old Tina and her mother. er-Punish) seemed to increase the probability of re-
Tina-Complain served as a dependent variable for currence. Notice, however, that the intersubject
a series of analyses. As a general case, the intra- component added nothing to the prediction of the
subject analysis showed that Tina’s mother was third target event.
disposed to continue her irritable reactions once As a next step, we analyzed the outcomes for
she initiated them to the child (.52). However, giv- patterns. For example, an aggregate of two child-
en that the mother’s reaction was prosocial, the Complains was accompanied by a likelihood of .75
likelihood was only .18 that she would continue that the mother’s reaction to the first and had been
into a second interval. As shown in Figure 8.8, noxious. Given an aggregate of three Complains
once Tina began to complain, the likelihood of a in sequence, then the likelihood of at least one of
second Complain was .33; the likelihood of a the mother’s intervening behaviors having been
third Complain, given a second, was .44. The in- noxious was .88. The claim was made that these
trasubject components for coercives were of high findings support the idea that extended chains
magnitude for both the mother and the daughter. may have a unique function; i.e., the child per-
Note that at each point in the sequence, Tina- sisted until the mother terminated her aversiveness
Complain, there was a strong likelihood of the and/or shifted to prosocial behavior. Was the
mother’s being prosocial! These probability values mother significantly more likely to react in a pro-
constitute a reliable increase over the base-rate social manner to the third Complain than to the
likelihood for a prosocial reaction from the moth- first? If so, then the sequence is not independent of
phase. Does a sequence of three Complains consti-
er.
Given that the mother punished Tina’s first tute an event whose function is demonstrably dif-

191
ferent from that served by a single Complain? A
series of discussions with Ewart Thomas convinced
me otherwise. By using base-rate information and Figure 8.9
probability summation, he was able to provide a Interaction Sequences of
reasonably close fit to our empirical findings for
Hyperaggressive Boys
likelihoods of the mother’s intervening. noxious
and prosocial responses as a function of the num- (from Raush, 1965, p. 495)

ber of Complains. I find his argument persuasive

and stand neutral on the issue of whether social in- 100
teraction does or does not fit a Markov model.
A more convincing demonstration of how social
interaction departs from a Markov chain has been 80
presented by Raush (1965). The reader should 2
3)
note, however, that the format was completely bi- <
lateral in that it assumed that the behavior of Sub- =|> 60
ject A was always followed by a reaction from =|
cv)
Subject B; i.e., Subject A’s second behavior never ‘=
Ruy
directly followed his first behavior. (See Thomas = 40
and Martin [1976] and Thomas and Malone )
)
q
[1979] for examples of more flexible models.) In 3
o)
the Raush study, six hyperaggressive boys were 20 @ Obtained
observed prior to, during, and following treat- O Hypothetical
ment. Their behaviors were classified as either
friendly (F) or unfriendly (Un). Early in treatment, 0
the base rates for these behaviors were p(F) of .70
t; t, G t, t;
and p(Un) of .30. The conditional p’s of most rele-
vance were p(F,|Un) of .25 and p(F2|F;) of .55. Sequential Position (lag)
These conditional p values plus the base-rate val-
ues can be used to describe the p that the process
will be in state F or Un at any given trial. A Reprinted with permission from “Interaction sequences,” Jour-
nal of Personality and Social Psychology. Copyright 1965 by
straightforward application of Baysean theory
the American Psychological Association.
would show the likelihood of the dyad being in an
unfriendly state at time 2, time 3, time 4, and so
on. As shown in Figure 8.9, for the data collected
prior to treatment, these boys were increasingly ers? Just how generalizable is it? These are inter-
likely to become engaged in unfriendly acts as the esting questions for which available code systems
sequence proceeded. Their observed behavior sug- and computer software offer a technology suffi-
gested an even greater likelihood than predicted by cient to provide some answers.
the Markov chain analysis. The contributing fac- There are many questions yet to be answered
tor lay in their great likelihood of launching “un- concerning the relative contributions of inter- and
provoked” attacks; the conditional was .08 for intrapersonal components and whether their posi-
Normals and .45 for the clinical sample. After two tion in a sequence makes any particular difference.
years of treatment, the conditional dropped to However, given our present incomplete under-
.19; this also means a lower p for Un in general. standing of these matters, can these concepts be
The Raush study is undoubtedly one of the most applied to the task of constructing a performance
important studies published in interactional psy- theory? The section that follows employs inter-
chology. It served as a prelude to the analyses and and intrasubject components to construct varia-
extensions by Cairns (1979), Suomi, Lamb, and bles that account for differences in the problem
Stephenson (1979), and Gottman (1979). For ex- child’s coercive performance level.
ample, here he presents a means by which the
reader can test for the question of whether social Irritable Exchanges
interaction follows a Markov chain or not. Does Earlier in this volume it was noted that socially
social interaction function as if neither party has a aggressive children live in aggressive families. At a
memory? Is the state in the next trial completely microsocial level, what is meant is that members
described by the state of the dyad at t—1? Does of these families are disposed to react irritably
the model hold for some interactions but not oth- with each other, particularly in their interactions

2
 with the problem child. There are four measures other family member reacts by nattering, then it is
of irritability: crossover, counterattack, punish- likely that the conflict will be extended. This reac-
ment acceleration, and continuance. Inter- and in- tion has been labeled counterattack. The hypothe-
* trasubject components that measure these disposi- sis is that members of families of antisocial chil-
tions provide the empirical basis for examining dren will be characterized by a greater likelihood
this process. The general idea is that problem fam- of counterattacks. The comparisons among nor-
ilies are more likely to start conflicts. Given that a mal and clinical samples (Table 7.6) showed that
conflict is started, the other members are likely to all of the trends were in keeping with the hypothe-
react in such a way as to keep the conflict going. sis. The disposition to counterattack was signifi-
This formulation is similar to the one presented by cant for some dyads but not for others.
Gottman (1979). His laboratory study showed In the clinical samples the mother and the father
that when a distressed spouse initiated an aversive, were significantly more likely than parents of nor-
the other was likely to respond in kind. His se- mal families to react irritably to coercive initia-
quential lag analyses of the husband and wife reac- tions by the target child. The problem child was
tions that followed the initiation were character- significantly more likely than normal children to
ized by an alternation. The couple seemed to “take react irritably to coercive initiations by the moth-
turns” being aversive. The net effect was to extend er. It is important to note that at this second point
the conflict. He labeled this process “negative-af- in the sequence it is the irritable reactions of the
fect cycle.” parents and the problem child that were signifi-
The first hypothesis is that members of families cant. This suggests that to understand the aggres-
of antisocial children are more likely than mem- sive child it may be necessary to study the parent-
bers of normal families to “start” conflicts. Given child interchanges with even greater care than sib-
that the other family member is behaving in a neu- ling-child interchanges. This hypothesis is exam-
tral or prosocial manner, there is a significantly ined in detail in Chapter 12. Chapter 11 summar-
greater likelihood that members of distressed fam- izes the covariations between measures of child
ilies will initiate an attack; this is labeled “cross- antisocial behavior and the dispositions of family
over.” A comparison of the likelihoods for samples members to react irritably to coercive initiations.
of Normals, Stealers, and Social Aggressors (Table Given an initiation by a family member and a
7.4) showed significant differences for all family counterattack, what is the likelihood that the aver-
members. The mothers, fathers, and siblings in sive exchange will continue? This next conditional
the clinical samples were about twice as likely as p value describes the punishment-acceleration in-
Normals to start a conflict with the target child. dex encountered in Chapter 6. In their interactions
When examining the likelihood for the problem with the target child, the mother, father, and sib-
child initiating a conflict, the differences among lings from families in the clinical samples showed
samples were significant only for his interactions nonsignificant trends for a greater likelihood of
with the mother (Table 7.4). In fact, the problem chain extension than did members of normal fami-
child’s favorite target for starting fights was the lies (Table 6.8). However, the problem child
mother by a factor of two to one. showed significantly greater punishment-accelera-
The importance of the crossover phenomenon tion indices than did normal children in his inter-
was first noted by Raush (1965). He found this to actions with mother, father, and siblings.
be the intersubject variable that most significantly The next measure describes the disposition to
differentiated between the interactions of aggres- persevere in one’s irritable reactions regardless of
sive and normal children. Given a friendly reac- the reaction of the other family member. Table 8.6
tion, the clinical sample was more likely to make summarizes the findings for the six comparisons
an unfriendly response. The crossover variable involving the target child and other family mem-
will be examined in Chapters 11 and 12 to deter- bers. The data were based upon samples of Nor-
mine the significance of its contribution to prob- mals, Social Aggressors, and Stealers described in
lem child performance level. Presumably families Appendix 6.2. As can be seen in the table, this in-
whose members are characterized by greater dis- trasubject component significantly differentiated
positions for starting fights would have problem normal from clinical samples for five out of the six
children with higher TAB scores. dyads. Members of distressed families were more
The next variable is also an intersubject compo- likely to persist in their irritable reactions regard-
nent. Given that either member starts a conflict, less of how the other person reacted. The findings
what is the likelihood that the other will counter- summarized in Chapters 11 and 12 show this par-
attack? If a fight starts but no one reacts, then the ticular variable to be a key to understanding chil-
fight is likely to stop right there. If, however, the dren’s socially aggressive behavior. The mothers’

193
 Table 8.6
Probability of Irritable Continuance for Dyads Involving the Target Child

Mean Probability for Samples

Normals Social Aggressors Stealers
Continuance of Irritable Reactions by: (N =38) (N=37) (N=38) F Value

Mother interacting with Target Child m2 .266 7 15.00="=

Target Child interacting with Mother 2125) 321 .238 13.867**
Sibling interacting with Target Child .208 Pei, wet, 1.50
Target Child interacting with Sibling .142 302 273 8.29"
Father interacting with Target Child anal .262 .203 9.04***
Target Child Interacting with Father .049 .283 .202 9/07

ee *D< .001

Figure 8.10
Likelihood of Irritable Reactions at Different Junctures for Mother-Child Dyads

p(crossover) p(counterattack) p(punishment p(continuance)

Saas

Mother ==
apes
Extended Coercive Exchange

~J44a p(crossover) p(counterattack) p(punishment

ie he, p(continuance)

and, to a lesser extent, the fathers’ disposition to tion with the mother as a key to understanding
continue their irritable reactions account for the what produces the problem. Figure 8.10 presents
largest portion of the variance in the performance the data for the Social Aggressor sample that de-
of antisocial child behavior. In another study, the scribe the disposition of one or the other to react
mothers’ irritable reactions during family problem in an irritable fashion. Given that they are inter-
solving correlated with poor problem resolution acting with each other in a neutral or prosocial
(Forgatch & Patterson, in preparation). fashion, the likelihood is .09 that one or the other
The material discussed in Chapters 11 and 12 will initiate an attack. Given ten exchanges per
emphasizes the role of the problem child’s interac- minute, the odds are very high that in a ten-minute

194
 Table 8.7
Intercorrelations Among Irritability Values for Mother-Child Dyad
Mother-Child Variables:
Child-Mother Punishment
Variables: Crossover Punishment Continuance Acceleration

Intercorrelations for Mothers

Crossover .50 .26

Punishment 74 56 4

Continuance Ey), .63

PunishmentAcceleration jl so 81

Intercorrelations for Children

Note: Any correlation over .20 is significant at p< .05.

period a conflict will start. Once a conflict begins, The child trait for aggression can be viewed in
both the child and the mother tend to persist in several ways. On the one hand, it is a stable dispo-
their irritable reactions, and there is about one sition to react to family members in an irritable
chance in four of an extended interchange. For fashion. The studies reviewed in Chapter 3
such a dyad, finding a peaceful interlude must be a showed that the child’s TAB score was stable from
precarious business. one week to the next and even from one year to
I think the four irritability measures (crossover, the next. We have come to think of this as a trait
counterattack, punishment acceleration, and con- measure for social aggression. It is also the case
tinuance) can be thought of as facets of a trait. that the child’s irritability p values define what is
The assumption would be that the disposition to meant by the TAB score, i.e., its component parts.
react in an irritable fashion would be stable across As shown in Chapter 11, each of the p values cor-
time and across settings. For both the mother and relates heavily with the TAB score. However, I be-
the child (and other family members as well) it is a lieve it is also the case that the child’s disposition
style of interacting with people and a means of to react irritably reflects the fact that others are re-
coping with problems. As a first step in testing acting to him in a similar manner. Social aggres-
such a hypothesis, it would be assumed that all sion, as measured by TAB scores, irritability
four measures of irritability would intercorrelate. scores, or Parent Daily Report, is a bilateral trait.
Table 8.7 summarizes the intercorrelations for It reflects the disposition of the child to react irri-
values that describe mother-child interchanges for tably; the magnitude also reflects the disposition
a sample of 112 families. The figures above the di- of family members to respond in kind. Note in Ta-
agonal describe the relations among the measures ble 8.7, for example, that with the exception of
for mothers’ behavior. Those below the diagonal crossover, the trait scores for mother and child in-
describe the relations among measures for the tercorrelate. Children with very high probability
child. The figures in the diagonal are the correla- of continuance scores are interacting with mothers
tions between trait scores for mother and child. who are similarly disposed.
There seem to be larger correlations among the A bilateral definition implies that if an individu-
measures for the child than for the mother. For ex- al performs socially aggressive behaviors at high
ample, the correlation between the likelihood of rates, then the members of his social environment
child crossover and child punishment was .74. are the major determinants. The child’s aggressive
The comparable value for mothers was .46. In ei- behavior can be changed by either altering his dis-
ther case the irritability values tend to covary. In position to initiate and react aversively and/or by
that sense, it can be said that the four measures altering these same dispositions for his family
form a kind of trait cluster for the mother and the members.
child.

195
 sponse to challenge made this phase of our explor-
Some Implications of Bilateral Emphasis ations a very productive one.
The focus upon inter- and intrasubject compo- I also wish to thank Alice Harris and Rolf Loe-
nents emphasizes the fact that much of each mem- ber for their critique of the earliest drafts of this
ber’s ongoing behavior is jointly determined. A chapter. John Reid provided his usual thorough
given event partially reflects the prior behavior of critique for a later version. Finally, there is the
the other person, as well as his own immediately kindly Dr. Hartmann, who sacrificed countless
prior behavior. An extended coercive interchange evenings of his social life to generate 18 pages of
may then be said to be the outcome of a bilateral critique for what I had previously thought of as a
exchange. As the interchange unfolds, both mem- flawless manuscript. My riposte required three
bers of the dyad alter each other’s reactions. Such weeks of writing time; the resulting chapter is, I
a bilateral point of view of parent-child interaction believe, considerably improved. And to you, Don
has, of late, become fashionable (Bell, 1968; Bell Hartmann, I say:
& Harper, 1977). This viewpoint is in marked
“John zegt dat je je moeder eens moet schrijven
contrast to the earlier child psychologists who,
en doe haar de groeten van hem.”
with some notable exceptions such as R. Barker
(1963), viewed the child as a passive-reactive or-
ganism. Parental attitudes and child-rearing prac-
tices had impact upon the child. Child behavior Footnotes
was thought to mirror the interplay between the 1. We conducted a computer analysis to deter-
forces of genetic endowment and environmental mine how much data would be lost by decision
exigencies. rules requiring up to 30 seconds of prosocial ante-
In the past decade the field of psychology has cedents. Informal analyses suggested that extend-
matured sufficiently to provide a more complex al- ing the interval beyond 18 seconds resulted in the
ternative. From that perspective the child is loss of too much data. Less than 18 seconds would
viewed as an active participant. He actively se- mean a lack of certainty about whether the the-
lects, stores, and organizes information; i.e., he is matic component had been disrupted. This analy-
both a reactor and a selector. As noted by Baltes sis should be repeated with larger samples and pre-
and Reese (1977) there are anumber of philosoph- sented with statistical analyses describing the out-
ical problems raised by the bilateral perspective. Is come for the various decisions.
the effect to be thought of as quantifiable and 2. The term base rate does not refer to a fixed
made up of component parts or as a qualitative number. It is always relative to the setting(s) in
emergence which can only be defined as a which the data were collected. By definition, sam-
“whole”? I have made a case for the idea that these pling across the population of (all) settings would
are mutual actions that are additive and unidirec- give lower base-rate values for all behavior. Even-
tional. As I perceive it, the child is an active parti- tually, the study of interactional psychology must
cipant whose behavior is a reaction to the behav- delineate base-rate values separately by setting.
ior of the other family members and also consti- 3. The magnitude of the conditional p values
tutes a stimulus for their behaviors. A behavioral obtained in these studies in no way matched the
event is an effect and a cause; in fact, one could precision noted by Hinde (1974) in his review of
describe several interchanges as a linear series of the findings from ethology, e.g., for agonistic dis-
cause-and-effect relations. These and the related plays in birds. For the ethologist, knowing wheth-
questions of fixity and independence of events in er the wings are raised or the nape is erect during
interactional sequences will be the focus for the body-horizontal conditions provides considerable
next generation of studies about structure and predictive power. The event mother-Ignore is, by
process. comparison, a much less powerful event.
4. The parents’ ability to make fine discrimina-
tions about their own or their child’s behavior def-
initely sets a ceiling on the complexity of behaviors
that can be studied in these studies. If the mother’s
definitions differ from those of the experimenter,
Acknowledgments then the study will not be successful. A case in
The writer is deeply grateful for the extensive ef- point involved our efforts to study reinforcement
forts of Katie Whalen and Marion Forgatch, who effects with Spring as the subject. It was assumed
served as research assistants for the experiments that if the sequence mother-Argue —>Spring-Ar-
reported here. Their patience and innovative re- gue was followed by mother terminate her Argue
 and interact in a prosocial manner, this would sequences R,|R, (chain extension) in the 1976
negatively reinforce Spring’s Argue. During base- study were considerably higher. I do not under-
line, the p(Spring-Argue|mother-Argue) was .52; stand the discrepancy between early and late sam-
the mean duration of Spring-Argue was 8.83 sec- plings for Normals except to note that the late ad-
onds. ditions were less deviant. Snyder (1977) found no
During the four days of experimental manipula- differences between normal and clinical samples in
tion, it became obvious that the mother could not the likelihood of chain extension (p values of .30
differentiate Spring-Argue from Spring-Complain, and .29 respectively). The findings for his clinical
-Disapprove, or -Whine. Not being able to clearly sample exactly match those for the OSLC sample,
discriminate that event meant that she was not but there was a major discrepancy in findings be-
successful in reinforcing Argue contingently. As a tween his normal sample and those for OSLC.
result, four days of effort produced no real 6. Computer programs for conducting sequen-
changes in the Aj—» R; bond. The mean condi- tial data analyses are available from Roger Bake-
tional p value for mother-Argue given Spring- man, Department of Psychology, Georgia State
Argue was .56. University, Atlanta, Georgia 30303 (see also Sack-
5. Note that our original estimates for response ett, 1977).

£97.
 Chapter 9
Abstract
Turn-taking is presented as a primitive building component for synchronous interactions. Synchro-
nicity refers to the functional relation between the behavior of one person and the immediate reaction
of another. It has particular relevance to extended episodes. The data indicate that the use of approv-
al by family members is not consistently synchronous; however, aversive events produce both
matched and nonmatched synchronous reactions.
Both positive events and aversive events produce “ripple effects,” i.e., increased probabilities for a
general continuance of the behavior. Distressed family members are thought to be characterized by a
greater general disposition to be aversive with each other and nondistressed family members by a
greater disposition to be positive with each other.
The concepts of reciprocity and exchange seem highly relevant to family distress and breakdown.
However, a review of the social learning studies to date suggests that analyses of social interaction
have not yet provided a suitable means for testing hypotheses relating to reciprocity.

198
 Chapter 9

Turn-Taking,
Synchronicity,
and Reciprocity
The functional relation between the behavior of terned sequence persists for a very long period of
one person and another is embedded in larger time. Cairns labeled such extended structures as
structures. The nature of these larger structures synchronous: the term “refers to a property of in-
varies a good deal. In some instances they are con- teractions which obtains when one person’s acts
ventions built into the culturally defined programs are coordinated with and supportive of the ongo-
which govern large segments of our behavior. In ing activity of another individual” (Cairns, 1979a,
fact, one might think of them as “rules” that stipu- p. 298). As I use the term, it refers to any kind of
late what is appropriate and what is not. One of functional relation between the behavior of one
the most interesting examples of this would be person and another, including extended inter-
“turn-taking” during talking. This basic social in- changes. These interchanges are often character-
teraction skill is taught during infancy. Gradually, ized by a shared content. Cairns notes that one of
turn-taking becomes so much a part of our reper- the properties of punishment is that it disrupts
toire that its presence is not noted. However, turn- these ongoing themes. Incidentally, it is interesting
taking does generate much of the structure identi- to consider punishment in this role of establishing
fied in sequential data, and marked disruptions in the “setting” for an event. One of the functions of
turn-taking may evoke the use of such labels as punishment may be that it redefines the setting;
“monologist” or “eccentric.” The research litera- i.e., it signals that this is the end of one theme and
ture relating to these topics will be considered in the beginning of another.
more detail in a later section of this chapter. To understand extended structures (or synchro-
Turn-taking tells us only that we should expect nous interactions as Cairns puts it) we need to de-
to find alternation when we study social interac- velop a means of cataloging episodes. Interaction-
tion. It is abackground theme against which inter- al psychology must eventually return to the task
actional melodies are played. It means that the set for us by Roger Barker (1963) and his col-
content of what one person does is often reflected leagues. They introduced the study of behavioral
in the behavior of the other person. For example, “episodes.” They carried out a series of methodo-
two elderly checkers players sitting at their table in logical studies which demonstrated that judges
the park engage in only two or three categories of could agree at a modest level in identifying the
behavior, but they alternate both in their verbal re- boundaries where one episode ended and another
actions and in many of their motor movements. began. Given a catalog for episodes, we could be-
Their behavior is functionally related. The pat- gin the search for determinants. Much of the in-

199
formation about these antecedents must be built then respond in kind. As Lieven viewed it, this was
into the setting itself. At high noon it is appropri- a basic communication skill.
ate for elderly men to play checkers in the park, One might think of this as a kind of modeling
but in the evening young people gather in the same effect made up of two components. First, there is a
setting to hold hands. matching of the other person’s behavior for con-
The microsocial data presented earlier showed tent; e.g., smiles produce smiles; vocalizing evokes
that given either the A; or the R;, we can.make vocalizing. Second, there is a rough matching for
some reasonably good guesses about which other duration of the behavior. In effect, the infant
events might have occurred in the same sequence. learns very early to take turns. Shields (1976) sees
A catalog of episodes and a search for their ante- this communication skill as essential for the acqui-
cedents could take us to the more interesting task sition of language. The child must learn to take
of predicting without relying upon information into account turn-taking plus a reflection and/or
about the prior or the following event. Such analy- extension of the meaning of what is being said.
ses would require a whole new generation of code Turn-taking is so commonplace that we no
systems. The codes which now exist limit the longer note its occurrence. With maturation, the
questions to much more mundane considerations. skill is elaborated by special cuing systems which
What do we know about turn-taking? Does syn- signal the end of an utterance and invite turn-tak-
chronicity hold for the immediate exchange of ing from the other person. In social interaction the
positives and aversives? pause between turn-taking is the point of maxi-
mum uncertainty. From the viewpoint of the
Turn-Taking speaker, his own behavior is highly predictable;
In the short run, there is for most of us a com- i.e., during this time there is a reduction in uncer-
fortable sense of continuity about ongoing interac- tainty. Much of his “sense” of continuity comes
tions in which we are involved. There is a vague from the fact that the connection between his con-
sense of predictability about it all. Most interac- secutive responses is eminently logical and predict-
tions do not contain many surprises. A novel able. What is surprising from his standpoint is
twist, a sudden humorous shift is the exception, how others react to this logical sequence. Some
not the rule. Those who are capable of producing persons, for example, might perceive a perfectly
such effects are highly valued (and well paid). A straightforward, supportive comment as being an
vaguely sensed comfort (almost boredom) is the attack! In this vein, Gottman (1979) found sig-
rule for most normal family interaction sequences. nificantly greater discrepancy for distressed than
Much of the sense of continuity comes from two nondistressed couples between the sending of a
structural components which are built into the in- positive communication and having it received as
fant’s repertoire even prior to the acquisition of such by the other. Even without such distortion,
language skills. These consist of an early disposi- awaiting the response of the other is the point of
tion to match the behavior of the other person greatest uncertainty. Which response will he
with regard to both the content and duration of make? Will he shift topics, or will he leave to talk
the response. As noted by Shields (1976, p. 315) to someone else? From the perspective of an indi-
“". . reciprocal activities between the child and its vidual, turn-taking in social interaction is experi-
caretaker in which each monitors and responds to enced as a sequence of regular pulsations from
the behavior of the other are well established be- greater to lesser uncertainty. As a dyad, they take
fore the first identifiable words appear.” Lieven’s turns in being uncertain. In long-standing relation-
(1976) tape recordings of 17- to 18-month-old in- ships the uncertainty should be of a lesser magni-
fants interacting with their mothers also showed tude than it is for newly formed relationships.
symmetry for their interaction. Given an infant The monologist chooses not to take any risk; he
vocalization, the likelihood that the mother would is absolutely certain of who will be talking five
vocalize was .76 for one mother, .51 for another, minutes from now. Very likely, he also knows
and .29 for a third. Given that the mother vocal- what he will be talking about. Gottman (1979, p.
ized, the likelihood.that the infant would vocalize 48) notes a pattern for distressed families in which
was only in the .20’s. As the study progressed, two turn-taking is replaced by monologue. Here, each
of the three infants became increasingly likely to person talks to the others; but he or she does not
synchronize their vocalizations with those of the react to what the others have said! It is not the rap-
mother. Although no data were presented regard- id alternation which implies, “Yes, I understood.
ing the mother’s description of these interchanges, Now here is how I react to it. What is your reac-
it sounded as if the infant would vocalize and then tion?” This disruption of turn-taking may be an
stop as if expecting the mother to vocalize and interesting but, as yet, largely neglected feature of

200
 familial conflict. One might think of distressed tration of this. This extremely high-rate person
families as having ritualized patterns of fighting can also be a borderline monologist. He speaks
which are highly predictable in both their linkage rapidly, then when it’s your turn, does he quietly
‘ and outcome. In that regard, Gottman makes the gaze and listen? No. No indeed! In your mid-sen-
following fascinating observations: tence, he begins rapidly nodding his head, inter-
jecting a “Yes, hmm, yes, yes.” All of this makes
“These findings are reminiscent of the second you feel as if you are moving much too slowly, and
law of thermodynamics that relates greater energy when you awkwardly pause trying to catch your
in a physical system with greater degree of disor- verbal balance, he charges in. He gets half your
der. The second law of thermodynamics also asso- turn. Etiquette prevents your saying anything. Be-
ciates high energy (and hence, less order and pat- sides, there is no label for what he has just done to
terning) with the capacity of a system to undergo
you.
spontaneous change. In other words, systems that Obviously dyads do not continue their synchro-
are highly patterned are more difficult to change nous turn-taking indefinitely. What determines
than more randomly patterned ones.” (Gottman, when a conversation will shift to a new topic or
1979, p. 48) stop altogether? It should be noted that empirical
work on this question has a very short history.
Signaling Turn-Taking Therefore, much of what we wish to know has not
Turn-taking could be a deadly bore. For exam- yet been examined. However, it seems that cues
ple, imagine if it were completely set by time con- which signal and support ongoing interchanges
straints. Then one could engage in conversations have both verbal and nonverbal components. For
that alternate speakers every two minutes, or per- example, Argyle and Cook (1976, p. 63) review a
haps engage in 7.5-minute parallel monologuing. series of studies in which visual cues were absent
The fact that social interaction is more flexible during telephone conversations. Here, one finds
than that suggests the presence of subtle cuing less synchronicity, i.e., more pauses and shorter
mechanisms. utterances. Verbal cues, such as “hmm” and “ah”
There have been a series of programmatic stud- occur more often.
ies by Exline at Delaware, then by Kendon, Ar- During social interchange the speaker uses gaze
gyle, and Cook at Oxford which have added to indicate the boundary of one utterance and as a
greatly to our understanding of this cuing process. prelude to his next statement. At this point the
This work consists of a series of elegant studies on other person may signal his interest with head
gaze as it is coordinated with speech and social in- nods and hmms. The content of what is being dis-
teraction. Much of the results from a decade’s cussed may eventually become of less interest to
work are summarized in Argyle and Cook’s Gaze one or both participants. If they become satiated,
and Mutual Gaze (1976). The combination of one or both persons may avert their gaze, and the
field and laboratory findings implicate gaze as an topic shifts through a rapid sequence until listen-
important source of information, particularly for ing, gaze, and turn-taking are resumed. This se-
the onset and offset of social discourse. For exam- quence of rapid satiation and quick channel shift-
ple, mutual gaze is an important prelude to initiat- ing is in full bloom in such settings as the airplane
ing a social contact. Avoidance of mutual gaze by or long bus trip. If both persons engage in it, then
the spouse, a colleague, or the committee chair- the shifts can cover entire life histories in very
man signals, “No, not now!” Typically, when lis- short order. The cocktail party is, of course, the
tening, we look at the person who is speaking; the classic example. Not only does channel shifting
person talking also tends to meet our gaze. At the occur by topics but, in addition, every several min-
end of an utterance, Person A gives a prolonged utes one must shift partners. Careful attention to
gaze to Person B. Person B then looks away and turn-taking and its accompanying verbal and non-
begins his utterance (see pp. 98-124 for a fascinat- verbal signal systems seems most evident during
ing account of these findings in Argyle and Cook, formal occasions. Greetings, introductions, and
1976). Gestural and other nonverbal cues also goodbyes seem to be carefully monitored ex-
provide signals which relate to turn-taking. For changes. Even here, however, the rules vary some-
example, a hand in mid-gesture at the end of an what as a function of social class and, of course,
utterance means, “Wait, more is coming!” Other by culture (Argyle & Cook, 1976).
investigators have noted shifts in body posture just If much of social interaction is indeed character-
prior to initiating an utterance. ized by turn-taking, then the data should take cer-
Turn-taking etiquette is often violated. The tac- tain forms. First, there should be a matching of
tics of the “conversation nibbler” are a good illus- category content. If Person A talks then Person B

201
 Table 9.1
Synchronicity and Symmetry for Positive Events
(adapted from Margolin, 1977, pp. 18-22)

Husband Wife
Behaviors D(Xy|Xw) p( Xi) p(Xw| Xx) p(Xw) xy

Possible Reinforcers
Approve .000 (.002) .000 (.004) Gh
Smile .022 (.006) .067* (.012) aD
Turn-Taking

Attention wool (.151) SOR? (.145) Dh

Talk 21967 (.058) 2205* (.051) 2555
*Comparison of conditional to base rate, z-score significant at p<.05.
**Comparison of conditional to base rate, z-score significant at p <.01.

talks and they continue to take turns, this should

be reflected in the conditional p value, p(A Talk|
Synchronous Matching
B Talk). Either this value or its converse should be for Reward and Punishment
greater than the base rate for Talk by either per-
Reinforcement
son. In other words, knowing what one person is
doing should provide a basis for “predicting” what There is nothing in the reinforcement theory lit-
the other member of a dyad is doing. The concepts erature which would lead one to expect a match-
of synchronicity and turn-taking lead us to expect ing of positive events on a moment-by-moment
people to alternate roles of speaker and listener. basis. Conventions of social intercourse dictate
The content of their behavior will also be func- that it is unacceptable to match positive events ex-
tionally related. The content may be similar, cept.in very special circumstances. “I like your new
matched, or the behavior of one may complement hairdo,” followed immediately by the reply,
the other, e.g., Tease-Laugh. Synchronicity refers “Thank you, I like your new dress,” is acceptable
only to the fact that the content is functionally re- behavior only for very young children or the most
lated. Notice, however, that it is possible to obtain superficial level of adult discourse (e.g., pro forma
high across-subject correlations from nonsynchro- greeting behavior, cocktail party discourse). Inti-
nous interchanges. Two children engaging in par- mates have very special rules about this: touch
allel play would not generate significant condi- may elicit touch; kiss may elicit kiss. But ordinari-
tional p values, but both would have high scores ly adults do not seem to be synchronous in their
on Play. This lack of symmetry for the two modes use of positive events such as approval and praise.
of analysis poses a special problem. We will return At one level, almost any behavior could be
to this problem later. thought of as a potential reinforcer. However, to
At this point it would be useful to design field provide a test for the matching hypothesis, I
studies to collect data on the synchronicity of posi- searched for data relating to a single event which
tive and aversive interactions for children of differ- most investigators would agree, a priori, is a rein-
ent ages and sex in several settings. To my know- forcer. Approval seemed to be a good example of
ledge these studies have yet to be carried out. such an event. There are two sets of findings
There is, however, a subset of findings which re- which provide data relevant to the concept of syn-
late to this issue and which are also of interest be- chronous matching for positive events.
cause of their relevance to coercion theory. It con- The data in Table 9.1 summarize the findings
cerns the synchronicity of rewards and punish- from the Margolin (1977) study of married cou-
ment in family interactions. Does synchronicity ples. The reactions were tabulated for 27 couples
extend to include a matching for the reinforcing or as they interacted in a 20-minute videotaped prob-
punishing behavior of the other person? lem-solving situation. The base-rate values are

202
 Table 9.2
Reactions to Target Child Behaviors by Male Siblings

Reactions by Male Siblings

Total
Frequency
Target Physical Base of Target
Child Approve Command Disapprove Negativism _ Negative Tease Whine Yell Rates TAB Behavior

Behaviors Dev. Nor. Dev. Nor. Dev. Nor. Dev. Nor. Dev. Nor. Dev. Nor. Dev. Nor. Dev. Nor. Dev. Nor. Dev. Nor. Dev. Nor.

Approve .000 .087 .000 .000 .059 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .002 .003 .059 .000 17° 23

Command 000 .000 .000 .000 .045 .000 .000 .000 .011 .000 .011 .020 .000 .000 .011 .000 .011 .007 .489 .360 88 50

Disapprove 000 .000 .010 .000 .198 .116 .010 .000 .046 .000 .010 .014 .005 .000 .005 .000 .026 .010 .360 .217 197 69

Negativism 000 .000 .018 .000 .109 .100 .027 .100 .009 .100 .000 .000 .000 .000 .009 .000 .014 .001 .236 .400 110 10

Physical
Negative .000 .000 .000 .000 .051 .375 .000 .000 .343 .125 .020 .000 .040 .000 .030 .000 .013 .001 .576 .500 99 8

Tease .000 .000 .035 .000 .221 .235 .012 .000 .081 .000 .047 .059 .012 .000 .070 .000 .011 .003 .558 .412 86 17

Whine 000 .000 .000 .333 .000 .000 .000 .000 .143 .000 .071 .000 .000 .000 .000 .000 .002 .000 .214 .333 14 3

Yell 000 .000 .000 .000 .155 .000 .000 .000 .048 .000 .012 .000 .000 .000 .012 .167 .011 .001 .369 .167 84 6

Base Rates 000 .001 .005 .003 .024 .008 .002 .002 .010 .001 .003 .002 .002 .001 .005 .002

TAB 000 .000 .011 .005 .111 .082 .007 .010 .071 .015 .012 .015 .008 .000 .016 .005 427 .303

presented in parentheses. The figures of greatest rather than subjects. The sample size is listed on
interest were the conditional p values describing the right side of the table. Given that the target
the immediate reaction of the spouse given that the child approved, the various reactions of his male
other spouse had approved. The average likeli- siblings are listed by columns. For the clinical sam-
hood of the husband’s approval, given an immedi- ple, the likelihood of an approval reaction by sib-
ately prior approval by the wife, was .000. The lings was .00 and it was .087 for the normal sam-
comparable conditional for wives was also .000. ple. For normals the conditional p value differed
Spouses in that situation were not synchronous in significantly from its base-rate value. It seems that
their use of this “reinforcer.” In fact, there was a synchronous matching of approval may hold for
trend for the conditionals to be Jess than the base normal siblings; it did not hold for siblings from
rate (i.e., approval by one person was related to distressed families. An examination of comparable
an inhibition for the use of approval by the other). data for mothers’ and fathers’ reactions to target
In my experience, a laugh or a smile seems to children revealed an interesting pattern. For nor-
elicit a like reaction in others. Bales (1953), in mal samples, given child-Approve the likelihood
fact, found very high conditional p values for such of a synchronous reaction by fathers was .160 and
behaviors in his observation studies of group pro- for mothers .077. In both cases the p values far
cess. Notice in Table 9.1 the sizeable increase in exceeded base-rate expectations. However, for the
the conditional probability for Smile given Smile clinical sample the likelihood of a comparable syn-
by the other spouse. Given that one would list chronous reaction for fathers was .00 and for
Smile as a potential reinforcer (and I do) it seems mothers it was .069. The latter probability repre-
that synchronous matching may apply to some re- sented a tenfold increase over the expected value
inforcers (Smile) but not others (Approve). for mother-Approve. It seems then that synchroni-
Also included are two examples of turn-taking city for Approve may hold for reactions to target
behavior—Attention and Talk. As expected, the children in normal families. For distressed families
conditional values were very high for both the reactions of mothers to target children may be
spouses. Given that a spouse attended, it was very synchronous but the reactions of fathers and male
likely that it would be matched by an attending re- siblings are not.
action by the spouse; i.e., A talks and B attends,
then B talks and A attends. Aversives
The second set of data were drawn from a sam- Conventional wisdom suggests that an attack
ple of 36 families of normal children and 37 fami- evokes an attack. Raush (1965) was one of the
lies of socially aggressive children. The data pre- first to provide field data which demonstrated this
sented in Table 9.2 describe a sample of events effect. His observations of normal and antisocial

203
 preadolescents in a residential setting showed a
similar disposition across the samples. Given an Ripple Effects
unfriendly initiation, the antisocial boy reacted in As noted earlier, synchronicity does not refer
an unfriendly manner .80 of the time; for normals, only to functional relations between adjacent
the comparable value was .77. While accepting the events. It is just that it is sensible to begin our ana-
general principle, one would wish to know more lysis by looking at these moment-by-moment co-
about the magnitude of this effect for family mem- variations. However, our understanding of social
bers as well. interaction will be advanced much further as we
The study by Margolin (1977) of married cou- focus upon questions concerning more extended
ples in a laboratory setting also showed significant episodes. For some time now we have believed
synchronous matching for aversives. Given a criti- that to understand the overall level of aggression
cism by the wife, then the likelihood the husband in family members, we must understand what it is
would criticize was .034 (this in contrast to his that determines extended coercive chains. By the
base-rate value of .007). Given that the husband same token, an understanding of extended proso-
disagreed, the likelihood that the wife would dis- cial behavior would tell us much about affection.
agree was .159 (her base rate for Disagree was What produces extended interchanges? The
.049). Synchronous relations were also found for simplest notion is that each event may have a “rip-
Interrupt (Husband and Wife), Complain (Hus- ple effect” (Rue Cromwell suggested this idea to
band), Not Tracking (Wife), and Put Down (Hus- me a decade ago). The occurrence of a single coer-
band and Wife). In this problem-solving situation, cive event might be accompanied by a slight in-
it was not only the case that an aversive elicited an crease in the likelihood of other coercive events
aversive, but it was likely to elicit the same kind of over the next few moments. Similarly, a single
aversive. hug, approval, or smile may increase the likeli-
An analysis of sibling interactions showed a hood for prosocial behaviors over a series of inter-
threefold to fourfold increase over base-rate values actions. Gottman (1979) was one of the first to
in the likelihood of a counterattack given an aver- examine this effect. Figure 9.1 summarizes inter-
sive antecedent (Patterson, 1980b). These in- actions across a series of laboratory tasks for dis-
creases in conditional p values were particularly tressed and nondistressed couples. Given that a
dramatic for male siblings in families of Social Ag- spouse responded positively at Trial One, then the
gressor children when attacked by another male lag-sequential analysis showed that the other
sibling (.49) or by a female sibling (.38). In that spouse was significantly more likely to respond
study, mothers and fathers were generally less like- positively at Trial Two. This, in turn, was fol-
ly to counterattack than were siblings. lowed by a trend for the initiator also to be more
The data in Table 9.2 tell a similar tale. Given likely to respond with a positive in later trials. In
that the problem child is coercive (high TAB fact, a positive event seemed to produce Crom-
score), then the likelihood of a counterattack by well’s “ripple” of positive effects, which endured
the male sibling was .43. Given a normal family, for at least seven trials.
this likelihood was .30. It is interesting to note A study of mothers and their 18-month-old in-
that these values were considerably lower than the fants showed that such effects were not limited to
figures obtained by Raush (1965) in the residential married couples in laboratory situations (Martin,
treatment setting. The remainder of the informa- Maccoby, Baron, & Jacklin, 1980). Given a
tion in Table 9.2 suggests a significant matching Mother-Positive (Talk, Touch, Hold, Smile, At-
effect comparable to the data from the Margolin tend, Comply), then the likelihood was .20 that
(1977) study of distressed couples. For male sib- the child would initiate a positive in the next 5-sec-
lings, Disapproval was matched by a synchronous ond interval, .16 at Trial Two, .13 at Trial Three,
Disapproval, Physical Negative by Physical Nega- and .11 at Trial Four. The comparable initiations
tive and so on. by mother, given a prior positive by the child,
There seems to be synchronicity for the ex- were .21, .18, .16, and .16 respectively. Only
change of aversives ‘among family members. This those sequences were included where the reactor
general effect has been obtained by many investi- had not been engaged in the behavior just prior to
gators (Snyder, 1977; Wahler, in preparation). As the criterion event; i.e., the reactor truly initiated
a general case there is a twofold to fourfold in- a prosocial behavior. This enhancing effect was
crease in the likelihood of an attack, given an traced out to 35 seconds (7 trials) in the Martin et
abrasive intrusion by another family member. Fur- al. (1980) study.
thermore, there is a tendency to match the type of I believe that the ripple effect for positive events
counterattack to the type of intrusion. will be shown to be of greater magnitude and

204
 Figure 9.1
Ripple Effect for a Positive Event
(from Gottman, 1979, p. 152)

e——® Nonclinic
@=-----+ Clinic

Z-Score

Reprinted with permission from Marital Interaction: Experimental Investigations. Copyright 1979 by Academic Press Inc. (London)
Ltd.

longer duration for normal families than for clini- generally positive manner, then it was likely that
cal families. The ripple effect for aversive events the other would respond in kind. Patterson and
should tend to be of a greater magnitude and long- Moore (1979) used sequential-lag analysis to iden-
er duration for families in the clinical sample. tify an event which produced a significant negative
Gottman (1979) found a significantly greater mag- effect and then identified the intradyad compo-
nitude of effects in his auto-lag comparison of nents which contributed to it. Given a positive re-
distressed and nondistressed couples when testing action by the mother, child-Complain was likely
for negative events. He did not find that the mag- to stop. Given mother-Aversive, then child-Com-
nitude of positive effects significantly differenti- plain was likely to continue. The longer the epi-
ated the two samples. Gottman’s finding of an sode, the more likely the final outcome was to be a
aversive ripple effect for distressed couples led me positive consequence.
to look for a similar phenomenon among the fami- Ripple effects imply structural underpinnings;
lies of antisocial children (see Chapter 8). but exactly what are they? The findings suggest
These ripple effects underscore the potential im- that one of the key differences between normal
portance of intradyad components. In fact, a rip- and clinical families lies in their relative disposi-
ple effect suggests an extended impact upon intra- tion to be positive or negative in their reactions
dyad components. Gottman (1979) found signifi- with each other. Members of families with antiso-
cant cross-lag (intradyad) relations in his analyses. cial children seem to declare a moratorium on eti-
This meant that, given that one spouse reacted in a quette and, placing malevolent interpretations on

205
positive, neutral, and negative behaviors alike, normal family members should demonstrate a
they constantly make ready for battle. Once they loosely organized cluster of values, recreational in-
begin, their battles are longer. terests, and personality traits as a result of sharing
the same settings. The intercorrelations among the
Shoppingand Mutuality shared traits may not be high, but they should be
There is yet another sense in which the behavior more similar to each other than they are to stran-
of dyad members can be said to covary. However, gers. Some settings within family life are associ-
these relations do not imply that the behavior of ated with high likelihoods for certain kinds of re-
Person A is currently functionally related to that of actions. For example, playing Monopoly or cards
Person B. Two persons may have similar disposi- together would increase the likelihood of synchro-
tions because each of them shopped for, and nous Laugh or Smile. Reading to a child would
found, a person with similar values, interests, and raise the level of mutual Touch and synchronous
preferences for play. They also display similar dis- Talk and Attend. It is thought that there are signi-
positions because, over time, each has modeled and ficant variations among family members in their
supported the development of this disposition in likelihoods of selecting such settings.
the other. Shopping and reinforcement-for-match- To understand the similarity between the aver-
ing are thought to be two variables which deter- sive behaviors of members of a dyad it is necessary
mine the similarity of behaviors found among to examine three factors: the relative frequency for
friends and family members. aversives may reflect the fact that the members of
the dyad selected each other because they were
Shopping similar; they may also have shaped each other to
It is assumed as a general case that each of us be more (or less) aversive; finally, many aversive
shops among settings, persons, and activities in events might be synchronous with aversives pro-
such a way as to maximize reinforcement. The vided by the other member of the dyad.
person most likely to respond to your behaviors in The problem of multiple determinants intrudes
a supportive manner is the person whose response when we attempt to analyze the relation between
hierarchy (of skills) is most like your own (Patter- what a person does to others and what he receives
son & Reid, 1970).1 Even in a nursery school the from others. Suppose that we study ten children in
“jocks” tend to affiliate with jocks, the popular a nursery school setting. These children are very
with the popular, the scholars with the scholars, different with respect to the general trait extraver-
and so forth. By a process of elimination, those sion. Some are very outgoing and initiate many
children with the lowest status in the peer group contacts with others. They also tend to be very re-
affiliate with each other. In effect, “likes” interact inforcing to peers. According to the shopping hy-
with “likes.” Rankings for the frequency with pothesis, these gregarious children will tend to se-
which two friends engage in various trait behav- lect others who are similarly disposed. Leiter
iors will tend to be similar. As a pair, they select (1977) studied free-play situations in a nursery
the same settings and often share the experience school setting. He found that children who made
together. They would also show higher intradyad frequent initiations to others received significantly
trait correlations. Child A and Child B both en- more initiations from others. Charlesworth and
gage in a good deal of rough-and-tumble play. No- Hartup (1967) studied children in four nursery
tice that while the two friends both select rough- school classes. They found a correlation of .79
and-tumble play, their behavior in that setting (p<.01) between the total number of reinforcers
does not have to be synchronous. On occasion, given and the total number received. Those who
each of them may enter that setting independently gave the most received the most; they also tended
and engage in a form of parallel play. Similarly, a to give to the greatest number of people. There
husband and wife may have selected each other was a correlation of .62 (p<.01) between the
because each required few social contacts. Both number of reinforcements given and the number
spend hours each day watching TV, but their be- of persons to whom they were given. Similarly, the
haviors may or may. not be functionally related. correlation between the number of reinforcers re-
Friends shop for and find similar settings. On ceived and the number of persons from whom they
those occasions when their behaviors in those set- were received was .70 (p<.01).
tings are synchronous, then they tend to reinforce I think the Charlesworth and Hartup (1967)
each other for engaging in behaviors that are ap- correlations might reflect differences in the rate of
propriate for that setting. Friends should become social interaction. Children who interact more
more similar over time, at least along the dimen- give and receive higher frequencies of social rein-
sions for which they selected each other. Similarly, forcers. For example, Reid’s (1967) study of 24 in-

206
 Table 9.3
Mutuality Among Family Members for Positive Consequences

Pearson Product Moment Correlations

Problem Male Female Median

Target Child Father Mother Sibling Sibling Correlation

Problem Child 44 EGO

Pes 45" .43* 44
(02) (02) (03) (02) (02) (02) (02) (02)
Father .09 .60* 41 43
(02) (02) (02) (03) (01) (01)
Mother 129 soGiun 44
(01) (03) (03) (03)
Male Sibling moles 48
(03) (02)
Female Sibling .47

*p<.05
bitods051
In 3% of interactions with the mother, the Ppproblem child received these consequences.
q In 2% of the interactions, he g gave the mother
these consequences.

dividuals from five families showed a correlation (1979) has sounded a similar note of caution.
of .56 (p<.01) between rate of social interaction Correlations of this kind are of interest, but any
and the number of initiations made by family single inference drawn from them is confounded
members to the target child. by the equally likely contribution of other determi-
nants (e.g., shopping, shaping similar behaviors,
Mutuality and shared settings).
Mutuality refers to the overall degree of similar- We are beginning to suspect that there may be a
ity between members of a dyad in their disposition characteristic “family constant” for positive and
to engage in a particular behavior “X.” It does not aversive consequences. Relative to other families,
necessarily refer to a disposition to react in a syn- Family A tends to react positively about .50 of the
chronous fashion (i.e., person A does X so person time regardless of content or person, while the lev-
B responds immediately in kind). Rather, it is the el for Family B is about .30. We would also as-
case that over an extended period of time person A sume that there is greater similarity among family
engages in X about as frequently as does person B; members than among individuals of the same age
mutuality refers to acommon level over time. This and sex from several different families.
could, for example, describe the dispositions to re- As an example of this mutuality among family
inforce other people. Mutuality would imply that members, Table 9.3 summarizes the data from the
you get what you give. Children who give the most first 22 families referred to OSLC. The positive
reinforcers receive the most. Family members who consequences, Approve, Physical Positive, and
initiate the most coercive behaviors receive the Laugh, were tabulated as outcomes of interactions
most. For example, Patterson and Reid (1970) ob- among dyads for each family. The frequency with
tained high covariations in rankings (in the .50 to which a given agent provided these consequences
.60 range) for giving and receiving both rewards was divided by the total number of interactions for
and punishments among family members. At that that agent (within that dyad). This provided a
time we labeled the relations as “reciprocity corre- control for differential selection among family
lations.” Similar correlations were noted for hus- members. It did not, of course, control for the fact
bands and wives (Patterson, Weiss, & Hops, that the dyad might mutually select preferred ac-
1976). In retrospect, I think it was a mistake to tivities. The data for the six to ten baseline ses-
apply the concept of equity, balance, or equality to sions in the home provided the data base. The
either inter- or intradyad correlations. Gottman data strongly suggest a mutuality in the level of ex-

207
 Table 9.4
Intradyad Covariations for Synchronous and Nonsynchronous Events

Nonsynchronous Exchanges Synchronous Exchanges

Number of
Physical
Family Sessions Approval Comply Volunteer Play Talk Work Positive

Tofu 20 —.05 —.07 —.25 .40 Wipe = Ai hae

Pumpkin 21 08 18 41 sil72 SoH sole) Hoi
Spring 20 .03 —.07 739 .45* .54* —.15 03 oa
Summer 15 36 162 .00 ee On 507 te (6Scg
Pluto 20 -.01 .07 .00 21 /0ne 39 Bay}e-"
Eclipse 20 .24 .07 —.18 .36 Shee 34 .40

p=. 05
=p —=<0l
=p 004

change for these three positive consequences.” A not a convincing set of findings for the across-
later analysis will demonstrate an even higher or- spouse covariation required for support of the
der of mutuality for the exchange of aversives. 1970 reciprocity-in-exchange hypothesis.
What do these mutuality correlations mean? Do The next study examined the same alternative
they reaffirm the fact that family members share a hypothesis using the data from six mother-child
common level, or do they describe more than that; dyads. They were observed in their homes for a
i.e., the members actually maintain a kind of equi- baseline of 15 to 21 sessions. The intradyad day-
ty or equality in the giving and receiving of rein- by-day covariations were calculated for four varia-
forcers, as suggested by Patterson and Reid bles thought to be positive reinforcers and for
(1970), Conger and Smith (1981), and others? An three turn-taking variables. The PPM correlations
expedient means for determining which is a more are summarized on Table 9.4. The findings sup-
reasonable alternative would be to test for covari- port the relation between synchronous behaviors
ations across time for a dyad. It would rule out and high positive values for intradyad correla-
the differences in familial level, which I think is the tions. Turn-taking behaviors, such as touching
prime determinant for the intradyad correlations. and talking, produce high values for intradyad co-
However, even such an intradyad correlation variations. The three events thought to be nonsyn-
would not control for shared selections of settings chronous but reinforcing did not produce signifi-
nor synchronous matching, which would build in cant intradyad correlations. It seems that the mu-
covariations. tuality hypothesis is the most conservative posi-
In the study by Wills (1971) the husbands and tion to take in viewing the usual correlations one
wives specified what it was about the behavior of obtains when analyzing the data from dyads
the other person that functioned as a Please (P) across families.
and what was for them a Displease (D). They were In the case of the interdyad correlations for
called at the end of each day for 14 days and aversive events, there are two major determinants.
asked, item by item, whether each of their prese- As noted earlier, aversive interchanges tend to be
lected P or D behaviors had occurred. Each spouse synchronous. I think that families are also charac-
was interviewed separately. The Pleases (reinforc- terized by differences in the /evel for aversive
ers) were analyzed Separately as to whether they events, and that families are more homogeneous
were instrumental acts (take out garbage, run an than a random.assortment of individuals matched
errand) or affectional (give a hug). First he for age, sex, or familial role. Taken together, this
searched for serial dependencies. There was an av- would suggest that when studying aversives, both
erage Lag, (1 day) correlation for affectional the inter- and intradyad correlations will be high
Pleases of .30; the average intradyad correlation and positive.
was .29. For instrumental P’s it was .25. This was Margolin (1977) studied a sample of videotapes

208
 Table 9.5
Mutuality for Coerciveness Among Family Members

Family Problem Male Female

Member Child Father Mother Sibling Sibling Median
Problem Child Bi a 24 Ole GS 62
Father no EOOn= oA 78
Mother .48* 1620 40
Male Sibling she 69
Female Sibling 70

*D<.05
**y< 01

of mildly distressed couples attempting to solve a reward for the behavior of the other; i.e., part-
some of their own marital conflicts. The tapes ners have reciprocal effects upon each other. We
were coded separately for the husband and the certainly were not the first to see its relevance to
wife. The interdyad correlations were very high family interaction (Burgess & Bushnell, 1969);
for aversive categories such as Complain, .74; however, it seemed to be an idea whose time had
Criticize, .81; Disagree, .89; Deny Responsibility, come for it quickly made its appearance in the
.75; Excuse, .65; Interrupt, .79; and Put Down, fields of marital therapy (Stuart, 1975; Azrin,
.74. These dyad correlations strongly emphasized Naster, & Jones, 1973) and in intervention with
the trait of mutuality for aversive events for hus- families of delinquents (Alexander & Parsons,
band and wife. 1973):
The data in Table 9.5 summarize familial mutu- The derivations from these premises seem to
ality for interchanges between dyad members. For have important applications to family life. Marital
each member, the total interaction with the other satisfaction, the duration of marital relations, and
member served as the denominator; the frequency love itself were said to relate to the exchange of re-
with which each was coercive to the other served inforcers (Hatfield, Utne, & Traupman, 1979). In
as the numerator. The same sample of 22 cases, some important sense intimates are interdepen-
analyzed in Table 9.3, provided the data for this dent; each relies upon the other as an important
study. It can be seen there that, with one interest- source of reinforcement. Each has discovered the
ing exception (mothers), there is a very high order other to be a unique source of rewards (Huston &
of mutuality. About twice as much variance is ac- Burgess, 1979). :
counted for by these correlations than the correla- Originally, the concepts were taken from the
tions for positive consequences (Table 9.3). The fields of sociology and anthropology, where they
intradyad data from mother-child pairs presented had a long history. The latter recognized the cru-
in Table 10.1 suggest a similar theme. On days cial role played by barter with the exquisitely bal-
when the child was particularly coercive, the anced rules for equity among primitives. Among
mother responded in kind. The median intradyad sociologists the idea of economic exchange theory
correlation for TAB scores was a substantial .49 had been applied to governments, groups, and
(range .41 to .61). Dyads are synchronous for families by theorists such as Homans (1961).4
aversive events. Given this, then it follows that Nord (1969) defined the value of a social reward
mutuality will hold when analyzing data for aver- in terms of its availability (market demand or scar-
sive events across dyads. city). Schaffer (1977) wrote, “The basic character-
istic of all interpersonal behavior is reciprocity”
Reciprocity (p. 172). Gouldner (1960) suggested that a norm
In 1970 we introduced the concepts of exchange of reciprocity may be universal and related to the
and reciprocity as two pivotal concepts necessary stability of some social systems and the contingent
for understanding families of antisocial children exchange of gratifications. He went so far as to
(Patterson & Reid, 1970). Exchange simply suggest that members of a dyad are reciprocal be-
meant that some behaviors of one person served as cause each member has a moral norm binding him

209
to reciprocity. Others, such as Homans (1961) crosocial data. We have already seen that the cor-
and Huston and Burgess (1979), assume that self- relations of, say, mother-child dyads across fami-
interest is best served by reciprocal exchanges. lies simply did not mean what we thought they
Each person learns that he wil! receive valued rein- meant. The covariation of dyad scores across ses-
forcers only if he reciprocates by giving them in sions controlled for some of the confounds. Ap-
equal measure. The early theorists were followed proval as a potential reinforcer did not survive
by enthusiastic translations of economic exchange these analyses. Even if it had, the more recent for-
theories to social interaction. mulation by Huston and Burgess (1979) points to
In that tradition, many writers note the relation one further complex issue. R. Burgess (personal
between an imbalance in the exchange of rewards communication) presents four different definitions
and a disruption in a relationship (Blau, 1964; to test for balance in social exchange. The first is
Emerson, 1972; Huston & Burgess, 1979). We direct exchange. A reinforces, and B responds in
found this to be an attractive idea and applied it kind. Reciprocity implies that in the exchange
directly to the families of antisocial children. It there is an equivalence in kind and amount. In his
was assumed that the deviant behavior of the child next two measures of balance, Burgess raises the
was produced in general by a low level of rewards consideration of individual differences in experi-
within the family amd an imbalance in exchange. ence. From this relative frame of reference, a
The same formulation was applied to distressed matching exchange requires calculation of the fre-
couples (Patterson, Weiss, & Hops, 1976). quency with which A reinforces B, divided by the
The problem with our early attempts to use ex- frequency with which A reinforces all other per-
change theory lay in the means by which we pro- sons. Relatively speaking, how much does A rein-
posed to test it. Our approach was to apply micro- force B? A match requires that the two relative p’s
social techniques to'!a problem which simply does for A and B be the same. His fourth measure, equi-
not lend itself to such a simplistic analysis. First of ty, stipulates that the ratio of p;|p2 should equal
all, the analyses were not focused upon the kinds P2|pi. With this in mind, the best that can be said
of rewards central to exchange theory. In Levin- is that the research literature based upon microso-
ger’s (1979) discussion, he lists the following: cial analyses is simply not relevant to the main ten-
money, goods, information, status, services, and ets of exchange theory. The key lies in our ability
love. While one can think of Approve or Physical to specify what it is that is reinforcing for each
Positive, as measured by the FICS, as a subset for member and to be able to design a method of
love, it is, at best, a minuscule component of the measurement that will adequately describe
broader concept(s) which were important to ex- “amounts” given and received. This, in turn, re-
change theory. Even if one accepts the belief that quires that the individuals’ cognitions be taken
the code category, Approve, is a relevant compo- into account.
nent, the measure leaves out the crucial dimension What is needed is a means of having family
played by the cognitions of the receiver. “Accord- members specify which aspects of their interac-
ing to equity theory, people become more com- tions with others are reinforcing. An experimental
mitted to relationships when they perceive that the manipulation in which one family member in-
value of each participant’s outcomes are propor- creased the frequency of these “perceived reinforc-
tional to the relative value of each individual’s in- ers” should be accompanied by a reciprocal in-
vestments” (Huston & Burgess, 1979, p. 11). crease in the frequency with which the other per-
Equity, then, is less in the eye of the observer son supplies events perceived as reinforcers by the
than it is in the mind of the receiver. Equity is a other. Both persons must be well aware of what is
relative term that requires some measure of the perceived as a reinforcer by the other. One such
cognitions of the participants. The final judgment experiment has been carried out in a field setting.
is a function of an intricate balancing of “goods” The study by Wills et al. (1974) involved seven
given “costs” and what is available outside of the married couples. Over the 12-day baseline, both
relationship (Huston & Burgess, 1979). As a re- spouses were called each day for a listing of how
sult, the appropriate means of study must include many “Pleases” they had received from their
some self-report information about costs, values, spouses during the previous 12 hours. The wives
and satisfactions. Such broad-gauge measures re- reported receiving an average of 1.56 Pleases per
late to satisfaction, duration of relationships, and hour during the preceding five days of baseline
conflicts in the manner predicted by exchange the- (Days 7-12). The husband was then told by phone
ory (Hatfield et al., 1979). what rate of Pleases he had been giving his wife
As noted earlier, the study of equity or equality and was instructed to double that figure for the
poses some unique problems when applied to mi- next two days. As shown in Table 9.6, for that pe-

210
 Table 9.6
Analysis of Changes in Affectional Behavior During Reciprocity Manipulation
(from Wills et al., 1974, p. 806)

Average rate/hour Average rate/hour t for correlated

Reported by: for Baseline for Manipulation means (df =6)
Pleasurable Behavior
Husbands 1.45 1.64 .64

Wives 1.56 Mepahs) 2:01

DispleasurableBehavior
Husbands 63 40 = ile)

Wives .66 .56 = il

*p<.05, one-tailed.
Reprinted with permission from “A behavioral analysis of the determinants of marital satisfaction,” Journal of Clinical and Consult-
ing Psychology, 1974, 42(6). Copyright by the American Psychological Association.

riod (Days 13 and 14) the wives reported a sub- antisocial children, a feeling that the individuals
stantial increase in Pleases received. Reciprocity are not “getting their fair share.”
would have it that over such an extended interval, My hunches about the concept of exchange and
the wife would have ample opportunity to provide balance remain pretty much what they were in the
an increase in Pleases (reinforcers) which she gave early 1970’s. These are important ideas that have
to her husband. Because of the manner in which great relevance for the treatment of distressed fam-
the study was designed, she would be well aware ilies. While committed to the relevance of the
of what constituted a Please for her spouse. ideas, I remain unconvinced that any of our cur-
As shown in Table 9.6, there was a nonsignifi- rent microsocial data apply to them. My hunch is
cant trend for the husbands to report an increase that it is here that some of our most interesting
in Pleases. There was also a nonsignificant de- contributions will be made in the next generation
crease in the magnitude of Displeases reported by of research on interactional issues.
both spouses. Incidentally, many of the wives
spontaneously reported that the increase in Pleases
received made them feel as if they were on a sec-
ond honeymoon.‘ Acknowledgements
This miniature study is a prototype of the kind The writer wishes to acknowledge the contribu-
of field study which could provide useful informa- tion made by Robert Burgess to this chapter. The
tion about exchange and balance in family struc- current text was written in response to his critique
tures. As things now stand, there is a very large of an earlier draft of this chapter. Many of the
gap between the microsocial analyses of structure ideas expressed in this revision were the outcome
and the complexity involved in questions concern- of a two-day seminar presented by Bob to the
ing the exchange of rewards. What is needed are OSLC staff. Certainly he cannot be held responsi-
studies of these issues that do not rely solely upon ble for the author’s current position. It was, how-
either self-report data or upon manipulations in a ever, his clearly enunciated position which led to
laboratory setting. improvements in this chapter.
The controlled studies described in Jacobson
and Margolin (1979) attest to the relevance of ex-
change theory to the newly developing techniques
Footnotes
for resolving many severe marital conflicts. The
sense of injustice characterizing spouses about to 1. D. Byrne’s early work suggested a corre-
separate is very keen. Their descriptions of what is spondence between attraction and similarity in at-
wrong are seldom stated in the language of ex- titudes (Byrne & Rhamey, 1965). His later work
change theory, but the fit for the ideas is often very on liking and similarity extended these findings to
close. There is also a sense of this in the families of demonstrate a relation between liking and rein-

p44
forcement. The general formulation is in general variable X and Y. I then began examining the raw
agreement with Hartup’s (1978) analysis of chil- data to calculate the different measures of equity
dren’s friendship patterns. Friends are more simi- suggested by Burgess. These new calculations did
lar along a number of traits, such as sociability, not differentiate distressed from nondistressed
physical attractiveness, perceived kindness, and samples. Others may well follow that same path,
friendliness. Children who are effective’social rein- but I, for one, could not find my way through the
forcers are also more likely to be selected as thicket of problems which surround it.
friends. 4. The writer believes that the models for eco-
2. As can be seen, the mean levels tend to bal- nomic exchange do not really fit social interaction
ance reasonably well for all dyads such that no one in the home setting. This metaphor and those
family agent stands out clearly as a victim. Dyads from game theory were both rejected because the
in each family were then examined to determine laws of cost, supply, and demand did not seem to
the proportion characterized by inequitable ex- provide a close fit to what could be observed in the
changes. For seven normal families there was an home. If the mother gave ten hugs to one child, it
average of one dyad in ten showing an inequity in in no way limited the number that she could give
which one person gave (at least) twice as much as to a second child. It didn’t seem that a zero-sum
he or she received. In the distressed families, the game described the exchanges which took place.
percent of inequitable dyads ranged from .20 to Nemeth (1970) reviewed studies which showed
1.00; the mean was .56. For the distressed fami- that in social interchanges characterized as zero-
lies, the correlations between percent inequity and sum games, one seldom gets either cooperation or
mean baseline TAB score (for the family) was reciprocity. Given a limited supply of rewards,
—.19 (n.s.) then each person attempts to maximize his own
3. Another means of parceling out the family- gain. In such a world, one would attempt to work
level confound would be to calculate the correla- out rules for an equitable exchange, but only as a
tion separately for each family. Several investiga- temporary expedient. In the long run, each person
tors have, in fact, calculated such correlations attempts to win at the other’s expense.
(Reid, 1967; Conger & Smith, 1981). I also car- 5. The across-couple trait correlation for the
ried out endless hours of such calculations and Wills et al. (1974) sample was .97. In a compara-
“found some” for both distressed and normal sam- ble study by Birchler, Weiss, and Vincent (1975),
ples. However, by their very nature, correlations the correlation for 12 distressed couples was .74
define a relation sans differences in mean level for and for 12 nondistressed couples it was .97.

Bs|
 rs ed Se pa

“A ane | i
Pi 2 ¥ ; i?
be ow a
i bat hh vt
‘ LSSy '
(We
f fie

y ‘
4 > : !

neo

'
jogs ’
i Pos

—
+
Gi. a t

tc

<e =i

>, 5| Kq

Bsr
ye
ce ee ¢ i } ¥
: ite te i
Mitel* '
aily» 24 bts, F i

Be tat, Peni Line

. till i id ld 4 “: 1

“igOrs N44,y%“|
;ai
tl Wetodq)wer
wattevienf i
aane?
; a nekDib
thfe
a8 wit
 Chapter 10
Abstract
Four family management (FM) variables are described that are thought to determine individual dif-
ferences among children in their rates of antisocial behaviors. These include: (1) house rules, (2)
monitoring, (3) contingencies, and (4) problem solving, negotiation, and crisis management. It is
thought that parents differ in the effectiveness with which they practice these skills. Furthermore,
their implementation varies as a function of crises and stressors that impinge upon the parents. Data
are presented that demonstrate a covariation between the mother’s TAB score and daily crises and
community supportive contacts. These four variables are thought to mediate the covariation between
antisocial child behavior and traditional variables such as broken homes, divorce, psychiatric condi-
tion of parent, marital discord, and lower social status.
It is assumed that the effect of disruptions in these FM variables is an increase in the rate of coercive
child behavior. Furthermore, measures of FM variables are thought to correlate with differences in
performance level for family members.

214
 Chapter 10

The Management and

Disruption of Families
Microsocial analyses provide a description of measured. At OSLC our hunches about FM vari-
family structure that, in turn, reflects a history of ables grew out of our clinical contacts with fami-
interaction. The hypothesis (untested as yet) is lies. Most of the process of variable selection was
that many of the coercion-related structures arise just common sense. Our first sessions in the home
as a function of disruptions in parental family convinced us that many of these parents did not
management skills. punish effectively (i.e., their punishment didn’t
Family management (FM) variables are thought work). Many of them lacked house rules that were
to account for a significant amount of the variance clearly specified. These ideas are eminently testa-
associated with differences among children in ble; both hypotheses point to behaviors that can
rates of performance of antisocial behaviors. This be counted. The list of variables we selected will
is being tested in a large-scale longitudinal plan- seem quite ordinary to the reader. Each has a fa-
ning study. It is assumed that measures of FM miliar, common-sense ring to it. This is reflected
skills will also demonstrate low-order positive cor- in the title the OSLC group has selected for a new
relations with the traditional measures of child- book about family intervention, Systematic Com-
rearing practices; i.e., variables such as laissez- mon Sense (in preparation). All of these FM vari-
faire, authoritative, warm, and punitive should ables would have had an immediate appeal to my
correlate with our measures of FM practices. Norwegian grandmother. I can imagine her read-
Measures of FM practices are designed to give a ing the list and saying, “Ja, but vy do you haff to
molar perspective on what the parent is doing that teach dis to people?”
relates to the maintenance of the family system. I As a matter of fact, it is not clear why, in this
assume that effective family management requires day and age, such simple skills must be taught.
a high order of skill, and that parents of antisocial Clinical experience suggests that many of the par-
children are lacking in their performance of one or ents of antisocial children did not have good pa-
more of these skills. rental models for the practice of family manage-
An interactional stance does not prescribe ment skills. Other parents “know” in a general
which variables should be studied at either the sense what the skills are but have decided for one
macro- or microsocial levels. That selection is up reason or another to put them into practice selec-
to the investigator’s intuitive hunches and his theo- tively. Others are overwhelmed by crises and can-
retical biases. Probably more than anything else, it not implement those skills that they do have. My
is a function of that which can be adequately general sense of it is that an increasing number of

215
young parents have had very poor modeling and erosion of family management skills. These dis-
even less supervision and support for the practice ruptions will, in turn, be accompanied by in-
of these crucial skills. creases in antisocial child behavior.
Disruptions in the implementation of family The impact of major crises on effective family
management skills are thought to be the major management is summarized in Figure 10.1. In a
mediating variable for antisocial behaviors in chil- path-analytic sense, the causal arrows move from
dren. These disruptions accelerate the rate at a crisis (such as divorce) to a disruption in family
which the child engages in out-of-control behay- management fo an increase in antisocial child be-
ior. If these disruptions continue, the process will havior. In each instance these two correlations are
involve siblings and parents, and eventually the expected to be higher than would be the case for
entire family will be disrupted. the covariation of crises with antisocial behavior.
The extreme conditions experienced in a ghetto As an alternative, it is conceivable that having an
place the child at risk for the development of a va- antisocial child could produce marital conflict,
riety of psychopathologies. Having one or more drinking, or psychiatric problems in parents.
parents who are psychotic also places the child at Not all parents going through a divorce will
risk. However, as Garmezy and Nuechterlein demonstrate a disruption in family management
(1972) point out, the majority of children survive practices, but those who do are likely to experi-
these conditions with few signs of adult pathology. ence an increase in antisocial child behavior. In ef-
In a longitudinal study (Rutter, 1979), given the fect, information on crises tells us when to expect
extremes of deprived neighborhood, plus parent an increased likelihood for disrupted family man-
criminality, plus bad child rearing, poverty, low agement and a commensurate increase in the like-
intelligence, and large family, over one-fourth of lihood of antisocial child behavior. The one excep-
the children showed no evidence of any kind of de- tion (that we know about) to this two-step pro-
linquency or antisocial behavior. Given that both gression is the bilateral relation between mother
parents are psychotic, three of five children will depression and child social aggression (see Chap-
show no major sign of pathology (Garmezy & ter 12). It is thought that the child’s extremely
Streitman, 1974). aversive interactions contribute to the mother’s de-
It is hypothesized that for children at risk, the pression. Given that an acute depression may also
practice of family management skills constitutes a disrupt family management, the relation is
set of moderating variables that will account for thought to be bilateral.
many of these false-positive predictions. It is the There is another empirical linkage for the func-
psychotic parent who does not have family man- tional relations presented in Figure 10.1. That is
agement skills that will most likely produce an an- the relation between parental crises and deviant
tisocial and/or schizophrenic child. The child liv- child behavior. Rutter and his colleagues analyzed
ing in extreme poverty, with low intelligence, will the data from the Isle of Wight and inner-city Lon-
survive the experience if the parents practice effec- don studies of 10-year-old children. They identi-
tive family management. Effective family manage- fied six variables that were reliably associated with
ment practices include the following: clearly stated child psychiatric disorders, including antisocial
house rules, monitoring, providing consequences problems. These variables were: (1) severe marital
contingently, and problem solving. discord, (2) low socioeconomic status, (3) large
family size, (4) father criminal history, (5) mater-
Crisis Disruption of FM Practices nal psychiatric problems, and (6) admission to
There are individual differences among parents care by local authorities (Quinton & Rutter,
in the skill with which they practice family man- 1976, cited in Rutter, 1979). Figure 10.2 summar-
agement. While their relative rankings would izes the findings when the families were categor-
show positive correlations over time intervals of a ized into those with none of these risk factors, one
few weeks or months, the effective implementa- of them, two, three, and four or more. Note that a
tion of these skills may be disrupted under condi- single crisis did not place the child at significantly
tions of extreme parental stress, e.g., severe ill- greater risk than was the case for peers who were
ness, unemployment, divorce, severe depression, living in a similar (but nonstressed) environment.
or a psychotic breakdown. As a corollary of this However, there was an exponential increase in the
hypothesis, it is thought that these skills may also likelihood of a child psychiatric problem as the
be disrupted by the cumulative impact of minor risk factors increased from one to three and three
crises or “hassles” that impinge upon all families. to four.
Given that these hassles occur frequently and/or It is only in the past decade that systematic em-
persist for a period of days, then there will be an pirical studies have been carried out to study the

216
 Figure 10.1
The Relation Among Family Management Practices, Crises,
and Antisocial Child Behavior

Parents Not Implementing

Family Management Practices:
1. House rules
Antisocial
2. Monitoring
Child
3. Contingent consequences Behavior
4. Problem solving, crisis
management, negotiating
compromises

Illness Marital conflict Parents overly Psychiatric disturbances

Drugs
Poverty Divorce committed to of parent(s), e.g.,
Alcoholism
Unemployment Broken home work, etc. depressed, psychotic

effect of stress upon human behavior. Much of the (Hetherington, Cox, & Cox, 1976).
contemporary work on these problems is based The idea that stress, crises, and familial con-
upon the work of Selye (1976) and Holmes and flicts may have a cumulative effect upon both the
Masuda (1974). The latter researchers found a mood of the caretaker and the quality of family in-
convincing correlation for a variety of samples be- teraction is certainly in keeping with the general
tween simple checklist measures of environmental perspective presented in sociology and anthropol-
stress and the incidence of psychosomatic disor- ogy. As mentioned earlier, the excellent descrip-
ders. In Chapter 4 it was demonstrated that their tion by Lewis (1968) in La Vida documents the
concept of crises as “stress” can readily be applied steady attrition of spirit and gradual sense of help-
to family interaction. Daily mood shifts for three lessness that accompanies extreme poverty. Tonge
of the mothers were shown to covary with daily et al. (1975) obtained descriptions of a compara-
fluctuations in conflicts and crises from outside ble environment during systematic interviews with
and from within the family. Systematic studies of multiproblem families from large metropolitan
families in divorce showed a similar covariation areas.
for an across-subject design. The self-esteem of Familial stressors that alter the caretaker’s
mothers and fathers became more negative follow- mood may also produce a disruption in his or her
ing separation. The divorce crisis was also fol- implementation of family management practices.
lowed by increases in oppositional child behavior The prediction is that day-by-day fluctuations in

217
minor hassles will produce day-by-day fluctua- positive (supportive). The daily crises score merely
tions in the caretaker’s mood and/or family man- summed across the number of items checked; no
agement practices. These, in turn, will covary effort was made to weight those that seemed obvi-
with changes in coercive child behaviors. To test ously more severe.
this, the data collected in the homes of the five The hypothesis was that measures derived from
observations of the coerciveness of the mother in-
teracting with her child would correlate with each
of the following: the aversiveness of the child with
Figure 10.2 whom she dealt, the daily hassles with which she
Multiplicity of Risk Factors had to cope, the daily contacts from her support
system, and her general mood. It should be noted
and Child Psychiatric Disorder
that all of these components are likely to be inter-
(from Rutter, 1979, p. 52)
actional; e.g., her mood and abrasive manner will
elicit coercive reactions from the child. The inter-
actional data for these mothers were analyzed to
determine the magnitude of the covariation of
S=wn
S)
ro)
these variables with the mothers’ TAB scores.

Likelihood
Psychiatric
Status It can be seen in Table
jects the composite
counted
the scores measuring
10.1
set of six or more variables
for a good proportion
the
of the variance
mothers’
that for three

coerciveness.
sub-
ac-
in

For all families the measure of the child’s coercive-

°
ness was the main contributor. The child’s behav-
ior would be the variable most involved in eliciting
and maintaining the mother’s coercive behaviors.
However, in terms of what accounts for the re-
mainder of the variance there are clearly different
patterns of relations for different mothers. For ex-
ample, for the mother of Eclipse it was her general
mood for the day and the number of crises that de-
termined how abrasive she was.
Number of Stressor Variables
As a general case, it can be seen that what goes
Reprinted from Social Competence in Children, edited by Mar-
on at a microsocial level reflects, in part, the im-
tha Whalen Kent and Jon E. Rolf, by permission of University pact of macro variables that impinge from outside
Press of New England. Copyright 1979 by the Vermont Con- the dyad. The relationships between the variables
ference on the Primary Prevention of Psychopathology. are outlined in Figure 10.3. As shown there, for
some families crises may disrupt family manage-
ment practices; this produces an increase in steal-
mother-child dyads (presented in Chapter 4) were ing and/or socially aggressive behavior in the
reanalyzed. Each involved a mother and preschool problem child (and siblings as well). For others,
child and occasionally a young sibling. Each fami- the crises may also increase the negativity of their
ly was observed in the home for a minimum of 20 reactions to the child; e.g., they are slightly more
sessions; each session lasted about an hour. The likely to be irritable and increase their punitiveness
variations in the mothers’ TAB scores served as the or launch “unprovoked attacks.” This, in turn, ac-
criterion variable. In a previous study the uncor- celerates coercive child behavior. At this point,
rected test-retest reliability for the child’s score both parent and child are likely to escalate the in-
was .78 (p<.01) for an interval of one week and tensity of their aversiveness.
.74 (p<.05) for 12 months. On each day the For the moment we can only speculate that
mother filled out three questionnaires: one of these relations exist. Currently we are carrying out
three versions of the Lubin Checklist (as a measure a study of children at risk for delinquency which
of mood), the Crisis Checklist, and a version of R. will provide data to test these relationships. The
Wahler’s Insularity Checklist. The Insularity across-subject design provides measures of crises,
Checklist generated one score each day for the fre- mood, irritability p values for all dyads, and mul-
quency of contacts from the community, an esti- tiple measures of each of the four FM variables.
mate of the number of minutes these contacts Multivariate analyses will tell us which of these
lasted, and an estimate of what proportion were variables are crucial and what the unique contri-

218
 Figure 10.3
Factors Contributing to Antisocial Child Behavior

Major Disruption of Family

Crises Management Practices

Increase in Antisocial
Child Behavior

Increase in
Irritability:
Caretaker’s . Crossove
Crossover
Depressed
Mood ; Counterattack
. Acceleration
Punishment
. Continuance

Table 10.1
Variables Which Covary with Mother’s Day-by-Day Fluctuations in Coerciveness

Correlations with Mother’s TAB Score

Minutes of
Community Frequency Positive Child’s Mother’s Frequency
Family Contact! of Contacts? Contacts? TABScore* LubinScore of Crisis® Mult Rt
Tofu =.01 .02 —.11 .49* .26 —23 00)
Pluto 32 42 30 41 .26 43 764
Eclipse =A) 412 .08 soon OE .40 824
Spring 43 01 .26 aul aoe 09 74
Pumpkin —.14 —.16 .44* 42 —.14 .07 Giile
Independent Variables
1. x number of minutes contact with community per day
2. x number of interactions with people in community
3. x number of positive interactions
4. x total aversive behavior (TAB) observed for child
5. Mother’s x Lubin score
6. x number of crises

ep=)S
epi = Ol
+The value includes the contribution of the six independent variables listed in this table.
tThe value includes the contribution of the sibling’s TAB score in addition to the six independent variables listed in this table.

219
butions are for each of them. skills that are necessary for survival in school. The
Most theories about family life assume that classroom studies reviewed earlier showed that an-
family management skills are a given condition. tisocial children lacked the simple, work-oriented
From that perspective, only neurotic conflict be- skills that are a prerequisite for survival in the first
tween the parents can interfere with the successful grade (Cobb, 1972; Cobb & Hops, 1973; Hops
implementation of family management skills. As a & Cobb, 1974). These rudimentary skills must be
result, therapy techniques that are based on these polished and altered by peers, teachers, and agents
theories focus upon resolving these neurotic con- from outside the family. If the initial phase of
flicts with the expectation that success here will be training in the home has been omitted, it is very
followed by changes in deviant child behavior. unlikely that the child will profit much from later
Perhaps antisocial children drawn from samples of stages of experience involving teachers and peers.
middle-class families might be described by such a If his primary mode of interacting with peers is ex-
process. These parents may “know” family man- plosive and immature, then they will not teach
agement skills but are too busy fighting with each him the subtleties of how one goes about interact-
other to apply them. However, when this formula- ing with equals.
tion is applied to the OSLC sample of working- It must be said at this point that this book is not
class and single-parent families, it is grossly inac- focused upon the prosocial aspects of family life.
curate. Most of these parents do not have the skills We simply have not studied the process of teaching
to avoid crises or to contain them when they do children prosocial skills, and few of our measures
occur. They have not learned the conflict contain- were designed for such questions. Our primary fo-
ment skills practiced by most middle-class parents. cus has been upon the process of controlling the
They do not seem to model or monitor the child’s child’s deviant behavior. All of the four FM varia-
development of prosocial skills. They do not bles listed in Figure 10.1 have as their prime func-
know how to manage a child. The model that pro- tion the containment of familial crises. Later,
vides the best fit to the OSLC clinical sample is one when we bring the development of prosocial sur-
that emphasizes impaired learning of family man- vival skills into focus, the list of FM variables will
agement skills plus stressors that further disrupt obviously have to be expanded.
the application of the fragments of skill that they The general assumption is that children do not
do possess. outgrow their early proclivities for antisocial be-
The family as a unit obviously has many func- havior. Unless taught otherwise, they will con-
tions. Of primary interest here is its function as a tinue to take what they want, and they will use
vehicle for teaching the child basic survival skills pain. control when it suits their immediate pur-
and controlling the child’s use of antisocial behav- pose. Parents must teach their children prosocial
ior. It is the function of the family to model and alternatives to these coercive techniques, and they
support the child’s learning how to relate to oth- must also punish their children when these coer-
ers, how to work, and how to survive in the aca- cive techniques are employed. The data are far
demic world. The school, the peer group, and so- from complete, but our best guess is that both pa-
ciety at large have much to do with the teaching of rental functions are necessary. Training children
these survival skills, but unless the family prepares to use prosocial skills is necessary but not suffi-
and then supports the child in polishing these cient; the parent must also monitor and punish, in
skills, he will not be in a position to profit from a reasonable fashion, those coercive behaviors
what these socializing agents have to teach him. that are “naturally” employed by all toddlers.
For example, parents must first teach the child the Only a parent loves the child enough to go
rudiments of work skills at home. This teaching through the hundreds of trials in which the child
takes place under the guise of doing simple chores learns when he can, or cannot, use pain control
and “giving” to others as part of the child’s role in techniques. Later, his wife or his psychoanalyst
the family. The child is also trained there to accept may give him love and support equal to that given
criticism. Hopefully such correction is presented by his mother, but they cannot teach him at a mo-
in a reasonable manner. The child learns to match lecular level those subtle skills that he should have
what he does to what he says he will do, i.e., to learned prior to age 6 or 7. Parental power, rela-
keep his word. The child learns the rudiments of tive to a preschool child, is simply overwhelming.
how to form close relationships with others by re- Social workers, friends, or therapists do not pos-
lating to parents and to siblings. The child must sess an equivalent means for punishing adolescent
learn how to be close enough to maintain a sense and adult deviant behavior. They may help the
of self-worth and yet be neither possessed nor dis- adolescent feel better about his antisocial behay-
tant. The family also teaches rudimentary work ior, but usually they cannot stop the behavior in

220
the sense that a parent can. In my more cynical tisocial child behavior and parental negligence
moments, I feel that only a parent (or a surrogate (Robins & Ratcliff, 1978).
parent) who lives with the antisocial child can Most families have a schedule for cleaning, re-
teach him to change. pairing equipment, buying groceries, paying bills,
Changing the antisocial behavior of a child re- and for doing chores. They have agreed upon
quires the application of FM skills. The supervi- times for getting up in the morning and for going
sion and support for the application of these skills to bed. Time is set aside for exchanging informa-
is the focus of the treatment program for these tion about the day’s activities. There are times for
families (Patterson, Reid, Jones, & Conger, doing things together; there are shared interests
1975). As noted earlier, each of these skills simply and activities. For members of middle-class fami-
restates the common sense practices which are em- lies, these agreements hardly need to be stated. In
ployed by most of us in raising children. the study by Wadsworth (1979), the home visitor
found the homes of delinquent children to be
Rule Setting poorly managed.
The setting of house rules by parents was so tak- Families also set specific limits upon each oth-
en for granted that it took years of clinical contact er’s behavior. For some families yelling is accepta-
before we even labeled this as a problem for fami- ble, but hitting is not; borrowing a comb is okay,
lies of antisocial children. Harbin and Madden but taking money from someone’s dresser is not.
(1979) noted the disorganized, chaotic lifestyle In many of the homes studied, there were no rules
that characterized violent families. I think that an about bouncing and playing on the furniture (so
absence of rules is likely to be found in families in long as the child didn’t step on a parent).
which physical violence is used. Like the other Whatever the rules are, they establish limits that
three FM components, the effective employment determine the point at which a particular behavior
of household rules is thought of as necessary, but will be monitored and punished. Some rules about
not sufficient, for producing change in antisocial aggression exist in almost all families; i.e., there is
behavior. While the setting of house rules should some point at which the parent will try to stop the
be negotiated with the child, in the last analysis it interchange. However, families differ on how
is the responsibility of the parent; parents and chil- clearly the rules are stated; they also differ in the
dren are not equal. As J. Haley (1978) points out threshold at which the implied limit is exceeded.
in Problem Solving Therapy, the question of who As we shall see, they also differ in terms of the
it is that sets rules lies at the very core of problems likelihood of an effective punishment being ap-
confronting many families. In antisocial families, plied for transgression.
the problem child does not grant the premise that One of the first steps in treating these families is
parents (or therapists) have the right to set house- to help the entire family outline one or two house-
hold rules. hold rules (others will be added). When the smoke
Household rules (some implicit and others ex- settles, it becomes apparent that the children em-
plicit) state the scheduling of time and effort phatically disagree as to the right of the parent to
among family members. More importantly, they institute household rules. The: efforts of these
spell out what is, and what is not, acceptable be- young tyrants are amazing to behold. When
havior. It came as a shock to find that in many of chores were being discussed, one young preadoles-
the homes we have studied at OSLC, such rules cent rose to the challenge, “I ain’t gonna do it; I
did not exist! For these families, for example, ain’t no nigger.” He assumed it was perfectly equi-
there was no set time for eating; it could occur at table for his mother to go on making his bed and
any time between 4:00 p.m. and midnight. What putting away his clothes! A 12-year-old broke into
was eaten, who ate it, and where it was to be eaten tears at the suggestion that she must be home by
changed from one day to the next. Sometimes it 2:00 a.m., “All my friends are there . . . they'll
was a can of beans, opened and left on the table laugh . . . you bitch!” The nature of the counterat-
with several spoons stuck in the can. At other tack varies, but it is characteristic of antisocial
times, one of the children would attempt to cook children that they know how to strike the parental
some meat but would find his meal picked clean Achilles’ heel. For some parents it is guilt; for oth-
by family members walking through on their way ers it is a basic uncertainty about what is fair and
to the street. For many children, even though sup- what is not.
per was cooked by a parent, there was no way of The field studies concerned with the application
predicting the quality of the food or who would be of reinforcement theory suggest that changes are
there to share it. These impressions are in accord brought about more rapidly by rule statement plus
with the findings that show a relation between an- contingencies rather than by manipulating rein-

221
forcing contingencies alone. As a treatment de- set of expectations held by the parent as to when
vice, setting up household rules is ot sufficient, in the child returns from school, whom he is with in
and of itself, to produce change. A well-controlled the evening, and when he can be expected to re-
laboratory study by Hobbs and Forehand (1978) turn. If the child does not appear at the expected
demonstrated that rule setting, per se, was not time, then the parent becomes concerned. He or
sufficient. Mothers of preschool children out- she may call friends or go out and search for the
lined their expectations that requests would be child. Most parents of out-of-control children are
met with child compliance. Each noncompliance unlikely to do either of these. Monitoring means
was met with, “You did not do what I told you taking the five minutes necessary to see that he did
to.” This was a relatively ineffective method for his chores and that they were done properly; e.g.,
producing compliance. Sixty seconds of time out check to see if the tools were put away. This is also
for noncompliance was significantly better. This the time to reinforce the child for a task well done.
replicates the effect obtained by Becker et al. At another level, it means noticing that he is tear-
(1967) for the classroom. There, too, the teacher’s ing the stuffing out of the chair. It means attending
statement of rules about disruptive behavior pro- to the rough-and-tumble play of siblings before it
duced no effect until contingencies for noncompli- gets out of hand.
ance were introduced. It is not the absence of There is another aspect of monitoring that is
rules, per se, that produces the problem. It is the much more subtle. In normal families there is a
absence of rules plus the absence of contingent regular time for sharing information. It includes
punishment that generate the problem. all family members; it is not an interview by a par-
For some parents, the lack of rules is no acci- ent who acts like a disapproving policeman, nor
dent; it is their means for avoiding a confronta- does it have the overtones of a cross-examination.
tion. When the child stays out all night, if there is It is, rather, a genuine expression of interest in the
no rule to the contrary, then it is unclear that the activities and well-being of other family members.
parent must punish. When these parents try to It begins with an expression of this interest; e.g.,
punish or impose sanctions, they must be prepared “How was your day?” When the question is given,
for a confrontation. Most of these parents wish to the parent listens to the answer. Some parents
avoid this at all cost. They may natter and scold, complain that their children never tell them what
but they do not confront—no rules, no confronta- they are doing. Closer inspection shows that they
tion. For others, their use of punishment is as un- do not listen when the child responds; worse yet,
planned as is their chaotic, unscheduled existence. they criticize many aspects of what the child does
They have several jobs and there is little time for report.
thinking of house rules or punishment. These information exchanges are sharing exper-
The hypothesis is that households that are dis- iences that make a significant contribution to
organized and/or characterized by poorly defined building emotional attachments among family
rules are more likely to have antisocial children. In members. When used by skilled parents (such as
the longitudinal planning study currently under the parents we studied in the summer of 1979), it
way, we have interviewed over a hundred children is clear that these exchanges are a major socializ-
and mothers and fathers of children from grades ing mechanism. The parents use examples from
four, seven, and ten. A subset have also been ob- the interchanges to teach their children about im-
served in their homes. For each of the three portant values and skills. These interchanges do
grades, those items that seemed, a priori, to relate not become lectures. Rather, they are brief and
to the house rules concept were correlated with done with good humor; e.g., “Yes, he is clumsy,
three criterion measures of antisocial behavior. but he had polio when he was a baby, so he can’t
The findings support the hypothesis that disorgan- help it. He is a good person in spite of how he
ized households with few rules are associated with walks.” The laughter, exclamations, and intense
more antisocial children. At the time of this writ- interest of the entire group is repeatedly made con-
ing the multivariate analyses have not been com- tingent upon certain kinds of prosocial behavior.
pleted. These analyses will tell us how much of the It is this focus, this warmth and involvement with
variance in each measure of antisocial behavior is each other, that makes a group of people into a
accounted for by a composite measure of house- family.
hold rules. Parents of antisocial children may wish to en-
gage in such interchanges, but when they attempt
Parental Monitoring to initiate these interchanges they quickly deterior-
At a basic level, monitoring implies a general ate into abrasive, sarcastic attacks. They do not,
awareness of the child’s whereabouts. It is really a therefore, know in a general sense what their chil-

rae
 dren are doing nor whom they are with. In addi- eral consensus among parents as to which child
tion, they do not monitor the child’s work chores behaviors are normal and which are deviant. For
- or homework study times. Most of these parents each distribution (normal and deviant), the events
have given up asking the problem child and/or are ordered from most to least preferred. The or-
siblings to do anything. “It is easier to do it my- dinates describe the expected frequency of occur-
self.” Indeed, it is easier. rence for the event in a population of children. As
At no other time of life will the individual find shown in Figure 11.2, there is an area of ambigui-
anyone who cares enough to monitor him 16 ty represented by the overlap between the distribu-
hours a day and who has the power to give him tion of deviant and prosocial behaviors. There are
feedback about what he is doing. If the parents do some prosocial behaviors, such as Cry or Whine,
not perform this act of love and commitment, who where it is not clear whether the behavior is an un-
will? What agency worker has that kind of time? If desirable prosocial act or a very mild deviant
he/she had the time, what contingencies would be event. There are other behaviors, such as attack
required in order to get the child to respond to the with a knife, that most adults would identify as
feedback? No monitoring means no punishment. deviant. Given that the deviant behaviors fall
No punishment means that the child slips even fur- along a dimension of acts directed against persons,
ther out on the deviancy dimension. then the assumption is that the parent of the prob-
Finally, the community and/or family forcibly lem child operates on a very strict criterion value.
calls parental attention to extreme events, such as He or she reacts in an irritable fashion to many
assault, firesetting, and stealing. In effect, the par- “minor” events that parents of normals would ig-
ent must be aware of it; the events are so salient nore (this is detailed in Chapter 8). The laboratory
that the parent must respond in some way. How- findings by Lorber (1981) are in keeping with this
ever, the parent who is uninvolved and/or fearful formulation. He compared the reactions of par-
of a confrontation with the child has one final re- ents of normal and antisocial children to video-
source. He or she can deny that the event really taped interactions of various families. They were
was an example of something to be punished. As to press one button when they viewed an antiso-
Reid and Patterson (1976) note, many parents of cial act and another for a prosocial act. Parents of
young stealers simply refused to label a given event problem children were significantly more overin-
as “stealing”; e.g., “I didn’t see him do it. That clusive in their categorization of socially aggres-
teacher is always after him; the kid said he didn’t sive acts and underinclusive when classifying pro-
do it, so what can I do? He said that a kid gave social behaviors.!
him that radio.” Effective treatment required a I think that parental criteria vary as a function
subtle relabeling of events so that the parent could of the kind of deviancy. My hunch is that Robins’
then (reluctantly) agree to take action. and Ratcliffs (1978) negligent parent, or the un-
motivated parents of stealers, would tend to set
Detection and Labeling their threshold value to minimize their response
Other investigators have noted that parents of cost. Therefore, when categorizing acts against
antisocial children were less likely to monitor their property (vandalism, firesetting, stealing), they
child’s activities. For example, Wadsworth (1979) would probably classify as deviant only those
found that parents of delinquents were less likely events that are extreme, e.g., setting fire to the
to check on their child’s school progress. All par- parents’ home. On the other hand, being accused
ents monitor and set sanctions for some deviant by the teacher of stealing would not necessarily be
child behaviors, e.g., attack with a weapon upon classed as deviant. Parents of stealers are thought
family members, setting fire to the house. Harbin to be characterized as having a lax criterion for
and Madden (1979) report exceptions even here. acts against property (see Figure 11.2). As yet this
They describe instances of brutal assault in which hypothesis has not been tested. For these parents
the parent was a passive spectator. However, it is one important function of treatment is to help
generally the case that all parents identify and la- them readjust their criteria further to the left so
bel some behaviors as deviant. Even though an act that they will accurately classify as deviant more
is categorized as deviant, it doesn’t necessarily of the child’s behaviors that are directed against
mean that the parent then implements effective the property of others.
sanctions. However, an absence of such a classifi- In the last decade a number of investigators
cation would mean an absence of negative sanc- have recognized the importance of the manner in
tions. Classification as a deviant act would, at which the parent categorizes behavioral events as
best, place the child at risk for a negative sanction. deviant or normal. These studies provide a prelim-
For the moment, let’s assume that there is a gen- inary base for understanding the accuracy of pa-

223
rental tracking. For example, one can understand favorable light? In the study by Erickson (1973),
why parents generally tend to underestimate fre- one group of observers tracked deviant child be-
quencies for almost any test event they are asked haviors, and a third group simply observed care-
to track. In contrast to a professional observer, a fully. Those tracking deviant behaviors formed a
mother is tracking a multitude of events, many of less favorable general impression than the observ-
which occur simultaneously; it stands: to reason ers in the other two groups. Observers who
that she would “miss” many. In the pilot study by tracked prosocial events formed more favorable
Peine (1970), three mothers and their preschool impressions. The latter finding was replicated in
children were observed in a laboratory setting. the study by Hines (1974), while the former was
The mothers and trained observers collected data not.
over a series of trials on a number of target events Before leaving this topic, there is one further
for many behaviors. The across-trial correlations question that needs to be emphasized. It seems
were in the .80’s for mothers and observers. How- conceivable that future studies will demonstrate
ever, on the average, the mothers underestimated that parents of problem children are highly selec-
the level of deviancy by 183%! The mothers sim- tive in their errors of classification. They might
ply saw less of it than the trained observer. Inci- correctly classify all deviant child behaviors as de-
dentally, if these findings were confirmed by addi- viant except stealing, lying, and firesetting. Are
tional studies, it would raise questions about the there patterns in what it is that parents classify as
efficacy of survey data in estimating base-rate val- deviant? How does this selectivity develop?
ues for deviant child behavior. These data may se- Currently, the monitoring-labeling FM skill is
riously underestimate rates of deviant child behav- being measured in a variety of ways (Patterson,
ior. Stouthamer-Loeber, & Loeber, in preparation).
The Peine study needs to be replicated and ex- There are items from the Observer Impression In-
tended. The author carried out one such study ventory that provide a rating of parental monitor-
with a group of Stanford undergraduates as ob- ing. Did the parents talk with the child about his
servers (and co-investigators). The data showed day or not? The Parent Daily Telephone Interview
the expected Peine Effect; i.e., the mothers under- provides a daily check on whether the parent
estimated the actual frequency of most behavioral tracked the child’s whereabouts over the preceding
events as compared to the rates recorded by ob- 16 hours. These various measures were used to
servers. However, when mothers served as observ- generate three clusters of variables that measured
ers for interactions of other mother-child dyads, parental monitoring. All of these composites con-
their rates were much closer to the rates recorded tributed to a multiple-regression analysis using
by the observers. Underestimation was primarily a four different measures of antisocial behavior. The
function of being a participant in the ongoing pro- multiple correlation with peer nomination scores
cess. for antisocial child behavior was .26 (F=2.85,
However, such a variable would not differenti- p<.05); with delinquency life-style (self-report)
ate mothers of problem and nonproblem children. the correlation was .46 (F=10.73, p<.0001);
What is the impact of repeated exposure to high and for Total Aversive Behavior (observation) the
rates of deviant behavior? Does the mother habit- correlation was .30 (F=1.90, n.s.). The discrimi-
uate to deviancy? A study by Lorber (1978) pro- nant function analysis produced a percent correct
vided a direct test of the habituation-exposure hy- classification of 82.7% between children with and
pothesis. He used college students to observe vid- without court records. These findings suggest that
eotapes of parent-child interaction. His findings monitoring may contribute more to our under-
demonstrated that prior exposure to high rates of standing of stealing and delinquency than to our
coercive child behavior was followed by signifi- understanding of social aggression.
cant decreases in the accuracy of tracking deviant
Parental Sanctions
child behavior. The findings are in keeping with
the habituation-exposure hypothesis. Does the “One of the striking features of most studies of
same process also relate to the parental disposition multi-problem families is the chaotic state of their
for overinclusiveness of socially aggressive acts? patterns of supervision and discipline. Moreover,
Finally, there are a series of studies that have in- poor supervision has been one of the common an-
vestigated the relation between tracking and label- tecedents of delinquency in most investigations.
ing. If the mother of the problem child were ... Good supervision and well-balanced discipline
trained (as she is in OSLC treatment programs) to can serve to protect children from a high-risk
more accurately track the minutiae of deviant be- background . . .. in conditions of chronic stress and
havior, would she see her child in a more, or a less, poverty, strict parental supervision of the child’s

224
activities was more effective in preventing delin- believe that it is impossible to change behavior by
quency than was a happy family atmosphere. . . .” the use of reinforcement. The effect of a contin-
(Rutter, 1979, p. 64) gency becomes apparent only after many trials.
Behavior change is a mystery. Because parents do
From the standpoint of coercion theory, parents not understand the determinants of their child’s
of antisocial children are more likely to be non- behavior, they tend to have poorly articulated fu-
contingent in their reactions to both prosocial and ture goals for their children. If you believe that
deviant child behaviors. Their inept use of punish- events are generally outside the control of the indi-
ment is a major determinant for antisocial child vidual, then there is little reason for acommitment
behavior. Their inept use of positive reinforcement to long-term goals for child rearing. This is in
is amajor determinant for many of the child’s skill keeping with the general time frame for these par-
deficits. These statements go far beyond the pres- ents, which is the immediate present. I think that
ently existing data base. They constitute a set of the emphasis upon immediate comfort and the rel-
working hypotheses that guide many of our ongo- ative lack of long-term goals may relate to a lax
ing investigations at OSLC. These hypotheses criterion-threshold score for both reinforcement
were generated primarily from our clinical con- and punishment confrontations; this may also co-
tacts with these families. vary with the lack of attachment of parents to the
In keeping with Lorber’s (1981) laboratory find- caretaking role, which seems to characterize par-
ings, these parents tend to “overlook” prosocial ents of stealers in particular. For others, it may re-
behavior. They forget to reinforce. Many of the late to the fact that they are terribly angry with the
reinforcers that are given tend to be presented con- child and feel it is inconsistent to reinforce a child
tingent upon no particular response; i.e., they are who does such awful things to them. For these
noncontingent. When disciplining is necessary, parents, reinforcing the child is hypocritical;
they tend to nag, scold, lecture, and natter, but “Why should I reward him for doing what he
they do not confront. They do not say, “No.” should do!”
They do not say, “Stop that behavior here and The criterion threshold for punishment may be
now.” In that they do not back up many of their equally complex. I suspect that for many families
threats, their punishment is noncontingent. When the determining variable is fear. It is an avoidance
they do confront the child, it tends to be in a of the pain that results from confrontation. A
weakened form that all parties know will not well-practiced antisocial child is a formidable ad-
stand for long. They often use “grounding”; e.g., versary in such a confrontation. For others, the
“You stay in this house and in this yard for the threshold may be set at a lower level because the
next month.” For emphasis, the father stomps out spouse interferes with attempts to punish behav-
of the house in self-righteous indignation. Three iors that are not extreme. Other parents seem to
days later the child slips away for an hour and pre- believe that their particular child deserves special
sents what seems to be a reasonable excuse, “My consideration. For example, both parents work
bike tire was flat.” Within a few days he is back on and feel guilty about its effect on the child, or the
the street again, and the incident is forgotten. child had a severe illness that ‘warrants special
As a general case, we might assume that paren- treatment.
tal consequences are forthcoming when the par- At a very speculative level, these parents seem to
ents’ expectations are exceeded. That seems rela- operate on an affect basis. They reward and pun-
tively straightforward in that it suggests that par- ish on the basis of how they “feel” rather than on
ents differ in the points at which they set their cri- the basis of some set of clear-cut criteria. They
terion thresholds for reward and punishment. Giv- natter because they are irritated. They physically
en this, it should be possible to assess these thresh- assault because they are angry. They reward when
olds and correlate them with the behaviors that they are very pleased, i.e., excited. This affective
the parents are observed to provide consequences change occurs only when the child has done some-
for. As yet there are no studies that directly test thing truly outstanding. Rewards are, therefore,
this hypothesis. In thinking about these criterion low base-rate events in the lives of these people.
thresholds, there are several hypotheses that come Ordinary prosocial behaviors are taken for
to mind. The first is that parents of antisocial chil- granted. Small improvements in behavior are “ex-
dren react consistently to maximize their immedi- pected” and, therefore, are not rewarded. Praise is
ate comfort and satisfaction. In the short run, it is reserved for Olympian levels of performance and,
aversive to be distracted from one’s own affairs in even then, may be accompanied by criticism. This,
order to track child behavior and set conse- in turn, works against the concept of “shaping”
quences. I also think that many of these parents behavior by providing positive reinforcement for

225
small increments of improvement. The natural There are data suggesting that the sex and person-
outcome of reinforcement-for-excellence is a large ality of the nursery school teacher significantly de-
set of skill deficits.? The implication of the affect termine the reinforcing contingencies provided for
base for punishment was detailed in Chapter 6. It boys and girls in that setting. The observation
was suggested there that nattering related to ex- study by Fagot and Patterson (1969) showed that
tended coercive chains and to increments in inten- female nursery school teachers reinforced both
sity. boys and girls for feminine behavior. On the other
There is one further implication the affect-base hand, they were unlikely to provide positive con-
concept has when applied to rewards. As noted by sequences for behaviors that were considered to be
Kahneman and Tversky (1973), the strategy of re- masculine. Male peers did reinforce masculine be-
warding only for excellence is likely to result in a haviors. Robinson (1976) compared 20 male care-
loss of belief in the efficacy of rewards. It is almost givers (nursery school teachers), 20 female care-
always the case that the next response will be infe- givers (teachers), and 20 male engineers. Judges
rior to the exceptional one; i.e., to the observer, from all three groups thought that boys should be
the effect of reward is to weaken performance! reinforced for masculine behaviors and girls for
There is yet another aspect of social reinforcement both masculine and feminine behaviors. However,
that leads many of these parents to doubt its effi- actual observations of their interactions in a nur-
cacy. Most of the so-called reinforcing events are sery school showed that both male and female
probably not, in and of themselves, automatically caregivers reinforced boys and girls more for femi-
activating. They are low-key and, therefore, not nine behaviors than for masculine behaviors! They
as readily noted as are aversive events. As Solo- also punished masculine behaviors more than fem-
mon and Corbit (1974) point out in their two- inine behaviors. Questionnaire data showed that
stage theory, both the anticipation and aftermath the male and female caregivers were more femi-
of aversive events are characterized by powerful nine in their personality scores than were the male
autonomic reactions. However, in the case of most engineers (i.e., less oriented toward endurance
positive social reinforcers, the affective reaction and achievement and more oriented towards sym-
occurs primarily during deprivation. At an affect pathy, affection, and emotional support).
level, the child understands about the familial sup- In effect, well-trained nursery school teachers
port system when he visits a friend’s house for the reinforce differentially as a function of their own
first time and discovers that he is homesick! The personal values rather than what they believe the
young couple that has grown accustomed to in- child needs; i.e., all teachers reinforce males for
tense mutual sharing and support feel intense love feminine behavior even though they agreed that
when the spouse is absent for a few days. The boys needed to be supported for being masculine!
writer believes that this delay of affect that is evi- Fagot (1978a & b), in her review of the rather ex-
dent during deprivation rather than when the rein- tensive literature that was based upon her first
forcement occurs leads many parents to steadfast- publication, concluded that the general effect was
ly refuse to believe in the power of positive rein- reliable across investigators.
forcement. In that sense, pain is obvious, and sup- In an analogous fashion, does the personality of
port (reinforcement) is not. the parent correlate with the likelihood that he or
she will contingently reward the child? Does it co-
Positive Reinforcement vary with his or her perspective on internal-exter-
The hypothesis is that parents of antisocial chil- nal control, or his or her emphasis upon the imme-
dren are less effective than parents of normal chil- diate present versus long-term strategies? We
dren in their use of positive reinforcement to shape know very little about the variables that determine
prosocial behavior. This breaks down into three the nature of parental reinforcement. We do know
hypotheses. First, the parents of antisocial chil- that in normal families events that seem to func-
dren tend to use fewer reinforcers for prosocial be- tion as reinforcers occur with some regularity.
haviors, e.g., approval. Second, they are less likely Physical Positive (touches) by mothers occurred at
to use social reinforcers contingently. The third a rate of .06 per minute and Approval at the rate
hypothesis describes the outcome for these over- of .10 (Reid, 1978). The rate of occurrence for
sights. Variables that measure these aspects of pa- these reinforcers was significantly lower for moth-
rental skill would be expected to show low-order ers of antisocial children (Patterson, 1980a). A
correlations with various measures of the child’s comparison of rates for mothers giving approval
prosocial survival skills. The more inept the par- to the target child showed a mean of .009 for a
ent is as a reinforcer, the more unskilled the child. sample of Normals, .007 for Social Aggressors,
What is it that adults reinforce children for? and .005 for Stealers. The samples of mothers also

226
 varied in terms of what they approved of. For ex- ever, in spite of all this punishment-like activity,
ample, given child-Comply, the likelihoods for children with conduct problems perceived their
- mother-Approve were .028, .018, and .006 re- parents as being unable to set limits (Goldin,
spectively. The conditional probabilities, given 1969). Even when limits were set, the children
child-Play, were .010, .005, and .001 respectively. perceived their parents as being unable to enforce
These findings are in keeping with the hypothesis them. Why is this? The observation data discussed
that the parents of antisocial children provide less earlier showed that parents of socially aggressive
positive reinforcement for the maintenance of pro- and abused-child families were significantly more
social behaviors. likely than parents of normals to respond aversive-
There is one further dimension along which par- ly to deviant child behavior. They punish more
ents differ in their use of positive reinforcement. frequently, and they are more likely to punish. If
This difference did not become apparent to us un- this is so, then why are their children more antiso-
til the summer of 1979. As a basis for constructing cial? In spite of all of this “activity,” these parents
the new code system (MOSAIC), we were spend- are unskilled in their use of punishment. There are
ing a good deal of time in the homes of “super nor- three components to this lack of skill.
mal” families. These families often engaged in The first problem is that their criterion thresh-
something we had almost never seen in families of old value for confrontation is too low (see Figure
antisocial children. The phenomenon might be 11.2). They threaten and scold very often, but
characterized as an intensive “How-was-your- they seldom follow through. As used here, con-
day” exchange. At this time, information was frontation means that the parent responds in such
shared about the day’s activities. These exchanges a way as to both stop the immediate deviant event
provided a rich basis for explicit “That’s good” and to reduce the likelihood of its future recur-
statements. In addition, it was a time during rence. Effective confrontation means both the im-
which the children received undivided and non- mediate suppression of ongoing behavior and
critical parental attention and interest. The child’s weakening the connection between the event and
report elicited both sustained interest and a discus- its controlling antecedent. My hypothesis is that
sion with the parents. I suspect that these periods the problem child often wins in such an out-and-
of shared information are a powerful mechanism out confrontation; therefore, these parents natter,
for shaping the child’s values and interests. Given but they do not confront. In their study of severely
that this was a frequent occurrence, one would ex- disrupted families from inner-city London, Quin-
pect that the members of normal families would be ton (1980) noted that parents of antisocial chil-
more homogeneous than distressed families in val- dren (more often than parents of the control sam-
ues, interests, and recreational activities. ple) did not bring punishment to some firm con-
These considerations are an important feature clusion.
of the treatment procedures for distressed families. When the parents of problem children do con-
In the beginning stages of treatment there is an em- front, they often respond with anger rather than
phasis upon parental tracking of prosocial behav- with some other more appropriate response. The
iors and an emphasis upon the use of both social result can be a severe beating. Reid, Taplin, and
and nonsocial (money, points, privileges) rewards Lorber (1981) found that roughly 30% of the
for these behaviors (Patterson, Reid, Jones, & OSLC clinical sample was comprised of abused
Conger, 1975). The prosocial behaviors typically children who had been referred for treatment be-
include doing chores, being cooperative, playing cause they were out of control. If an angry parent
well with siblings, doing homework, or bringing punishes extremes of deviant behavior, then in
home a positive report from the teacher. terms of the Kahneman and Tversky (1973) for-
mulation, this would have several interesting im-
Punishment plications. As applied to antisocial children, it is
Parents of antisocial children punish more often unlikely that a series of two coercive episodes
than do parents of normal children. This is true re- would both be characterized as extreme; i.e., the
gardless of what type of punishment is considered. second should tend to be less extreme than the
Their children may engage in three or four times first. Therefore, given that the extreme behavior
more deviant behaviors, and these parents re- was punished, and that the second was less ex-
spond by engaging in considerably more scolding, treme, then the logical conclusion is that physical
threatening, and hitting. I’m not sure whether violence works! The. same syllogism that con-
these parents use deprivation of privileges, with- vinces the parents of antisocial children that re-
holding allowance, or work as punishment more ward does not work may also provide support for
frequently than parents of normal children. How- the idea that physical punishment accompanied by

B27
anger does. In addition, the beating would also studies reviewed in Patterson and Fleischman
have short-term effects that are reinforcing to the (1979):
parent, i.e., the child behaves. The third type of parental deficit lies in a fre-
However, the child’s extremely deviant acts are quent use of commands, many of which are poorly
embedded in a pattern of related acts that occur at stated. This may be thought of as a form of natter-
high rates. The child who sets fires also steals, lies, ing, but it has some special properties that lead me
and spends a great deal of time unsupervised (see to think that it should be considered separately. A
Chapter 11). An occasional beating for a particu- parental command can start up the coercion pro-
lar theft does not change the problem. The child cess in an otherwise quiet interval. The fact that
continues to come in at all hours of the night, he the children have been poorly trained means the
lies, his chores are not done, and he is failing in parent must issue more commands. The data from
school. The parent cannot beat him for each oc- the Observer Impression Inventory showed a trend
currence of these high-rate acts. None of these are for parents of Social Aggressors and Stealers to be
extreme enough to exceed the criterion threshold more likely to issue commands that were unclear
value; they are, therefore, neither labeled as devi- (.30 and .26) than were parents of Normals (.11).
ant nor punished. While the parents are irritated The F value was .66 (df= 2:41, p<.20).
by these recurring problems, they do not necessar- Given that the parent is irritable, a request can
ily perceive that these child behaviors need to be quickly become a command. Many of the com-
changed (confronted). As noted in earlier discus- mands are accompanied by overtones of things
sions, their nattering elicits further coercive inter- that happened in the past, a litany of yesterday’s
actions and is associated with extended coercive sins and omissions. These negative accompani-
interchanges. As shown in the Loeber (1980) ana- ments increase the likelihood of noncompliance
lysis, such extended chains were correlated with (this hypothesis is untested). Acommand is quali-
hitting. tatively different from a request; it is the difference
Nattering makes things worse. Members of an- between the drill sergeant’s “Get in here!” and
tisocial families are more likely to natter than their “Could you come here for a minute, please?” The
normal counterparts. This second skill deficit parent of the problem child is not only more likely
seems to relate to the parents setting their criterion to command but, in addition, to tack on further
threshold value too strictly. Almost anything the aversive baggage, “Damn it, you never do what I
child does can irritate the parent and elicit natter- tell you to; now get in here!” For many parents of
ing. Sallows’ (1972) function analysis showed that problem children, one cannot be certain of just
not responding was the most effective way of re- what it is that they wish done; what is communi-
ducing the likelihood of extended coercive se- cated is anger and impatience. On the other hand,
quences. If the parent behaved as if the coercive re- the skilled parent is more likely to express even a
sponse never existed, it was unlikely to continue. request in nonabrasive terms, “I think your toy
Our observations in normal homes gives support will get broken if you leave it there; what do you
to this theme. Effective parents often “over- think?”
looked” a coercive child behavior and simply de- The observation study by Halverson and Wal-
flected the behavior. They might use humor or drop (1970) showed that some mothers tend to
they might respond verbally in such a way as to al- have a consistently abrasive style whether they are
ter the meaning of the entire interchange. When coping with their own children or other children.
they did choose to punish. it was not accom- The correlations were .52 and .48 between likeli-
plished by a casual threat, scolding, or sarcasm hood of negative statements delivered by the
(nattering). They stopped whatever else they were mother to her own children and other male and fe-
doing and persisted in their punishment until the male children respectively. They also found that
behavior stopped. The OSLC treatment program boys who were rated as impulsive elicited more
teaches parents of problem children either to de- commands from their mothers. This relation be-
flect the sequence or to punish it effectively (Pat- tween rates of deviant child behavior and mother
terson, Reid, Jones, & Conger, 1975). They are commands is also a consistent finding. Delfini et al.
also taught to use nonphysical forms of punish- (1976) compared homes of aggressive children to
ment, which can be applied to each occurrence of homes of nonaggressive children. Parents of prob-
deviant behavior. Effective parent training would lem children not only gave significantly more com-
require pairing stop commands with an appropri- mands, but for the clinical sample, child deviancy
ate back-up punishment. This should increase the and mother commands correlated .63 (p <.01).
effectiveness of parental punishment. This was, in The Lobitz and Johnson (1975) studies provided
fact, demonstrated and replicated in the outcome further confirmation for these findings. They also

228
 demonstrated the causal relation between in-— cur. For example, Hicks and Platt (1970) showed
creases in parental commands (for both normal that in normal families the kind of problems en-
- and deviant behaviors) and deviant child behavior. countered did not change over time. However, I
The assumption is that measures of these three think that in distressed families these unresolved
skill deficits will covary. Parents of antisocial chil- crises accumulate. For example, 50 normal and 50
dren should be less likely than parents of normals distressed couples were compared in the study by
to confront their children for serious deviant be- Birchler and Webb (1977). Their response to an
havior. They should also be more likely to natter area-of-change questionnaire showed an average
about minor deviations they might better ignore. of 28.5 areas thought to require change by dis-
Finally, they should issue more commands and tressed couples. This is in contrast to 6.9 areas for
commands of poorer quality. These hypotheses nondistressed couples. Presumably, families of
will be tested by the assessment measures con- conduct-problem children would demonstrate a
structed for the study of children at risk for delin- similar accumulation of unsolved problems. In the
quency. That study includes a structured interview Birchler and Webb study, the distressed couples re-
that contains nine items relating to these ques- ported a mean of 3.4 conflicts per week; this was
tions, a revised Observer Impression Inventory in dramatic contrast to the mean of 1.0 for nondis-
with 11 items that are relevant, and a daily Parent tressed couples. Quinton (1980) noted that dis-
Telephone Interview that contains yet more items tressed families had twice as many unresolved con-
relevant to punishment and disciplinary practices flicts and crises as did matched controls from the
(Patterson & Stouthamer-Loeber, in preparation). same area of inner-city London.
The analyses produced two rather different sets of The accumulation of unsolved conflicts and cri-
composite scores. The first set, which we labeled ses relates in part to the fact that little time is given
Obedience, consisted of four composite scores to problem solving, and even when time is given,
based upon interview and questionnaire data. The members of distressed couples demonstrate a lack
multiple regression analysis showed a correlation of the skills that are necessary to put the time to
of .36 (F=5.07, p<.001) with delinquent life- good use. Videotapes of their problem-solving ses-
style (self-report), .47 (F=9.42, p<.0001) with sions showed that during problem-solving situa-
peer nominations for antisocial behavior, and .21 tions distressed couples engaged in significantly
(F=0.61, n.s.) with TAB scores. The discriminant more coercive behavior than did nondistressed
function analysis showed a correct classification of couples (Birchler, Weiss, & Vincent, 1975). A
69.9%, differentiating children with court of- similar phenomenon was observed in a compari-
fenses from those without. The second set of items son of families of normal and antisocial children
was based upon observers’ global ratings follow- (Forgatch & Patterson, in preparation). As one
ing each observation in the home. This set of four observes these interchanges, the desperation of the
variables, labeled Discipline, correlated .56 people involved can be seen very clearly. A discus-
(p<.001) with TAB scores for fourth grade chil- sion about an area of change has the quality of
dren but were nonsignificant for measures relating having been repeated many times before; the
to stealing and delinquency. thrust and repartee have a well-practiced quality.
The outcome is anticipated, and the forthcoming
Problem Solving impasse is viewed with mounting frustration. As
It is assumed that parents of antisocial children the impasse is reached, the language becomes in-
are less skilled in solving family problems than creasingly sharp and argumentative.
parents of normal children. This skill deficit is These problem-solving tasks may take place in a
thought to covary both with rates of antisocial be- laboratory setting, but the interchanges are far
havior and with the number of crises that impinge from sterile. The most salient features are the co-
upon the family. erciveness of the participants. The comparison
Effective family management requires a resolu- study showed that the mother’s rate of aversive be-
tion of the crises that constantly impinge upon havior during familial problem solving correlated
families. Some crises arise from within the family, negatively with observers’ ratings for the quality of
such as an unresolved conflict between two family the solution reached by the family.
members. Some crises arise from outside the fami- Families of conduct-problem children have a
ly, such as debts, unemployment, and conflict at long history of failure in solving problems. Their
work. Ideally, it is the shared responsibility of the efforts to solve crises. and conflicts are not fol-
husband and wife to anticipate many of these cri- lowed by perceivable changes in anyone’s behav-
ses before they arise and to resolve those which do ior. Even if a discussion results in some type of an
occur. Most of us have some crises that tend to re- agreement, the changes are either temporary, or

Dee
simply never occur. Talking and negotiating were treatment program for families of antisocial chil-
probably never very popular modes for producing dren. This brief description is not intended to fully
behavior change with these people. Failure makes describe the process but only to illustrate some-
these activities even less attractive. Discussion and thing that normal families do with little fanfare.
requests quickly degenerate to demands. Demands Probably the best description of the clinical proc-
are accompanied by negative verbal statements, ess is to be found in the recent publication by Ja-
“You are a drunk just like your old man. Every cobson and Margolin (1979). Problem solving be-
payday you take off and drink up half of your pay- comes most evident when it is absent. It is only
check.” Here, the wife does communicate in a gen- when the debris of unsolved problems is every-
eral way the area that needs to be changed; i.e., where that this omitted mechanism comes into fo-
the husband should first bring his paycheck home cus (see Illustration #3).
(and then go off with the boys). However, he is It is asad commentary that even in normal fami-
more likely to respond to one of the two insults lies there is often sufficient disruption in the proc-
than to the issue itself. In fact, he is most likely to esses of problem solving that family members en-
counterattack by emphasizing some aspect of the gage in higher rates of coercion than should be
spouse’s perceived failures. “You are such a bitch; necessary. Bales (1953) noted that the most abra-
you drive me to it.” First one, then the other, be- sive small groups encountered in his laboratories:
comes angry. When the discussion ends, the prob- were families! Halverson and Waldrop (1970)
lem solving has been effectively sidetracked. It is noted a similar lack of tolerance in dealing with
added to the accumulation of unsolved crises, one’s own family members. In that study, mothers
which will probably be rerun on the next payday. of preschool children were observed to be five
This contributes to the attrition in esteem for both times more abrasive in interactions with their own
self and others. It is further proof that things will child than they were in interactions with other
never change. Similar disruptions in communica- children in the same setting! In many instances it is
tion and inept technique have been identified in not so much the'case that family members do not
other clinical samples as well. For example, R. Li- know how to cope with each other in a pleasant
berman and his colleagues at Camarillo work with fashion, but they do not try to do so. It is not
family members of schizophrenic patients (Liber- learning, but performance, that is the issue. The
man, et al., 1979). Their training in problem-solv- study by Vincent, et al. (1979) was a classic in this
ing skills for patient and family members heavily regard. They arranged for distressed and nondis-
emphasizes the concepts of receiving skills, proc- tressed couples to solve problems in a laboratory
essing skills, and sending skills. Their approach situation. On counterbalanced trials, each spouse
draws heavily from the work of Spivack, Platt, engaged in problem solving with a spouse and a
and Shure (1976). nonspouse of the same sex. Problem-solving proc-
The treatment program for family members of esses were significantly disrupted when distressed
antisocial children emphasizes the following skills: spouses interacted. These disruptions were not ob-
(1) stating the problem in neutral terms; (2) speci- served when the same individuals interacted with
fying clearly what changes are desired; (3) accom- nonspouse adults of the same sex! It can be in-
panying such requests with a positive statement ferred that problem-solving skills were available to
about some other aspect of the person that is ap- the distressed spouses. It was as if family mem-
preciated (this communicates to the other that bers, particularly those in distressed relations, de-
he/she is a good person, but there is this one spe- clared a moratorium on etiquette and rational at-
cific behavior that is a problem); (4) training the tempts to change a problem situation. They en-
recipient to paraphrase the request (it is amazing gaged, instead, in something that seemed analo-
how many of these family members have learned gous to nattering. They express irritation and an-
to effectively “tune out” requests for change. It is noyance, but this does not produce change in each
interesting to note that this also is a key portion of other’s behavior.
the Liberman et al. (1979) social skills training for Typically, what is missing for the caretaker is a
schizophrenics and their family members); (5) support system for problem solving. As noted in
brainstorming alternative solutions; and (6) conse- an earlier discussion, it is not the presence of more
quences are also set because it is expected that the external crises that determines the psychiatric sta-
agreement will occasionally be broken. Typically, tus for an individual; it is, in addition, a function
these agreements are written down and involve of the availability of a support system. Is there a
two or more persons stipulating areas of change. friend, a relative, or a husband who can provide
Clinically, this training process is extremely del- daily assistance in dealing with unsolved prob-
icate; it is, in fact, at the very core of the OSLC lems? In family sessions, is there one other adult

230
 _—_—-_
oe ee
———

Illustration #3: The Rising Tide of Unsolved Problems

who will join a coalition on the crucial issues? Is and the support systems are severely disrupted.
there one other member in family meetings who The studies of families in the process of getting a
can be counted upon to join in brainstorming for a divorce showed a rising tide of antisocial behavior
new alternative to solving an old problem? Is there for the children (Hetherington, Cox, & Cox,
another adult present who can stay calm when the 1978; Wallerstein & Kelly, 1976, 1980). I think
caretaker commits some unspeakable breach of that such severe stress disrupts problem solving
problem-solving etiquette? Wahler (1979) noted and other family management practices as well.
lower rates of aversive behavior on those days There are several other studies that suggest that
when the mother received positive support from when parents are in severe conflict, the children
outside the family. The modest correlations in Ta- are more likely to be out of control. It is not clear
ble 10.1 showed a similar covariation for four of whether the relation is due to the parents interfer-
five mothers. Were problems being “solved” dur- ing with each other’s child management attempts,
ing these contacts, or was it the emotional support or whether the parents are just not making the ef-
that was involved? At this point, we don’t know. fort. To what extent is disrupted problem solving
There is a related issue that also requires further implicated as well; e.g., crises, internal or external,
investigation. It is assumed that extreme stress can accumulate that, in turn, reduce the efforts of the
temporarily disrupt the performance of normal caretaker to manage the children? The Johnson
problem-solving skills. For example, Forgatch and and Lobitz (1974a) study found a correlation of
Patterson (in preparation) found a significant cor- —.48 (p<.01) between parental reports of mari-
relation between frequency of crises and mothers’ tal satisfaction (on the Locke-Wallis) and the
negative verbal behavior during familial problem child’s observed rate of coercive behavior in the
solving. One can think of the dissolution of a mar- home. Similarly, Oltmanns, Broderick, and
riage as being a time when there is a severe stress, O’Leary (1977) found correlations in the —.3 to

231
 —.4 range using parent self-reports as criterion aversive events leaves out those aspects of family
data. They also cite several other studies with sim- that are of greatest interest to most of us. Where is
ilar findings. They note the general tendency for the sense of shared purpose, values, interests, and
parents of children referred for treatment to report recreational activities? Where is the sense of be-
less satisfaction with their marriage. Reid, Taplin, longing and caring? As suggested in Figure 10.4,
and Lorber (1981), in their study of child-abusive these are the aspects that become the casualties of
families, noted a significant tendency for these a disorganized family. If crises and conflicts are
parents to display high rates of coercive behavior not properly managed, then there will be lowered
when interacting with each other. These studies rates of shared activities, lowered responsiveness,
reiterate the idea that FM variables may serve as and an increased sense of powerlessness and isola-
mediators between antisocial child behavior, on the tion.
one hand, and those variables identified in the re- The ratio of positive to aversive components in
search literature that often show low-level correla- the interaction might index the family’s progres-
tions with it, e.g., marital discord, broken homes, sion toward disruption in functioning. In Lennard
large family, low income, and criminality of par- and Revenstein’s (1969) review of the Bales stud-
ents. Presumably it is the interaction of one or ies, the ratio of positive to negative events during
more of these at-risk variables accompanied by a small-group tasks was 0.77:1.00 for the least sat-
disruption in family management practices that isfied group and 6.20:1.00 for the most satisfied
would increase the predictive efficiency. Not all group. Normal control families had scores of
children who live in the ghetto become delinquent. 1.05:1.00, and families with a schizophrenic
Not all homes in which there is severe marital dis- member had scores of 0.61:1.00. Birchler et al.
cord produce an antisocial child. Given that fami- (1975) found a significant difference between dis-
ly management skills are practiced by either par- tressed and nondistressed couples using a similar
ent, then the at-risk child may make a perfectly measure. In disrupted families there were fewer
normal adjustment. positive events and considerably more aversive
Measurements of family problem-solving skills events.* It should be noted that at OSLC no sys-
are obtained from four different modes of assess- tematic work has been done in exploring the use of
ment. A structured intake interview contains 25 such a ratio; it might, for instance, differentiate
items relevant to this issue; the Observer Impres- Stealer and Social Aggressor families from Normal
sion Checklist has four; and the daily telephone families. It may also describe families who have
call may provide additional information. The as- moved further along a dimension of disorganiza-
sessment procedure of key importance is thought tion and anarchy.
to be a videotape of efforts during family interac- As the process of disorganization continues,
tions. Here the members are asked to solve one of most of the family members avoid/escape each
their own previously identified problems. In the other’s presence. It is simply unpleasant to be
second episode, they are asked to plan a family around other members of the family. As a result,
outing. The interaction is coded by a complex sys- the rates of social interaction are reduced, as are
tem, PANIC (Forgatch & Wieder, in preparation), shared activities. The comparison of distressed to
which contains 40 content codes, 13 measures of nondistressed couples by Birchler et al. (1975)
nonverbal behavior, and ten qualifier categories. showed lower rates of interaction for the clinical
The studies of children at risk for antisocial behav- sample. For them, unsolved arguments suddenly
ior have demonstrated that several of the compos- emerge in the context of an otherwise pleasant
ite scores from the PANIC correlated significantly Sunday drive. This phenomenon is so prevalent
with the child’s TAB score, as well as his status as that marital therapists advocate a “problem-solv-
delinquent or nondelinquent. In the preliminary ing time and place” by appointment; distressed
analyses of the data for 85 families, the construct couples are “not allowed” to bring up conflict top-
measuring mother irritability during problem solv- ics at any other time. One logical solution is to re-
ing correlated .32 (p<.01) with the child TAB duce the amount of social interaction. This proba-
score and .31 (p <.01) with peer nominations for bly also happens to families of socially aggressive
antisocial behavior. The covariation with court re- children. Picnics, visits with relatives, vacations,
ported delinquency and with a self-report of delin- camping trips, and Thanksgiving dinners all be-
quent life-style score were nonsignificant (For- come battle grounds. In these families, coercive
gatch & Weider, in preparation). processes cannot even be turned off when relatives
or observers are present in the home. The classic
The Family is Disrupted studies by Johnson and Lobitz (1974, 1975) re-
A single-minded focus upon the management of viewed in Chapter 3 showed that families of coer-

232
 Figure 10.4
Outcomes for Disrupted Family Management Practices

Reduced social
interaction

Reduced responsiveness to
modeling, reinforcement,
and punishment

Sense of helplessness,
powerlessness

Loss of self-esteem
Impaired family Increase in level
for caretaker
management of Family Aversiveness

Increasing negative
Antisocial child
attribution to other
family members

Few shared activities,

recreation, etc.

Family isolated

cive children could not “fake good” even when work with distressed and nondistressed married
asked to do so! couples also supported this concept (Patterson,
Many of the parents in the clinical families had Weiss, & Hops, 1976). Birchler and Webb (1977)
to give up trying to go out without the children— showed that significantly more happy couples
baby-sitters simply refused to tolerate the chaos. (56%) engaged in shared activities than did un-
The parents simply stay home, turn up the TV, happy couples (46%). This agrees with earlier
and escape as best they can. Interactions are un- findings that suggested that marital satisfaction
pleasant, and one way of handling this is the sim- correlated positively with the amount of shared
ple expedient of no longer making plans for shar- recreation-leisure time (Birchler et al., 1975; Bar-
ing activities. For most of these families it also nett & Nietzel, 1979). Blood and Wolfe (1960)
means not doing anything for recreation. Our and Tonge, James, and Hillam (1975) found fewer

233
shared activities in their casework study of multi- lihood of attack. These two hypotheses from
ple-problem families. The Fagan et al. (1977) lon- Dodge’s work may well relate to a phenomenon
gitudinal design showed parents of younger delin- identified by David Littman as “crossover” (Litt-
quent-prone boys were characterized by lower man & Patterson, 1980). In a normal household
rates of shared leisure and recreational activities. there are occasional unprovoked attacks upon
Again, one cannot isolate cause and effect from other persons. These attacks are crossovers in that
such covariations. These findings do not offer di- the attacker shifts the interactional state from a
rect support for the writer’s hypothesis that high prosocial to a coercive one. In fact, as noted ear-
rates of coercive family behavior cause reduced lier, this characterizes about two-thirds of a nor-
rates of shared recreational activities. mal child’s coercive behavior; i.e., the antecedents
The imbalanced ratio, reductions in social inter- are prosocial or neutral. The data in Chapter 7
action, and the reduction in shared activities are showed that families of antisocial children were
accompanied by more subtle shifts that, them- more likely to engage in crossover. It seems rea-
selves, carry ominous implications for the family. sonable to suppose that such behaviors would pro-
The hypothesis is that members of such families duce negative attributions.
become less responsive to each other’s reinforce- These shifts in the level of coercive inter-
ment and become more likely to make negative at- changes, reduced interactions and shared leisure
tributions. The latter produces an increased likeli- time, reduced responsiveness, and increasing hos-
hood for unprovoked attacks; this, in turn, in- tile attributions take their toll at the interactional
creases the general sense of uncertainty in dealing level. There also seem to be accompanying shifts
with other family members. As detailed in Chapter at a molar level. The family itself is, or becomes,
12, the net effect of these changes is reflected most isolated from the support systems provided by the
graphically in the decrease in self-esteem for the community. Wahler, Leske, and Roberts (1977)
person who occupies the role of caretaker. noted that mothers of antisocial children who
The battles over ancient issues and the accumu- lived in the ghetto area had an average of less than
lation of new unsolved problems continue, and the 2.6 social contacts per day; most were initiated by
increasing level of aversive interchanges and de- people outside the family who were not friends,
creasing positive experiences demand an explana- e.g., agency workers. By contrast, the mothers of
tion. Each member must explain to himself how antisocial children from nonghetto areas had an
this could be; each can see that he has tried to re- average of 9.5 daily social contacts, most of them
late to the others. Certainly the individual’s good from friends. Other investigators have also noted
intentions are obvious to everyone, particularly to the.insular quality of distressed married couples
the individual. Why others would behave so (Birchler et al., 1975) and for families of delin-
badly is the matter of concern. Each family mem- quents (Fagan et al., 1977).
ber becomes adept at reading the malevolent in- The isolated quality of these families is of inter-
tentions of the others. These negative evaluations, est. It suggests that those families most in need of
in turn, feed the process of disruption. This im- community support systems are least likely to
portant facet of the coercion process is poorly un- make use of the opportunities that are offered.
derstood. In an extended baseline observation of a These families seem to function like an island of
preschool child, there was a correlation of .64 poorly organized and slightly xenophobic inhabi-
(df=18, p<.01) between the child’s rate of aver- tants. For example, there are a number of survey
sive behavior (TAB score) in the home and the studies that note the tendency for parents of ag-
mother’s ratings on a scale that included, “He is gressive children to be noncooperative when asked
being naughty just to get my goat.” On days when to fill out questionnaires or to participate in stud-
he was more aversive, she increasingly attributed ies (Eron, Walder, & Lefkowitz, 1971, Rutter,
malevolent intentions to his behaviors. It is as- Tizard, & Whitmore, 1970; West & Farrington,
sumed that the attributional process is generalized 1973). Longitudinal studies, such as the one by
to include all family members. Each tends to per- Eron et al. (1971) also note that these same fami-
ceive him- or herself as essentially powerless to do lies tend to be highly mobile. D. Elliott (personal
anything about what is going on in the family. communication) believes that most of their moves
As noted earlier, the studies by Dodge (1980) are within a limited area; that is, they move from
showed that in otherwise neutral situations, the one place to another within the city in order to es-
antisocial child was significantly more likely to cape from crises that build up as a result of their
attribute hostile intentions to the other and to at- futile efforts to cope with an assortment of irate
tack. Similarly, if the other person had been exper- neighbors, bill collectors, and agency personnel.
ienced as aggressive, this also increased the like- West and Farrington (1973) made a similar obser-

234
 vation. individual differences in stealing, social aggres-
Caretakers occupy center stage for this process sion, assault, and child abuse is the focus of the
- of family disruption. Confrontation with failure is current children-at-risk study. Those family man-
their daily fare. They can perceive the rising tide of agement variables that survive these multivariate
coercive interchanges, the failure of family mem- analyses will be incorporated into a longitudinal
bers to form positive attachments, and the failure study planned for 1982. I think only a longitudi-
of their children in academic and peer-related ac- nal design can answer the questions that must be
tivities. Many of them are also in the position of raised about the OSLC analyses of FM variables.
negotiating with agency personnel in order to satis- Even though the analyses demonstrate that this or
fy day-by-day needs. As noted in Chapter 4, these that pattern of FM variables differentiates Social
experiences take their toll by generating a sense of Aggressors from Stealers, or from abused chil-
helplessness and reduced self-esteem. The correla- dren, we will not know what is the cause and what
tion between mothers’ depression and general neg- is the outcome of a process. Is it the disruption in
ative self-evaluation with antisocial child behavior practice of a particular FM variable that, early on,
will be examined in detail in Chapter 12. initiates a process that eventually produces a Steal-
My clinical experience suggests that the disor- er? I assume that there are such differentiating pat-
ganized family moves in a steady progression to- terns and that FM variables will be shown to play
ward increasing anarchy (see Chapter 12). In this a key causal role for antisocial child behavior.
progression, children come increasingly to occupy
elements of the parental role. Pain control more
often escalates to high-intensity techniques, in-
Footnotes
cluding physical violence.
1. These results should be accepted with cau-
Implications tion in that they are in sharp disagreement with
Disrupted families can change. During the past the findings by Bogaard (1976), who carried out
decade we have become convinced that to treat an- the first study in this series. Both the larger sample
tisocial children one must change the social envi- size and automated procedures in the Lorber
ronment in which they live (Patterson, Reid, (1981) study lead me to give greater credence to
Jones, & Conger, 1975). The family, school, and the findings from this study. Finally, a replication
peer group constitute the major components of the and extension by P. Holloran (in preparation) pro-
child’s environment. The key concept is that it is vide data that are in close accord with those in the
the parent who must change the problem child. To Lorber study.
do so, he or she must learn how to be an effective 2. Many parents seem to use such contingen-
parent. If one can be taught to improve in his or cies very skillfully to produce children who are
her application of family management skills, then high achievers. However, I believe that a careful
this dismal process of disruption, avoidance, and study would show that they carefully track and re-
lowered self-esteem can be stopped. The role of inforce intensively at the early stages of learning.
the therapist is that of teacher and facilitator for Only later do they seem to shift to the extreme,
such change. It is his/her role to help the parent “Tll be pleased only if you get straight A’s.” The
alter values and norms concerning deviant child parents of antisocial children leave out the initial
behavior and to overcome the fear or hesitation in block of training trials.
confronting the child on these issues. Even if the parent is successful in reinforcing
The family management variables outlined here high achievement behaviors, he may end up with a
are the core of the treatment program. They are, child who detests the whole academic enterprise.
in fact, an outgrowth of the therapy process itself. The parent has been reinforcing contingent upon
Each of these FM skills must be carefully role- the child’s being “better-than,” i.e., for successful
played and supervised. Simple as they are concep- competition. The programmatic work by Lepper
tually, they prove extremely difficult in their appli- and Dafoe (1980) and others showed very con-
cation. The problem is, how does one provide an vincingly that, even when successful, the child pre-
empirical base for our sense of deep commitment fers not to engage in such activities.
to these FM variables? The success rates in teach- 3. Some parents may be able to effectively use
ing them to parents have been gratifying; the out- aversives to shape prosocial skills in the child,
come studies are briefly reviewed in Chapter 13. e.g., negative reinforcement for skilled perfor-
However, a successful application of the FM vari- mance and punishment for inept performance. An
ables does not, of course, establish the causal sta- adequate prediction of child skill level would re-
tus that we ascribe to them. Their contribution to quire the inclusion of measures of parental skill in

235
using such pain control techniques. imbalance. The problem there would be that of
4. What I have in mind here are families of an- knowing when an aversive had occurred. It is my
tisocial children. Although I have never studied impression that in these families the aversives are
them systematically, I wonder if families of “neu- very subtle and, perhaps, idiosyncratic to each
rotic” children would not display a comparable family.

236
 Chapter 11
Abstract
The hypothesis that there is a continuity between early and later forms of antisocial behavior is ex-
amined. The corollary for the continuity hypothesis is that deviant-child behavior covaries with dis-
ruptions in family management practices. Antisocial children are said to have the following in com-
mon: a tendency to be extremely noncompliant, a tendency to maximize short-term gains, relative
nonresponsiveness to social reinforcers and punishments, attentional deficits, arrested socialization,
poor survival skills, and poor prognosis for adjustment as adults.
Antisocial behavior is defined by two subsets, Stealer and Social Aggressor. By definition, the sub-
sets are characterized by different symptoms ascribed to them by parents at intake. The discriminant-
function analyses for samples of Normals, Stealers, and Social Aggressors show that mothers and, to
a lesser extent, fathers of Stealers are more distant and unfriendly. Parents of Social Aggressors, Lee
ticularly the mothers, are more coercive.
The covariation of microsocial measures of irritability with performance level for Social Aggres-
sion is examined. In a multiple-regression analysis, the likelihood of Continuance for mothers and
fathers and the likelihood of Crossover for siblings accounted for significant amounts of variance for
differences in performance level, as measured by TAB and PDR scores. These findings support a bi-
lateral definition of aggression; i.e., the reactions of family members are a major determinant for dif-
ferences in the target child’s performance level.
Multimethod assessments for family management variables are correlated with four criterion meas-
ures of Social Aggression. For two samples, the scores for parental Discipline and Obedience corre-
late significantly with the criterion measures. The family management variables, Monitor, Obedi-
ence, and Problem Solve, covary significantly with four different measures of protodelinquent and
delinquent behavior. Stealers and Social Aggressors seem to be characterized by disruptions in differ-
ent aspects of the family management process.

238
 Chapter 11
Similarities and
Differences Between
Social Aggressors and
Stealers

One function of this chapter is to pull together earlier, most of the antisocial children referred for
the diverse themes and scattered findings that have treatment were said by their parents to be disobe-
been presented in preceding sections. The findings dient. These disobedient children are unpleasant to
from both the microsocial and the family manage- be around. They take, they are abrasive, and they
ment variables will be summarized in terms of the give little in return. Secondly, each subset of anti-
amount of variance accounted for in measures of social children is thought to be an example of ar-
antisocial performance. The other goal of the rested socialization. Each child, in his own way,
chapter is to pursue the topic of similarities and maximizes his immediate gains/ pleasures at some-
differences between Stealers and Social Aggres- one else’s expense. The third characteristic shared
sors. As presented thus far, children with these by antisocial children is a reduced responsiveness
problems are said to belong to a larger class, anti- to social stimuli. They are less responsive to ordi-
social; the implication is that they share some nary social reinforcers and to threats and scolding.
characteristics that differentiate them from normal They are characterized by attentional deficits that,
children. The first section of this chapter outlines in turn, may relate to an insensitivity to some as-
the elements that they have in common. When we pects of the behavior of other persons. The fourth
began treating these families, it gradually became characteristic thought to be shared by antisocial
apparent that families of Stealers and Social Ag- children is skill deficits in the crucial areas of
gressors differed in some important respects; the work, peer relations, and academic achievement.
family processes that produced Stealers differed In the following section, each of these shared char-
from those that produced Social Aggressors. Both acteristics will be discussed in detail.
processes seemed to differ from what was ob-
served in normal families. The second and third Noncompliance
sections of this chapter will focus upon family- The one thing shared by all antisocial children
process variables that differentiate among Social referred for treatment is noncompliance. During
Aggressor, Stealer, and Normal samples the intake interview, when asked to describe the
difficulties they have with their child, 89% of the
Characteristics Shared parents say, “This child does not mind.” This, of
by Antisocial Children course, is also noted as a problem even for parents
The most salient element shared by the Stealer of “normal” children. Rutter, Tizard, and Whit-
and Social Aggressor is noncompliance. As noted more (1970) found that 31.5% of the parents of

239
normal, preadolescent children perceived disobe- children. The observed rates per minute for moth-
dience as a problem in their own families. ers of Normals, Stealers, and Social Aggressors
During the preschool years, noncompliance is were .280, .423, and .413 respectively. The F val-
the first indication of a lack of effective parenting ue was 4.95 (p< .01). (See Appendix 6.2 for de-
skills. Teaching compliance becomes an issue gen- tails about the samples.) In addition, parents of
erally at about 2 to 3 years of age. In this training, problem children often accompany commands
the parent can positively reinforce the child when with expressions of irritation and/or threats. This
he complies with a request, and/or the parent can disposition is assessed by the FICS category Com-
firmly punish noncompliance. For some reason, mand Negative; these are commands accompanied
parents of antisocial children are not successful in by sarcasm or threats. For normal mothers, the
teaching compliance. It may be that they forget to mean rate was .002; for mothers of Stealers it was
reinforce compliance, that they neglect to firmly .020; and for mothers of Social Aggressors it was
punish noncompliance, and/or their children are .013. The F value was 6.33 (p<.001).
simply of a more difficult temperament. For what- Parent commands have several functions in rela-
ever reason, the data generally show antisocial tion to child deviant behavior. On one hand, ma-
children to be about half as compliant as are nor- ternal commands may initiate a conflict; i.e., they
mal children of the same age. may serve as crossovers. On the other hand, many
In addition to the inept use of contingencies, it parent stop-commands are probably a reaction to
may be that the mother of the problem child lacks coercive initiations by the child; i.e., they are a
skill in her use of commands and requests. Fore- type of counterattack. Because of this dual role, it
hand, Wells, and Sturgis (1978) observed 18 would be expected that rates of parent commands
mother-child dyads, both in the home and in the would covary with rates of child deviancy, e.g., as
laboratory. The children had been referred for reported in Johnson and Lobitz (1974a) and Lo-
treatment of noncompliance. The likelihood of the bitz and Johnson (1975).
noncompliance observed in the home served as the Noncompliance to parent requests is a kind of
criterion variable in a multiple-regression analysis. rule breaking that occurs in the presence of the
The key variable accounting for most of the vari- parent. It is assumed that the level of noncompli-
ance for the multiple-regression coefficient of .74 ance covaries with more general measures of child
was a measure of the mother’s tendency to give deviancy. In keeping with this hypothesis, Griest,
vague commands. Stephen Johnson and his group Wells, and Forehand (1979) found that the corre-
were among the first to differentiate among the lation between observed rates of noncompliance
various kinds of parent commands (Johnson, and.observed rates of deviant child behavior was
Wahl, Martin & Johansson, 1973). He intro- .50 (p<.05). A study based on OSLC samples
duced two dimensions. He noted first that some produced comparable results. In the comparison
commands were “stop” commands, e.g., “Stop involving 34 Social Aggressors, 36 Normals, and
teasing your sister.” Others were initiating com- 32 Stealers, the mean rates of noncompliance for
mands, e.g., “Please clean up your room.” The the target children were .148, .033, and .084 re-
other dimension had to do with the clarity of the spectively. The F value of 19.11 (p <.001)showed
communications to the child. He noted that moth- that these differences were significant. Note that
ers of distressed families tended to give vague com- while the Stealers were more noncompliant than
mands for which compliance might be difficult Normals, they were less so than Social Aggressors.
even if the child were determined to be obedient. The finding that antisocial children were signifi-
I think that high rates of parent-command and cantly less compliant than Normals has been repli-
accompanying child-noncompliance are the first cated by a number of investigators (Johnson &
step in a progression that produces an antisocial Lobitz, 1974a; Forehand, King, Peed, & Yoder,
child. It can mean two things. First, the parent has 1975).
been unsuccessful in teaching many prosocial, self- The level of compliance varies as a function of
management skills. Given this deficit, the parent age and sex of child and sex of parent. Of course,
substitutes many commands: “Clean up your many parents feel that children mind their fathers
room,” “Hang up your coat,” “Take out the gar- better than their mothers. Data have been re-
bage.” Second, it could mean that the child’s rate ported confirming this impression. Children
of coercion is very high, and the high rate of stop- showed significantly more compliance to fathers’
commands is the parent’s reaction to this situa- (mean .56) than to mothers’ (mean .49) positive
tion. commands (Hetherington, Cox, & Cox, 1976).
As expected, mothers of antisocial children use Similarly, girls were significantly more compliant
more commands than do mothers of nonproblem (mean .57) than boys (mean .47).

240
 For a given age range, the research findings meant by arrested socialization.
show a rather surprising consistency across set- In addition to skill deficits, he may even engage
_tings in the likelihood of compliance. The likeli- in the whining, clinging, dependent behavior of
hood of compliance or p(Child-Comply|Parent- the immature child when interacting with adults.
Command) for normal 3- and 4-year-olds was In a very real sense, the antisocial child may be
.72, as reported by Johnson et al. (1973) for inter- thought of as immature. However, I think the an-
actions in the home, .60 in the Johnson & Lobitz tisocial child has in his repertoire a wider range of
(1974a) laboratory study, .69 for structured labor- pain-control techniques that are applied to a wider
atory interactions in the study by Terdal, Jackson, range of people. The antisocial child coerces sib-
and Garner (1976), and .62 in a free-play situa- lings, peers, teachers, and parents.
tion in the Forehand et al. (1975) study. The latter Infants, Stealers, and Social Aggressors maxi-
found no difference for social class. It seems, mize short-term payoffs. By definition, they do
then, that the normal range of compliance for not delay gratification, and it goes without saying
3-and 4-year-olds is 60% to 70%. This range that these immediate payoffs are usually obtained
holds fairly consistently across settings. at the expense of other persons. This is part of
At a very speculative level, I think even in the normal development; it is what normal infants do.
preschool years that parents of Stealers and Social If they want a toy, they take it. If they do not like
Aggressors may differ in the kind of difficulties the food being offered, they throw it on the floor.
they have with child disobedience. Parents may be If their mother refuses to give them candy at the
highly selective as to what kind of disobedience supermarket, they have a temper tantrum. Infants
they will tolerate. For example, the parent of a and antisocial children are the original “here and
Stealer may tolerate rule violations concerning now” people.
vandalism and stealing, but will not tolerate non- An antisocial child takes what he wants and
compliance to a direct request. In other words, avoids that which is unpleasant or boring. Rather
children who steal are involved in a different type than learn the socially reinforcing skills leading to
of noncompliance. They disobey all rules when cooperation and acceptance, he uses pain control
there is no adult to enforce them, i.e, rules about to produce compliance. Rather than join in, he
other people’s property, rules about when to come disrupts. Rather than learn to anticipate crises, he
home, rules about which friends to play with, and avoids them as long as he can and then explodes.
rules about stealing. These children are perceived The antisocial child can be thought of as someone
as disobedient by their parents. On the other who has never quite grown up. However, unlike
hand, parents of Social Aggressors will tolerate Peter Pan, he is not a child’s delight. There is also
noncompliance to direct commands, but will not a price for continuing to employ these infant-like
tolerate violations of rules about property damage techniques. The metaphor that nicely describes the
or theft. Social Aggressors are also perceived by position in which the antisocial child finds himself
their parents as disobedient. has been drawn by the economist Hardin (1968)
We will return to parent selectivity about devi- and is further elaborated by Platt (1973). The
ancy in a later section. The hypothesis presented Hardin essay, “The Tragedy of the Commons,”
here is that the kind and magnitude of disobedi- points out a basic dilemma posed as the outcome
ence permitted during the preschool years has a for a competitive strategy. In his essay each shep-
great deal to do with the form of antisocial behav- herd increases the size of his flock to maximize his
ior expressed during school-aged years. short-term gains. But as his competitors increase
the size of their herds, the outcome is the inevita-
Arrested Socialization ble destruction of the village commons. The
It was implied in earlier discussions that the an- shared grasslands that had once supported them
tisocial child functions somewhat like an over- all are lost.
grown infant (see Chapter 2). He shares the in- In a similar vein, social interactions have both
fant’s philosophical position, i.e., to maximize short-term and long-term payoffs. The problem
short-term payoffs and ignore the long-term costs. child who utilizes coercive events with skill pro-
Like the infant, he makes frequent use of coercive duces increased likelihoods for certain kinds of
means to get what he wants and to avoid doing payoffs by family members. However, in so doing,
anything displeasing. Like the infant, he has not he fails to note that there is an increase in the like-
yet been taught to comply to the majority of re- lihood of the family members’ punitive reactions
quests made by adults. Like the infant, his proso- and their future surprise attacks upon him. For ex-
cial skills are, at best, rudimentary. It is this con- ample, the correlation between the child’s overall
figuration of characteristics that defines what is coercive performance (TAB scores) and the moth-

241
er’s likelihood of a counterattack in coping with miliar Figures (MFF) Test, and DePree’s Delay of
him was .72. His TAB score correlated .82 with Gratification task were administered to two clini-
the mother’s likelihood of Crossover. In other cal samples and a normal sample. He correlated the
words, the higher the rate with which the child five dependent variables that measure impulsive-
(Normals, Stealers, and Social Aggressors) em- ness for the total sample. The median correlation
ployed pain-control techniques, the more likely, among variables was .53; this provided strong
in the long run, that the mother would react to support for a general trait of impulsiveness. On all
him punitively. She also becomes more likely to of these variables, the antisocial clinical samples
launch unprovoked attacks upon him. This is a showed significant differences in their scores when
nicely coordinated dance in which a price is ex- compared to Normals or problem children who.
acted for maximizing immediate gratification. were “internalizing.” The antisocial children
Other family members respond in a similar fash- showed less of a tendency to delay gratification
ion to the child. The comparable correlations for and a greater likelihood for impulsive behavior.
fathers between child TAB and paternal punitive-
ness and Crossover were .68 and .79 respectively.
For siblings the correlations were .44 and .58. Responsiveness to Social Stimuli
In effect, the problem child acts impulsively, of- I think there are disruptions in the development
ten producing immediate compliance or attention of several interrelated mechanisms that further ac-
but, in so doing, increases the likelihood of future celerate the drift toward both increasing deviancy
unpleasant and unpredictable experiences. This, and greater deficits in prosocial skills. The first is
in turn, contributes to his sense of outrage; it also concerned with learning to respond to social stim-
provides a justification (if, indeed, he requires uli. It is during the preschool years that much of
any) for initiating future attacks of his own. It the training for responsiveness to social stimuli
seems as though these children have unwittingly takes place. Just as the antisocial child is signifi-
invented a perpetual motion machine. Rather than cantly retarded in his learning of obedience, he is
levers, belts, and cogs, they substitute rage, imme- also retarded in his development of learned re-
diate gratification, and a lack of awareness of sponsiveness to social stimuli. This reduced re-
long-range consequences. In retrospect, it is easy sponsiveness applies to social reinforcers and to
to see why various treatments for aggression have threats and scolding. As a result, the antisocial
focused on draining off rage or teaching self-con- child may fail to make subtle discriminations re-
trol and delay of gratification. These characteris- garding the behavior of the people with whom he is
tics of the aggressive child are obvious to even the interacting. It is important to emphasize that the
most casual spectator. The question is, however, issue is relative nonresponsiveness. Being some-
are these causes or concomitants for the processes what less attentive will, I think, lead to the slower
producing antisocial behavior? development of some key social skills. This, in
The antisocial child is a person caught up in the turn, has a profound impact upon the eventual ad-
immediate moment. In samples of younger out-of- justment of the antisocial child. For example, giv-
control children, it has been deemed of sufficient en reduced responsiveness, it might take him long-
importance for a number of experienced research- er to learn how to form close relationships. Given
ers to focus on teaching children techniques de- that he is also coercive, then he may never learn.
signed to give greater internal control over impulses A child is involved in literally thousands of in-
(Meichenbaum, 1979; Spivack & Shure, 1974). terchanges each day. Parents, siblings, peers, and
This rush for immediate gratification is also char- teachers present him with cues to which he reacts.
acteristic of older delinquents, as noted in the liter- In these interchanges he must learn to make subtle
ature reviewed by Quay (1965). There is also an discriminations between this or that social cue,
empirical base for this hypothesis. For example, and then react in an appropriate fashion. What if
Riddle and Roberts (1977) reviewed the literature the subject finds the reactions of other persons to
on scores from the Porteus Maze Test. The find- be relatively neutral in regard to their reinforcing
ings provided consistent support for the concept value? Worse yet, as in the case of Gottman’s
that delinquents are significantly more impulsive, (1979) distressed couples, what if the other per-
have less foresight, and are less able to delay grati- son’s positive reactions were experienced as neu-
fication in solving mazes. The careful study by tral or as punishing? It is likely that these fum-
Weintraub (1973) not only provided additional bling, inept approaches would be experienced by
support for these inferences, but compared several others as aversive and lead them to label the child
direct measures of delay of gratification as well. as socially handicapped. Being perceived by peers
The Porteus Maze Test, The Kagan Matching Fa- as “different” or “deviant” could, in turn, initiate a

242
 new process having unfortunate implications for the four experimental conditions. The finding
the child. most relevant to the present discussion was that
_ What is the evidence for antisocial children the Externalizers and the children of schizophrenic
making faulty discriminations in situations involv- mothers were less effective than Normals under
ing social reinforcers? In the last decade, there the cognitive-facilitation conditions. However,
have been a series of well-executed laboratory when the antisocial sample was provided an exter-
studies relating to this issue. Prominent among nal reinforcer, they were no different from Nor-
these is the series of doctoral dissertations at the mals! The author concluded that for the antisocial
University of Minnesota directed by N. Garmezy. group the problem of attention was instead a
Phipps-Yonas (1979) compared antisocial delin- problem in motivation. A similar effect was ob-
quent children, normal, and hyperactive children tained in the instrumental conditioning studies by
to children of normal, schizophrenic, and (non- Johnston (1976) and Levin and Simmons (1962a,
psychotic) psychiatrically disordered mothers. b). They found that antisocial children learned as
The procedures were designed to test for selectivi- well as Normals when nonsocial reinforcers were
ty in attention and for the ability to shift focus; re- introduced.
action time served as the dependent variable. The These consistent findings raise some interesting
Antisocials differed from all the other groups in questions. Does a monetary reinforcer enhance the
that their reaction times were significantly slower saliency of the cues, produce a higher state of
and more variable than the other groups. The next arousal, or actually strengthen A;—» Rj connec-
study in the series was by Driscoll (1979), who tions more effectively than social reinforcers?
worked with the same four samples on a set of However one may conceptualize the difference be-
procedures measuring the amount of incidental tween monetary and social reinforcers, the as-
learning occurring during two different discrimi- sumption is that both Stealers and Social Aggres-
nation tasks. The antisocial samples were shown sors would display comparable deficits on almost
to have learned significantly less than the other any set of learning tasks involving social reinforcers
groups on both the required and the incidental or social punishment as consequences. When ap-
learning tasks. The results from these first two plied to the learning of subtle discriminations in-
studies are supported by findings from other inves- herent in social interaction, then even a slight re-
tigators. For example, the Camp (1977) and duction in responsiveness could have a profound
Camp, Zimet, Doomineck, and Dahlm (1977) ser- effect. The question is, just how solid is the evi-
ies of studies compared younger aggressive and dence for such an assumption? While there is over
normal boys on a variety of laboratory and aca- a decade’s work relating to this issue from labora-
demic tasks. The findings showed a general im- tory and field studies, it must be said that the qual-
pairment for antisocial boys on those tasks, re- ity of the studies leaves much to be desired. For
flecting an impulsive style of responding. this reason, I entertain the hyporesponsiveness
For me, the key study was the first of the Min- hypothesis with great interest but with an equal
nesota group’s investigations of attentional defi- measure of skepticism.
cits. Marcus (1972) tested 240 children in a simple In the early 1960’s, during the first blush of en-
reaction-time procedure, using an auditory signal thusiasm for verbal conditioning studies, Johns
involving both regular and irregular preparatory and Quay (1962) demonstrated that antisocial
intervals (1, 2, 4, 7, and 15 seconds). Two condi- adults did not condition as well as neurotic adults.
tions were used. In the cognitive-facilitation pro- The general effect was replicated by Quay and
cedure, the children were informed about the Hunt (1965). Ruenster (1970), cited in Sallows
length of preparatory intervals to be used in the (1972), used a simple discrimination task with re-
next trial. In the motivational-facilitation condi- action time as the dependent variable and ob-
tion, the children were told that a short reaction tained a reliable difference between delinquents
time could win a desired toy. Each of the four clin- and normal children. Zylstra (1966) also obtained
ical groups was matched with a group of normal a significant effect for institutionalized delinquents
children. Based on the work of Achenbach (1966), compared to Normals; he employed the Taffel
referred children were classified as Internalizers or conditioning procedure. Johnston (1976) demon-
Externalizers. Children labeled as Externalizers strated diminished responsiveness for delinquents
were said by their parents to lie, steal, fight, and on both a Taffel and an autokinetic procedure.
so on. The two other clinical groups were children For both procedures, the institutionalized delin-
of schizophrenic mothers and children of de- quents were less responsive than Normals to both
pressed mothers. The Internalizer sample was not peer and adult reinforcing agents. They were also
significantly different from Normals under any of less responsive to disapproval. What was of par-

243
ticular interest was the fact that when money was and to include nonsocial back-up reinforcers such
used as a reinforcer, the delinquents were slightly as money, treats, and special privileges (Wahler,
more responsive than were Normals. A similar 1968). Each of us found it clinically expedient to
finding was demonstrated for the marble-box ap- introduce these back-up reinforcers and punish-
paratus by Levin and Simmons (1962a, b). Fifteen ment as well. If we did not use them, we produced
hyperactive-aggressive boys showed ~minimal little change in behavior.
change when reinforced by an adult for shifting The studies reviewed in Chapter 6 suggested
their preferences on a two-hole apparatus. How- that antisocial children may be less responsive to
ever, in a subsequent study, they showed signifi- control by parental nattering or disapproval. The
cant changes in preferences when food was served studies reviewed here suggest that these children
as a reinforcer. There have also been a number of may also be less responsive to social reinforcers.
studies employing similar procedures providing in- There is much yet to be done before either set of
conclusive results. Sarbin, Allen, and Rutherford evidence is entirely convincing. For the present,
(1965), Hedlund (1971), Orzech (1962), Bryan the main support for the hyporesponsiveness hy-
and Kapche (1967), and Bernard and Eisenman pothesis is from clinical experience rather than
(1967) all failed to find differences between anti- from definitive experimental studies. Given these
social and normal children. qualifiers, it would be useful to speculate briefly
The difficulty with this series of studies lies in about the possible determinants for this condition.
the nature of the procedures. This entire genre of Two general possibilities come to mind: tempera-
studies was flawed by their failure to deal ade- ment and the ratio of contingent to noncontingent
quately with the problem of subject awareness. As reinforcement.” Each will be briefly considered be-
noted in Chapter 5, the reviews by Parton and low.
Ross (1965), Dulany (1962), Spielberger (1962), It is conceivable that some children, early on,
and Patterson and Hinsey (1964) concurred in are characterized by a hyporesponsiveness to posi-
identifying “methods” as making greater contribu- tive and aversive social stimuli. For example,
tions to scores measuring reinforcement effects Mednick and Christiansen (1977) and others hy-
than did reinforcement contingencies. However, pothesize a genetic predisposition for hyporespon-
given these possible confounds, the studies did siveness to aversive stimuli (see Chapter 6). The
serve the function of forcibly introducing the re- loosely organized study of temperament by Thom-
sponsiveness hypothesis. They also introduced the as, Chess, Birch, and Hertzog (1960) is also in
idea of selective responsiveness. keeping with such speculations. The difficult chil-
A decade later, the results from behavior modi- dren*they studied could have achieved that status
fication studies carried out in the home and the because they were hyporesponsive to social stimu-
classroom extended these earlier findings. A large li. For the moment, the Thomas et al. (1960) ex-
number of studies showed that, for extreme cases, planation is not appealing because of the difficul-
extinction procedures were not effective for man- ties in measuring what is meant by temperament.
aging oppositional children’s coercive behavior For example, does the fact that the 2-year-old
(Corte, Wolf, & Lecke, 1971; Laivgueur, Peter- child is difficult to manage reflect a constitutional
son, Sheese, & Peterson, 1973; Herbert, et al., difference and/or differences in the reaction of the
1973; Budd, Green, & Baer, 1976; Walker & parent to the child?
Buckley, 1973). Of even greater relevance for the The contingency hypothesis has been presented
present discussion was the finding that for opposi- by Cairns (Paris & Cairns, 1972; Warren &
tional children, even in well supervised situations, Cairns, 1972) and by J.S. Watson (1979). They
adult attention and approval were not effective in have discussed it particularly as it relates to the
increasing such prosocial behavior in the home problem of hyporesponsiveness to positive rein-
(Herbert et al., 1973) or in the classroom (Hops, forcement. Simply stated, the idea is as follows: to
Walker, & Greenwood, 1977). In fact, in both the extent that social reinforcers are initially sup-
studies the contingencies produced a reduction in plied in a noncontingent fashion then they will be
task-oriented behavior.! This corresponded with less effective in shaping behavior when used in lat-
our clinical experience in treating these families. er trials. Two functional relations define what is
We began by training parents to use only social re- meant by this statement. The first includes the
inforcers, such as approval or attention, to shape likelihood of a positive consequence given a target
competing prosocial behaviors (Patterson & Gul- behavior (p[C*|Rj]). The second is the likelihood
lion, 1968). However, within a few months, those of the same consequence given all nontarget be-
of us working with extreme cases found ourselves havior (p[C * |zonR;]. Given that parental approv-
teaching parents to use time out as a punishment al or attention is forthcoming for deviant behavior

244
 as well as for task-oriented behavior, then at- Skill Deficits
tempts to use these same events in a contingent
During preschool years, normal children come
. fashion will have little impact. For example, if a
under the control of social reinforcement. They
great deal of positive regard is provided by the
well-intentioned nursery school teacher (no matter also learn to delay gratification and to acquire the
rudiments of basic social survival skills. Even at
what the child does), then when she attempts to
use these reinforcers contingently, they will pro- the preschool level, the normal child has acquired
a complex set of skills relating to self-help, work,
vide little control. I think this process may also de-
and to initiating and maintaining social interac-
scribe mothers of antisocial children, where posi-
tion. These skills serve as substitutes for the child’s
tive reactions to coercive child behavior and posi-
earlier coerciveness. The basic training for these
tive reactions for prosocial behavior have similar
skills takes place in the home prior to the child’s
probabilities of occurrence.
entrance into public school. It is here he learns to
Cairns and his colleagues have carried out pro- sit still and listen, comply with reasonable re-
grammatic studies demonstrating such an effect. quests, and carry out simple assignments on a reg-
In the first of a series of studies, they demonstrated ular schedule. It is also here that he learns to be at-
in six classrooms of educationally retarded chil- tentive to other people’s behavior, know when he
dren that an average of 38% of the teachers’ posi- has hurt their feelings, when he is boring them, or
tive evaluations occurred independently of any when he has pleased them.
discrete response made by a child. In fact, only As noted in Chapter 2, when predicting adult
28% of such teacher behaviors were clearly con- adjustment, the level of the child’s social skills is
tingent (Paris & Cairns, 1972). Then, on a dis- thought to function as a moderating variable.
crimination task, children from similar classrooms Young children who are socially aggressive but so-
were shown to be relatively nonresponsive to cially skilled have a better prognosis for adjust-
praise. They were, however, responsive to a stim- ment than young problem children lacking in so-
ulus signifying “correct.” In a third study, 56 chil- cial skills. It is assumed that the antisocial child
dren from educationally-retarded classrooms par- who is also unskilled is most at risk for chronic de-
ticipated in a discrimination learning task. One linquency.
group received praise when correct; others were In middle-class families, training the child to
told when they were wrong (i.e., they were pun- work begins in the preschool years. The child is
ished). The punished group showed a significant given a series of tasks analogous to adult work sit-
change in their behavior, while the rewarded uations, e.g., cleaning up his room, simple chores
group showed only a weak effect. The fourth that are carried out while an adult is present, then
study (Warren & Cairns, 1972) established the re- chores that are not continuously supervised, and
lation between prior exposure to noncontingent finally part-time jobs for neighbors. In these situa-
social reinforcers and reduced responsiveness on a tions, the child is told clearly what is expected and
subsequent task. The experiment provided the what the rules are. He is given feedback if the task
necessary controls by comparing the effects of pri- is not properly carried out (i.e., he is monitored)
or contingent social reinforcement and prior non- He is also expected to carry them out on a consis-
contingent social reinforcement. The experiment tent basis. Later on, he may receive his allowance
also provided for various levels of exposure to contingent upon successful completion of his
these prior conditions. Those children exposed to chores. Presumably, these work skills will also
high levels of noncontingent reinforcement generalize to other settings such as school.
showed less behavior change on the subsequent The studies reviewed in Chapter 2 suggested
task. Those with prior exposure to high levels of that antisocial children were less likely to be ex-
contingent social reinforcement showed a facilita- pected to do chores at home or to hold weekend or
tion effect; i.e., they were more responsive. summer jobs. There are some fascinating observa-
Antisocial children receive significantly fewer tion data available from the Whiting and Whiting
positive social reinforcers, such as Approval (Pat- (1975) study of six cultures, suggesting that the
terson, 1979b). Their assumed hyporesponsive- more simple the culture, the more likely its mem-
ness to positive reinforcers, such as Talk or At- bers are to assign chores to young children. The
tend, may relate to the fact that the likelihoods for mean number of chores varied from 2.4 per day
these consequences are similar for both deviant for a more simple culture to a low of 1.0 for the
and prosocial behavior. Presumably, in normal most complex culture (a small community in New
families the likelihood of a positive reaction is England). In the simpler agrarian cultures, the
higher for prosocial than for deviant behaviors. mothers assigned more tasks and were also more

245
likelyto emphasizeobedience.? ther effective models nor skilled at tracking proso-
The studies reviewed in Chapter 2 also showed cial skills, result in the child’s skill deficits becom-
that peers consistently rejected antisocial children. ing increasingly obvious. If one assumes that anti-
Rutter et al. (1976) underscored the magnitude of social children have attentional deficits, then the
the problem with data that showed fully half of the lack of social skills becomes more understandable.
antisocial boys were not liked by peers! The recent Maintaining an extended interchange with a peer,
investigations of the determinants for peer rejec- particularly an older one, requires good timing
tion agree in identifying coerciveness as the major and careful attention to subtle variances. If, for
variable (Gottman, 1977; Hartup, 1977). For example, the child is talking in a monologue, he
samples of younger children, rejection covaried may ignore the shift in gaze signaling that the peer
with child aversiveness. On the other hand, the is ready to take his turn. By plunging in and mo-
use of positive reinforcement covaried with accep- nopolizing the peer’s attending behavior, he may
tance by peers. then miss the extended, averted gaze signaling
My clinical experience with these children leads boredom. On the occasion of his next meeting he
me to believe that the antisocial child also lacks an may encounter a slightly cool reception; repeated
awareness and appreciation for how others feel. trials may produce more obvious distancing. An
The work by Feshbach and Feshbach (1969) and effective interaction must provide support 'to both
Radke-Yarrow and Waxler (1976, 1979) suggests members. This means taking turns; one person
that during the preschool years the normal middle- talks, the other listens. At intervals the roles are
class child learns to empathize with other persons’ reversed. These shifts require the ability to make
feelings. Antisocial children behave as if the pain very precise discriminations. These discrimina-
they inflict on others is of little consequence. They tions are the basic processes underlying social in-
seem to portray the general attitude outlined by teraction. The writer believes that it is in this sense
Ryan’s Blaming the Victim (1976). He examined that the antisocial child is flawed.
aggression in different cultures and found that, What is the long-term outcome of a child’s fail-
generally, the victim was denigrated and placed in ure to notice that he just hurt his best friend’s feel-
the position of readily deserving that which he re- ings? If he missed classifying the hurt expression
ceived. My clinical experience with antisocial chil- even one-third of the time, what would that mean?
dren suggests that they, too, perceive their victims How will peers label such a child, especially if he
with a kind of self-righteous contempt. The chil- also has a tendency to fly off the handle easily?
dren they attack deserved the assault because: What is the likelihood that such a child will be a
“They are always after me; I just get ’em first. Ev- skillful competitor in sports? At a speculative lev-
ery day they say this stuff to me, so I just slug el, it is, I think, the antisocial child’s lack of basic
em.” Ryan (1976) also notes that the penchant for social survival skills that contributes as much to
blaming the victim characterizes society in gener- his marginal adjustment as an adult as does the an-
al. tisocial behavior itself. These are very unskilled
Only a few studies compare normal and antiso- people.
cial samples for social skills. Achenbach and Edel-
brock (1978) found that children categorized as The Differences Between
delinquent scored lower on a measure of social
skills than did Normals. Children classed as Social Social Aggressors and Stealers
Aggressors scored lower on social skills than did While antisocial children may share some traits
those categorized as delinquent. Cox, Gunn, and in common, it is hypothesized that Stealers and
Cox (1976) also found that behavior-problem Social Aggressors differ from each other in two
children were significantly less skilled than were important respects. First, they are thought to dif-
Normals. fer in the kinds of parenting practices associated
In light of the earlier discussion, antisocial chil- with them. A general formulation will be pre-
dren may not find ordinary social behaviors to be sented that delineates these different processes. A
very reinforcing or very punishing. This could later section presents the empirical findings that
mean that the child is less motivated to learn the relate to these assumptions. By definition, Stealers
subtle discriminations required of the adolescent and Social Aggressors differ in the kind of prob-
and adult. Given that this process continues, then lems or referral symptoms described by parents.
these children may function like social sleepwalk- However, it is assumed that these referral symp-
ers. What they do in social interaction is often toms may also reflect different underlying progres-
“not quite right.” This reduced motivation, ac- sions. In these progressions, the performance of
companied by the fact that their parents are nei- one problem behavior is related to the performance

246
 Figure 11.1
Parental Processes Relating to Three Kinds of Antisocial Behavior

Parent
Under inclusive Failure
DISTANT Pure
categories for to punish
and property damage STEALER
stealing
unattached

MIXED
Stealer and Social Aggressor

Parent
Over inclusive Failure Pure
IRRITABLE
categories for to punish SOCIAL
and f
coercive acts coercion AGGRESSOR
enmeshed

of yet more “extreme” but lower rate problem be- ther the target child or from his siblings. I suspect
haviors. The evidence for such progressions is that they are motivated only enough to teach the
briefly reviewed in this section. child a bare minimum of survival skills. For exam-
ple, they are probably not motivated enough to
Parenting Practices monitor his school work and support high-level
Two aspects of the parents’ behavior comprise achievement.
the key mechanisms thought to differentiate Steal- The second outstanding feature of these parents
ers and Social Aggressors from each other and lies in their lack of concern for property viola-
from Normals. It is assumed that parents differ in tions. They tend to ignore minor violations that
what it is that they classify as deviant (e.g., some occur in the home; e.g., taking money from Moth-
parents believe temper tantrums are normal but er’s purse is not categorized as stealing, nor is tak-
that stealing is deviant). The deviancy-categoriz- ing an item that belongs to a sibling. Given that
ing process, in turn, determines which child be- the child can explain why he took it, then no pun-
haviors will be monitored and which will be pun- ishment is forthcoming for these protodelinquent
ished. Parental monitoring and punishment prac- behaviors. A similar pattern characterizes what
tices directly determine the rate at which antisocial happens when neighbors, storekeepers, and police
behavior will be performed. officers accuse him of theft. He “explains it,” and
As shown in Figure 11.1, parents of Stealers are consequently the event is not categorized as a devi-
thought to be relatively uninvolved in the role of ant act, and, of course, no punishment occurs.
caretaker. They are not motivated to extend them- It is hypothesized that parents punish what they
selves in the service of the child. They are suffi- believe to be “deviant.” The parents of “pure”
ciently skilled in child management so that they do Stealers tend to be overly exclusive in their classifi-
niot permit face-to-face coercive behavior from ei- cation of property violations that are thought to

247
 _ Figure 11.2
Applications of Signal Detection Model
to Parent Classification of Deviant Child Behavior

Normal Overlyexclusive
Parents of Stealers threshold criterion threshold

Normal
Child
Behavior

Child Behavior

a :% oe
é 4 o% x% Ke
% 4% g%
iS Ss i) fe ain
4 % ie %* a oA ° 4 ra o?¢ ”
oe . % % 4a % % 4% %,
©
ae e % 2 2 ~% > e
LD () % % %
LS 4 © )%

; Overlyinclusive
Parents of Social Aggressors critctioneureckold

Normal
threshold

Normal
Child
Behavior

Deviant Child Behavior

4 S: Ss.
% al al
ae) 4.
Q
e if
* 4
% &

be deviant. As shown in Figure 11.2, they classify detection model to this problem was suggested by
as deviant only the most extreme acts directed D. Littman and P. Holleran. Both have completed
against property. These same parents may classify laboratory studies to test the possibility of apply-
encopresis, hitting, or wetting the bed as deviant. ing it to parent deviancy classification. As yet,
Almost all families train their children not to light however, there are no studies that test the hypoth-
fires indiscriminantly. The idea of applying a signal esis that parents of Stealers are overly exclusive;

248
i.e., compared to parents of Normals, they classify acts directed against persons. As illustrated in Fig-
many acts against property as normal. For the ure 11.2, they would tend to be very reluctant to
present, the hypothesis is based upon clinical ex- classify most acts of stealing or vandalism as being
perience. deviant. However, if the child’s behavior is un-
Given that the hypothesis is true, it is not at all pleasant for the parent, then the reaction is likely
clear how such misclassification might come to be irritable. These parents are generally un-
about. Do these expectations about what is nor- skilled in their use of punishment. They are com-
mal and what is deviant precede the problem pletely committed to here and now. If the child
child, or do they arise as an outcome of the par- steals from a neighbor, let the neighbor take care
ents’ interactions with him? These questions are of it himself, but do not bother the parent. If the
the crucial ones for the next phase in the develop- child is making noise, yell at him until he stops.
ment of a coercion theory. I suspect that in the These children steal, and they are socially ag-
case of parents of Stealers, their expectations pre- gressive; they are also at very high risk for being
cede the problem child. These expectations may physically abused. In a retrospective study, Reid et
relate to prior criminal and quasi delinquent his- al. (1981) showed that OSLC cases who were
tories of the parent. However, for the moment this physically abused had very high TAB scores. This
is sheer speculation. The data reviewed in a later would classify them as Social Aggressors. In a
section of this chapter will address the issue of the more recent analysis, Reid found that almost all of
breakdown in parental monitoring and punish- these abused children were also Stealers! This sug-
ment as it relates to stealing and delinquent behav- gests that a mixed sample that both steals and is
ior. socially aggressive must have very unusual par-
As shown in Figure 11.1, the parent characteris- ents. I suspect that they are not only at grave risk
tics that relate to producing Social Aggressor for abuse but for a variety of adult adjustment dif-
symptoms are thought to be quite different from ficulties as well. I also suspect that the mixed cases
those that produce Stealer problems. Rather than are most at risk for becoming delinquent adoles-
being distant and unattached, the parent of the So- cents. If this is true, then the child known to the
cial Aggressor tends to be overly enmeshed and juvenile court as delinquent will be inadequately
very irritable. It is hypothesized that when differ- monitored and punished for rule breaking by his
entiating between coercive and normal acts, these parents. He will also reside in a household com-
parents will tend to be overly inclusive; i.e., they prised of distant and irritable people.
tend to classify many behaviors as deviant that
parents of Normals would classify as borderline Symptom Progressions
acceptable. Normal parents let a good deal go by; There are two assumptions about the problems
they ignore borderline events. The mother of the for which antisocial children are referred. First, it
Social Aggressor becomes an impassioned partici- is assumed that Stealers and Social Aggressors dif-
pant, scolding and threatening at the slightest fer in terms of the kinds of symptoms that are of
provocation. Even though she becomes readily in- concern to the parent. Second, it is assumed that
volved, she fails to confront the child and punish the symptom patterns describing these two sam-
effectively. As noted earlier, the irritable reactions ples also define some underlying progressions. In
make the situation worse. these progressions, there is an orderly sequence
Neither type of parent directly confronts the from higher rate to lower rate symptoms. In the
child. Parents of Stealers and Social Aggressors are present context, it was thought that the progres-
similar in their failure to punish the relevant anti- sions would also move from less extreme to more
social behavior. One parent misclassifies stealing extreme forms of deviancy. As used here, the term
and avoids the necessity of punishing it. The other progression implies transitivity; children who ex-
parent substitutes nattering for punishment. hibit the more extreme low base-rate symptoms
Our analysis of referral problems suggests that a will also have performed all of the preceding prob-
substantial number of children are socially aggres- lem behaviors in the sequence.
sive, and they steal. In Figure 11.1, these cases are To examine these questions, we must obviously
labeled “Mixed.” They are thought to be produced begin by asking the adults who interact with the
in homes that are characterized by the worst fea- child to describe the kinds of problem behaviors
tures of both processes. These parents are poorly that he presents. The literature reviewed in Chap-
motivated to do anything for the child, and they ter 3 showed that parent reports about child devi-
are irritable in their face-to-face contacts with ant behavior should be accepted with a good deal
him. These parents have a faulty mode for cate- of caution. Two studies showed agreement of only
gorizing both acts directed against property and 36% between independent reports of mothers and

249
 _ Table 11.1
A Symptom Structure for Stealers

Parent-Reported Symptoms:
Disobedient Lie Steal Firesetting
Variables (.89) (.68) (.56) (23)
Likelihood of p; progression p(Lie|Disobey) p(Steal|Lie) p(Fireset|Steal)
TP? .70* 32m
Likelihood of pz alternate path p(path other than p(pathother than p(pathother than
Disobey|Lie) Lie|Steal) Steal|Fireset)
.06 «4 we

The figures in parentheses refer to base-rate values.

*Significantly greater than its base-rate value.

fathers as to the nature of the child’s symptoms impact on low base-rate events found later in the
(Dreger, et al., 1964; Lapouse & Monk, 1958). progression. Such progressions would also be of
The latter study also showed poor agreement be- interest in that they might identify the sequence
tween parent and child) reports about deviant be- children follow in performing deviant behaviors.
haviors. Reid and Patterson (1976) studied a sam- Given such a sequence, then a longitudinal study
ple of children referred at the request of communi- might show that, first, children become high-rate
ty agencies because the boys were known for some liars, then some of these will begin to steal, and
time to have been high-rate Stealers. A 30- to 60- then a certain proportion of them may begin to set
day baseline study showed that each child actually fires.
stole at least twice a month. However, during the I began by following the procedures described
intake interview, 21% of these parents failed to by Guttman (1944) and identified scalogram pro-
even mention stealing as a problem. The sample of gressions for both Social Aggressor and Stealer
38 Stealers described in Appendix 6.2 showed a symptoms. For both progressions it was possible
similar pattern. Each child in this carefully se- to obtain reproducibility coefficients around .90
lected sample was known by the court and/or and to replicate the effect. However, a conversa-
family to be a high-rate thief. At the intake inter- tion with John Robinson and a careful reading of
view, 15% of the parents failed to mention steal- his critique of scalograms (Robinson, 1973) led to
ing as a problem when going over the symptom a decision to reconsider this means of analyzing
checklist. It seems reasonable to conclude that the the data.4
false-negative error for parent reports of stealing The parent-reported symptoms for the scalo-
at intake is somewhere around 15% to 20%; i.e., gram progression were reanalyzed using a condi-
the child steals often, but the parents do not label tional-probabilities format. For this, there are two
it as a problem. pieces of information needed at each juncture.
Clearly one must entertain many reservations What is the likelihood if the subject performs A (a
about data based upon parent reports of symp- high-rate symptom), that he will also perform B (a
toms. However, as noted in Table 2.9, our own lower rate symptom) or (p[B|A])? Secondly, given
studies of symptom clusters by Chamberlain that he performs Symptom B, what is the likeli-
(1980), and the earlier work by Jenkins and Hew- hood that the structural “path” was through A
itt (1944), and Achenbach (1978) reported in Ta- (p[A|B])? A comparison of each conditional to its
ble 2.8, clearly identified differences in symptoms appropriate base rate (p[B] and p[A] respectively)
for Stealers and Social Aggressors. It occurred to can provide a basis for testing the significance of
me that it might be useful to know if these two each conditional p value. A significant p value
clusters of symptoms were really made up of tran- would mean that the adjacent symptoms are a part
sitive progressions; i.e., all children who set fires of the same structure. A series of such values de-
also steal and lie, but not all liars set fires. If such scribes a “path” that leads to such infrequent
progressions could be shown, then one option for events as firesetting. This is of particular interest
treatment might be to alter high-rate symptoms to the student of low base-rate events.
occurring early in the sequence and thus have an In Table 11.1 there are three sets of information.

250
 First, there is the base rate for each symptom. The Y-statistic also showed that it did not correlate
data were based upon intake information from the well with the other symptoms (Robinson, 1973).
~ 114 cases referred to OSLC. The symptoms are The Guttman coefficient of reproducibility was re-
ordered by base-rate values. As shown there, the calculated for the four symptoms. The CR was a
likelihood of occurrence in that sample for Dis- very acceptable .95.
obey was .89, for Lie it was .68, and so on. In The next task was to search for a similar pro-
passing, it should be noted that the analysis would gression for social aggression. The symptom clus-
be facilitated if these values did not exceed .50 for ters identified in Tables 2.8 and 2.9 were analyzed
any of the symptoms. This could be accomplished in an attempt to form Guttman Scalograms for
by calculating these values based upon a general Social Aggressors. The most promising candi-
population of cases referred to child guidance clin- date was the sequence: Noncomply —>Quarrel
ics rather than just a population of antisocial chil- with Siblings—» Hit—» Tease—» Temper Tan-
dren referred to OSLC. trums. The CR values were a borderline .84 for
The second row of figures contains the condi- one sample and the same for a replication sample.
tional p value for the lower rate symptom given However, the probability analyses showed no sig-
the presence of the higher incidence symptom; nificant linkages at any juncture in the progres-
e.g., given Disobey, what is the likelihood of Lie? sion.
It can be seen that 72% of disobedient children What does it mean when one cluster of symp-
were also said by their parents to lie, and 32% of toms forms a transitive progression and the other
all Stealers were also said to set fires. The appro- does not? My notion is that there zs such a pro-
priate next step was to compare each conditional gression for social aggression that describes a drift
to its appropriate base-rate value. The linkages towards increasingly high-amplitude behaviors.
Steal given Lie and Firesetting given Steal were sig- What is necessary is to change the intake checklist
nificant. The findings imply that a study of the to focus upon symptoms that would tap into such
higher incidence problems, such as Lie and Steal, a progression, e.g., kicks others, hits with object,
may tell us much about the low-rate problem, attacks with sharp object, and threatens to kill
Firesetting. One might, for example, study the others.
stimuli that control these higher rate symptoms as There is a very appealing implication of these
a means for understanding firesetting. By defini- findings that should also be noted. The hypothesis
tion, low base-rate events are almost impossible to is that parent perception of child problem behav-
study. However, the analysis of progressions of ior is not correlated so much with the onset of the
this kind may make the study of low base-rate problem as with the frequency of its occurrence. If
events feasible. One might also study which family the problem has been occurring at a high rate for
management or microsocial variables determine an extended period of time the parent will eventu-
whether the child will move from A to B, or from ally perceive it as a problem. The data for the
B to C. Why do some children lie but not steal? Stealer progression suggest that first the child be-
The third row in Table 11.1 provides a rather gins to lie at noticeably high rates. Then, at a later
different kind of information. Functionally, the p2 point in time, the parent may become aware of
value is independent of either the base-rate value stealing events that occur in the home and/or the
or pi. The metaphor I apply here is that of “alter- neighborhood. At a still later point, the parent
native paths.” Conceivably a child could move to may come to perceive firesetting as a problem.
Firesetting by a number of alternative paths, only
one of which is Disobedient -» Lie —»Steal. In Interactional Variables
keeping with the metaphor, p> describes the likeli- It is hypothesized that when using social interac-
hood of an alternative path. Only 14% of children tion variables, Social Aggressors will be differenti-
accused of stealing were not accused of lying. ated from Normals and from Stealers by the high-
Only 22% of Firesetters arrived at this juncture by er levels of coerciveness for all family members. It
a path other than Steal. These p values describing is also expected that the family interaction pat-
the likelihood of alternative paths suggest that if terns for children who steal will not be significant-
we understood the determinants for Lie and/or ly more irritable than those for families of normal
Steal, we would know much of what we need to children.* However, the pilot study by Reid and
know about Firesetting.° Hendricks (1973) suggests that parents of Stealers
The Guttman Scalogram had identified the may be differentiated from parents of Normals by
symptom, Wander, as part of this progression. variables measuring distance and involvement in
However, the probability analyses showed both social interaction. Data will be presented that ad-
the p; and pz values were nonsignificant. Yule’s dress this issue.

251
 The general formulation presented earlier’ sug- showed that Stealers were significantly more coer-
gested that the key differentiating variables lie in cive than Normals but significantly less coercive
the parents’ tracking and punishment of antisocial than Social Aggressors. This was almost an exact
acts that occur within the family and outside the replication of the Reid and Hendricks (1973) pilot
home. It is expected that measures of parent moni- study.
toring and child-obedience will covary significant- The next question is, how well can the three
ly with measures of protodelinquent behavior samples be differentiated by the observed behavior
(stealing, firesetting) and, for older children, with of the target child? The samples were first com-
measures of delinquency. On the other hand, fam- pared for each of the 29 FICS categories. The re-
ily management variables most characteristic of sults of the ANOVA summarized in Appendix
social aggression are thought to be measured by 11.1 showed that 15 of the comparisons were sig-
house rules, discipline, and problem-solving skills. nificant. These variables were then used in a dis-
The relevant correlational findings will be pre- criminant-function analysis that generated two
sented in later sections of this chapter. However, a functions. The canonical coefficient for the first
related set of hypotheses concerns the effect of function was .608. It was defined by a large coeffi-
these disruptions in management upon social in- cient (.88) for Noncomply and .73 for Disapprov-
teraction patterns. It is assumed that irritable, hos- al. The canonical correlation for the second func-
tile interactions are correlated with disruptions in tion was only .286, defined primarily by a nega-
family management. The distant, uninvolved in- tive coefficient for Command Negative. Taken to-
teractions supposedly found in families of Stealers gether, the two functions accurately classified
are assumed to correlate with inept parental moni- 61.5% of the cases from the three samples. The F
toring and punishment. The latter is measured by value for the Wilks’-Lambda was 10.95
the family management variable, Obedience. (p< .001).
The first step was to determine whether families The two clinical samples were nicely separated
of Social Aggressors were significantly different in from the Normals by the first function, which was
interaction patterns from families of Normals and labeled “Negativistic.” The F values were signifi-
families of Stealers. The analyses proceeded in a cant in showing the clinical cases to be more nega-
sequence of two steps. First, three samples were tivistic. The second function, labeled “Threat” or
selected from OSLC files (see Appendix 6.2 for de- “Put-down,” differentiated between the Stealers
tails). Each case met several criteria in order to be and Social Aggressors, as shown by an F value of
selected. For the normal sample, the parents 7.87 (df=3:104, p<.001). The Stealers were
claimed that the child was normal, the juvenile more likely to enploy Command Negative than
court records showed no police offenses, and the were the Social Aggressors. Together, the two
TAB score was in the normal range. Stealers were functions accurately classified 94.6% of the Nor-
identified by parent complaint and/or court rec- mals and 60.7% of the Social Aggressors. Howev-
ords of police offenses. In addition, the PDR rec- er, Stealers were classified only at the chance level;
ords showed three or more stealing events at some 45% of the Stealers were misclassified as normal.
point during the study at OSLC. The Social Ag- It seems, then, that at an interactional level the
gressors were described as antisocial during the in- Stealer child may be indistinguishable from Nor-
take interview, and parent complaints did not in- mals, but Social Aggressors may differ from both.
clude stealing. In addition, their baseline TAB However, a plot of the clinical-case TAB scores
score was equal to, or greater than, .45. employed in the preceding analyses showed a very
The goal was to use “ideal” samples as a basis interesting pattern. The TAB scores for one-half of
for constructing the first phase in testing for ob- the Stealers were an almost exact match for the
served differences in family interactions. The man- distribution of scores for Social Aggressors. Half
ner in which the samples were selected insured sig- of the Stealers were “pure,” and the other half
nificant differences in the TAB scores for normal were mixed cases. This means that for referrals to
and Social Aggressor target children, but the ques- the OSLC clinic (as a rough approximation) 25%
tion was whether the TAB scores for Stealers should be pure Stealers, 33% pure Social Aggres-
would be intermediate between the scores for Nor- sors, and 42% mixed cases. It seemed that there
mals and Social Aggressors, as found in the Reid were not two, but three, subsets for antisocial chil-
and Hendricks (1973) study. In the current study, dren: “pure” Stealers, “pure” Social Aggressors,
the mean TAB score for Normals was .17, for and “mixed” Stealers and Social Aggressors.
Stealers .65, and for Social Aggressors .98. The The proper basis for investigating the problem
differences were highly significant (F=25.12, would be comparisons from samples of pure Steal-
p<.0001). The Duncan Multiple Range Test ers (who are not Social Aggressors), pure Social

252
 Table 11.2
Variables Differentiating Among Samples of Antisocial Children

Mean for Samples

Stealer Social Aggressor Mixed
Code Categories (N =10) (N=17) (N =16) F Values

Child Behavior

Command Negative .00 .01 .04 4.93*

Disapproval .04 16 pi Dee
TAB Score .20 aol ooh 543375

Father Behavior

Command Negative 01 .02 .06 Sty?

No Response oil .08 .16 Seo
TAB Score .28 .62 sit 4.09*

Mother Behavior

Disapproval .18 26 34 2.71

No Response 13 .09 18 SFO
TAB Score 55 78 1.06 Sales

*p<.05
-*p=.01

Aggressors (who are not Stealers), and mixed for the Wilks’-Lambda was 6.91 (p< .01). The ca-
cases of children who are both Stealers and Social nonical correlation for the first function was .76.
Aggressors. Simard (1981) made such a set of A very high TAB Score placed most Social Aggres-
comparisons based upon 18 families of pure Steal- sors at one extreme for this function, while high
ers, 12 families of pure Social Aggressors, and 13 use of threatening commands (Command Nega-
families of boys who represented both sets of tive) placed the mixed cases at the other extreme.
problems. The samples were subsets of the clinical The canonical correlation for the second function
samples described in Appendix 6.2. The ANOVA was .63. The distant father (No Response), who
comparing mothers, then fathers, with target chil- made few threatening commands (Command Neg-
dren on all code categories produced eight signifi- ative), placed the pure Stealers in the upper range
cant variables. The general pattern was for family for this function.
members in the mixed group to be slightly more The large number of variables, the number of
coercive than members of Social Aggressor fami- analyses, and the small samples involved hardly
lies and a great deal more so than members of provide grounds for confidence in these findings.
Stealer families. The family members of mixed However, the data required for replication have
cases also tended to engage in more No Response already been collected as part of the longitudinal
than did families from either of the other samples. planning study. The new analyses have yet to be
In effect, families of Stealers behaved more like done, but it does seem reasonable to summarize
Normals, while families of mixed cases seemed the hypotheses generated thus far. The findings
slightly more distant and more coercive than did from the pilot studies suggest that the parents of
families of Social Aggressors. pure Stealers are more distant and uninvolved in
The eight variables identified by Simard were their interactions with family members than are
then used in a discriminant-function analysis. The parents of normal children. They are also more
means and F values for a slightly different subset distant than are parents of Social Aggressors.
of samples than Simard’s are presented in Table Mothers and siblings of Social Aggressors are sig-
11.2. Two functions were derived that correctly nificantly more coercive in their interactions with
classified 81.4% of the clinical cases. The F value family members than their counterparts in either

253
 Table 11.3
Factor V: Social Aggression
(from Carlson, 1981a, b)

Factor loading Coded behavior Definition

.536 NE NEGATIVISM: A statement in which the verbal message is neutral, but

which is delivered in a tone of voice that conveys an attitude of, “Don’t
bug me; don’t bother me.”
423 NO NORMATIVE: A person is behaving in an appropriate fashion and no
other code is applicable.
— .633 PL PLAY: A person is playing, either alone or with other persons.
459 PN PHYSICAL NEGATIVE: A subject physically attacks or attempts to
attack another person with sufficient intensity to potentially inflict pain.
.487 TE TEASE: Teasing another person in such a way that the other person is
likely to show displeasure and disapproval or when the person being
teased is trying to do some other behavior but is unable to because of the
teasing.

HA SA NOR
.064 .907 —1.142

families of Normals or Stealers, It is also hypothe- of homogeneous classes of responses occurring

sized that families of mixed cases will be more co- during their interactions with family members. If
ercive and the parents more distant than either of such classes were to be found, it would imply a
the other two clinical samples. My hunch is that it common process. The remainder of the section
is this pattern that will be found to characterize summarizes the findings relating to a performance
most of the families of delinquent adolescents. theory for Social Aggression. How much of the
It is also assumed that these microsocial varia- variance can be accounted for by microsocial vari-
bles will covary with disruptions in family man- ables measuring the irritable reactions of family
agement practices. Presumably, some family man- members? How much variance can be accounted
agement practices will be shown to covary with for by measures of family management variables?
distance and others with irritability in family inter-
action. The sections that follow detail the availa- Classes of Observed Responses
ble findings that contribute to our further under- Three different techniques of constructing re-
standing of Social Aggressor and Stealer differ- sponse classes have been followed. Factor analy-
ences. sis, scalogram analysis, and functional analytic
procedures have been used to identify the covaria-
Social Aggressors tions in observed behavior that might characterize
Essentially, the Social Aggressor is a child who Social Aggressors.
directs his attacks against people rather than Carlson (1981a, b) carried out a factor analysis
against property. The symptoms identified by the for a combined sample of normal and clinical
factor-analytic studies that relate to this class are cases observed at OSLC. The analysis for the 29
summarized in Tables 2.8 and 2.9. Some of these code categories produced six factors. The one la-
symptoms were demands attention, screams loud- beled Social Aggression is summarized in Table
ly, fights, argues, disobeys, is destructive, yells, 11.3. As the second step in his analysis, he used
runs around, whines, and is hyperactive. The clin- the factor scores to differentiate a subset of Social
ical labels that Social Aggressors might receive Aggressors from a group diagnosed as Hyperac-
would vary from conduct problems, to immature, tive. The two samples were significantly differenti-
to hyperactive. However, the Social Aggressors of ated from each other (as well as from Normals) by
particular interest in the present context are those their score on this factor.” Children who met the
who fight, hit, and have temper tantrums. The OSLC requirements for the label, Social Aggres-
first question examined concerned the possibility sors, were observed to Hit, Tease, and to be more

254
 Negativistic than were antisocial children who had deviancy level.
previously been diagnosed as also being Hyperac- First, it was of some interest to examine the re-
” tive. lation between the child’s performance level and
The Social Aggression factor identified by Carl- his own irritable reactions. This study could be
son was similar to one obtained by a scalogram thought of as a kind of component analysis of
analysis of code categories. AGuttman Scalogram TAB scores. Each of four irritable reactions
analysis of the 14 coercive code categories (Patter- (Crossover, Counterattack, Punishment Accelera-
son & Dawes, 1975) produced a progression for tion, and Continuance) was correlated with the
six behaviors. Rank ordered by base rates, the pro- child’s own TAB score. The data are from the
gression was: Noncomply, Negativism, Tease, combined sample of Stealers, Social Aggressors,
Physical Negative, Negative Command, and Hu- and Normals described in Appendix 6.2. These
miliate. For the clinical sample, the coefficient for heterogeneous samples could produce a spurious
reproducibility was .92; for a sample of Normals, inflation in magnitude for these correlations, so
it was also .92. Note that both the factor-analytic they should be interpreted with caution. Each of
and scalogram techniques identified Noncomply, the child’s four irritability scores was calculated
Tease, and Hit as members of the same response separately for interactions with his mother, father,
class. and siblings. As shown in Table 11.4, all four
Two of these responses were described in Chap- measures of his interactions with each of the fami-
ter 8 as defining a class of functionally related re- ly members correlated with his TAB score. Gener-
sponses (Patterson & Cobb, 1973; Patterson, ally speaking, his disposition to start up conflicts
1977a). The initiation of Tease and Physical Neg- and counter the attacks by others seemed to be the
ative was produced by a shared network of con- major components.
trolling antecedents. This class was also shown to The first test of the bilateral hypothesis requires
share a common set of accelerating consequences that the child’s irritable reactions to others ac-
(Patterson, 1977a). The latter study also showed count for some of the variance in a measure of his
that the family members most involved in eliciting performance level. In this context, the use of his
Physical Negative and Tease were siblings. Sib- own TAB score as the dependent variable would
lings also provided the consequences most likely to generate a hopelessly confounded state of affairs.
accelerate the responses. However, the use of TIRO (Telephone Interview
It is interesting that children characterized by Report Observation) as a measure of deviancy
such a broad spectrum of antisocial problem be- would not present such a problem. The correla-
haviors demonstrate such consistent interaction tions between the child’s irritable reactions and the
patterns that the same pattern of responses have TIRO scores are also summarized in Table 11.4.
emerged from three different analytic techniques. As shown there, it was the child’s disposition to re-.
At a more speculative level, I think that all of these act irritably to his mother that correlated with par-
problem behaviors share a common coercive func- ent-reported levels of child deviancy. The multiple
tion. The way in which it is manifested varies regression for the measures of child reactions to
somewhat as a function of what the parents will the mother and her TIRO reports produced a cor-
accept. Why is it that if parents permit Tease and relation of .502. The F value was 5.30 (p< .001).
Noncomply, they also permit Physical Negative? The findings offer strong support for the first part
of the bilateral hypothesis.
Microsocial Analyses for a It seems, then, that the problem child’s irritable
Performance Theory exchanges with his mother may function as a
The concept of a bilateral trait implies that the prime determinant for her daily reports of his level
child’s performance of coercive behavior is the of deviancy. However, the bilateral perspective re-
outcome of two dispositions. Each member of the quires that one must also analyze the reactions of
dyad brings his own disposition to react irritably; family members to the child. Measures of the
the performance level for the dyad is a function of mother’s irritable reactions to the child accounted
the product of these two dispositions. It was hy- for over 40% of the variance available in the
pothesized that measures of the irritable reactions child’s TAB score as a measure of his coercive per-
of other family members would account for signif- formance (see Tables 12.3 and 12.5). Similarly,
icant amounts of variance in measures of the the measures of father and sibling irritable reac-
child’s aggressive performance. It was also as- tions to target children also accounted for signifi-
sumed that measures of the child’s irritable reac- cant variance. Combining the key measures from
tions would account for significant amounts of the analyses for mothers with those from fathers
variance in the parent’s daily reports measuring his and siblings provided a multiple regression of

255
 Table 11.4
Covariation of Child’s Performance Level and Child’s Irritability p Values

PPM Correlations of Irritability Variables

Be pues Punishment
When child ss Crossover Counterattack Acceleration Continuance
interacting with: N TAB TIRO TAB TIRO TAB TIRO TAB TIRO
Mother 112 ASS
ree. Ted
Ora ceegeDn RAO
Ee Bae NS ROUa oe
(69)
Father 68 an OD .68*** 04 .46** ails} onlnist ies
(38)
Sibling 104 38 08 Aare Ne.05 iil: 2 CASee oy)
(62) |
The figuresin parenthesesare the N’sfor TIRO data.
*p<.05
**p<.01
py < 001

Table 11.5
Coercion Levels by Family Members

Mean TAB Score by Samples

Family Members SocialAggressor Stealer Normal F Value Duncan

TargetChild 98 65 ely DiS ai SA>ST>N

(34) (37) (36)
Mother .87 .89 45 LOS SA & ST>N
(34) (37) (37)
Father .68 .50 .34 4 Sis SA>N
(14) (25) (26)
Siblings .66 53 Bi 5.00** SA>N
(31) (33)
The figuresin parenthesesare N’s.
pO
p< 001

.581. The F value was 5.90 (p< .001) (Patterson, all) members of his family.
1981b). To study the first question, the TAB scores were
These findings support the idea that the prob- calculated for each family member in the samples
lem child’s performance level is related in some of Normals, Stealers, and Social Aggressors from
fashion to the manner in which other persons react Appendix 6.2. Between-sample comparisons were
to him. It follows, then, that most aggressive chil- made separately for each family agent, including
dren must live in families of aggressive people; problem child, mother, father, and the mean for
i.e., in part, the child’s aggression is a reaction to siblings. As shown in Table 11.5, Social Aggres-
their noxious intrusions. If this is true, it would sors were five times more coercive than were Nor-
follow that if the problem child is more coercive mals, and Stealers were performing at about three
than the normal child, then the mean level of coer- times the level of Normals. The mothers, fathers,
civeness must be higher than normal for some (or and siblings of Social Aggressors were significantly

256
 Table 11.6
Enmeshment with Target-Child Coercive Episodes

Normal Social Aggressor F value by:

Antecedent Younger Older Younger Older Group Age Interaction
Parents
Overall ods) WS) .89 .82 2.00 0.74 0.59
Aversive 66 69 94 85 7.54** 0.13 0.63
Neutral 81 90 .96 .84 0.42 0.05 2.09
Positive .84 58} 78 76 0.05 0.75 0.36
Siblings
Overall 64 .62 oot 58 0.81 0.02 0.06
Aversive 54 .38 45 47 0.00 0.68 1.14
Neutral 90 .89 81 A 176 0.04 0.00
Positive 66 .70 2. 48 5233" 0.00 0.34
*p<.05
a ies10)

more coercive than were those of Normals. Moth- Comparable analyses of problem-child interac-
ers of Stealers were also significantly higher than tions with siblings showed that they were involved
were mothers of Normals. These TAB scores for in roughly 60% of the interchanges with problem
family members summarize each person’s coercive children (Patterson, 1979a). This suggests an
initiations plus coercive reactions to the coercive enormous overlap between siblings and parents.
behaviors of all other family members. However, Both must have been involved in the same inter-
these scores do not tell us which family members changes about 40% of the time! Note, too, the cu-
are actually involved in the interactions with the rious fact that the overlap seemed to be highest for
identified problem child. How often are parents or conflicts that had neutral antecedents. What is
siblings a part of his coercive episodes? there about such an antecedent that defines it as a
The data in Table 11.6 summarize the findings kind of vortex drawing several family members
from two separate analyses of the samples of 30 into a confrontation with the problem child?
Social Aggressors and 20 Normals described in Obviously, much coercion in normal and dis-
Appendix 5.1. In the first analysis, if either or tressed families is a group enterprise. Any given
both the mother and father were involved at any coercive episode is very likely to involve three (or
point in the problem child’s coercive chain, it was more) persons! This bears an interesting relation
tabulated as “involved.” For example, the problem to a finding from studies of assaultive behavior
child and a sibling might have been fighting; if the (Loeber & Janda, in preparation). They found
mother intervened, it would be counted as “in- that in the normal sample, hitting tended to occur
volved.” It can be seen that for normal and clinical during altercations between siblings. However, for
families alike, roughly 80% of the time parents the families in the clinical sample, hitting was
were involved in target-child episodes. There was more likely to occur when the mother became in-
a nonsignificant trend for parents in distressed volved in conflicts among children.
families to be generally more involved than par- The findings support the hypothesis that Social
ents of Normals. Given that the problem child was Aggressors live in aggressive families. As a general
counterattacking (aversive antecedent), there was case, many of the conflicts seem to involve more
a .90 likelihood that a parent was involved at than two persons. However, it is entirely possible
some point! For normal families, this likelihood that the parents and siblings in these families are
was significantly less. Clearly, parents are in- coercive only when they interact with the identi-
volved in a surprising proportion of all coercive fied problem child. A study by Patterson (1979a)
episodes involving the problem child. To change compared parents and siblings from normal and
the problem child requires that one change the clinical samples under two conditions. One com-
parents. parison examined the interactions among family

257
 -, ‘Mablets7
Correlations Among Family Management Variables and Child Aggression

Lack of Family Correlation with Child Parent Report of Peer Rating of

Management Skills (MOSAIC) TAB Score Physical Aggression Physical Aggression
Monitor .49* toa .06
Obedience 2 [6S a es
Discipline ATT or | .40*

ProblemSolving .28 .06 AZ

pie 05
=op<. Oil
***p<.001
tp < .10

members including the problem child, and the oth- out numerous questionnaires, and also responded
er excluded interactions with him. The analysis re- to brief daily telephone interviews. Only small
vealed the curious fact that (with one exception) subsets participated in some of the more costly
when the parents interacted with siblings, or with procedures.
each other, their behavior fell within the normal Separate studies were made for each of the four
range of coerciveness. The exception involved fa- family management constructs. In each case, sev-
ther and male sibling interactions. The parents of eral dozen variables were identified from the vari-
Social Aggressors seem to allow all children to be ous assessment batteries that seemed (a priori) to
coercive in their interactions with each other. Even relate to the construct. Following this, each item
in the absence of the problem child, the siblings was intercorrelated with all other items, as well as
were two to five times more coercive than were with each of the four criterion measures of antiso-
normal siblings! Apparently, siblings in families of cial behavior. These correlation matrices were
Social Aggressors are generally disposed to be irri- constructed separately for each of the three
table in their dealings with one another. However, grades. Composites or clusters of variables were
for these parents, the identified problem child be- identified that intercorrelated significantly at each
comes a kind of storm center. It is in the presence grade level and correlated with one or more crite-
of the problem child that the parents become max- rion measures of antisocial behavior at two or
imally coercive. The Social Aggressor occupies a more grade levels. A given family management
very special niche within the family; his deviancy construct might be defined by one to four such
status is tied directly to his ability to elicit coercive composite scores. At the time of this writing, the
reactions from his parents, particularly the mother. analyses have been completed for Monitor, Disci-
pline, Obedience, and Problem Solve. The con-
Family Management Variables struct, House Rules, is still being analyzed.
From 1979-1981 an effort was made in the lon- Table 11.7 summarizes the findings for the rela-
gitudinal planning study to translate the family tion between the family management scores and
management variables into assessment proce- three criterion measures of aggression (Loeber,
dures. A sample of over 200 normal children from Patterson, & Dishion, in preparation). Only the
grades 4, 7, and 10 were measured by a battery of findings for the youngest sample (the 4th graders)
over 18 assessment procedures. The children and are presented here. The criterion measures con-
their families participated in home observation sisted of a TAB score based upon data from the
sessions using the new code system (MOSAIC by MOSAIC code,.a composite from several items
Toobert, Patterson, & Moore, 1981), and video- from a parent-report questionnaire that measured
taped problem solving (PANIC by Forgatch & the child’s physical aggression, and items from
Wieder, 1981). Laboratory measures of parent peer nomination that also measured physical ag-
perception of child deviancy (Holleran, in prepa- gression. All three criterion measures intercorre-
ration) were also employed. The children and late (range .41 to .46).
their parents were interviewed separately, filled It can be seen that the pattern of correlations

258
 varies somewhat as a function of which criterion both retrospective (Robins, 1966) and longitudi-
measure serves as the focal point. This can most nal designs (Shea, 1972) showed early antisocial
- easily be put in perspective by comparing the find- patterns put the individual at greater risk for crim-
ings to the patterns obtained for the sample of 7th inal behavior, institutionalizaton, and general
graders. In general, it is the disruption in Obedi- marginal adjustment as an adult. Indeed, a read-
ence and Discipline that relates most closely to the ing of the preliminary findings suggests the hy-
widest range of criterion measures in the two pothesis that extremely antisocial children may be
grade levels. The two variables are indirect meas- at greater risk for later adult pathology than are
ures of the parents’ use of punishment for deviant children who have one parent diagnosed as schizo-
behavior (Patterson & Stouthamer-Loeber, in phrenic (Garmezy & Devine, 1982).
preparation). The Obedience score was based Our current speculations would have it that
upon four composite scores that describe parent young Social Aggressors, who are also socially un-
and child perceptions about the extensiveness of skilled, will grow up to be marginally adjusted
the child’s rule breaking. The data sample ques- adults. We are beginning to suspect that it is the
tionnaire data from fathers and mothers, as well mixed case (the young Social Aggressor who
as interview material from the mother and child. steals) who is most at risk for adolescent delin-
The Discipline score was based upon global rat- quency and status as a career offender. However,
ings made by observers following each visit to the this is all very speculative. The purpose of the dis-
home. The composite describes how effectively cussion that follows is to review the findings relat-
the parents use punishment in coping with the on- ing to the long-term implications of symptoms
going coercive behavior of the child. that would be categorized as Social Aggressive.
The assumption is that patterns of family man- The findings from the West and Farrington
agement variables will differentiate pure Stealers (1973) study showed that boys rated by teachers
from pure Social Aggressors. As yet, no formal as aggressive at age 9 had a .14 chance of being
test has been made for this hypothesis. However, later identified as violent adolescent delinquents.
the set of findings summarized in Loeber, Patter- In fact, 70% of the violent adolescent delinquents
son, & Dishion (in preparation) suggest that the had been rated as aggressive at age 13 years; 48%
Obedience and Discipline scores account for the had been so rated at age 9. McCord’s (1977) well
larger share of variance in measures of Social Ag- known follow-up study of the Cambridge-Som-
gression. The analysis by Patterson, Stouthamer- merville sample collected teachers’ ratings of boys
Loeber, and Loeber (in preparation) suggests that when they were about 7 years of age. Official rec-
it is the Monitoring, Problem Solving, and Obedi- ords were then obtained for the men who were
ence scores that account for variance in measures now in their forties. The teachers’ ratings showed
of attacks against property. a cluster of relations among temper tantrums, bul-
While the findings in Table 11.7 are based upon lying, and getting into fights. Of these, only tem-
very small samples, they are strongly supportive of per tantrums were related to later criminality.
the hypothesized relation between family manage- A retrospective study by Loeber, Janda, and
ment practices and socially aggressive behavior. In Reid (in preparation) identified adolescents with
this normal sample, the disruption of family man- assault experience, who had been studied earlier at
agement procedures covaried with increases in an- OSLC. Their analyses showed significant differ-
tisocial child behavior. We plan to replicate the ences in family processes for the sample later iden-
findings using a longitudinal design, where it will tified as Assaulters. The children studied earlier,
be possible to demonstrate the predictive power of who were at risk for later assault, had a higher
the family management variables. We also assume TAB score than did all other family members. Rel-
that pre- and post-measures for these variables atively speaking, children at risk for later assault
would show significant improvement in family were the most coercive members of their own fam-
management practices following treatment. The ilies. In the majority of families from normal and
third set of studies will search for the covariation clinical samples, it is the mother who has the high-
between disruptions in management processes and est ranking TAB score.
changes in microsocial variables such as Crossover
and Continuance. Stealers and Delinquents
There are two assumptions that are central to
Prognosis for Adjustment as Adults this section. The first is the continuity hypothesis.
The findings reviewed in Chapter 2 showed Robins (1966) emphasized the continuity between
consistent support for the stability of antisocial be- early forms of antisocial child behavior and later
havior across long periods of time. Findings from antisocial behavior. In the present context, we em-

259
phasize the continuity between early protodelin- which things belong to which other people. Then
quent behaviors such as lying, stealing, firesetting, he may be taught to ask permission before using
vandalism, and adolescent delinquency. The sec- things that belong to someone else.
ond assumption is that there is a related continuity The data in Figure 11.3 attest to the eventual
in family process, from the young child engaged in success achieved by most parents. This longitudi-
protodelinquent activities, to the older adolescent nal study of an essentially middle- and upper-class
chronic offender. Specifically, it is assumed. that sample showed that with girls the vast majority of
both early and later forms of antisocial behavior parents were successful by the time their daughters
are accompanied by disruptions in parent moni- reached the age of 5 or 6; i.e., they did not steal
toring and a failure to punish attacks against pro- often enough for it to be a concern of the parent
perty. Given that a child remains antisocial over a (MacFarlane, Allen, & Honzik, 1954). Parents of
long period of time, then the family processes that boys reported more concern about stealing than
support these behaviors must also remain stable did parents of girls. From ages 3 to 10 there was a
over time. These key assumptions can only be steady decline in concern.
tested in longitudinal designs. The literature relat- The Isle of Wight survey by Rutter, Tizard, and
ing to the continuity for protodelinquent and de- Whitmore (1970) showed that 5.7% of, parents
linquent behavior will be reviewed in a later sec- were concerned about their sons stealing at ages
tion. As yet, there are no data testing the related 10 to 13, and 2.6% expressed similar concerns
hypothesis about continuity in family process. about their daughters. The comparable percen-
Lacking such data, we can only proceed with a tages reported by teachers were 3.4 and 1.6 re-
more indirect test and use a cross-sectional design. spectively. Based upon a psychiatric interview,
Here, the assumption is that the pattern of family 30.2% of the boys identified as Stealers were diag-
management variables relating to protodelinquent nosed as “disturbed.” The comparable percentage
measures for a younger sample will be similar to for female Stealers was 13.5. For those samples
that which covaries with measures of delinquent the base rates for psychiatric disturbance were
behavior for an older sample. Those findings will 4.4% and 2.4%. Sometime between the ages of 5
be summarized ina later section. and 10, taking things is no longer perceived as
normal and acceptable behavior. At that time, the
Normal Stealing episode will be relabeled as stealing and is associ-
Almost everyone has stolen. In the study of self- ated by adults with “emotional disturbances.”
report data by Hood and Sparks (1970), many The data from cross-cultural, self-report studies
normal adults reported stealing regularly. During reviewed by Hood and Sparks (1970) strongly em-
the toddler and preschool ages, it is perfectly re- phasized the fact that many persons do not stop
spectable to “steal”; however, it is not labeled as stealing at the age of 4 or 5; they simply reduce
such. It becomes a “symptom” only if the child their rates of stealing and are discreet enough
persists in doing it at high rates when he is of about it to reduce the likelihood of being appre-
school age. If it persists through age 7 or 8, then hended. For example, adolescents’ self-report data
the community is likely to become concerned. covering a three-year period showed a mean inci-
I assume that it is entirely “natural” for an infant dence of 1.1 events of petty thefts ($2.00 or less)
to reach for, and take, objects that are of interest (Elliott & Voss, 1974). The mean for major theft
to him. Early on, however, the parent quickly in- ($2.00 to $5.00 value) was 3.4 events. The self-re-
tervenes when he reaches for objects that represent port rates for females were roughly 30% less than
a potential danger, e.g., scissors, lighted cigar- the comparable figures for males.
ettes, a heavy lamp. At some later point (probably If the young child regularly steals from family,
much later) the child is trained to “respect the neighbors, and peers at school, he is likely to be
property rights of others.” That this early training caught eventually, or at least be suspected of being
is, for a time, rather tenuous is witnessed by Har- a Stealer. This is my “frequency labeling” hypothe-
tup’s (1974) observation of nursery school chil- sis. The children most at risk for being labeled are
dren. He found that 78% of their quarrels were al- those engaged in the higher rates of deviant epi-
tercations about “possession.” It may also be the sodes. For our purposes, a Stealer is a child be-
case that many early disputes among siblings con- tween the ages of 6 and 12 who is “caught” steal-
cern possession of objects and/or territory. By the ing about once every three or four months. The
age of 3 or 4, children are generally trained to un- hypothesis is that, if permitted, the young Stealer
derstand what belongs to whom and which objects will persist in maximizing short-term gains. If the
are jointly shared by family members. The parent family permits a 5-year-old to steal, the chances
may begin this training by teaching the child are very good that they will be equally permissive

260
 Figure 11.3
Parents’ Reports of Stealing as a Problem Behavior
(adapted from MacFarlane, Allen, & Honzik, 1962, pp. 66-67)

@ Boys N 60
O Girls N 50

Sunioday
syusieg
U29I9g
==

Age

261
about stealing when the child is 10 or 12. Given a form to these rules; i.e., the rules are “internal-
continued lack of monitoring and punishment for ized.” The cultural-strain theorists assume that the
stealing, then the child may drift into the- pattern knowledge is present, but the original bonding of
of a career offender. The hypothesis is that any the individual to cultural norms has been dis-
time up to about age 15, the parent can bring this rupted by factors such as extreme deprivation or
problem behavior under control. Beyond that age, frustration. Deviant acts occur as a result of the
it may be that other forces, such as the peer group disruption. The culture-deviance theorists, on the
or work place, may have a greater impact than the other hand, would assume the rules that have been
parent upon the youth’s behavior. internalized reflect cultural values that are, them-
selves, “deviant.” From this perspective, the indi-
Speculations about the Families of vidual is conforming, but to a deviant subculture.
Chronic Offenders Control theory represents yet another perspec-
The parents of young children at risk for be- tive on this problem. The most coherent statement
coming chronic offenders have some special char- of control theory, together with a description of its
acteristics that were noted in earlier sections. empirical underpinnings, has been provided by
Many of the parents of Stealers were not attached Hirschi (1969): |
to their children (Reid & Patterson, 1976). They
“In the end then, control theory remains what it
did not seem to care what happened to their chil-
has always been, a theory in which deviation is not
dren; an apt label for this subset would be “unmo-
problematic. The question, ‘Why do they do it?’ is
tivated parent.” There was another group that
simply not the question the theory is designed to
seemed attached to the child but could not identify
answer. The question is, “Why won’t we do it?’
with the role of parent. They cared for the child;
There is much evidence that we would if we
they fed him; but they did not want to be responsi-
dared.” (Hirschi, 1969, p. 34)
ble for training him. They particularly did not
wish to monitor and confront the child’s antisocial As I understand control theory, attachment or
activities going on outside the home. The parents bonding is assumed to be a necessary (but not suf-
of Stealers shared a distaste for using the term ficient) condition for conformity. The accompany-
“steal” (Reid & Patterson, 1976) when describing ing commitments, involvement, and belief systems
their children. It seemed that any and all “explana- are also thought to be necessary (but not suffi-
tions” given by the child were accepted if they con- cient) conditions for conformity to cultural norms.
tributed to the parents’ belief that their child did I would tend to agree with Hirschi (1969) that
not steal. In fact, it took us months to figure out a there is not much of a bond between the child who
procedure to get around the fact that many of steals and his parent(s), or between the child who
these parents could not believe that the problem steals and his peer group for that matter. To this I
even existed! To them, the term “steal” meant would add that many of the parents of Stealers are
something analogous to proven in a court of law. not attached to the role of parent. Their lack of
Often they would not accept the word of a neigh- motivation has been a major problem during our
bor, a teacher, or a storekeeper as “proof.” efforts to specialize in the treatment of Stealers. It
I suspect that part of the difficulty these parents was almost impossible to get parents of Stealers to
had in labeling stealing related to the fact that come in for treatment (Reid & Patterson, 1976).
many had delinquent careers of their own. Parents After being referred, they tended to drop out of
with criminal histories would be less likely to the baseline study at alarming rates. A recent ana-
monitor and punish antisocial behavior, such as lysis of the factors determining early dropout from
stealing. The correlation between antisocial be- treatment showed that almost invariably these
havior of the child and criminality of the father were Stealer families. Those that did participate in
has been supported in studies by McCord (1980), treatment seldom carried out their assignments.
Robins (1966), and Mednick and Christiansen The problem was of such magnitude that it
(1977). prompted us to design and carry out a study to de-
There is, of course, an extensive literature about termine the impact of providing “parenting sala-
the child’s role in acquiring or rejecting cultural ries.” The publication by Fleischman (1979b) dem-
norms. The child’s internalization of cultural onstrated a significant increment in parent partici-
norms has been a major concern for theorists in pation in the training program if salaries were
sociology, developmental psychology, and person- made contingent upon their carrying out their as-
ality. Generally, it is assumed that children and signments. From the perspective of coercion the-
adults alike know what society’s rules and expec- ory, neither parental motivation nor parental skills
tations are and, secondly, that they desire to con- are thought to be a given condition.

262
 The assumption presented here is that most
chronic adolescent offenders begin their careers Family Processes Relating to the
‘very early; i.e., by the time the child reaches the Continuity Hypothesis
age of 6 or 7 a discerning observer would note the Regardless of when the antisocial behavior of
pattern. However, our clinical experience suggests the child begins, it is produced by a disruption in
that some chronic offenders begin their careers in parental monitoring and punishment for antisocial
early adolescence. We don’t know what the rela- attacks. For the adolescent, this disruption could
tive occurrence of early and late starting processes occur during a time of severe marital conflict be-
might be. In our experience, there is a ratio of tween the parents, a major illness, a move to a
about 2 to 1 (early to late) for our sample of new neighborhood or unemployment for the par-
chronic offenders. My hunch is that the late start- ents. It is also assumed that if the antisocial behav-
ers are the product of a breakup in family struc- ior continues over time, parental monitoring and
ture, e.g., divorce, severe illness, or unemploy- punishment remain disrupted. If the preadolescent
ment. We also have no data on the relation be- terminates his antisocial career, this would likely
tween early and late starting and the risk for career be produced by the parents reinstating control of
adult offenses. However, the research literature their child.
consistently relates early starting of police offenses The first study of the relation of family manage-
with risk for adult offenses, so I will pursue the ment variables to the continuity hypothesis used a
formulation about early starters a bit further. cross-sectional design for two age groups. One
Recent discussions with R. Loeber led to a for- sample consisted of 60 families of 4th grade chil-
mulation about early starters that is viewed as a dren, and the other combined a sample of 100
kind of transitive progression beginning with the families of 7th and 10th grade children. For each
disruption or absence of parental monitoring of sample, parent-report data identified children
the child’s protodelinquent attacks against proper- thought to lie and steal (Patterson & Loeber,
ty. The 5-, 6-, and 7-year-old steals with increas- 1981). For the older sample, these two criteria
ing frequency from family, neighbors, and stores. plus police-offense data were used as dependent
He is also engaged in frequent acts of vandalism. variables.
On both counts, he quickly learns to lie as ameans As can be seen in Table 11.8, for the older sam-
of covering up his activities. In fact, it is his lies ple it is the Monitor score and, to a lesser extent,
that may be noted first by parents and other Obedience and Discipline scores that correlate
adults. The third step in this progression would be heavily with measures of delinquency based upon
that he receives much more unsupervised street official records. The adolescent child with police
time than normal children of this age would be offenses is characterized by parents who tend not
given. Fourth, during his wanderings, he will very to track him closely. The child is disobedient both
likely encounter older and more experienced anti- at home and at school. Even when the child is dis-
social children. This peer support base may ruptive in the presence of his parents, their disci-
emerge slowly, but we see it as an important step pline techniques are erratic and ineffective.
in the child’s increasing commitment to an antiso- The next hypothesis was that (for the older sam-
cial career. The current OSLC study involves a ple) parent reports of protodelinquent behaviors,
sample of 109 chronic delinquent offenders. Half such as lying and stealing, would covary with
of their police offenses occurred in the company of court records of police offenses. As shown in the
peers or siblings. About one youth in five com- study by Patterson and Loeber (1981), the correla-
mitted at least one offense in the presence of a sib- tion between lying and police offenses was .31
ling (usually an older sibling). (p<.001), and for stealing and police offenses it
It is assumed that the early starter is engaged in was .46 (p<.001). It was assumed that the same
frequent acts of theft and vandalism; these occur family management variables that covaried with
at a steady rate throughout his school years. The police-offense data correlate with the measures of
high density of offenses eventually leads to his be- protodelinquency. As shown in Table 11.8, the
ing labeled as antisocial. This high density also pattern of the correlations was quite similar. The
places the child at risk for increasingly “serious” measures for Monitor, Obedience, and Discipline
crimes, such as arson, assault, burglary, and so on. all correlated with parent reports of Lie and Steal.
The hypothesis is that density covaries with label- These findings offer strong support for the hy-
ing and with the seriousness of the crime. It is also pothesized continuity from measures of protode-
assumed that the early starter is most at risk for be- linquency to measures of police offenses and also
coming an adult offender. for the assumption that they share a common fa-
milial process.

263
 Table 11.8
Covariations of Family Management Variables
with Measures of Protodelinquent and Delinquent Behaviors
(fromPatterson
&Loeber,
1981)

Lack of Family 4th Grade Sample 7th and 10th Grade Samples
Management Skills Lie Steal Lie Steal Police Offense

Monitor bs aad .48*** A4tet 58t**

(59) (92) (89) (100)

Obedience 239*% 54t** 35*** 31**
(46) (87) (81) (90)
Discipline .45** .34* .37* .36*
(10) (35) (33) (36)
ProblemSolve 5350)
(22)
The figures in parentheses are N’s.
tThese correlations have not been calculated. More tapes are being scored to increase the N’s.

*p<.05
SMO
et SOO

The next feature of the continuity hypothesis to as antisocial. It was also assumed that the more
be tested was that the same pattern of family man- frequent his attacks, the more likely they would be
agement variables would also correlate with meas- to eventually escalate to more serious episodes.
ures of protodelinquent behavior for the younger Wolfgang, Figlio, and Sellin’s (1972) extensive
age group. As shown in Table 11.8, the measures Markov analyses showed that the movement from
for Monitor, Obedience, and Discipline all covar- one type of crime to another seemed random.
ied significantly with parent reports of Lie. How- There was no systematic progression or path to-
ever, the findings for Steal only partially supported ward any particular crime. The child did not first
the hypothesis. Only the covariation between Steal steal cars, then progress to burglary, and finally
and Discipline was significant. The findings pro- become assaultive. The systematic analysis of self-
vide general support for the continuity in family reported delinquency from the National Probabili-
process patterns that maintain earlier and later ty sample by Elliott and Huizinga (1980) also
protodelinquent behaviors. failed to find evidence for either a progression or
the concept of specialization. The most that can be
The Continuity of a Delinquent Career said is that the high-rate offender was more likely
The hypothesis is that young children who steal to engage in all forms of serious crime. The career
are at risk for adolescent delinquency and for delinquent self-reports higher frequencies of status
adult adjustment problems as well. The magni- (less serious) offenses, as well as crimes against
tude of the adjustment problems is thought to re- persons (theft and property damage). This relation
late to the level of work and relationship skills is not, however, symmetric; i.e., not all children
possessed. Presumably, children characterized as who report frequent status offenses go on to more
high-rate Stealers, who are also lacking in social serious crimes. I interpret these studies to mean
skills, are at greater risk for becoming career of- that we should be concerned primarily about the
fenders. Both the deviancy status and the lack of frequency with which a child engages in status and
social survival skills are thought to relate to paren- nonstatus offenses. For the young child, this could
tal skills in family management. The career of- mean a general focus upon rule-breaking behavior
fender is more likely to be an early starter. Presum- at home, in the school, and in the community.
ably, it is the high density of his attacks against There are no studies that test the assumed co-
property that leads to his eventually being labeled variation between amount of unsupervised time

264
 and lack of punishment with increases in depreda- alike that many of these children know each other.
tions against property. Nor are there tests for the Occasionally, they “work” together. More often
. assumed relation between density and labeling. As than not, they meet casually on the street.
they now stand, these assumptions are based pri- The young, high-rate Stealers treated at OSLC
marily upon our clinical experience in working do not seem to have close ties to family members.
with these families. If a young child has a great Contrary to popular press, he is not an angry
deal of unsupervised, on-the-street time, he re- shark cruising in a dedicated search for prey.
sponds to targets of opportunity. The unguarded Rather, he is an aimless wanderer. He drifts
bicycle is appropriated, then discarded. A building through school and across the community. In his
site is vandalized; tires are slashed. The teacher’s wake he leaves a trail of petty, and sometimes not-
purse left on the floor by her desk is examined. so-petty, crimes. He drifts through relationships in
Most of the depredations are unplanned. The the same meaningless fashion. He does not think
chance that his activities will be detected is slight. of himself as good or bad. Above all, it is not he
However, if his rate is high enough, then eventual- who is responsible for what happens; he is only
ly he will be caught. When caught, the first, sec- sometimes lucky and sometimes not. Unless
ond, third, or even tenth instances are usually of helped, he becomes a piece of psychological flot-
little moment. “He is just a little wild; did you hear sam, drifting on a current of offenses that gradual-
what he. . . ?” Over time, the offenses become ly increase in frequency and seriousness. When his
more serious, and/or the sheer weight of an accu- drift has achieved chronic status (five or more of-
mulation of these events eventually leads some fenses), his prospects for adult adjustment are not
adults to take a rule violation as being serious. At good. While still a juvenile, he may proceed to
that point, he may be referred to the court and la- adult crimes. In his follow-up study of Philadel-
beled as delinquent. The label did not cause the phia delinquents, Wolfgang found that 76% of the
behavior. His frequent escapades increased the chronic juvenile delinquents later became adult
odds of his eventually being caught, and his persis- criminals (Wolfgang, 1977). Wolfgang’s follow-up
tence led to his being labeled as deviant. Even study of Philadelphia delinquents takes a provoca-
though he is labeled as delinquent, this child prob- tive stance on the criminal career: “. . . at whatev-
ably does not perceive himself as deviant, crimi- er age the chronic offender begins his fourth or
nal, or delinquent. As noted in the review by EI- fifth offense he will commit further offenses with
liott, Ageton, and Canter (1979), there is substan- very high probability . . . the probability that an
tial empirical evidence that many embezzlers, auto offender after his fourth offense will recidivate is
thieves, check forgers, and persons involved in as- about .8” (p. 17). As Wolfgang points out, given
saults or rape do not view themselves as criminal such continuity, the juvenile/adult statutory di-
or delinquent! In their study, delinquent children chotomy makes little sense. It is not the age of the
who were caught and labeled suffered no loss in offender that should be of concern to us but, rath-
self-esteem. Interestingly enough, those who were er, the number of offenses. As the number of of-
caught, labeled, and counseled did show a more fenses increases, so does the likelihood for commis-
negative evaluation of self. sion of yet another offense (Wolfgang et al., 1972).
Most investigators are convinced that the peer I find myself in accord with Wolfgang’s position
group plays a critical role in the development of but would urge that the continuity is apparent at a
the career offender. While we share this convic- much earlier age. I believe that by the age of 7 or
tion, it is also the case that the necessary data are 8, high-rate stealing places a child at great risk for
lacking for a test of the process described earlier. If chronic delinquency as an adolescent. Gersten et
the young child has a great deal of unsupervised al. (1976) factor analyzed the data from a large-
time, does this relate directly to a high density of scale survey of Manhattan children. Over a five-
antisocial acts, or is it mediated by his finding an year interval, the stability for the delinquency di-
antisocial peer group? Survey studies have consis- mension was .44. The authors believe that these
tently shown that rural and small-city delinquents delinquent behaviors stabilize after the age of 10,
tend not to be gang members (Call, 1965; Lentz, while antisocial behaviors, such as fighting and
1956). For example, Lentz (1956) reported that conflict with parents, may stabilize as early as 6
only 22% of rural delinquents were members of years of age. The most relevant study testing the
gangs, as compared to a rate of 87% for metropol- continuity between juvenile and adolescent steal-
itan delinquents. In his study, the majority of rural ing was carried out by Moore, Chamberlain, and
delinquents were “loners.” This also characterizes Mukai (1979). They examined the court records
most of our clientele; however, it is evident in for 60 antisocial children studied earlier at OSLC.
working with chronic delinquents and Stealers At the time of this follow-up study, all were 14

265
years of age or older. During their initial contact,
the 21 Social Aggressors were, on the average, 8.4
Footnotes
years of age; the 25 Stealers were, on the average, 1. Before leaving these studies, it might be
9.8 years of age. Slightly more than half (.57) of noted that the introduction of punishment (such as
these children were both Stealers and Social Ag- time out) did bring the deviant behaviors under
gressors. The mean age of the normal sample was control in the studies by Corte et al. (1971), Laiv-
8.8 years. Community records indicated that all of gueur et al. (1973), Budd et al. (1976), Wahler
them were in residence in the community during (1968), and Walker and Buckley (1973).
the period studied. The data show that boys with 2. Probably the simplest idea would be that
stealing as a complaint were at grave risk for being prolonged exposure to aversive interaction pro-
picked up at least once by the police. The likeli- duces a reduced responsiveness to either positive
hood was .77. In fact, 52% of the young Stealers or aversive reactions. One could also imagine pro-
became chronic offenders (four or more offenses). longed aversive interactions being correlated with
Nonstealer clinical cases (“pure” Social Aggres- the likelihood of a negative set, but there are no
sors) were no different from Normals. studies that relate to either of these ideas.
The findings for the Stealer sample were partial- 3. The data from the Whiting’s study also
ly confounded due to the fact that at the time of re- showed the amount of infant caretaking correlated
ferral 42% of these young Stealers already had at significantly with the child’s observed nurturant-
least one police contact! This represents a very responsibility behavior. It may be that more “ad-
early start indeed. It is of particular interest be- vanced” cultures do the child a double disservice
cause in the research literature such early begin- by not expecting him to be responsible in some
nings are often associated with increased likeli- fashion to other family members, i.e., to give as
hoods for later chronic offenses and recidivism fol- well as to take. In addition to making the child rel-
lowing incarceration. In the OSLC study, the fol- atively “functionless,” modern society has de-
low-up records showed that even if those with pri- prived the child of that which he can learn from
or police contacts were excluded, the likelihoods being responsible for the well being of another.
were .56 for Stealers, .13 for Nonstealers, and .15 This, as noted by many modern writers (such as
for Normals. It seems reasonable to conclude that A. Adler), seems to be something lacking in our
young boys referred for stealing were at great risk modern culture.
for at least one police contact by the age of 14. 4. The concept of transitive progression is cer-
Given that many of the families who moved tainly a useful one. However, the adequacy of the
from the community may have contained children Guttman coefficient as a means for describing it
at risk, and that the peak years for delinquent ac- has, from the beginning, been open to serious
tivity are not 14, but 15 through 16, then it seems questions (Festinger, 1947, cited by Robinson,
reasonable to conclude that the Moore et al. fig- 1973). As Menzel (1953) and others have pointed
ures are probably underestimates. By age 18, out, the value for CR is affected both by the item
probably a much larger proportion of young Steal- marginals and the number of items. The efforts by
ers become chronic offenders. Given that crimes Menzel (1953), Schooler (1968), and others to
against persons are more likely to peak at about correct these limitations still leave unanswered the
that age, then it seems too soon to conclude that problems detailed in the Robinson critique (Rob-
(nonstealing) Social Aggressors are not at risk for inson, 1973). I found Robinson’s arguments suffi-
delinquency. Would nonreferred, high-rate Steal- ciently compelling to lead me to search for a differ-
ers be at risk for chronic delinquency? That is the ent way of summarizing information about pro-
topic for a longitudinal study that is currently be- gressions.
ing prepared. 5. R. Vreeland (in preparation) carefully re-
checked all of the OSLC records for any mention
of firesetting and identified 36 children referred to
OSLC. Ninety-three percent of this more loosely
defined group of Firesetters were also Stealers.
This in contrast to the comparable figure of 78%
from Table 11.1.
In passing, the interview study by Nielsen and
Acknowledgments Gerber (1979) has just a hint of data suggesting
The writer wishes to thank Alice Harris, Tom that we should look for vandalism (at school) and
Dishion, and Rolf Loeber for their careful cri- cruelty to animals as lower base-rate symptoms
tiques of earlier versions of this chapter. that are part of this same progression.
 6. Based upon the Reid and Hendricks (1973) lem. In his factor-analytic study, Carlson (1981b)
study, there is a hypothesis that families of Stealers showed that Noncomply and Disapproval (and
‘are more distant and/or uninvolved than either other items like them) loaded on a factor that he
families of Normals or families of Social Aggres- labeled Verbal Assertion. High scores on this fac-
sors. The measures of these concepts (based upon tor significantly differentiated the Hyperactive
the FICS) were not thought to be effective transla- from both Social Aggressor and the Normal sam-
tions of either distance or parent involvement. The ples. Both the Hyperactive and Social Aggressor
newer measures from the MOSAIC and PANIC children performed at about the same coercive lev-
codes may prove to be more useful in this regard. els, at least as shown by their TAB scores. They
At the time of this writing, the appropriate analy- differed, however, in the kinds of deviancy that
ses have yet to be completed. made up the high levels of performance. The Hy-
7. Carlson (1981a) showed that roughly one- peractive child tended to be more verbally coer-
fifth of the antisocial children referred to OSLC cive, and the Social Aggressor more oppositional
had been previously diagnosed by a physician as (including physical) in his interactions.
hyperkinetic and were medicated for that prob-

267
 Chapter 12
Abstract
The requirements of the caretaker role are reviewed. Particular emphasis is given to training the
caretaker to use effective family management skills and noncoercive reactions to irritable child behav-
iors. The effect of disruptions in family management is compared to the effect of permissive, laissez-
faire child rearing.
The father is cast as a key member of the mother’s support system, with adjunct functions in child
management. Fathers’ and mothers’ irritable reactions to problem children are compared in terms of
variance accounted for in the criterion measure of child performance. The analyses emphasize the key
role played by the mothers’ irritable reactions. Acomparison of mothers and fathers from intact fami-
lies further emphasizes the differences in parental role. Acomparison of mothers’ and fathers’ self-re-
port data from clinical and normal families reiterates the fact that family pressures and/or lack of ad-
equate support produces depressed mothers. As compared to Normals, mothers of antisocial families
show significant elevations on almost all of the clinical scales for the MMPI. Fathers of normal and
antisocial families show no significant differences on their MMPI scores. It is hypothesized that the
caretaker bears the brunt of crises that impinge from outside the family and the conflicts occurring
within the family itself.

268
 Mothers: Everyone
Loves Them, But...
“In the United States today, parents rather than tremely coercive children a very permissive and/or
the state have primary responsibility for socializ- unskilled teacher would preside over chaos. It is
ing their young. Socialization is an adult-initiated conceivable that a child could be trained to high
process by which the younger person through edu- levels of coercive performance at school or within
cation, training, and imitation acquires his culture the neighborhood by peers operating in unsuper-
as well as the habits and values congruent with vised settings. Once trained, the child might gener-
adaptation to that culture. There is no way in alize his “disposition to be aggressive” to the
which parents can evade having a determining ef- home; but, again, it is the parent who determines
fect upon their children’s personality, character, at what level he will be allowed to perform his
and competence. Children are not the originators new-found skill. As a general case, I believe the
of their own actions in the sense that their parents primary training for extremely coercive children
are or should be.” (Baumrind, 1978, p. 129) takes place in the home; it is generalized from
there to other settings.
While some children may, by temperament, be If our society has antisocial children, it is the
more difficult to control than others, in any given adults who are responsible. The mother and the
setting it is the adult who is the major determinant father (if he is present) allow these behaviors to oc-
of how much aggression will occur there. The par- cur. It may be that they have good reasons for
ent, the teacher, and the playground supervisor their omissions in child management practices,
determine mean coercive performance level for the e.g., overwhelming crises, ignorance, misguided
peer group. In well-engineered, token-culture expectations, or an extremely difficult child. Re-
classrooms, even groups of extremely aggressive gardless of the reason for the omission, it is a par-
children perform at almost zero levels of aggres- ent who must be taught, supervised, and sup-
sion (Walker & Buckley, 1972). If they choose to ported to undertake the responsibility of socializ-
do so, adults can set the level at which aggression ing the child. Training the parent in child manage-
will occur for young children. In actual practice, ment skills is a necessary component for produc-
the mean performance for a setting is determined ing long-term changes in extremely antisocial chil-
by some interaction of the adult’s child manage- dren. The emphasis upon parental responsibility is
ment skills in relation to aggression and the gener- a means of keeping our attention upon the locus of
al disposition to be coercive that is brought to the the problem. The explanation for both the exis-
setting by the children. In conjunction with ex- tence and the solution of the problem may be

269
found in the parent(s). old) are demanding, complaining, and aversive.
The purpose of this chapter is threefold: first, to Infants produce what may be the highest density
speculate briefly about the caretaker role; second, of aversive events of any role in our culture. Per-
to examine the differential contributions of the sons hospitalized for illness are expected to com-
mother and father to the deviancy process; and plain and criticize; the nurse in the role of caretak-
third, to look at the evidence for the assumption er is trained to accept such querulous behavior as
that the caretaker is at risk for psychological prob- part of the job. Similarly, some theories emphasize
lems. the necessity for the nursery school teacher, the
play therapist, and the social worker to be uncon-
The Role of Caretaker
ditionally open, accepting, and warm. Uncondi-
The contemporary view is that the infant does tional positive regard is viewed as a necessary
not require the continuous monitoring of a single (and, by some, a sufficient) means for helping the
adult caretaker. Rather, his needs may be well child move beyond his or her vicious attacks to
served by several adults who perform this func- more pleasant modes of interaction. The mother
tion. However, regardless of who fulfills the role and these other professionals have one thing in
of caretaker, there are certain components that common: all are participants in inequitable ex-
must be present for effective child management. changes. They are expected to receive a great
The infant and preschool child provide the care- number of aversive events but to give none in re-
taker with a very high density of aversive events. It turn. Few of their clients give positive reinforce-
is important that the caretaker be trained not to ment directly to the caretaker. The caretaker must
react in a consistently irritable fashion, or it seems then rely upon a support base that can redress the
likely that the child will be at risk for increasing imbalance of positive to aversive experiences.
antisocial behavior and physical abuse. The care- In most cultures, the training for the caretaker
taker also needs the skills necessary to model pro- role begins early and is, I think, directly associated
social behavior and to employ the painfully slow with sex role development. Two cross-cultural
process of positive reinforcement to teach proso- studies employing observation procedures have
cial skills. Over a period of months and thousands shown that after the age of 2 or 3, preschool girls
of trials, the child is trained to ask for what he are less aggressive than boys (Omark, Omark, &
wants rather than to scream or cry. Later, he Edelman, 1973, cited by Maccoby & Jacklin,
learns to wait and to ask at an appropriate time. If 1974; Whiting & Whiting, 1975). Maccoby and
the caretaker is disposed to react irritably to in- Jacklin (1974) point to the possibility of a biologi-
fant-dispensed coercive behavior, then this crucial cal basis underlying the general sex differences in
beginning of the socialization process may be de- rates of aggression. Nurturance and warmth are
layed. I think that training in prosocial skills and also an expected facet of the female sex role. The
reducing infant coerciveness are two separate par- cross-cultural study by Whiting and Whiting
enting skills, but when one is absent, the other (1975) showed that, in general, girls offered help
may also be deficient. and support more often than boys. These differ-
It is my impression that most cultures are fully ences, however, were not significant until ages 6
cognizant of the necessity for the caretaker to react or 7 but were then maintained at ages 8 through
to infant coerciveness in a nonirritable fashion. 11 years. Across all cultures there was a tendency
Most cultures are also aware of the necessity for for older girls to reflect more nurturance. The cor-
the parent or some surrogate to spend thousands relation between age and nurturance for girls was
of hours modeling or reinforcing the child for his .51; the covariation also held within each of the
slow acquisition of social skills. I think that this six cultures. The authors stressed the fact that ob-
awareness is reflected in the cultural insistence served nurturant behaviors covaried with experi-
that the future caretaker be trained very early to ence as a caretaker. Typically, it was the young girl
function in a nonaggressive fashion. Even during who was assigned the task of caring for a young
preschool years, the future caretaker is encour- infant. Incidentally, boys who were assigned such
aged to relate to other children in a nonaggressive tasks tended to be more nurturant and less aggres-
mode. Their play activities focus upon the details sive. The review by Maccoby and Jacklin (1974)
of infant care and housekeeping activities. showed that in free-play situations with preschool
The emphasis placed upon warmth and the ac- children, there were no consistent differences fa-
companying prohibitions upon counteraggression voring girls as being more nurturant than boys.
are necessary because of the client populations However, as they point out, an appropriate test of
served by caretakers. These clients (the very the caretaker role would require data collected in
young, the very ill, the handicapped, and the very settings in which a child younger than the target

270
 was available. boys and six girls were observed in their (middle-
During socialization, both the family and the class) homes. There were 46 categories of child be-
‘peer group teach the individual to substitute pro- havior; six differentiated between boys and girls.
social skills for the coercive techniques employed Girls engaged in more doll play, dress-up, and
by young children. Given that a mother had not asked more frequently for help; boys were more
participated in such a process, then her first infant likely to play alone and/or with blocks (Fagot,
would be at risk for neglect and abuse. The series 1974). In the next study of toddlers and their par-
of primate studies reviewed in Ruppenthal et al. ents, Fagot (1978b) again found that both parents
(1976) showed an appalling lack of skill for moth- were more likely to reinforce and criticize girls
ers who had been reared in isolation. Observation than boys. The parents were observed to be more
data showed 100% effective maternal care (for likely to provide boys with praise for play with
first-borns) for feral-reared monkeys, 95% for la- blocks; girls were more likely to be punished for
boratory-reared monkeys, and 24% for monkeys this behavior. Girls, on the other hand, were more
reared in isolation or with surrogate (cloth or likely to receive positive consequences for playing
wire) mothers. It was interesting to note that the with dolls and other kinds of soft toys; boys were
latter were more adequate if their first offspring more likely to be punished for engaging in this
was female (39%) rather than male (13%). Those kind of play.
monkeys reared with peers and isolated from In summary, the Fagot studies support the idea
adults were observed to be 75% adequate as that preschool boys and girls are differentially re-
mothers. As the isolated mothers reared their sec- inforced for caretaking-related behaviors. That is
ond and third infants, they improved significantly not to say that a well socialized young father could
in their caretaking skills. At the level of primates, not serve effectively in the caretaker role, but I sus-
then, it seems that the culture must at least arrange pect that he would require some training. The
for the future caretaker to have the normal social- most important among these acquired skills would
izing experiences. be a disposition to react in a nonpunitive fashion
I believe that this situation is somewhat analo- to coercive toddler behavior. Like the young
gous to the unskilled mothers of antisocial chil- mother, he would also require a support system.
dren. It is not that the young mother has been The function of the support system is detailed in a
reared in isolation; rather, it seems likely that the later section of this chapter. Suffice it to say here
mother of the Social Aggressor may have been that the absence of such a support system places
only partially socialized and displays a general the caretaker at risk for depression and perhaps
proclivity to react in an irritable fashion in a wide other psychological problems as well.
variety of settings. Coerciveness serves as a substi- Lacking the necessary longitudinal data, we
tute for the mother’s prosocial skills. know very little about the first stages in the devel-
I think that the training for caretakers begins opment of the coercive process. My best guess
early and is rather subtle in its form. The pro- would be that it is characterized by one or both of
grammatic studies by Beverly Fagot and her col- two variables. There is some likelihood that the
leagues demonstrate clearly defined sex differences young caretaker is coping with a difficult infant
in the reinforcement contingencies supplied by (see Chapters 4 and 6). However, most “difficult”
adults in the nursery school and the home (Fagot, infants do not become antisocial children. In keep-
1973, 1974, 1978a, b). Fagot (1978a) first empiri- ing with the findings from the longitudinal study
cally demonstrated the differences in play prefer- by Werner and Smith (1977), I think the deciding
ence of boys and girls in the nursery school setting. variable lies in the prior level of socialization of
A series of studies carried out by Fagot and other the caretaker. Their findings showed that, given a
investigators showed that nursery school teachers difficult infant and an unskilled mother, then the
tended to reinforce both sexes for feminine-pre- child is indeed at risk for later antisocial behavior.
ferred behavior about 80% of the time. Male The amount of external stress/crises would func-
peers tended to reinforce for masculine-preferred tion as a mediator here. The greater the number of
behaviors. She then began a series of studies inves- life crises and the less the skill of the caretaker, the
tigating differential reinforcement for sex role be- greater the risk for the child. However, the key
haviors in the home. In the first study, she estab- variable is the presence of a well socialized care-
lished six sex-typed behaviors based upon ratings taker. Either parent may serve, but they must be
of 38 traits in children as to their perceived mascu- functioning at a coercive level that is normal for
linity and femininity; e.g., aggression and rough- adults. Their level of socialization must also in-
housing were perceived as masculine behavior in clude a commitment to societal norms about theft,
toddlers (Fagot, 1973). In the follow-up study, six vandalism, and violence, together with a commit-

271
ment to the caretaker role. My guess is that most could one find a sense of commitment to the chil-
parents of antisocial children would not measure dren.” (Segal & Yahraes, 1978, p. 101)
up to these criteria. Prior to forming a. family,
This, of course, is an example of extreme ne-
many would be classified as unskilled, disposed to
glect. The formulation presented thus far would
react irritably, and only partially committed to so-
suggest that these neglected children are at risk for
cietal norms regarding criminal acts.
Stealer and/or Social Aggressor problems. The
Parental Neglect and Permissiveness amount of risk depends upon whether the parental
neglect is accompanied by an effort to monitor
It is the parent who is chiefly responsible for the
and punish theft and/or a disposition to respond
first stage of child socialization. If, for some rea-
irritably to child coercive behaviors. The field ob-
son, this responsibility is not met, then the nor-
servation studies by Burgess (1978) found that
malizing efforts of other socializing agents, such as
child-neglect families were significantly more coer-
peers and teachers, will be hampered. I think it is
cive in their interactions than families of Normals.
important, particularly when the child is young,
It seems, then, that neglected children are at great-
for the parents to be in charge. They make the ma-
er risk for social aggression. The correlation be-
jor decisions and set the standards for conduct.
tween neglect and stealing has yet to be tested.
From my own personal bias, the parents would
strengthen their role by using more positive rein- Permissive Parenting
forcement than punishment. When punishment is
One can selectively fulfill the requirements of
used, it should be firm but nonphysical. What
the caretaker role by accepting some responsibili-
concerns us in this section is what happens when
ties but rejecting others. One can, for instance,
the parents relinquish the responsibilities of their
satisfactorily discharge responsibilities regarding
role. This rejection may be manifested as parental
food and shelter, but leave matters of conduct,
neglect and/or permissiveness. The literature in
achievement, and leisure time more or less up to
developmental psychology will be briefly exam-
the child. As a general case, the extent to which
ined for findings relating various types of antiso-
the young child is in control of his social environ-
cial child behavior to parental neglect and permis-
ment places him at risk for antisocial behavior.
siveness.
Permissive child rearing places the burden of con-
Neglect trol upon the child too early. As a general thesis,
the control of antisocial behavior is learned as an
In our clinical contacts we often encounter
outcome of the parent providing firm, consistent,
adults who have children but who do not wish to
negative sanctions for such behavior. Typically,
be parents. As noted in Chapter 11, children who
such punishment is not a part of permissive, lais-
were high-rate Stealers tended to have parents
sez-faire child rearing.
who were unmotivated. Some of them also ne-
Sears, Maccoby, and Levin (1957) interviewed
glected the child’s physical needs, e.g., meals were
mothers to determine both the aggressiveness of
sporadic, and laundry and cleaning were haphaz-
the child and the parental recollection of the child
ard. Others have also commented upon such fami-
rearing practices that brought it about. While the
lies. For example, Rothchild and Wolf (1976),
covariations with child aggressiveness were slight,
cited by Segal and Yahraes (1978) detailed in a
a pattern of relations emerged that is very similar
dramatic fashion the lack of commitment to the
to a pattern that is suggested by coercion theory.
role of parent on the part of parents in counter-
They found that mothers of aggressive children
culture communes.
were more likely to be permissive about aggres-
“
. what they found were communities in siveness, to use physical punishment, and tended
which virtually all children were neglected and to be lacking in warmth. These findings were sup-
abused, victims of treatment that would be judged ported by the observations made by Diana Baum-
cruel and inhuman by most parents. The authors rind at Berkeley. In her first study she compared
were struck by the boredom, apathy, and melan- three small groups of normal, middle-class nursery
choly of the children . . . the parents seemed bent school children.. She found that the assertive, self-
on keeping the children out of their way, even if it reliant children had firm, controlling, and loving
meant denying them everything. Themselves still parents (Baumrind, 1966; Baumrind & Black,
children, they continued to seek gratification with- 1967). The discontented, withdrawn children had
out sacrifice, as if it were actually possible to deal relatively controlling (but detached) parents, and
with the young without fully accepting their pres- the children lacking in self-control had relatively
ence in our lives. . . . Nowhere in this new utopia permissive parents. In her second study (Baum-

272,
 rind, 1971), 95 preschoolers and their parents sample, and one or two in his second sample. An-
were studied. Again, she found that firm discipline tisocial children and their families may be further
‘was associated with competence in the child; per- out on a dimension of deviancy than the four sam-
missiveness related to a lack of self-control. In a ples studied by Schuck (1974) and Olweus (1981).
longitudinal design, she and her colleagues are I call this hypothetical dimension Anarchy; we will
studying 134 white and 16 black preschool chil- discuss this dimension in detail in a later section of
dren from middle-class families. this chapter. For the moment, suffice it to say that
Schuck (1974) and Olweus (1981) have updated its definition includes several variables not expli-
the work on the relationship between parental per- citly referred to by the Olweus triad.
missiveness and aggressiveness in the child. Both In addition, there are two tenets central to coer-
have analyzed the correlational findings by using cion theory that are not included in the Olweus tri-
causal models; both find low-level support for the ad. First, I assume that the parents of antisocial
covariation. Olweus (1981) used peer nomina- children are, generally speaking, unskilled people.
tions for aggressiveness by 6th- and 9th-grade As reviewed in Chapter 10, the studies showed
peers as criterion measures. He used a structured that antisocial children viewed their parents as be-
interview, similar in format to that designed by ing less skilled; interviewers also perceive the par-
Sears et al. (1957), to evaluate child rearing prac- ents of deviant children as less skilled (e.g., Sher-
tices. In each sample he particularly focused upon man & Farina, 1974). They lack specific skills for
four main variables: the child’s temperament, the punishing deviant behavior and supporting proso-
mother’s negativity, the mother’s permissiveness cial behavior. Secondly, I assume that the parents
for aggression, and the tendency for the mother of Antisocials differ from those of Normals in
and father to use power-assertive tactics in coping their threshold values for categorizing behaviors
with the children. In the first sample the causal as deviant (see Figure 11.2). Are these parental
analyses showed the mother’s permissiveness for views about child deviancy shaped by the process
aggression, the mother’s negativism, and the of living with these children, or are they the deter-
child’s temperament all contributed directly to the minants for it? I don’t know. However, I do know
criterion measure of aggressiveness. The PPM cor- that these views about what is deviant and what is
relations of these three variables with peer-nomi- not hamper clinical efforts to alter the course of
nation scores for aggression were in the range of the process once it has begun. Changing these pa-
.30 to .45. The multiple R testing for the causal rental perceptions is one of the most difficult as-
model was .579. A similar pattern of relations was pects of the treatment process.
also found for the second, older sample. Here the Ultimately, I think there comes a point when the
range of PPM correlations was from .23 to .34, parent begins to see the child as being different
and the multiple R was .460. and to experience strong negative feelings about
The studies by Olweus (1981) have done much him (if he is a Social Aggressor, but perhaps not if
both to correct the difficulties with the earlier he is a Stealer). Is it parental negative attribution
studies and to extend the pattern of findings. The and anger that make it so difficult to reverse these
difficult child in conjunction with an overly per- processes during treatment? Again, we do not
missive, highly irritable mother is a familiar theme know. There are miniature examples of this that
indeed. Note also that these three variables ac- also merit study. It’s my impression that, follow-
count for a substantial 20% to 30% of the vari- ing hospitalization for severe, nonpsychotic ill-
ance in the children’s aggressiveness. I believe that ness, many children return to their homes and are
the coercion formulation overlaps to a considera- permitted to become highly coercive by well-inten-
ble extent with the Olweus temperament-permis- tioned parents. The parents explain the increase in
siveness-negativity triad. The OSLC definition of coerciveness as an emotional upset caused by the
the family management variables may simply be a child’s separation from his family. But some chil-
powerful extension of the last two variables in the dren do not return to their prehospital baseline.
Olweus triad. However, I think that the differ- Sigal (1974) followed previously hospitalized chil-
ences involve more than that. dren for five years. Both the target child and his
The studies by Schuck (1974) and Olweus nonhospitalized siblings were ages 7 to 12 years at
(1981) have been based upon samples of normal, the time of the follow-up. The former patients
middle-class children. However, the antisocial were rated as significantly higher on the Petersen-
child (who is the topic of the present volume) is at Quay conduct problem scale. Why is this? I think
or beyond the 90th percentile on measures of ag- one possibility is that as the coercive performance
gression. Using that standard, Olweus may have level climbs, the increasing skill and commitment
had three or four antisocial children in his first of the child to this process means that it will be

2/3
more painful to the adult who tries to change the Figure 12.1. The normal coercion level for a fami-
child’s behavior. Most parents of antisocial chil- ly member is indicated by a blank square; as the
dren avoid confrontations with their children. To coercion level rises, the square is darkened.
a lesser degree this may also be what happens with The second hypothesis is that as more dyads
the previously hospitalized child. In addition, the within the family become more coercive than nor-
illness itself may give the child a “special” status to mal, the risk for physical assault increases com-
parents who feel that he has already suffered mensurately. The connecting lines in Fugure 12.1
enough (see Chapter 13). indicate that the dyad is characterized by higher-
The coercive process may begin because the par- than-normal rates of coercion. The data for the
ent has a neglectful, laissez-faire attitude toward figure were taken from several published studies
child rearing. However, the extremely antisocial (Patterson, 1981b; Reid et al., 1981; Reid, 1978).
child probably has a parent who is also relatively One begins on the left side with normal amounts
unskilled in teaching prosocial skills. An extended of coercion. The TAB scores for the normal sam-
participation in the coercion process also implies ple showed a mean value of .21 for fathers, .26 for
that the parent has (or soon acquires) a selectivity mothers, and .24 for siblings over the age of 7
about which child behaviors are categorized as de- (Reid, 1978).
viant and which are not. These cognitions, in Unfortunately, even among normal families
turn, relate to parental hesitation to punish antiso- there is some risk for physical assault. In a nation-
cial behavior. If the coercive process is permitted al probability survey the odds for a physical attack
to run its course, then it is no longer the caretaker upon a wife were .038 (Straus, 1978). The odds
who is in charge of family interactions. The reign- for a child being hit with a fist or an object are
ing monarch is the socially aggressive child; he and probably somewhat higher. For the sample of
his entourage maintain their hegemony by their “normal neighbors” interviewed by Gelles (1972),
skillful use of pain-control techniques. the odds that an assault had ever occurred in their
families were a very high figure of .077. He sam-
Anarchy pled families from lower socioeconomic levels; I
Given a rejection of the caretaker role, there are assume that the odds for an entirely normal sam-
several outcomes that may result. The material in ple would be considerably less than this. The point
Chapter 10 detailed the reduction in social interac- is that for a sample of “normal” families during
tion, shared leisure time, and increasingly negative any given year, there is a non-zero likelihood for
perception of other family members that occur physical violence. As a conservative estimate,
when the caretaker fails to use effective family about one family in 20 will include one or more
management practices and the family becomes dis- persons who have been beaten.
rupted. There is also a collapse of the support and The data show that for families from the Stealer
problem-solving functions of the family. Given sample, three of the four roles are noticeably more
that these disruptions have occurred, there is an coercive than normal. The increases are most
accompanying process that is of clinical concern. I marked for the deviant child and the mother. The
have dramatized this process by labeling it Anar- connecting lines show that the mother —® target
chy. It represents a transitive sequence. Families at child and mother -» sibling interchanges are sig-
an advanced stage of Anarchy fulfill all of the re- nificantly deviant. As noted in an earlier discus-
quirements for the earlier stages. Each stage de- sion, there is a strong likelihood that the mixed
scribes an increasing likelihood for the use of high- cases (Stealer plus Social Aggressor) are particular-
amplitude pain control, e.g., beatings and physi- ly at risk for child abuse. I assume that these fami-
cal assault. lies would also show a high percentage of dyads
The Anarchy progression rests upon the as- enmeshed in coercive interchanges.
sumption that the density of coercive interchanges To illustrate what the next step in the disorgani-
covaries with intensity (Reid, Patterson, & Loe- zation would look like, data are presented that de-
ber, 1981). The analyses that support this assump- scribe a sample of families of Social Aggressor
tion indicate that family members with high TAB children. The increase in anarchy is portrayed by
scores are more likely to Hit. Higher TAB scores the fact that all four familial roles are implicated,
are also characterized by increasing proportions of and the level of coercion is now three to four times
extremely aversive behaviors. Given that one greater than that for Normals. The family is large-
family member increases his or her coerciveness ly (but not completely) controlled by the children.
beyond the normal range, then the other members The disorganization is not complete in that the
become increasingly at risk for being hit. Amodel role of father is only minimally in conflict with sib-
for violence among family members is presented in lings and with the mother. However, even a casual

274
 Figure 12.1
A Progression for Violence

Social Child
Normals Stealers Aggressors Abusive
Siblings T. Child Siblings T. Child Siblings T. Child Siblings T. Child

(gag ol
wy J
Mother Father Mother Father Mother Father Mother Father

coerciveness is normal
a dyad of borderline significance

coerciveness is twice the normal level

significant dyad

coerciveness is three times the normal level

id coerciveness is four times the normal level

observer would be struck by the fact that in these while the children physically disable the mother. It
homes something is wrong. These are families in is hard to conceive of such a group; it is even more
disarray. I assume that during any given year, one difficult to think of it as a “family.” Here, anarchy
or more family members are at risk for being beat- reigns and coercion is king.
en. As yet there are no data that directly test this The Anarchy progression dramatizes what we
hypothesis. see clinically. It may be that these structural
The next step toward anarchy is exemplified by changes are indeed accompanied by increasing risk
the abused-child sample studied in Reid et al. for physical assault. That remains to be seen. The
(1981).1 As shown in Figure 12.1, all dyads are question is why do adults allow this to happen to
now committed to coercive conflicts. The Reid them?
analysis showed that even the mother and father
Mothers as a Storm Center
are significantly engaged in open warfare. The
parents in these families were observed to hit two As noted earlier, young children are totally com-
to four times more often than were parents of mitted to maximizing short-term payoffs. If left to
Normals. By definition, at least one child in each their own devices, a cluster of siblings would
family was being beaten. No data are given for the quickly shape each other to very high levels of co-
number of beaten wives in this sample, but, clini- erciveness. The effectiveness with which the care-
cally, it seems that it would be a sizeable number. taker practices child management skills determines
Even in these families anarchy is not complete. I the level at which antisocial behaviors will be per-
think in most of the families, if the father were formed. The caretaker’s selectivity in categorizing
present, he would not permit the children to physi- behaviors as deviant determines which antisocial
cally injure the mother, nor would the mother be behaviors will be permitted and which will be pun-
an idle spectator to attacks upon the father. How- ished. As noted in Chapter 11, there seem to be
ever, there are now appearing in the literature re- several paths to antisocial behavior. One involved
ports of attacks in which the father sits idly by the adult (unattached to the caretaker role) who

275,
 Table 12.1
Comparisons of Mothers from Three Samples

Mean rate per minute for mothers of:

Social
Code Stealers Aggressors Normals Duncan Planned
Categories (N=37) (N =34) (N = 36) F Values Comparisons

Approve .0S5 .068 .097 BASS)

Attend .709 .670 .786 .64

Command .413 .423 .280 A OS ST & SA>N

Command Negative 059 .044 -O11 4.07** ST & SA>N

Comply .028 .028 .010 Sa ST & SA>N

Disapprove 298 =o3 131 12D ee ST & SA>N \

Laugh .096 .108 .185 fey 24Oiae N>ST & SA

Noncomply .019 .012 .007 SIPs ST>N

Negativism .007 .008 .002 EH,

Normative 1.718 1.409 .320 129

No Response mloi4 087 .075 SESSage ST>SA & N

Physical Negative 015 .029 .004 2.90

Physical Postive .070 065 .090 3

Talk 2/93 3.046 3.586 4.09* N>ST

Tease O11 .001 .003 Tis

Touch .011 022 .019 .64

Work 3.518 3.213 3.490 Foul

Yell .023 .006 .001 4.14* ST>SA & N

*p<.05
erp.)
tp O01

selectively categorizes coercion within the family these differences are presented in Table 12.1. Gen-
as deviant but not acts committed outside of the erally, the mothers of problem children were more
caretaker’s presence. As John Reid put it, “Out of coercive (more Command, Command Negative,
sight, out of mind.” These parents punished coer- Disapprove, and Noncomply) and distant (more
cion within the family but permitted stealing, fire- No Response, less Talk and Laugh) than were
setting, and vandalism as long as it did not occur mothers of Normals.
in their presence. It is the second path that will be At this point we encounter a mystery. It has to
of concern for most of the following discussion. do with the fact that the parents of Social Aggres-
On the second path, the irritable, unskilled sors are highly selective about which child the co-
caretaker (usually the mother) could simply func- ercive involvement is focused upon! In one analy-
tion as a spectator to sibling-initiated chaos, but, sis, the contribution of the problem child was re-
typically, this is not the case. It is much more likely moved from the TAB scores for mothers and also
that the mother becomes deeply enmeshed in this for fathers and siblings (Patterson, 1981a). When
process. She and/or the father are directly in- this was done, there was little real difference be-
volved in 70% to 80% of all episodes involving tween parents of normal children and parents of
the target child (see Table 11.6). The percentages Social Aggressors. For mothers, the mean likeli-
were in the 90’s for episodes involving a younger hood of coercive behavior was .050 and .051 re-
problem child reacting to aversive and/or neutral spectively. The comparable values for fathers were
antecedents. .030 and .042. Given that the contribution of the
It would follow, then, that mothers of problem problem child was included, then the means for
children should differ in their general interaction mothers were .049 for Normals and .065 for So-
style from mothers of Normals. The details of cial Aggressors. This increase over the prior value

276
 Table 12.2
Mean Irritability Scores for Three Samples of Mothers

Samples
Normals SocialAggressors Stealers
Irritability Variables (N = 37) (N = 37) (N =38) F Values

Mother’s Reaction to Target Child

Crossover .021 .047 041 11.00***

Counterattack 150 .228 .164 4.12*
Punishment Acceleration .288 .406 .299 2.45
Continuance 122 .266 217 15.00***

Target Child’s Reaction to Mother

Crossover .007 .040 .027 SOMES
Counterattack .093 274 AGE 20.82%***
PunishmentAcceleration .148 .416 2337, 11.64***
Continuance 225 naval 238 13.86***
sp s.05
***5<.001
Fir pa.0001

was significant at p<.001. The comparable differ from one child to the next? These are crucial
means for fathers were .033 and .068 respectively. questions to be answered by the next round of re-
The increase for fathers of Social Aggressors was search on coercion processes. They relate to the
also significant at p< .001. It was true that, with more general issue of how it is that one member of
or without the contribution of the problem child, what is essentially a deviant system receives the la-
the siblings in socially aggressive families were sig- bel “deviant,” while the others do not.
nificantly more coercive than were siblings of Nor-
mals. Caretaker Irritability
It seems, then, that there is something “special” The hypothesis is that mother-Irritability is a
about the identified problem child. Why is it that key variable for the understanding of children’s so-
the parents look deviant primarily as a function of cial aggression. Mothers provide the reactions that
their inept handling of this child? They seem to be are crucial to the maintenance of high levels of co-
able to control the coercive initiations directed at ercive performance. The data that served as the
them by siblings; why don’t they cope equally well base to test this hypothesis were drawn from sam-
with the problem child? The parents permit sib- ples of Normals, Social Aggressors, and Stealers
lings to be coercive with each other but not with (see Appendix 6.2 for details). The mean values
them. The point is that parents of Social Aggres- for each irritability variable are summarized in
sors are highly selective about their areas of in- Table 12.2. The data for mother-Crossover (given
competence. Why is this? In his review of the de- an interaction with the problem child) showed sig-
velopmental literature, Bell (1968) noted very low nificantly higher values for the clinical samples
correlations (in the .20’s) between sibling reports than for the normal sample. Comparing values for
of their home environment. He also emphasized mothers to those for fathers and siblings (Table
the lack of maternal consistency in affectional be- 7.4) showed that in all samples there was a trend
haviors across children. To these trends we would for parents to initiate more conflict with the prob-
add that there may also be considerable imbal- lem child than did siblings. Examination of the
ances in the irritability of parents toward children, child’s disposition to start up conflicts with the
particularly in distressed families. Do these par- mother showed a significantly greater effect for
ents have different deviancy criterion thresholds the clinical samples than for the normal sample. In
for the problem child than for the other children? the clinical samples the child was more likely to
Does the likelihood for parental irritable reactions start fights with the mother than with any other

277
 Table 12.3
Irritable Reactions as Covariants for Child Coercive Performance Level (TAB)

Irritable Reactions to Problem Child

Punishment
Member Crossover Counterattack Acceleration Continuance
of dyad N rxy Bt rxy B fey B iy B Multiple RFValues

Mother 93 ALPS FA8 Pal -—.13 Gly = 22) 465 ee S58 .506 7 AL***
Father 68 Ba
by -—.12 14 —.27 .22 = (2) 42 6G .471 3.41*
tThese values are standard partial-regression coefficients. The asterisks describe the level of significance for the tests of significance
for the beta.
05
7 p= Ol
py < 001

family member. mothers of antisocial children were overly inclu-

The mothers of antisocial children were signifi- sive in labeling deviant child behavior. On the la-
cantly more likely than those of Normals to pun- boratory task they were significantly lower in their
ish coercive child behavior (counterattack). The threshold for confronting and punishing the child.
Duncan Multiple Range Test showed that this ef- Their criterion threshold measure for Punish cor-
fect was primarily due to the contribution of related .42 (p<.05) with their likelihood of pun-
mothers of Social Aggressors. There did not seem ishing deviant child behavior in the home. These
to be major differences among family members in are important findings. While we are just begin-
their disposition to counterattack when coping ning to explore what they mean, they do suggest a
with the problem child (Table 7.6). relation between mothers’ cognition about devian-
It is of particular interest to note that two labor- cy and punishment and their observed behavior.
atory studies have now demonstrated a relation As shown in Table 12.2, mothers of antisocial
between the mothers’ likelihood of counterattack children were not significantly different from Nor-
in the home and their system of categorizing child mals in their disposition to accelerate when pun-
deviancy (Lorber, 1981; Littman, Freund, & ished by their child. Similar comparisons for sib-
Schmaling, in preparation). In comparing 24 lings and fathers (Table 6.8) also showed nonsig-
mothers of Normals to 24 mothers of antisocial nificant differences. However, the problem child
children, Lorber (1981) found the latter were was significantly more likely than the normal child
more inclusive in differentiating negative from to accelerate when punished by his mother.? As
positive behavior; acts classed as normal by nor- shown in Table 6.8, he was no more likely to ac-
mal mothers were perceived as deviant by mothers celerate in his interactions with mothers than with
in the clinical sample. Measures of the laboratory fathers and siblings.
tracking variables were also shown to correlate The mother’s disposition to continue in her irri-
significantly with the home observation data mea- table reactions regardless of the child’s reaction
suring the likelihood of mother-Punish given coer- was significantly greater for the antisocial samples
cive child behavior. Mothers who tended to be than for the normal sample. The comparable dis-
overinclusive in categorizing were also more likely position for the child was also significantly greater
to punish. The second study employed a different for the clinical samples than for the normal sam-
series of videotaped family interaction for a sam- ple. The Duncan Multiple Range Test showed
ple of 27 mothers of Normals and 27 mothers of each sample to be significantly different from the
problem children. These parents were first asked other two. It comes as little surprise to find that
to categorize the child behavior; on their second this disposition to continue is highly correlated for
task they were asked to indicate when they would the mother and the problem child. As shown in
punish child behaviors (Littman et al., in prepara- Table 8.7, the correlation was .58 (df=91,
tion). The findings agreed with those from the p<.001). As shown in Table 12.3, mother-Con-
Lorber (1981) study. As compared to Normals, tinuance (given her interaction with the problem

278
 Figure 12.2
Caretaker’s Irritability, Problem-Solving Skills, and Antisocial Child Behavior

Caretaker’s Problem-Solving
Skills Reduced
(N=56)
Crises Clarity of Problem Definition
Impinging Increase in Caretaker’s Extent of Resolution
on Irritability During eee
Caretaker Problem Solving Quality of Solution
(N = 50) eooe
Discussion was Chaotic

Target Peer
Child Nomination
TABScore | as Aggressive
(N=85) (N=91)
Increase in
Antisocial Child Behavior

*p<.05
Pep 200
***pb<.001

child) accounted for a major portion of the vari- ior, she natters and threatens. I believe that, in
ance in the child’s baseline TAB score. The father’s part, families train caretakers to be irritable. Irri-
disposition for continuance was also the major tability becomes a mode for dealing with children
variable correlating with the child’s performance and others as well. I suspect that for some mothers
level. Interestingly enough, this was not the key irritability is generalized to persons outside the
variable for siblings; irritable reactions by siblings family; but, for present purposes, the key idea
accounted for only half as much variance in prob- concerns the extent to which it generalizes across
lem child performance level as irritable reactions settings within the home. Again, I suspect that for
by parents (Patterson, 1981a). The fact that par- mothers of antisocial children (Social Aggressors
ent-Continuance also served as a major variable in and Mixed), irritability becomes highly general-
accounting for variance in Parent Daily Report ized. Thus far we have only two tests for the hy-
measures of child deviancy further emphasizes the pothesis.
importance of this particular measure of irritabili- Reid et al. (1981) found that in families of
ty (Patterson, 1981b). abused children the observed conflict between par-
The mother’s disposition to react irritably is ents was several times higher than that for Nor-
viewed as an alternative to problem solving. She mals. Forgatch and Wieder (1981) constructed a
reacts in such a way as to make the pain stop im- code system to categorize videotaped familial
mediately; as noted in Chapter 7, she is the victim problem solving in a laboratory setting. The con-
of a reinforcement trap of her own making. Rath- struct, mother-Irritability, was shown to correlate
er than train the child, she scolds. Rather than with an impressive array of criteria. As shown in
confront the child and change his deviant behav- Figure 12.2, irritable mothers reported themselves

272
 _ Figure 12.3
A Feedback Loop for Caretaker Depression

Increase in
Caretaker’s Feelings
of Depression and
Helplessness

Change in Ratio of Increasingly Negative

Positive Events to Attributions about
Aversive Events Self and Others

Increase in
Increase in
Coercive Family
Parental Irritability
Interactions

to have high frequencies of daily hassles. The ob- tion to be irritable or to be delinquent, place the
servers’ global ratings showed that families with child at risk for antisocial problems. In keeping
irritable mothers tended to be less effective in solv- with our interactional stance, we also suspect that
ing their problems. Finally, the coded measure of some parental dispositions are associated with
mother-Irritability correlated with two different their membership in a coercive system. The hy-
criterion measures of antisocial child behavior. pothesis is that prolonged immersion in coercive
This study demonstrates that the maternal dis- interchange places the individual at risk for de-
position to react irritably interferes, not only with pression. Constant conflict plus the lean schedules
her child management efforts, but also with famil- of positive reinforcement provided for mothers in
ial problem solving. It does seem to be a general- these families place them at grave risk for depres-
ized coping response that may serve some short- sion. For parents of antisocial children, the ratio
term purpose, but, in the long run, I think it in- of positive to aversive events is very low. In part, it
creases the likelihood that more serious child man- is a lack of parental skill that determines this ratio
agement problems will occur. The unsolved crises of positive to aversive experience. However, once
will be added to the accumulation of past failures. the imbalance (with its concomitant feelings of de-
I think, too, that this irritability reflects the pres- pression) occurs, there are two additional out-
ence of internal processes. The irritable mother comes. First, it is thought that there are increases
sees herself differently than do other mothers. She in the likelihood of parental irritable reactions ac-
feels angry about her family and the uncontrolla- companied by an increase in negative attributions
ble situation in which she finds herself. The fol- about other family members. As shown in Figure
lowing section explores these hypotheses. 12.3, both increases feed back into family proc-
esses to produce further increases in levels of aver-
Erosion of the Caretaker’s Mood sive events. In effect, depressed reactions are
Parental personality traits, such as the disposi- viewed as the outcome of a process. Here, the de-

280
pressed reaction feeds the process to further exac- found that maternal measures of depression were
erbate an already bad situation. significantly higher for families referred for treat-
ment, as compared to nonclinical families (Griest,
Impact of Crises upon Mood Wells, & Forehand, 1979). In that study, mater-
As a general case, it is assumed that caretaker nal depression correlated with measures of child
skills partially determine the intensity and fre- deviancy; mothers of deviant children were more
quency of crises occurring within the family, e.g., likely to describe themselves as depressed.
conflicts among children or conflicts between par- Which variable was the cause, and which was
ents. The data from a recently completed OSLC the outcome? Does mother depression cause anti-
study showed that in normal families the mean fre- social child behavior, or are both variables an out-
quency of crises reported by the parents covaried come of the same process? There is only one study
with the child’s level of coerciveness. The greater that relates to this issue, a short-term longitudinal
the number of crises, the higher the child’s TAB study of families during separation and divorce
score. For the combined sample, the correlation (Hetherington et al., 1980). During the separa-
was .31 (df=84, p<.01). The correlations were tion, both parent ratings and observation data
higher for 4th graders (.42) than for 7th graders agreed in identifying increasing out-of-control be-
(.34). The family’s problem-solving skills correlate havior for the children, particularly for boys. The
with their ability to anticipate crises and to con- aggression peaked at one year after separation and
tain the impact of those crises that do arise. The then dropped. Ratings of mood from the mothers’
general level of social skills also determines the diaries showed significant correlations with rat-
amount of positive reinforcement from leisure ac- ings of child aggression (from that same source)
tivities, friends, and so forth. In this formulation, and mothers’ checklisting of child aggression. Of
then, social-competency variables play a key role. most interest were the cross-lag correlations that
The literature reviewed in Chapter 4 detailed showed that child aggression early on correlated
the frequency with which aversive events occur in with the later measures of the mothers’ mood. The
normal family life and the higher rates that were findings suggested that early disruptions in child
obtained for families of antisocial children. The behaviors were causally related to mothers’ later
question is, do these aversive events covary with depressed reaction. However, these correlations
mood shifts? The data from the extended baseline lend themselves equally well to some alternative
studies presented in Table 4.3 showed such a co- explanations, so these findings are not compelling.
variation for three of the five mothers. Days char- Even so, it seems reasonable to assume that mater-
acterized by many crises were also days in which nal depression may be produced by child antiso-
the mother described herself as more depressed. cial behavior.
The Hetherington, Cox, and Cox (1980) study of Traditionally, it has been assumed that dis-
families getting a divorce also provided data that turbed children have disturbed parents. For exam-
relate to this point. During the early stages of the ple, Rutter (1966) found that about one child in
separation, the mothers reported in an interview five with a behavior problem had parents with
that they felt overwhelmed and disorganized by psychiatric problems. This was significantly great-
the demands placed upon them asa result of the er than the comparable percentage obtained from
separation. The interviewer ratings of economic a matched control group. On the MMPI clinical
stress correlated significantly with the mothers’ re- scales, however, parents of children seen at child
ports on the Rotter I-E scale of increased sense of guidance clinics did not score as high as parents
helplessness, i.e., things being out of their control. who were actually psychiatric patients (McAdoo
The stress ratings also correlated with a depres- & Connally, 1975). Goodstein and Rowley
sion score obtained from ratings of mothers’ dia- (1961) compared MMPI profiles from parents of
ries during this same interval. Normals to small groups of parents who had chil-
The hypothesis presented earlier emphasized the dren referred for treatment with four different
imbalanced ratio of aversive to positive events for types of problems. The mothers of antisocial chil-
caretakers in families of antisocial children. If this dren tended to score significantly higher on de-
is true, then one might expect these mothers to be pression, hysteria, and psychopathic deviate scales
more depressed than mothers of Normals. OSLC than did mothers of children with personality trait
self-report data (MMPI) showed that mothers of and/or neurotic problems. Wolking, Dunteman,
antisocial children were elevated about one stan- and Bailey (1967) compared MMPI profiles for
dard deviation above the mean normal level on the parents of six different groups of problem children
depression scale. There are also a series of studies referred for treatment. The mothers of antisocial
by Rex Forehand relating to these issues. He, too, children scored higher than mothers in the other

281
 Table 12.4
MMPI Data for Several Samples of Mothers

OSLC Samples Samples from Other Settings

Social Social
MMPI Aggressors Stealers Normals F Tests?
Normals? Normals* Aggressors‘
Scale! (N=25) (N =34) (N=20) tii(df=2, 78) Duncan (N = 50) (N =29)

IE SUES 48.57 50.15 1.03 — — 50

F 57.00 56.57 50.60 2.20 — -- 50
K 54.34 49.20 57.05 SYeSe
la! SA & N>ST — — by
Hs Sess) 54.83 47.10 8.475 SA & ST>N 53: 52 Sy?
D 57.54 58.83 SIE 7S Sy ST>N 51 by 57 |
Hy 5227.3 58.54 58.90 1.10 56 56 59

Pd 63.92 61.49 39-89 3.94 SA & ST>N 54 ois) 56

Mf 45.04 45.49 43.20 84 49 49 47

Pa 62.73 56.43 52.45 SeSOr a SA>ST & N 52 54 56

Pt 57.50 56.54 50.90 Sets SA & ST>N SSy/ 51 538)

Sc 59.31 56.69 52.10 3. 46m SA & ST>N 54 bp) 54

Ma 56.08 56.23 49.10 3.84* SA & ST>N 49 48 45

Si S723 Seal 51.90 314 _ — 51

1AIl scales are K corrected.

?The F Tests were based upon data for raw scores.

3Based upon 50 nonproblem families matched with families referred to the Child Psychiatry Service at lowa Psychopathic Hospital
(Goodstein & Rowley, 1961).
‘Based upon a sample of nonproblem families matched with a sample referred to a child outpatient clinic for treatment (Liverant,
1959).
‘Data from 29 families referred for treatment because of an aggressive child (Anderson, 1969).
*p<.05
ep <.01
aap —2001

groups on hysteria and psychopathic deviate scales ies. In fact, if one converts the raw scores on the
and lower on the masculinity-femininity scale. scales to T-scores (with K corrected), it is apparent
The study by Anderson (1969) compared mothers that this is the peak score on the MMPI profiles
of normal and neurotic children to mothers of an- for mothers in the OSLC sample. The implication
tisocial children. The mothers of antisocial chil- is that these are essentially angry, nonconforming
dren had significantly lower scores on the mascu- women. That is extremely interesting because of
linity-femininity scale. the key role the mother’s general irritability plays
The data in Table 12.4 summarize the findings in determining the child’s level of coercive perfor-
for OSLC samples of mothers of 26 Social Aggres- mance. The next step will be to determine if this
sors, 35 Stealers, and 20 Normals (see Appendix scale and the depression scores covary with mea-
6.2 for description of samples). The results are sures of mother irritability.
consistent with the earlier analyses (Patterson, It was of some interest to compare the MMPI
1976, 1980a). Mothers in the clinical samples (as scores for mothers of Stealers to the scores for
compared to mothers of normal children) showed mothers of Social Aggressors. Mothers of Stealers
consistent elevations on almost all of the clinical scored significantly lower on the K and Pa scales,
scales, including depression. but other than this, the differences between moth-
The findings that the clinical samples scored sig- ers in the two clinical samples were slight. The fact
nificantly higher on the psychopathic deviate scale that the two groups scored equally high on the de-
replicates the finding from both the Goodstein and pression scale did not confirm an earlier finding
Rowley (1961) and the Wolking et al. (1967) stud- for treated samples (Patterson, 1980a).

282
 significant when tested in three different across-
Mood, Attribution, and Family subject analyses (Johnson & Lobitz, 1974b; Pat-
Management terson, 1981a; Forehand et al., 1980). It seems
‘The assumption is that the depressed mood of that the intradyad format, with its extended base-
the caretaker may serve as a determinant for three line design, may be a more appropriate design
components that further contribute to the disrup- when investigating the impact of caretaker mood,
tion of the family. First, there is an increased likeli- crises, and so on upon child management practices
hood of negative attributions about other family (Patterson, 1981b).
members, particularly the problem child. The There are no studies directly showing that on
mother begins to perceive her child as more devi- days when the mother is more depressed she is also
ant than he really is. Second (and this may be re- less effective in her practice of family management
lated to the first), she becomes increasingly more skills. If that mood was shown to covary with co-
likely to be irritable in her reactions to others. ercive child behavior (Table 10.1), one might
Again, this is particularly true for her interactions assume that family management practices were
with the problem child. Finally, she is likely to be- disrupted, but that, too, remains to be demon-
come increasingly disrupted in her practice of fam- strated.
ily management skills, which in turn relates to a
commensurate increase in the child’s antisocial be- Increased Family Management Skills
havior. The material to be reviewed in this section Produce Change in Depression
relates directly or indirectly to these three hypoth- It is assumed that the inept practice of family
eses. management skills produces increases in both cri-
It is assumed that depression is accompanied by ses and family conflicts, and that these, in turn,
increased negative attributions, not only about produce an increase in caretaker depression. If this
self, but about others as well. At this point, the is true, then successfully training the caretaker in
caretaker begins to see other family members as the application of family management skills
“not caring.” She might even perceive her problem should reduce family conflicts, crises, and the
child as intending to make her feel bad. These clin- caretaker’s depression. The treatment program at
ical impressions are supported by the interview OSLC is focused upon the task of teaching parent-
data from the Weissman and Paykel (1974) study. ing skills to members of families of antisocial chil-
Depressed women reported themselves to be gen- dren (Patterson, Reid, Jones, & Conger, 1975).
erally more hostile to others than nondepressed The treatment outcome data reviewed in Chapter
women. The Novaco (1975) studies also showed a 13 demonstrate that the parents have been suc-
covariation between low self-esteem and the likeli- cessful in applying these skills; i.e., the deviancy
hood of attack and negative attributions about the levels are significantly reduced for almost all fami-
victim. In the Forehand, Wells, and Griest (1980) ly members. I assume that the dramatic reduction
study of oppositional preschool children, the in familial coercion, per se, is sufficient to produce
mother’s self-report score on depression correlated a significant reduction in caretaker depression. |
much better with her perceptions of child deviancy think the frequency of crises is also reduced, but as
than it did with observed child deviancy. This is in yet there has been no test of this idea.
keeping with Chamberlain’s analysis of OSLC To test this hypothesis, the MMPI questionnaire
data (in preparation). In a replicated design, she was administered at baseline and again at termina-
demonstrated that more depressed mothers re- tion for a small sample of mothers of Stealers and
ported higher frequencies of deviant child behav- mothers of Social Aggressors (Patterson, 1980b).
ior (Parent Daily Report). The comparison showed a significant F value for
The second hypothesis is that increases in the trials for the decrease in the F scale (F=10.53,
mother’s dysphoric moods are associated with in- p<.003) and for the increase in the K scale (F=
creases in her irritable reactions to the child. Thus 7.98, p< .009). On the clinical scales the decrease
far, this hypothesis has been tested in only one pi- in depression (F=3.41, p<.076) was of border-
lot study in which the covariation held for some line significance, with a highly significant reduc-
dyads but not for others. The findings (Table 10.1 tion in social introversion (F=13.40, p<.001).
and Patterson, 1981b) are suggestive but hardly For both of the latter the bulk of the decrease was
compelling. contributed by mothers in the Social Aggressor
Incidentally, mothers who are more depressed sample. My confidence in these findings was con-
are not necessarily more coercive with their chil- siderably bolstered by the Forehand et al. (1980)
dren. The correlation of mother-depression scores replication. Fifteen mothers of oppositional pre-
and her observed rates of coerciveness were not school children were trained by student therapists

283
to bring the child’s noncompliance within normal her ability to rear competent boys and girls. The
range. This was accompanied bya significant re- function of social support systems for the single
duction in the mothers’ depression scores. The parent should be to help her or him to perform
parents’ perception of the children also became well both instrument and expressive functions.
significantly more positive. A more positive per- ...” (Baumrind, 1979, p. 15)
ception of the problem child has also been a con-
sistent outcome of social learning approaches to Baumrind, in her lucid style, then goes on to
parent training (Patterson & Fleischman, 1979). emphasize the importance of what she labels as
The findings from these studies point to an in- “symmetric” or shared child care. Chodorow
teresting difference between coercion theory and (1978) also calls for “equal parenting.” Both of
traditional formulations. Most clinicians writing these writers speak from a feminist position; how-
about child guidance clinic populations have ever, neither would disagree that at least one per-
noted the high incidence of anxiety, depression, son should be committed to the role of caretaker.
and somatic complaints reported by mothers of There seems to be a general consensus in the litera-
problem children. They assumed that these mater- ture that the commitment does not have to be
nal traits were the cause of the child’s problems made by the biological mother (as was once
and advocated therapy for the mother’s neurosis thought to be the case). Rather, it may be a shared
as a means of helping the child. For the mothers of role. It may also be assumed by professional care-
antisocial boys, we can see that these maternal takers, as demonstrated in the Israeli Kibbutz sys-
problems may be the result of, rather than the tem. There a trained adult cared for a small group
cause for, the process. This being the case, it of infants and children, who were then returned to
makes sense to teach her how to manage a difficult their working parents late each afternoon. These
child as one means of reducing her depression. children received warmth and affection from
Mothers of Social Aggressors are unskilled, irri- teachers, peers, and parents alike. They also re-
table, and mildly depressed people. The irritability ceived skills training and supervision from both
and lack of skill seem to be highly generalized, at parents and supervisors. The studies of these ex-
least within family settings. However, she is often periments have consistently shown no ill effects re-
not the only adult present in these families. What, sulting from sharing the caretaker role among a
then, is the role of the father in this process? group of committed adults.
However, whether or not the caretaker role is
The Father’s Role in the shared, those who serve in this capacity do so only
Support System at a very high cost to themselves. The details of
this cost have been presented in Chapter 4 and at
For a preschool child, there is a need for the
other points throughout this volume. Even with a
continuous involvement of a caretaker. As noted
normal infant or preschool child, the cost to the
in Chapter 4, children of this age require a com-
caretaker is high. Given a difficult child plus a dif-
mand or request from the parent at least once ev-
ficult marital and/or economic situation, then the
ery three minutes. Given that the caretaker does
cost may be overwhelming. By itself, there is little
more than just tell a child to stop doing some-
about the caretaker role, per se, that generates
thing, or to do something he has not done, then
positive reinforcers. It is, rather, the function of
the demands upon caretaker time increase com-
the support system to provide positive experiences
mensurately. For example, teaching simple self-
that will keep the ratio of aversive to positive ex-
help skills requires hundreds of training trials. De-
periences in balance. I believe that the father
viant child behavior can and does increase very
makes his primary contribution to family manage-
quickly in rate. These increases (trial runs) must
ment by providing a support system. The caretak-
be monitored and dealt with. The child must be
er (usually the mother) is primarily responsible for
monitored in terms of his chores, self-help skills,
the child and family management. The father pro-
academic progress, and relationship skills with
vides the back up (or support) necessary to main-
peers and family. If neither parent functions as a
tain the skillful implementation of these manage-
caretaker, then the child begins his drift toward
ment practices.
both deficits in prosocial behavior and increases in
The caretaker support system can be put togeth-
antisocial behavior.
er in a number of ways. It may include a spouse,
In some families the caretaker task is shared; in
friends, relatives, or neighbors. However it is con-
most instances the role is occupied by the mother.
structed, these are the people who provide outside
“In the case of the single mother, whatever helps contact, almost on a daily basis. It is these people
her to be agentic as well as nurturant will facilitate who support the caretaker in their daily telephone

284
calls, the coffee session, and in the quiet talk after participated in laboratory tasks measuring the rel-
dinner. | think the spouse can play a key role here, ative dominance of the mother, father, and child.
partly because he usually has more time to invest Regardless of social class, child dominance of ei-
than friends and relatives (or agency workers)! At ther parent was associated with less perceived
this point, I am not referring only to his reinforc- marital happiness. The negative correlations were
ing the caretaker for child management skills. Iam higher for middle- than for lower-class families
referring to his role in bringing friends over, and in and were also higher when the mother was domi-
arranging pleasant weekend activities. This indi- nated by the child than when the father was domi-
rect support serves the important function of re- nated. I think that in most normal families the
dressing the imbalanced ratio brought about by children perceive the alliance between the caretak-
the presence of children and daily crises. Who is it er and the father as being more powerful than any
that plans the pleasant activities for single-parent coalition of children within the family. Sibling co-
families living in isolation from relatives? alitions generally become effective only when the
The second function that can be served by the alliance between parents is disrupted. When the
father is that he can take direct responsibility for parents are in conflict, sibling-based coalitions
some aspects of child management. When he is have an increased likelihood of controlling the
home, and a sibling conflict arises, he can moni- family.
tor, set consequences, and solve problems. He can Aside from the psychological effect of having
also question the adolescent as to his/her where- the father present and in agreement, he is often an
abouts and negotiate the appropriate time for a re- important contributor as a strategist in handling
turn in the evening. Most fathers can and do carry crises, as well as child management problems.
out some adjunct child management activities, but Most young parents find it necessary to spend
the actual amount of time probably varies with the hours checking strategies with other parents, rela-
age of the child, social class, and employment sta- tives, and friends. “What do you do when... .”
tus of the father. It is probably the case that even The popularity of how-to manuals suggests the
the most committed father actually handles only a necessity for this kind of brainstorming of alterna-
small fraction of the daily round of child manage- tive ways to manage children. Who brainstorms
ment problems. For example, the Fagot (1974) alternative coping strategies with the adolescent
study of normal, middle-class toddlers showed single parent living in isolation?
that roughtly 70% of child care was carried out by
mothers and only 30% by fathers. It is my impres- Disruptions in Parental Alliance
sion that in distressed families the fathers are even The hypothesis is that disruptions in the care-
less involved than this. In a later section of this taker’s support system are likely to produce two
chapter we will examine the research findings re- outcomes. First, the impact of crises will be en-
lating to differences in interactive style between hanced and perhaps lead to depression. Second,
mothers and fathers. the caretaker is placed at risk for a disruption in
The third function served by the spouse, or oth- her skillful performance of family management
er members of the support system, is to provide a practices; this, in turn, produces an increase in an-
reserve force when the caretaker confronts the tisocial child behavior.
children on some crucial issue. If the caretaker sets A study by the anthropologists Goody and
up a house rule or negotiates an agreement, then it Groothues (1979) provides data in keeping with
should be the case that the spouse supports and, if the first hypothesis. Field notes were collected on
possible, monitors and enforces it. In this fashion 20 West African couples working in London as to
the parents present a united front. Their cohesive the frequency and intensity of crises. The couples
stance communicates the message that in this were also categorized as to whether the spouses
household the parents are in control. jointly shared in caretaking responsibilities or
Most children (rightly or wrongly) perceive whether their roles were segregated in relation to
their fathers to be the more powerful parental fig- domestic chores, finances, leisure, parental roles,
ure. For example, a survey study by Kagan (1956) and future planning. The analysis showed that if
showed that 70% of elementary-school-aged girls the couples retained their traditional segregated
and 83% of boys said they were more afraid of roles, they were more likely to be rated as under
their fathers than their mothers. Roughly 60% re- high stress. A jointly shared role structure, on the
ported that if they were bad, they were most likely other hand, was related to lower stress. These
to be punished by their fathers. Kolb and Straus findings are in keeping with the study of families
(1974) obtained ratings from the children as to involved in divorce (Hetherington et al., 1976,
their parents’ marital happiness. Then the families 1980). The separation produced a marked in-

285
crease in stress reported by both husbands and They do, however, run contrary to some ideas that
wives. Each reported difficulty in coping with the are currently popular, e.g., marital fighting is in-
functions and duties previously fulfilled by the evitable, necessary, and desirable (Charny, 1969).
other person. Incidentally, in the 1980 report, On the contrary, severe and repetitive marital dis-
Hetherington showed that for both intact and di- cord is indicative of a disintegrating family struc-
vorced families there was less deviant child behav- ture. The fights may serve a short-term cathartic
ior when the parents agreed upon child discipline function for the spouses, but in the long run it
than when they disagreed; i.e., when the alliance means a disruption in the support system required
regarding child management was still intact. by the caretaker and commensurate increases in
If there is severe marital conflict, the support child deviancy.
system is likely to be disrupted. This does not hap-
pen for all couples in conflict; there are undoubt- Empirical Definitions of the Father’s Role
edly some who continue to cope with crises and It was stated earlier that fathers served a dual
family management problems even though they role in maintaining the family as a system. One
are in severe conflict over some aspect of their re- role is direct and the other is indirect. When the fa-
lationship. However, I think that for most parents ther is present, it would be expected that he would
the effects of their conflict generalize and reduce carry out some child management responsibilities,
their efficiency in the performance of child and as well as support the caretaker in her efforts. Are
family management practices. As yet there is no the observed behaviors of fathers discernibly dif-
study that directly shows a breakdown in family ferent from those of mothers? Are the interactions
management skills during marital conflict, but of fathers in clinical families different from those
Rutter et al. (1976) cites several studies showing a of fathers in normal families?
relation between marital discord and children’s con- The data in Table 12.5 summarize the compari-
duct problems. Similarly, the review by Oltmanns, sons for two samples that would tend to show
Broderick, and O’Leary (1977) showed that, in maximum differences in parental behaviors.
general, parents of children referred for treatment Mothers and fathers were compared for samples
were less satisfied with their marriages. In their of intact Social Aggressor and intact normal fami-
study of cases referred for treatment, they found lies (see Appendix 6.2 for details). The 2 x 2 ana-
negative correlations in the —.3 to —.4 range be- lysis of variance was done separately for each code
tween parental ratings of child deviancy and mari- category. Six of the 29 comparisons were signifi-
tal satisfaction. Johnson and Lobitz (1974b) took cant. Fathers in both samples performed the fol-
the analysis one step further by providing indepen- lowing behaviors more often than did mothers:
dent measures of the two variables. The mothers Attend, Normative, and Tease. They were signifi-
and fathers of 36 young, antisocial children re- cantly less involved than mothers in the use of
ferred for treatment filled out the Locke-Wallace Command, Touch, and Work. As a pattern, these
measure of marital satisfaction. These scores were findings suggest the role of a somewhat playful
correlated with the referred child’s observed rate spectator. When the observers were present, the
of deviant behavior. The correlations showed that fathers tended to spend more time than mothers
marital disruptions covaried with increases in engaging in No Response (NR), or they quietly
child deviancy. watched (AT) what was going on. Fathers also
Burgess and Conger (1976) observed family in- Tease and Play more often; there was a borderline
teraction in the homes of normal and abused fami- trend for lower TAB scores. It may be stretching
lies. Among other things, they found that mothers things a bit to label the father as a resident “good
from the abuse sample complied with spouse re- guy,” but he does leave the major child manage-
quests only 20% of the time, as compared to 64% ment category (Command) up to the mother. It is
compliance for women in the control families. also the mother who engages in the necessary
One suspects marital conflict, but it was not di- housework, while the father Attends, or sits and
rectly measured in that study. In their comparison reads. The teasing may represent a kind of verbal
of Abused Child, Social Aggressor, and normal playing analogous to findings for fathers in infant
families, Reid et al. (1981) found that in families and toddler research. Both Lamb (1976) and
of abused children, the rates of parental coercive Clarke-Stewart (1978) observed fathers to be sig-
interchanges were shown to be 18 times higher nificantly more playful than mothers. It is interest-
than those found for either of the other samples! ing to note that in the latter study the fathers’ in-
These findings provide only indirect support for volvement in play and the duration of his interac-
the assumed relation between the disruption of tions with the toddler correlated with measures of
family management practices and marital discord. infant IQ at the age of 3.

286
 Table 12.5
A Comparison of Parents from Normal and Clinical Samples

Mean Rate for Samples F Values by:

Social Aggressors Normals
Mother Father Mother Father
Code Categories (N= 16) (N= 16) (N = 26) (N = 26) Sample Parent

Approve -085 .041 stil .085 Sil 3.08

Attend .845 1.072 SL 1.029 .30 4.07*

Command .433 .264 .289 .190 ees DAdies

Command Negative .027 .020 .012 .017 1.14 .02

Comply .024 .020 011 .018 2.32 233

Disapprove .274 e2ou se NOH Le ee .28

Dependency .000 .000 .000 .001 1.65 1.02

Humiliate .007 .009 .002 .001 4.94* .00

Ignore .027 .030 .010 .003 2.02 83

Laugh Ng? .102 .193 .003 2292) DOU

Noncomply .007 .014 .007 -005 1.05 .07

Negativism .002 .004 .003 .004 .04 3)

Normative 13383 SEOS2 .243 2.645 2.18 SES

No Response 095 .087 .067 .076 1.61 .03

Play .587 1.190 .982 ie pil eps)

Physical Negative .007 .006 .004 .001 3 1.49

Physical Postive .086 .107 .104 .047 23 SHG

Receive .008 .012 .007 .014 .06 2.16

Self Stimulation .008 .033 .003 .008 3.40 2.38

Talk 32299 2.819 3.462 3.265 1.54 1.63

Tease .008 .032 .004 .013 2.24 4.23*

Touch .063 011 .026 011 1.66 3.995

Work 3.310 .716 3.667 1573 2.84 ADS

Yell .006 .007 .001 .000 5.46* .08

TAB .800 .637 452 342 18.02 SS 3.16

*p=.05
**p< 01
"42 p= .001

The data in Table 12.6 summarize the mean were a weak reflection of those that were obtained
rates of occurrence for each of the FICS categories for mothers. Fathers of Stealers were more distant
for three samples of fathers of Stealers, Social Ag- in contrast to the other two samples. Fathers of
gressors, and Normals (see Appendix 6.2). While Social Aggressors and Stealers were more coercive
there were only a few significant findings, the pat- than were fathers of Normals.
tern was similar to that which was obtained for a
similar comparison for mothers. Fathers in the Fathers’ Irritability
clinical samples tended to be more distant (NR), The data summarized in Table 12.7 suggest that
less talkative (TA), more aversive (DI), and less there are differences in the fathers’ irritability as
approving (AP) than were fathers of Normals. It they react to the target child for samples of both
should be emphasized that the findings defining normal and distressed families. The fathers in the
this pattern are few in number. This reduced set of clinical samples were signficantly more likely than
variables did successfully discriminate among the fathers of Normals to crossover and to continue.
groups in the discriminant-function analysis that Note, too, that the fathers were more likely than
appeared in Chapter 11. The significant functions the problem child to start up a conflict.

287
 Table 12.6
Comparisons of Fathers from Three Samples

Mean rate per minute for fathers of

Social
Stealers Aggressors Normals
Code Categories (N=25) (N=14) (N =26) F Values Duncan

Approve .047 .039 085 SACO N>ST & SA

Attend 927 1.040 1.030 alg
Command 220 LE .190 1.80
Command Negative .037 .021 .017 1.47
Comply .020 .020 .018 .04
Disapprove 165 2/2 .107 4.94** SA>ST&N |
Laugh .090 .100 .136 1eS
Noncomply .012 015 .005 hats:
Negativism -005 -004 .004 .09
Normative sheys) 3.309 2.645 1.34
No Response .136 .092 .077 eco ST>N
Physical Negative .008 .007 .001 Maps:
Physical Positive .039 ai Uihs) .047 2.46
Talk Beis: 2.852 3.265 S33 N>ST

Tease .010 .036 .014 2.18

Touch .006 .009 .011 34

Work 1.585 .686 1573 1.64

Yell .004 .007 .000 1.56
*p<.05
Ey DSGAU

iPaplestZie7,
Mean Irritability Scores for Three Samples of Fathers

Samples
Normals SocialAggressors Stealers
Irritability Variables (N =26) (N=14) (N=25) F Values
Father’s Reaction to Target Child
Crossover .013 .038 .022 6.90
Counterattack .078 143 .103 2.33
PunishmentAcceleration wo .496 .308 Gig!!!
Continuance 2lsfa .262 .203 9.04554
Target Child’s Reaction to Father
Crossover .004 .019 .016 1.58
Counterattack .079 .184 131 4.67*
PunishmentAcceleration .038 .414 244° 13.50mae
Continuance .049 283 .202 CAV oie
*p<.05
ap .01
"p< 001

288
 The multivariate analysis (Table 12.3) showed that fathers perceived fewer problems than did
that the fathers’ irritable reactions were signifi- mothers. On self-report questionnaires the fathers
cantly correlated with the child’s coercive perfor- were also less depressed, anxious, irritable, and
mance level. The multiple correlation between the unstable than were the mothers. The mothers, on
fathers’ p values for irritability and the children’s the other hand, described themselves as being se-
TAB scores was .471. The F value was 3.41 verely stressed; in fact, their self-statements pro-
(p<.05). While Father’s irritability adds some- duced profiles that were similar to those for moth-
thing to our understanding of the child’s perfor- ers of antisocial children. The findings are in keep-
mance level, it can be seen from Table 12.3 that ing with the hypotheses being presented here. The
the measures of fathers’ behavior account for less caretaker seems to be the one who bears the full
variance than the comparable measures of moth- brunt of familial stress. The father is only secon-
ers’ behavior. A further analysis showed that when darily involved. In addition, I suspect that the pos-
the measures of p(Continuance) for fathers and itive experiences from his work are sufficient to
mothers were combined, the measure of mother ir- keep his ratio of aversive to positive experiences in
ritability accounted for all of the unique variance fairly good balance.
(Patterson, 1981b). Do fathers of antisocial children describe them-
In terms of understanding the families of antiso- selves differently than fathers of nonproblem chil-
cial children, my general sense is that the informa- dren? Earlier studies suggested few differences in
tion about fathers’ interaction is only a weak re- MMPI scores for two small samples (Patterson,
flection of what we already know from analyses of 1976, 1980a). Table 12.8 summarizes the findings
the mothers’ data. This reiterates the theme that from OSLC samples of Social Aggressors and
the burden of the caretaker role falls primarily Stealers (Appendix 6.2). At this point, we have
upon the mother. What the father does in the way data for a sample of only eight normal fathers;
of child management is similar, but he does less of these data were not included here. For this reason,
it. His interactions with the children determine the OSLC clinical samples must be compared with
less of their behavior. The question that continual- the Goodstein and Rowley (1961) and Liverant
ly presents itself is, why don’t the fathers in clinical (1959) samples from fathers of nonproblem chil-
families perceive that the caretakers’ child man- dren. The OSLC samples were of modest size, but,
agement procedures are not working and intro- even so, none of the differences exceed .5 SD be-
duce new procedures? Why does he imitate the tween the sample of fathers of children referred as
techniques used by the mother when they are inef- Aggressors from the Anderson (1969) study, or
fective? the two OSLC clinical samples.
The most noteworthy finding was that none of
Fathers’ Self-Report the clinical samples differed markedly from either
As noted earlier, when the family is severely sample of fathers of normal children. Unlike the
stressed by conflicts from within or crises from mothers of antisocial children, fathers of antiso-
without, it is the caretaker who bears the brunt. cial children did not report themselves as being
The accompanying hypothesis is that fathers are more angry and depressed than normal parents.
usually only slightly affected by such stressors. The fathers of antisocial children, like the fathers
The feelings of anxiety, depression, anger, confu- of handicapped children, seem to have some kind
sion, and isolation that characterize caretakers in of acquired immunity that protects them from fa-
distressed families are not a part of the self-percep- milial stressors. Given that the family is disrupted,
tions of fathers. I think this is partially related to how does the father protect himself from the
the fact that they do not see family management as stress?
being a significant feature of their responsibility.
Earlier pilot studies showed that their self-percep- Single-Parent Families
tions were like those of fathers in nondistressed As already noted, in most families the function
families (Patterson, 1976, 1980a). of the father is to carry out some child manage-
The study by Tavormina, Boll, Dunn, Lus- ment practices and to provide a back up and sup-
comb, and Taylor (1975) was very revealing. The port system when needed. To this I would like to
sample involved physically handicapped children add one further function. I think in normal fami-
whose very presence engenders extreme stress for lies the father also serves as a corrective factor for
parents. However, in this case the child’s problem the caretaker when she begins to drift into inept or
was unlikely to have been produced directly by deviancy-producing management techniques. |
any psychopathology in the parents. Their careful think that in the normal course of events most of
study of physically handicapped children showed us drift toward the use of techniques that maxi-

289
 _ Table 12.8
MMPI Data for Several Samples of Fathers

Mean T scores for MeanT scoresfrom Other Samples

OSLC Samples!
Normals
Aggressors Stealers Aggressors Goodstein & Normals
MMPI Scale (N=12) (N=21)
Anderson (1969) Rowley (1961) Liverant (1959)

L 51.00 49.81 46 _ _—
F 53.08 53.43 53 — -
K 56.69 54.33 ay) — _—

Hs 54.46 53.19 54 53 SZ

D 56.62 54.62 56 ol! 53

Hy 59.77 57.43 56 56 56
Pd 60.54 60.00 64 55 56
Mf 59.00 58.88 54 58 57

Pa 57.00 52.05 53 Sy 52

Pt 58.62 52.05 54 52 52.

Sc 57.92 53.90 55 51 52

Ma 55.69 57.48 55 51 53

Si 48.50 54.29 52 _ —_

1These scores were K corrected.

mize short-term payoffs, i.e., we become more co- expected to be accompanied by increases in out-
ercive. It is the corrective impact of feedback from of-control behavior.
a spouse, friends, and relatives that keeps this drift The clinical literature has emphasized the rela-
in check. If the relation is disrupted, and/or the tion between broken homes, delinquency, and an-
spouse neglects this corrective function, then I tisocial behavior (see Chapter 2). For example,
think the caretaker becomes increasingly at risk to Rutter et al. (1970) found that 25% of the antiso-
drift into irritable reactions and to be inconsistent cial children studied resided in father-absent
in tracking and punishing antisocial behavior. By homes, as compared to a figure of 14.6% for
definition, single parents would be more at risk for matched controls. M. Hetherington engaged in a
such a drift. series of studies that further elucidate the process.
There are several aspects of our contemporary In her first study, she compared intact, father-ab-
culture that contribute to disruptions in caretaker sent-due-to-death, and _father-absent-due-to-di-
performance. The population is highly mobile, as vorce homes (Hetherington, 1972). Relatively
shown by the estimate that about one family in large samples were compared using an impressive
seven moves each year (Coates, 1978). Even if the variety of assessment procedures. The findings
move were only to another residence within the suggest that for girls, father-absent-due-to-divorce
same community, it would still mean a disruption was associated with greater loss of self-esteem,
in the support system normally provided by neigh- more heterosexual activity, and more conflict with
bors. A shift to another city would mean an even mothers. Many of these effects were exacerbated if
larger disruption in the support network of friends the separation had occurred earlier rather than lat-
and relatives. The trend for increases in the num- er in the girl’s development. In her next studies she
ber of single-parent families is even more alarm- compared samples of normal families with fami-
ing. As noted by Coates (1978), approximately lies in the process of separation and divorce (Heth-
45% of children born in 1976 will live with a sin- erington, Cox, & Cox, 1980). The observation
gle parent for some period of time. Earlier, the hy- data from both the home and the laboratory con-
pothesis was presented that a severe crisis such as curred in demonstrating an increase in child devi-
separation or divorce is itself associated with pro- ancy for the divorced group. For example, during
longed disruptions in family management and is the first year following separation, there was a sig-

290
 nificant increase in child noncompliance for boys have these skills. If, under temporary stress, things
and girls. The Wallerstein and Kelly (1980) clini- get out of control, they can combine forces to halt
cal studies of children from families getting a di- the increase in rates of coercive behavior. Howev-
vorce showed a similar outcome. The parents re- er, in the single-parent family, the person who
ported that their lives were disrupted for two or ends up as caretaker may be the more unskilled
three years during the process. Again, there was a parent. What M. Hetherington calls the “buffering
general increase in aggression (including stealing) effect” is no longer in operation. The corrective
for the children involved. Her case studies give feedback, the back-up punisher, the sympathetic
some fascinating details on the difficulties encoun- colleague, and fellow negotiator is no longer avail-
tered by some of the mothers in effectively manag- able. The single parent reacts in an irritable, non-
ing their children. problem-solving manner and loses battle after bat-
Single-parent families do indeed seem to be at tle as the coercion levels increase in frequency and
greater risk in terms of the likelihood of producing amplitude. Here we understand the contribution
an out-of-control child. It would be useful to of the father by observing what happens when he
know what proportion of single-parent families is no longer present.
do, in fact, have antisocial children. I believe that
it is also the case that in these families, when the
child is out of control, there is a greater likelihood
that his performance will be more extreme, i.e.,
Footnotes
his rates are higher than those found for out-of- 1. John B. Reid pointed out a fundamental
control boys from intact families. Oltmanns, oversimplification in the Anarchy progression.
Broderick, and O’Leary (1977) noted a trend of Many of the child-abuse cases now being treated
this kind in their analysis of 62 cases referred for at OSLC consist of parents who have “won.”
treatment. At OSLC we have noted that single- Their children are relatively low-level coercers
parent families seemed to produce problem chil- who are definitely controllable by normal means.
dren who coerced and/or stole at extremely high The parent, however, when irritated (which is of-
levels (they also seemed difficult to treat). Horne ten), expresses it, not by nattering, but by brutal
(1980) corroborated this with his analysis of the attacks. It seems likely, then, that there may be
first few dozen cases referred to OSLC. He com- several paths or progressions that move to assault.
pared intact and father-absent families for both The Anarchy progression is only one.
normal and out-of-control samples (see Chapter 2. The reader should keep in mind that during
2). For the normal sample, father presence or ab- a session each family member, in turn, served as
sence did not significantly relate to increased coer- the target for observation. If two different dyads
cive behaviors for any family members. However, were in conflict simultaneously, then the observer
given a sample of out-of-control families, those would record only the behavior that involved the
with absent fathers tended to have coercive rates target subject. Given that this occurred seldom
twice as high as those for intact families. For the (and I think this is the case), then the percent of
problem child the mean TAB scores were .98 and mother involvement used here is a slight overesti-
.41, for mothers .68 and .36, and for older sisters mate of the true state of affairs.
.99 and .13 respectively. 3. The reader should recall that this index is
In intact families there is at least a possibility somewhat confounded in that it reflects the dispo-
that if one parent is inept in the practice of one or sition to extend coercive chains and the disposi-
more family management skills, the other may tion to extend when punished.

291
 Chapter 13
Abstract
This chapter is amore phenomenological approach to the antisocial child and his family. An effort
is made to describe different types of families who are referred for treatment because one or more
children are antisocial.
There is also a brief review of the OSLC studies evaluating the outcome of family intervention.
This includes replication, follow-up, and the four comparison studies completed thus far.

292
 Clinical Contacts and
Treatment Outcomes

Over the last decade and a half the staff at reiterates a dual theme of anger and unpredictable
OSLC has treated and studied over 250 families of explosions. The following captures the flavor of
antisocial children. The children in these families such interactions:
(both boys and girls) ranged from 3 to 15 years of 6

age. The referral problems for these children in- *, . . abstractedly he undressed himself and began
cluded stealing, firesetting, chronic delinquency, to soap and wash himself, still retaining the kind
conduct problems, and child abuse. At a clinical of somnambulistic facial expression that had first
level, we know more about these families than we attracted my attention. Suddenly, in a savage out-
have been able to express within the empirical pouring of verbalization, he began to curse with
framework of coercion theory. There is a sense of the most primitive swear words at nothing at all.
repeating patterns that suggest we should consider When I asked what was wrong, he shifted to his
constructing a typology for these families. As a mother, saying she was a no good bitch, a fucker,
staff, we have discussed this possibility but have was never any good, he hated her, his goddam
not yet been able to find the time to initiate the ap- brother was mean and wouldn’t help his father
propriate studies. Nevertheless, I think it is useful who was going to die and he, Bill, had to go home
to include the outlines of what the clinical staff and help out.” Entry: Paul Deutschberger) (Redl
perceives as a typology. It may prove useful in de- & Wineman, 1957, p. 105)
signing different treatment strategies for different Other investigators showed that the treatment
families. Better yet, it may motivate other investi- program designed by Redl and Wineman pro-
gators to systematically study the problem. duced significant changes in the behavior of these
It doesn’t seem appropriate that the reader boys in the residential setting (Raush, 1965).
could complete a volume about antisocial children However, when the boys were returned first to
without being exposed to what it is like to interact foster homes and, within a few months, to their
with such children. The classic series of articles own homes, these gains were quickly lost.
and books by Redl and Wineman (1951, 1957)
provide a literate and, I think, accurate descrip- 6
*. . . thus our ‘children who hate’ went back into
tion of the behavior of socially aggressive children. limbo of ‘the children that nobody wants.’ The
Their viewpoint is that of therapists deeply in- spectacle of their retraumatization of strengths
volved in trying to help the child adjust to a resi- that had been so painfully, if incompletely, im-
dential treatment setting. Their graphic account planted in their personalities being literally wasted

293
in a battle with a hostile environment, is one that The violence of his temper outbursts was frighten-
fades slowly, if at all, from our minds. ing and seemed to be triggered by relatively minor
“And we are still having trouble in recovering provocations. At school, a simple request to turn
from our amazement that . . . it would remain im- in his homework, a mild rebuke, or a suggestion
possible to create adequate treatment channels to that he had erred in his work could lead to shouted
rescue these five lives.” (Redl & Wineman, 1957, obscenities, overturned desks, or attacks on other
pp. 556-557) children with a pencil held as a dagger. The ob-
servers commented that in the home he ruled
During the 1950’s and 1960’s, others of us who whatever territory he occupied.
attempted to work with such children experienced At home, the behaviors coded showed extended
a similar sense of bewilderment and frustration. bursts made up of Whines and Yells, interspersed
The fact that some of our best psychotherapists with shouted Disapprovals, Noncomplies, and
working with a small number of clients in well de- physical attacks upon family members and furni-
signed residential treatment settings could not pro- ture. During the intervals when he was absent
duce lasting change underlined the necessity for from home, telephone calls would often mark his
coming up with an alternative formulation. We progress through the neighborhood, e.g.; he left
needed a new treatment for aggression. Redl and school two hours early, stole candy from a store,
Wineman’s program was the best of its kind, but it and appropriated a toy from a neighborhood
conceptualized the problem as being within the child.
child. This is in stark contrast to current perspec- No baby-sitter would brave this storm center, so
tives (including our own) that view the problem as the parents had long ago given up the idea of a pri-
being in the family. This means that effective treat- vate life, movies, or weekends together. Both par-
ment must make provision for altering the reac- ents worked. The mother (not yet 30 and physical-
tions of family members fo the child and his reac- ly attractive) looked as if she was in the throes of a
tions to them as well. One cannot just place the severe illness. The family physician provided med-
child in a residential setting, change his behavior, ication for her chronic depression and accompany-
then return him to his family and expect the treat- ing fatigue. Work was a reprieve from her morn-
ment effects to persist. The family members’ well ing and afternoon bouts with her son, Don. Our
established patterns of interactions, expectations, treatment regime provided two therapists in atten-
and feelings will quickly reinstate the child’s old, dance in the home whenever Don was present.
familiar habits. Halfway houses and residential Typically, her day began at 7:00 a.m., rousing
centers give a welcome respite to the community, him from his wet sheets (which she changed), then
but they do not produce lasting changes in behav- scolding until he went sullenly to his tub. Once
ior (cf. review by Patterson, 1980b). there, she washed and dried him as if he were an
Regardless of one’s persuasion about this or that infant or visiting royalty.
form of treatment, most of us experience the anti- He often dawdled while dressing, which pro-
social child in the same way. The following case is duced a stream of prompts and commands from
introduced to give a sense of what the clinical “ex- his mother. Suggested items of clothing were re-
perience” is like. fused; this led to bitter exchanges with the now
When I met him, he was 61%years of age. There thoroughly exasperated mother. He emphasized
was nothing about his appearance that identified their disagreements by kicking the door and
him as the boy who had set the OSLC project rec- throwing things around the room. Through all of
ord—his baseline TAB score was an Olympic cali- this the mother hovered about, helping to get him
ber 3.7 coercive responses per minute! A trim dressed. She alternately cajoled and scolded,
four-footer, he had a sleazy look about him, like a wheedled and glared.
postcard carried too long in a hip pocket. He sat in She stood in attendance while he dined. Not
the reception room, slouched down in the chair only did she serve, but she finally fed him when-
and cooly looked me over as I approached. While ever he deigned to open his mouth. Through it all
his mother hardly looked up at my approach, his ran a steady cacophony of yells, cries, and argu-
father seemed friendly enough and said he was ments about whether his mother had any right to
glad the treatment was finally about to begin. force these unreasonable requests upon him. The
We then walked down the hall, renewing the re- mother alternated between patient and antagonis-
lationship that was begun at the initial interview tic answers to his arguments and threats. At one
but was disrupted by the two weeks of baseline point, she brought a stick from behind the refrig-
study. From the beginning the parents thought it erator door. Her menacing demeanor left little
might be necessary to institutionalize their son. doubt that she regularly employed this weapon. In

294
 the face of this ultimate threat, Don showed tem- the theme of coercion (and variations on this
porary compliance and moved forward in his gla- theme) is recognizable. The anger of the child
- cial progress toward leaving for school. and the deference with which he is attended raise
In the afternoons Don returned from school to the question of why these parents allow them-
pick up the morning refrain. His 4-year-old broth- selves to be defeated in this way? Experience with
er was also available as a partner. The latter (a these families suggests that there is no single an-
Machiavellian of considerable stature) knew when swer to this question. There are a number of paths
to probe, when to attack, and when to withdraw to this same outcome. What follows is a series of
with tearful protestations to the protection af- descriptions of the different family types that pro-
forded by his parents. For example, as Don sat duce antisocial children. The accounts are based
eating his ice cream (with his fingers), the younger partially upon what the parents have reported
brother surreptitiously slipped a more efficient and, in part, reflect our reconstruction of how the
spoon into the mess and ran triumphantly down process might have begun. Even though the result-
the hall to hide behind the door in his bedroom. ing typology is purely descriptive, it might serve
Don ran shrieking after him, grabbed the door, several useful functions. For one thing, it offers a
and repeatedly slammed it into his younger broth- set of hypotheses that might be tested in longitudi-
er. The screams brought both parents to the scene. nal studies that attempt to predict which families
The father listened for amoment to their shouted will produce problem children. Secondly, we be-
claims and counterclaims. After a brief pause, he lieve that the family types might covary with the
simply began to slap both children. With that, the amount and/or kinds of resistance generated by
mother turned, walked quietly back into the kit- the therapists’ attempts to help these families
chen and sat staring out the window. change themselves (Patterson, Reid, & Chamber-
Later the family was to go for a ride in the car. lain, in preparation). In either case, it seems ap-
Both parents began shouting commands. In the propriate to present these untested hypotheses as a
rush of the moment, they often overlapped in their first step in constructing a typology of problem
targets; e.g., the mother said, “Don, wash your families.
face right now,” while the father ordered, “Put on
your jacket, Don. Hurry up now.” A steady Speculations about Family Typology
stream of commands was given as they moved to- It should be kept in mind at the outset that the
ward the car. The children moved at their own vast majority of the cases studied have been from
pace, largely ignoring both parents. welfare, blue-collar, and working-class families. I
During the day, the observers noted periods am sure that what these families have to teach us is
where the interactions seemed warm and positive. different from what might be learned from study-
For example, on numerous occasions one parent ing middle-class families with older adolescents.
would read to the children, who would often sit For this reason, the typology to be presented (and
for long periods of time entranced with the story. many aspects of coercion theory) may not fit what
At these times they seemed to be the prototypical a private practitioner experiences in working with
loving family unit. families of antisocial children.
During treatment, the mother expressed consid- At one point we tried to conceptualize the main
erable doubt that Don could be changed by their dimensions that could account for why families
efforts. The father, on the other hand, seemed produce antisocial children, and then use them to
more sanguine. It soon became apparent that the construct a typology. The resulting structure sim-
mother was incapable of participating in a pro- ply did not adequately reflect the full range of fam-
gram that required her to punish either of her chil- ily types. We then set about simply listing them
dren. The children had perfected a simple means and ignored the fact that they do not fall along any
of controlling their mother. It also worked reason- particular set of dimensions. Some of the types re-
ably well on their father. If a parent tried to use flect particular roles played by one or more par-
time out, their efforts were met with howls of out- ents. Others refer to variables that had something
rage and an inevitable retreat by the parent. When to do with starting the process and/or maintaining
encouraged by the therapist to persist, the mother It.
would often break into tears and disappear into We believe that these types relate to different ex-
her bedroom. The cost of confrontation was sim- periences in therapy. Most families who receive
ply too high. training in child management do not proceed in a
Not all of the children we treat are so unremit- textbook fashion through the various compo-
ting in their coercive efforts. Not all parents are so nents. For many of them (particularly for families
thoroughly defeated by their children. However, with abused children and chronic delinquents),

295
there are major difficulties. The child and the par- simply fends for himself and, in some extreme
ents may be extremely resistive to doing what they cases, comes home only to sleep. The parents as-
have agreed to do. They may also resist coming to sume that if the child is left free to express his own
an agreement. In some of these families, up to wishes and desires, that gradually inner controls
30% or 40% of their behavior has been coded as will emerge (a modern expression of the Rousseau
resistive (Patterson, Reid, & Chamberlain, in thesis and familial analogue to the Summerhill
preparation). The hypothesis is that the family ty- philosophy).
pology may tell us something about the amount In their clinical studies of families in divorce,
and kind of resistance that is encountered. Wallerstein and Kelly (1980) noted a very similar
pattern. Some of the women in that study had
The Parent-Sibling functioned well as caretakers when their marriages
When you enter the room for your first meeting were intact. The husbands were the rule setters
with this single-parent family all of the family and disciplinarians. With the dissolution of the
members are dressed in Levi’s, and all are slouched marriage, these women found themselves unpre-
down in their chairs. It is not at all apparent which pared to serve in the role of disciplinarian, and, in
one is the parent (see Illustration #4). Clinically that study, their children tended to become in-
speaking, there is no parent here. This is a typical creasingly antisocial. The authors also described
single-parent family, where the mother has given many of these mothers as being afraid to set limits
up the unpleasant features accompanying the role for fear of losing their children’s affection.
of caretaker. In fact, this becomes a central issue in At OSLC, treatment of these families was often
the treatment of these families. The Parent-Sibling difficult. The Parent-Sibling was usually over-
pattern characterizes perhaps one in ten of the whelmed by outside crises. Many of them were
families of preadolescent, antisocial children re- isolated from other adults and depended upon re-
ferred for treatment. It was one of the first family inforcers provided by their children. Treatment
types that we noted (Patterson, Cobb, & Ray, often included assisting the parent to form satis-
1973), factory relationships outside the family. Gradually
As much as possible, the mother functions as an they learn that needing to be loved all the time by
equal, a friend. Sometimes the position is based on their children makes them terribly vulnerable to
a misinterpretation of a laissez-faire, egalitarian manipulation. Although hesitant and terribly dis-
philosophy. Many of these mothers seem cut off organized, they eventually learn that all is not lost
from other adults; the children are their primary when rules are set and punishment is employed.
source of support. There are few house rules, few Their. move to establish the parent role is difficult
schedules, few assigned roles, and no single person but satisfying, as they learn that the children actu-
punishes antisocial behavior. Each family member ally become more warm and loving when structure
defends his or her own territory. Sometimes a tem- is imposed.
porary vigilante committee may be set up to re-
The Unattached Parent
dress a wrong. Sometimes an older sibling will oc-
cupy a very tentative role as housekeeper, but usu- There is another type of adult who has also dis-
ally this does not include rule setting. In doing carded the role of parent. They, too, impose few
away with the disciplinarian features of the parent limits or structure, but they do not wish to be a
role, the mother buys friendship from her chil- friend or colleague of their children. This group
dren. This can often become a major problem in has been noted and labeled “Disengaged” by other
treatment. If she now attempts to set rules and en- investigators (Minuchin, Montalvo, Guerney,
force them, the mother fears the children may re- Rosman, & Schumer, 1967). These parents occu-
ject her. The mother’s concern about such a rejec- py no philosophical position; they simply do not
tion is overwhelming. care to be parents. These parents are difficult to
If some Parent-Siblings also hold delinquent val- engage in treatment because they are only vaguely
ues, then the therapists may find themselves in the motivated. If their children are socially aggressive,
interesting position of trying to convince the chil- then a cessation of pain is a possible motivator. If,
dren and the mother that stealing is a reprehensi- however, the children steal or set fires outside of
ble act. The Parent-Sibling and the children may the household, then the parents are unlikely to be
form a xenophobic enclave that serves to protect motivated to retrain their children.
its members from outsiders. In counterculture Control theory seems to reflect the biases of its
families we sometimes find both parents adopting middle-class investigators, who emphasize the im-
this role. They assume little or no responsibility portance of the attachment or bonding of the child
for any aspect of the child’s behavior. The child to the parent as a necessary condition for sociali-

296
 vist pte ies

4ED
ole 0 Oz

&

Lo)

Illustration #4: The Parent-Sibling

zation (Hirschi, 1969). However, the theory over- When we began studying high-rate Stealers, in-
looks the possibility that some parents are not at- creasing numbers of unattached parents were en-
tached to their children. They are not committed countered. They presented two problems to the
to the role of caretaker. In fact, it was these fami- therapy team. First, it was almost impossible to
lies that led us to introduce the idea of a “parent- get them to come for treatment; they were not
ing salary” (Patterson, Cobb, & Ray, 1973). The really motivated to change their child. The use of a
motivator for the parent to participate in treat- parenting salary helped to bring them into the clin-
ment could be money, points toward an appoint- ic, but once they arrived, they provided us with a
ment at the hairdresser, driving lessons, and so on. new problem. They were not only noncoopera-
The experiment by Fleischman (1979b) showed tive, but very resistive. Some of these parents were
that the contingent use of as little as $10.00 per very critical and/or irritable in their face-to-face
week significantly increased parental cooperation interactions with the therapist and their children.
in the training program. Like rebellious adolescents, they actively refused

297
to cooperate and disagreed at very high rates dur- In our clinical experience, it represents an attempt
ing the treatment sessions. Others wanted to con- by trapped parents to deal with an impossible situ-
trol the situation at all costs and were adept at ation. Effective training in child management will
keeping the discussion away from topics relating often help the parent and the child to risk being in-
to intervention. They tended to miss appoint- timate with each other and to express affection
ments, come late, and failed to carry out their again. A sizable majority of the parents thus
agreed-upon tasks. trained report feeling more affection for their
These are difficult parents to have as clients. child. They generally like being parents. One does
Their lack of concern for their children is an af- get a definite sense of family here. The therapist’s
front to my more middle-class values. The fact problem is how to convince the parent that the
that many of them also have delinquent values fur- normal rules and sanctions should apply to the
ther exacerbates the situation. It seems, at times, special child. This must be done in such a way that
that there is no common meeting ground. I do not the parent does not lose face. It should be noted
believe that we know how to completely alter this that some of these families are easier to work with
condition. If we had the means for accurately clas- in this regard than others.
sifying the unattached parent, | think it would cor-
relate with recidivism during follow-up. Overwhelmed
This pattern is another example of parents who
This Child Is Special know about child management but do not imple-
These families present another kind of paradox. ment it effectively. The parent is simply over-
They know what to do for their children, but they whelmed by circumstances, e.g., a single parent
choose not to do it! It seems a frequent pattern, es- working and raising a family, both parents work-
pecially among families with younger problem ing, a very large family reared in poverty, a pro-
children. longed illness, or extended crises. If both parents
Some of these parents had previously demon- are working, neither of them really has the energy
strated skill in raising children who functioned to sit down and quietly talk with their child each
well. For some reason, these demonstrated skills day. There are long blocks of time when neither
were held in abeyance when dealing with the prob- parent is home and the child is left with large
lem child. For him, certain house rules were set amounts of unsupervised time. For most children,
aside, tracking was very selective, and sanctions this in and of itself is a sufficient condition for the
were seldom imposed. The reasons for this selec- child to drift into stealing and vandalism. It also
tivity were varied. Sometimes it was the case that sets the stage for the child to seek out a set of unsa-
the child had been diagnosed as special, e.g., an vory and probably similarly unsupervised com-
early diagnosis of retardation or minimal brain rades.
damage, or an extended hospitalization for illness. The problem here is not a neurotic conflict be-
The parents were told that structure and limits tween spouses. It is not an authoritarian parent,
should not be imposed upon their child because it nor even a parent wishing to be loved too much. It
would exacerbate his condition. Whatever the cri- is simply that the caretaker is preoccupied. Many
sis was, the parents seemed to try to make it up to of these parents genuinely care about their chil-
him by removing the constraints that still applied dren. When they are at home they may even be
to his siblings. His resulting hyperkinetic and anti- said to have good relationships with them. How-
social behavior was then attributed to the prior ill- ever, they are more deeply committed to activities
ness and constituted proof of the necessity for yet that are unrelated to the caretaker role. In some
further reductions in constraints. cases, there may be crushing debts to be paid, or
This special status, however, does not serve the the father may be a long-haul truck driver, and the
child well. In spite of the parents’ wish to be more mother may be preoccupied with family crises.
loving to this special child, they cannot help but Many of these parents can eventually give up
feel angry and puzzled in the face of his on- some of their outside commitments, or arrange
slaughts. This further adds to the guilt engendered them in a more balanced fashion, so that one care-
by the original crisis. Many of these parents resort taker is always “on duty.” There are, however,
to a “refrigerator” stance, in which their icy de- some parents who must remain absentee caretak-
meanor communicates neither anger, guilt, nor ers and arrange for after-school care, or monitor
love. They minimize their contacts with the child the child by telephone. There is a smaller group of
in order to neutralize as much hurt as possible. To these parents who cannot even arrange this much.
a therapist, it may seem reasonable to assume that These people communicate all the right feelings,
the icy demeanor caused the aggressive behavior. attitudes, and motivations in treatment. The prob-

298
 lem is, they do not totally cooperate in carrying tempted to set rules or punish deviant behavior,
them out. The therapist is often temporarily left in the grandparent served as the child’s advocate and
‘the position of arranging for a surrogate caretaker was sufficiently adept to cause confusion in the
(including him- or herself). ranks of the parents’ forces. The grandmother
There is another sense in which parents may be served as baby-sitter for more than a decade and,
overwhelmed. Some families seem to live in a per- during that time, effectively prevented the parents
petual state of crisis. We encounter them particu- from setting consequences for stealing, even when
larly in samples of families of chronic delinquents the child stole from her!
and, to a lesser extent, in the abused-child sam- There is a more frequently occurring variation
ples. The effect of the stressors they experience is of this theme; it is the My-Child and Your-Child
to badly disrupt the already precarious balance of pattern. This is often found where a new step-par-
their family management practices. The parents ent joins the family. The wife is not allowed to
are simply too preoccupied with the problem of re- punish her husband’s child; he, on the other hand,
ality to conscientiously practice their parenting is not allowed to punish her child. The children
skills. As a therapist, one has the definite impres- are, of course, eager participants in this game and
sion that these people could be good parents if feed mildly distorted reports of yesterday’s battles
only . . . I have the feeling of being slightly out of to each parent. In one blended family of five chil-
breath when working with these people. In their dren, when the stepfather objected to his wife’s
attempts to manage their child, one can observe children using obscenity in his presence, the chil-
them using punishment, rule setting, ignoring de- dren simply repeated his objection to the mother.
viancy, and so on. However, these things are tried She launched into a vituperative attack upon him
in quick succession, with no effort to pursue a pro- for attempting to punish her children. He, in turn,
cedure to its conclusion. The behavior of these assiduously defended his children from her efforts
parents is as chaotic and disorganized as is their to set limits. Both parents wanted the option of be-
lifestyle. ing overly permissive with their own brood. As a
result, all five children were out of control and
Sadomasochistic Arabesque were in frequent contact with the local police. Lat-
From the problem child’s perspective, another er, when these parents separated, it was possible
label for this family type would be Divide and for them to effectively apply child management
Conquer (see Illustration #5). This ballet requires procedures to their own children.
the committed participation of three partners. The In our treatment of multiple-offending, adoles-
parents never dance together, but each dances sep- cent delinquents, we encounter this pattern much
arately with the child. When this happens, the more frequently than we do with younger antiso-
abandoned partner tries to stop the music. The re- cial children. For either age group, it causes prob-
actions of each parent to the child are delicately lems for the therapist. No matter where treatment
balanced. Each tracks the behavior of both the begins, at least one parent will temporarily feel ali-
child and the other parent. One parent occupies enated from the process. The child is an active
the role of the harsh, punitive martinet. When the participant in the arabesque and knows just how
martinet is present, the behavior of the child is in to exploit the rift between the parents. Julius Cae-
moderate control. The other parent attempts to sar had some knowledge of how to play one tribe
balance the perceived severity of the other by be- of Gauls against another; these children have up-
ing noncontingently warm and permissive. As a dated his technique and know how to use it effec-
result, when the despot is not present, the child’s tively.
behavior is out of control. Deviant behaviors oc- It is difficult to do justice to the feelings of the
curring in the despot’s presence are punished, participants. These are interchanges in which peo-
sometimes brutally. The despairing despot sum- ple are likely to get physically injured. The rage of
marizes the situation: “But he acts okay when I’m the punisher and rule setter is frightening to the
home. She is just too easy on him, that’s all.” Oth- child, spouse, and therapist alike. In the punisher’s
er investigators, such as McCord and McCord world, there is order, there are rules, and there are
(1961), have noted a similar pattern for families of sanctions. Warmth and affection are mistrusted,
antisocial children. though, in private, the despot will sometimes ad-
There are apparently a large number of varia- mit to a terrible loneliness. When the rule setter is
tions on this theme. In one instance, it was the res- home, the child behaves most of the time. There
ident grandparent who functioned as the secret are sporadic beatings, presumably to remind the
agent, actively sabotaging the efforts of the par- child who is boss. As the child matures, the beat-
ents to set limits upon their child. When they at- ings merge into physical battles, and the injuries

202
 SNe

NS fo
iN Jae

CP
VA

Illustration #5: Divide and Conquer

increase commensurately. After each confronta- cover for the child. This is the time for the child to
tion, the other spouse withdraws psychologically, ask for money, the use of the family car, or an ex-
and this increases the despot’s resentment toward tended weekend with friends. The rule setter sus-
the child: “If that kid weren’t here we would get pects the entire sequence, and his or her rage is
along fine.” When the despot leaves the house, the boundless when later confronted by a police re-
rules are changed. The other parent is warm and port of the child’s most recent escapade. Both par-
giving (particularly after a major battle) and will ents have suffered defeat. The rule setter has

300
 failed, and the “good” parent’s efforts to be warm transgressions of house rules, they do not say,
have failed. The child wins the battles but often “You took my car without asking, and that makes
- loses the war because many of these children are me mad.” Such confrontations are avoided at all
eventually removed from their homes. cost. The direct expression of anger is also
Treatment of these families is difficult. Both the avoided as long as possible. The communication
mother and father are resistant. Even if progress is comes out more like, “Mommy and Daddy feel
made in getting one to shift positions, the other re- there is still some misunderstanding about the car.
mains mistrustful. Neither is willing to change un- As you know, you are still under age, and accord-
less the other does so first. Any alliance achieved is ing to the law. . . .” They lecture their child, but
shaky at best and is easily disrupted by the re- they do not confront him. Sweet reason prevails.
newed acting out of the child, producing an “it The child sits with downcast eyes, awaiting the
didn’t work” reaction from both parents and per- lecture’s end. If the child appears to listen, and if
haps a beating by the martinet. In one family, the he claims to feel sorrow and repentance, then the
author managed to simultaneously shift the behav- parents are satisfied. The parents believe in rea-
ior of both parents (the result of endless negotia- son, information-giving, and in the redeeming vir-
tions and monumental compromises). Neither tues of guilt. The child’s weeping is accepted as a
parent trusted the other, nor did they have much sign of guilt. The point is, at no time do the par-
confidence in this first therapeutic step. At this ents punish him for his behavior; instead, they lec-
point, the father discovered the boy had stolen and ture. During their discourse, one or both parents
sold his father’s family jewels (literally). Treat- bring to bear much of their store of understanding
ment ended here, and the boy was sent to an insti- about why children do what they do. They read,
tution. then reread, the child’s intentions. When they are
finished, not a blemish in the child’s character re-
Perfect Parents
mains unvarnished.
Two variations are encountered on this same These are people who are trying to be rational.
theme, the most notable being Mommy and Dad- They don’t usually hit their child, but sometimes
dy, with Gotcha as a minor variation. The pattern their facade crumbles, and a beating ensues, or the
emerged primarily in the context of our efforts to child is ejected from the house. A major battle pro-
work with the families of chronic delinquents. It duces profound guilt reactions in the parents.
may have characterized some families of younger This, in turn, is deftly utilized by the child to ob-
Social Aggressors and Stealers, but we failed to tain even more freedom from parental control.
note this. The general picture presented is of a per- The parent who must be thought well of by the
fect middle-class family. It is as if one were observ- child at all times is vulnerable to a variation on the
ing a TV show; their lines are carefully chosen, the general theme. This subtheme is called Gotcha.
voices and feelings are carefully modulated. The For such parents, being accused of being unfair or
parents speak of love and responsibility. The chil- illogical is painful enough to halt their attempts to
dren are neatly dressed and are usually well be- punish a wrongdoing. The child may also interject
haved during the sessions. guilt as an even more powerful means of neutraliz-
The parents have usually read a number of ing the parental role. Guilt induction may involve
books on the subject of child psychology or parent accusations that the child is unloved, e.g., because
training. At first sight, the therapist may be en- the mother works, or because the child is adopted.
chanted, thinking: “Here, at last, are parents in- The possibilities are endless. For example, after
terested in our theory; this will be fun.” If the ther- much urging by the therapist, a mother told her
apist has a middle-class orientation, he or she may son that coming in late at night would cost him
feel they can identify with this family. Some of the part of his allowance. On the videotape, one can
parents were subjected to brutal conditions in see the trepidation with which she made this per-
their own childhood experience, and they want to fectly reasonable statement. Amid the acclaim and
make sure that this does not happen to their chil- applause by the two therapists present, the son
dren. After assiduous study, they set out to do murmured very quietly, “Yeah, like I always
what they believe middle-class parents do. They knew.” “What?” asked the mother. “That money
want to be Perfect Parents and raise perfect chil- means more to you than I do,” hissed her delin-
dren. quent son. The mother looked as if she had been
A salient variation on this theme is that of struck in the face. Needless to say, she did not en-
Mommy and Daddy.' Some of these parents tend force the rule.
to drop the first person pronouns and substitute Guilt induction is the main mechanism em-
Mommy and Daddy. In talking to the child about ployed by children who are playing Gotcha. It can

301
be done with threats; e.g., “Okay, I’m going to run succession, and the child’s antisocial behavior in-
away. I know you want me to anyway.” A success- creases dramatically. At the peak of the experi-
ful manipulation has had a prior history of years ence, the parents quietly announce that they have
of trial-and-error training. The child keeps trying arranged to place the child outside of the home. A
new combinations until he or she finds one that moment’s reflection brings to mind the fact that
works. The parents’ mistake is in allowing it to they had raised this possibility early in treatment.
work. In our clinical experience, families of this The therapy experience was a means of proving to
type are surprisingly difficult to treat. The prob- themselves that their child was impossible to cope
lem in each case is to get them to stop lecturing with.
and openly confront their child and to be consis- There is another family type I have heard about
tent in confronting even the less deviant proto- but have not yet encountered. The protagonists
types of extreme behaviors. are two parents who agree on punishment and
house rules and have a clear grasp of what is devi-
Misattribution ant, protodeviant, and normal. However, these
This type of parent emerged during our recent parents are extremely authoritarian and oppres-
study of abused children. At one level, they seem sive when dealing with their child. To keep from
terribly misinformed about what a 2- or a 5-year- being crushed, the child must engage in a kind of
old can do or needs to do. The behavior of the neurotic acting out.
child is evaluated against standards that are more The problem with typologies is that it is very
appropriate for an older child. For example, one difficult to find a “pure” type of anything. The
mother explained that she had to beat her child be- same is probably true here. As a staff, we have
cause he refused to do what she asked him to do. agreed on the importance of classifying families,
She would make a request, then repeat it. The but, as yet, we have not begun to do so. I suspect
child she was beating was 13 months of age! that when we do begin, we will find that most
Another aspect of the problem is that the dis- families fit into several categories. It may also be
crepancy between the child’s behavior and the par- the case that one parent fits one category, and the
ent’s expectations is viewed as a reflection of the other would best be described by a different pat-
child’s malevolent intentions. The parent can be- tern. However that may be, the time has come to
come so preoccupied with negative attributions begin the construction of an empirical base for a
about the child that even his most innocent behav- typology of families.
ior is viewed as hostile. For example, when asked
why she persisted in slapping her child’s face, one Some Speculations
mother explained that when her 18-month-old It is an interesting exercise to use a few simple
pursed his lips and blew, it meant that he was in- concepts from coercion theory to construct a tax-
sulting her. She slapped him to teach him to be onomy of families. The earlier analyses suggest
more respectful. that there are two general dimensions of parent
While many abused children have been ob- behavior that are of particular interest. The first
served to be highly coercive, some of them interact one describes the extent to which the parent is ei-
well within normal limits. Even within this subset ther distant or irritably enmeshed. As this dimen-
of abused children, the parents seem to function sion is presented in Figure 13.1, it is somewhat
with a kind of thought disorder analogous to pro- distorted, for it suggests that a parent is either dis-
jection. They take many aspects of their child’s be- tant or enmeshed. As indicated earlier, we suspect
havior very personally. This, combined with a that some parents of delinquents may be both.
profound ignorance of the intellectual and motor Keeping that possibility in mind, for the moment
abilities for children of different ages, sets the let’s view it as a type of bipolar dimension. As
stage for repeated misunderstandings. It is our such, it is analogous to the bipolar factor for par-
clinical impression that many of these families can ent behavior that was so carefully delineated by E.
be helped. Schaefer (1971). His bipolar factors were labeled
Psychological Autonomy (detachment, emancipa-
Miscellany tion, indifference) at one end and Psychological
In our clinical experience, we have encountered Control (intrusive, intolerant) at the other. These
another repeated pattern that does not necessarily bear more than a passing resemblance to my a pri-
fit a particular type of family. We call it Last Stop. ori dimension of Distance and Enmeshment. No-
These parents seem to be trying to carry out their tice that in Figure 13.1 normal parents do not fall
assignments, but, for some mysterious reason, within either extreme of this dimension (i.e., they
nothing works. Failure follows failure in quick are neither Distant nor Enmeshed).

302
 Figure 13.1
Hypothetical Relationship Between Parent Behavior and Child Pathology

Parents
Distant and
Unmotivated

Withdrawn Child
Child Steals
@ ®

Parental Parental
Punishment Punishment
Overly Effective Ineffective

@
Child Anxious, Aggressive
Depressed, Child
Anorexic
Parents
Enmeshed and
Irritable

As used here, distant parents would be charac- cisms too subtle for the FICS to measure. The dis-
terized by lower rates of observed prosocial inter- pleasure and dissatisfactions of the parents do not,
action with children (i.e., less talking, laughing, however, go by unattended by the child.
and playing). These parents would also score sig- The second a priori dimension concerns the
nificantly higher on categories such as No Re- ability of the parents to use punishment to change
sponse. In addition, these parents probably share behavior. Assuming, for the moment, that we do
less leisure time with their children than parents of not have an effective measure of punishment effec-
Normals. tiveness, it may be premature to introduce such a
The enmeshed, irritable parents tend to interact concept. Nevertheless, I think it is a key concept
at high rates with their children. The factor that for understanding several very different types of
differentiates them from Normals is that a dispro- child pathology. The a priori dimension of punish-
portionate share of their interactions are negative. ment effectiveness is similar to the bipolar factor
In some families, this would mean very high values of lax and firm control identified by Schaefer
for the irritability measures (crossover, counterat- (1971). My dimension differs in that it emphasizes
tack, punishment acceleration, and continuance). punishment as the method of control.
In other families, this might be expressed by criti- As shown in Figure 13.1, some parents are Jess

303
effective than Normals in their use of punishment,
while others are significantly more effective. In the Treatment Outcomes
latter homes, the parents “win” almost every con- At this point, we have treated over 200 families
frontation. Furthermore, if they are also en- of extremely antisocial children. One of the most
meshed, they tend to be assiduous trackers of child heartening features of our work relates to the fact
behavior. No transgression is insignificant, and al- that these parents have been able to change the
most everything the child does arouses the anger problem child’s behavior as well as their own. The
of the parent. Even successful child performances focus in this treatment has been to reprogram the
are not quite good enough. As suggested in Figure social environment in which the child lives. The
13.1, I am guessing that the neurotic child is the most efficient means for bringing this about was to
outcome of these parenting practices. train all of the family members, particularly the
The enmeshed, irritable parent who is ineffec- parents. The focus of training was upon the skills
tive in the use of punishment produces the Social these people seemed to lack (Patterson, Reid,
Aggressor. The relation was detailed in Chapter Jones, & Conger, 1975). As noted earlier, it was
11. It is also supported by the Olweus (1981) during these clinical contacts that we gradually
causal modelling analysis of parent and child be- identified the four areas of family management de-
havior. tailed in Chapter 10.
The child who steals is depicted as a member of It is assumed that family management skills are
a household in which the parents are distant and “teachable.” If the parents of antisocial children
inept in the use of punishment. These are parents can learn to perform these skills, then they should
who do not track rule breaking that occurs outside be able to significantly reduce the problem child’s
the home (see Chapter 11). Transgressions that deviant behavior. In fact, the coercion levels for
are brought to their attention go unpunished. the entire family should be reduced. If the out-
The last quadrant in Figure 13.1 describes a come evaluation data reflected such changes, it
highly speculative guess. The parents are distant would give indirect support for the central status
but extremely effective in the use of punishment to of family management variables as they relate to
control children. It is suggested that the child’s antisocial behavior; i.e., changing family manage-
withdrawn behavior is a coping device for avoid- ment variables produced changes in antisocial be-
ing parental criticism, a kind of last-ditch defense havior.
by a child who has already lost most of the battles In the OSLC treatment program, the teaching
with the parent. The only data even remotely and supervision of effective family management
bearing upon this hypothesis has been collected by skillsis viewed as a necessary (but not sufficient)
H. Hops (personal communication). His analyses component. There is a second component that is
were based upon observation in homes of normal also thought to be necessary (but not sufficient).
and withdrawn children. The findings showed The therapist must be skilled in coping with the re-
that the fathers of withdrawn children were signif- sistance to change that characterizes the majority
icantly more likely than fathers of Normals to re- of the families referred for treatment. Ordinarily,
act punitively to neutral and prosocial behaviors! this level of clinical skill requires several years of
Hops’ findings do not test the hypotheses that supervised clinical experience. In fact, the mea-
these fathers were more distant or more effective surement and analysis of this second component
in their use of punishment. I am assuming, howev- (clinical skill) is the focus of our current research
er, that both hypotheses apply to these fathers. on treatment.
In constructing the four quadrants of Figure By the mid 1960’s, I believed that the new fami-
13.1, I have used only two concepts to classify ly intervention technology was sufficient to change
most of the functional disorders of childhood. Ob- families of antisocial children. Moreoever, I
viously this cannot be. However, the discussion thought it might be possible simply to give the par-
does serve the useful purpose of illustrating the po- ents a book to read that would tell them how to
tential of these two simple concepts. I do think accomplish these changes (Living with Children,
that the Distance-Enmeshment and Punishment 1968). We quickly learned that treatment for these
Effectiveness dimensions relate to more than just families was much more complex than we had an-
antisocial child problems. The extent to which ticipated. At that time, there were five other cen-
they covary with other child problems such as ters developing social learning procedures for anti-
withdrawal, anxiety, or depression is a matter for social children. All of these groups collected obser-
further empirical analyses. vation data prior to, during, and following treat-
ment. After a decade of work, the data showed no
significant behavior change (as measured by TAB

304
 scores) for three of the six groups! It is my impres- thought necessary. The families are assessed at
sion that the successful and unsuccessful groups baseline, termination, and for 12 months after
. differed in two respects. The unsuccessful efforts treatment. In addition to parent global ratings, the
used time-limited treatment procedures and were parents provide daily telephone reports for se-
carried out by graduate student therapists. lected symptoms (PDR). Home observations pro-
By the early 1970’s, we were beginning to be- vide TAB scores for each family member at base-
lieve that perhaps there was more to treatment line, termination, and follow-up. The first studies
than just the technology described in our treat- demonstrated significant reductions in both PDR
ment manual (Patterson, Reid, Jones, & Conger, and TAB scores from baseline to termination (Pat-
1975). A systematic study by Fleishman (1979a) terson, Cobb, & Ray, 1973). This effect was rep-
convinced us that clinical skills must be added as a licated using essentially the same therapy staff
second, necessary component. In his study, half (Patterson & Reid, 1973). A series of detailed
the line staff from three community agencies were analyses followed that included the combined
trained and supervised in the application of social samples from these earlier pilot studies (Patterson,
learning components. They were carefully moni- 1974a, b).
tored to see that they applied social learning tech- The next question concerned the stability of
niques. In fact, the analysis of therapy tapes these treatment effects. Patterson & Fleischman
showed the trainees to be “on task” much of the (1979) studied 106 families of antisocial children
time. The tapes were then rated on such global treated at OSLC. Eighteen percent of them
measures of clinical skill as warmth, humor, and dropped out during the baseline assessment, a fig-
smoothness of flow. Compared to the OSLC staff, ure in keeping with the 24% to 30% dropout rates
the trainees were found to be cold, mechanical, for traditional child therapies. Of the 86 families
and nonfluent. When that study was done, clinical who received four or more weeks of treatment,
skills were not thought to be particularly impor- 33% refused to participate in the 12-month fol-
tant, so the majority of the on-site monitors were low-up. The 50 families who participated both in
not even trained as family therapists. Their focus treatment and in follow-up were observed in the
while on site was not on the intricacies of working home and contacted by phone (PDR) each month
through resistance but, rather, on the minutiae of during follow-up. Figure 13.2 shows that, at ter-
applying social learning technology. The dropout mination, 76% of the problem children had TAB
rate for treated families was very high. Observa- scores within the normal range. At the end of the
tion data were collected for 12 families of socially follow-up phase, the comparable value was 84%.
aggressive children that were treated by the agency The ANOVA for repeated measures showed sig-
staff. The mean baseline TAB score of the children nificant reductions from baseline to follow-up for
treated was 1.00; at termination, the score was both the TAB and PDR scores. Findings were also
.96 (Fleischman, 1979a). The Fleishman study summarized demonstrating significant reductions
and the experiences at other parent training cen- in TAB scores for siblings and for mothers. Parent
ters treating similar families emphasized the need perceptions of the problem child also showed sig-
to study the second component (clinical skill). nificant improvement.
Our research strategy has focused upon a se- If training families in family management skills
quence of five questions: (1) Using consecutive, effectively produced a reduction in the coercion
treated cases, can a significant reduction in ob- levels for these families, then one might expect the
served rates of coercive behavior be demonstrated treatment effects to persist (Patterson & Fleisch-
for the child and for other family members? (2) man, 1979). When crises and conflicts arise after
Can this effect be replicated by the same therapists treatment, if the parents reinstate the family man-
using more or less the same procedures; i.e., is the agement procedures, they should “work” for them
effect a reliable one? (3) Do the treatment effects again. In effect, each success is an example of neg-
persist over a 12-month period? (4) Is it possible to ative reinforcement, increasing the likelihood that
train other therapists to produce the same effects? in the future they will apply family management
(5) Are the effects specific to the treatment? The procedures. In a sense, the mechanism for persis-
following is a brief review of the studies that relate tence is “built in.” It should be noted, however,
to these issues. that this same mechanism does not seem to hold
The general format for the OSLC studies has for families of Stealers. Reid and Patterson (1976)
been to employ professional therapists trained at described the unmotivated parents of high-rate
the center and supervised during weekly staff Stealers. Apparently, these parents did not find
meetings in which videotaped treatment sessions stealing to be particularly aversive. The majority
are viewed. Treatment continues for as long as is of the Stealers studied were not coercive in their

305
 Figure 13.2
Follow-up Data for Treated Children
(from Patterson & Fleischman, 1979, p. 179)

60 3.0

e@

50 2.5

rv)
-
o
oi)
A” 40
=
se
Problems
PDR
Mean
Day
per 2.0

>
o
E30
3 @
°
e
= 20
S
o
a

10

Baseline Termination 6 18 &9 12 Baseline Termination 6 8 12

Follow-up Months Follow-up Months

Phase Phase

interactions with their parents. While they could tacts is lengthened. Termination does not occur.
be trained to significantly modify the child’s steal- The data for the first 36 delinquent cases shows
ing in the family and the neighborhood (Reid, Hi- that the effects persist over a three-year period
nojosa-Rivero, & Lorber, 1980), in the long run (Patterson, Reid, & Chamberlain, in prepara-
these parents did not continue to apply family tion). The reduction in police offense rates was
management practices. The community follow-up greater for the treatment group than for a random-
study showed that by the time the Stealers were 14 ly assigned control group. At present, the OSLC
years of age, 84% had committed at least one po- staff is treating a second sample of young Stealers
lice offense (Moore, Chamberlain, & Mukai, using the modified techniques that have been de-
1979). In fact, 64% were chronic offenders! For veloped in the last three years. Long-term follow-
the parents, the problem (stealing) was not a prob- up of these cases will determine whether these
lem. | assume that for the parents of Stealers, after modifications are sufficient for the task.
invervention there was nothing analogous to a The next question concerned the possibility of
negative reinforcement mechanism that could teaching the OSLC treatment procedures to other
come into play and maintain the family manage- therapists. Three new therapists were trained at
ment practices. : OSLC over a period of a year or more; each served
Our current work with the chronic adolescent as a co-therapist with the OSLC staff and partici-
delinquents presents a similar problem. Many of pated in the weekly (intensive) videotape feedback
the parents are unmotivated and/or extremely re- sessions. The replication team worked in their
sistant to changing their behavior. For these cases, own treatment setting; it had its own staff of
when they have brought the stealing and truancy trained observers, secretaries, and receptionists.
under partial control, the interval between con- Weinrott, Bauske, and Patterson (1979) compared

306
 Table 13.1
Outcome Data for the Treatment of Social Aggressors

(from Weinrott, Bauske, & Patterson, 1979, p. 343)

Terminated families Terminated families with

without follow-up follow-up data (minimum 4 months)
N Baseline Termination N Baseline Termination Follow-up

Original sample N6ixs 128 -475 axe .842 .461 -453

(Patterson, 1974a) SD .924 283 SD .618 ele .454
Replication sample 18 x .826 361 14x -900 .307 B23.
SD .458 .383 SD .493 349 193

Reprinted with permission from “Systematic replication of a social learning approach.” In P.O. Sjoden, S. Bates, and W.S. Dockens,
editors, Trends in Behavior Therapy. Copyright 1979 by Academic Press, Inc.

the outcome data for the 16 socially aggressive blocks of three to experimental and placebo
children treated in the original study to the 18 So- groups. The analysis showed the samples to be
cial Aggressors treated in the replication study. As matched for the number of children in the family,
can be seen in Table 13.1, both the original and occupational status of parent, and age of problem
the replication sample showed significant reduc- child. There were no significant differences be-
tions in TAB scores for the identified problem tween the groups on baseline TAB and PDR mea-
child. The time-series analyses for each case’s PDR sures. After four weeks, families in both groups
data showed significant reductions for the majori- were observed in their homes. As shown in Figure
ty of problem boys. Fleischman (1981) analyzed 13.3, deviant behaviors that were targeted in
the data for the entire set of Social Aggressors and treatment were significantly reduced for the exper-
Stealers and demonstrated that the outcomes for imental group. Comparable scores for the placebo
the replication sample were similar to those that group showed nonsignificant increases for the
were obtained in the earlier studies. The findings same time period. The PDR data showed compa-
showed that whatever it was that the OSLC thera- rable effects, with a significant reduction for the
py staff did in working with families, it could be experimental group and no change for the placebo
taught to others. group. These investigators introduced the interest-
Are these changes specific to the treatment? A ing feature of adding a measure of expectancy for
adequate answer to this question requires a com- improvement at each “treatment” session for
parison design and random assignment. The first members in both groups. Their analyses showed
pilot study of this type was designed to control for no significant differences between groups or
the passage of time. Six consecutive referrals were across trials. Both groups maintained high expec-
placed for two weeks in a waiting-list control tations for a successful outcome.
group (Wiltz & Patterson, 1974). The next six re- An adequate comparison study would involve
ferrals received the “standard” parent training. At longer time periods that are more representative of
the end of the 5-week period, the experimental the typical duration of OSLC treatment. On the
group showed a significant decrease in the obser- average, families receive from three months to one
vation codes measuring those child behaviors that year of treatment and 20 to 30 hours of therapy
were targeted during treatment. There was no time. In the study by Patterson, Reid, and Cham-
change in comparable scores for the control berlain (in preparation), 21 socially aggressive
group. However, this study did not use random families were randomly assigned to an experimen-
assignment; it was also flawed in that there were tal group or a comparison group. The comparison
marked differences in the baseline measures for group was told that they were on a waiting list to
the criterion variables. receive treatment and were given the option of be-
The next pilot study partially corrected both of ing referred for treatment to alternative communi-
these defects (Walter & Gilmore, 1973). Twelve ty agencies or to alternative therapists. All but one
consecutive referrals were randomly assigned in family in the comparison group chose to exercise

307
 p<.05. It showed the changes to be significantly
greater for the experimental group than for the
Figure 13.3 comparison group. The study showed that the
Changes in Targeted Deviant Behaviors outcome effects were specific to the treatment it-
self.
for Two Groups
This long series of studies brings us to the most
(fromWalter&Gilmore,
1973,p. 371) interesting questions of all. Why does the treat-
ment work? What is it that some parents do that
makes teaching them parenting skills so difficult?
5 Which clinical skills are most relevant for neutral-
| izing their resistance? These are reasonable ques-
.30 Deas tions; they will occupy center stage for our next
five years of research.
ome ~~
S
&
= .20
Set
o
Acknowledgments
Sus 2.15
2 The content of this chapter was drastically al-
ho] tered several times following staff meetings in
“ 10
which the family typology was discussed. My
05 Treatment thanks to John Reid, Patti Chamberlain, Marion
-=-=-=— Placebo
Forgatch, and Kate Kavanagh. In many respects,
this chapter is a summary of their clinical contacts
0
with these families. I also wish to thank Marion
Baseline 4 Weeks
Forgatch for her careful editing of this chapter.
Observations

Reprinted with permission from “Placebo versus social learning Footnotes

effects in parent training procedures designed to alter the be-
haviors of aggressive boys,” Behavior Therapy, 4, 361-377. 1. The Mommy and Daddy family type arose
Copyright 1973 by Academic Press, Inc. from the intensive discussion among the research
team headed by Marion Forgatch, Deborah Too-
bert, and Gary Wieder. It emerged as they were
the option. The OSLC staff then arranged for the studying the videotapes of family problem solving.
transfer and also arranged to pay the fee if one was 2. Given that the data show a significant reduc-
required. When an OSLC family was terminated, tion in deviant behavior, then it will also be neces-
their matched control family was also contacted sary to study the problem child’s acquisition of
and arrangements made for home observations prosocial skills relating to work, peer relations,
and daily telephone calls for PDR data. and academic survival. Such studies have yet to be
The mean TAB scores for the treated group at done. Another key area will be the demonstration
baseline were .92 and at termination were .32. that treatment changes family management prac-
The comparable values for the comparison group tices. It was only in 1980 that assessment devices
were .89 and .74. As predicted, the ANOVA for were introduced that tap into these variables.
repeated measured produced an F value for inter- Therefore, it will be another year or two before we
action of 4.63 (df=1:17) that was significant at have data relevant to these questions.

308
Appendices
 _Appendix 1.1
The Evolution of Social Learning Theories: Some Significant Contributors
(fromCairns,1979a,p. 334)

Antecedents First generation Second generation Third generation

(1900-1938) (1938-1960) (1960-1970) (1970- )

Interactional
analysis
G.R. Patterson
Psychoanalysis M.R. Yarrow
S. Freud R.Q. Bell
Learning theories Social learning i learning
Social i W.W. Hartu P
I.P. Pavlov R:R: Sears and personality

E.L.Thorndike J.W.Whiting oe haa hota

J.B. Watson N.E. Miller
R.H. Walters cognitive
C.L. Hull J. Dollard
reinterpretation
E.R. Guthrie J.B. Rotter
W. Mischel
E.C. Tolman
Behavior E. Maccoby
analysis J. Aronfreed
Cognitive theories Operant
S. Byou
J.M. Baldwin conditioning
D.M. Baer
J. Piaget B.F. Skinner
J.L. Gewirtz Social-
environmental
Field theory
K. Lewin
structure
H.L. Raush
R.D. Parke
U. Bronfenbrenner

Reprinted with permission from Social Development: The Origins and Plasticity of Interchanges by Robert B. Cairns. Copyright
©1979 by W.H. Freeman and Company.

310
 Appendix 3.1
Definitions for the 29 FICS Code Categories
(fromReid,1978)
AP Approval HU Humiliate PP Physical Positive
AT Attention IG Ignore RC Receive
CM Command IN Indulgence SS Self-stimulation
CN Command Negative LA Laugh TA halk
CO Compliance NC Non-compliance Er welease
cm. Cry NE Negativism TH Touch
DI _ Disapproval NO Normative WH Whine
DP Dependency NR _ No Response WK Work
DS ___Destructiveness Pie Play Vevey cl
HR High Rate PN Physical Negative

AP (APPROVAL): Approval is a clear indica- CN (COMMAND NEGATIVE): A negative

tion of positive interest or involvement. It is more command differs from the reasonable command in
reinforcing than Attend (AT). AT is a neutral or the manner in which it is delivered. This kind of
non-directive response whereas AP has reinforcing command must be characterized by at least one of
characteristics. Approval can be gestural or verbal the following: (1) immediate compliance is de-
in nature and need not be elaborate or lengthy, but manded; (2) aversive consequences are implicitly
should be used to indicate even the smallest posi- or actually threatened if compliance is not imme-
tive gesture. Approval is directed at behavior, ap- diate; (3) sarcasm or humiliation is directed to-
pearance, or personal characteristics of an individ- ward the receiver. Implicit use of aversive conse-
ual. It does not include the granting of permission quences is indicated by the tone of voice as well as
to carry out an activity. That is coded TA. the statement.

AT (ATTENTION): This category is to be used CO (COMPLIANCE): Use this category when a

when one person listens to or looks at another per- person does what is asked or indicates verbally or
son. Attending behavior may either be initiated by behaviorally that he will. Compliance need not
a person or may be in response to another person’s follow the CM, CN, or DP immediately; other be-
behavior. Sometimes, when listening is used as a havioral sequences can intervene. However, the
reason for coding AT, it may be difficult to tell if indication of compliance must occur within 12
the person is, in fact, listening. In general, unless seconds of a behavior coded as CM or CN. Delay
eye contact or some form of verbal recognition is of compliance beyond 12 seconds is NC. Com-
offered by persons supposedly listening to another mands which require compliance after a period of
person, the behavior of the respondent would be 12 seconds would not be coded CM or CN, nor
coded NR. Some form of non-verbal recognition is would the agreement to comply be coded CO.
necessary before a person’s behavior would be Both the request and response indicating compli-
coded AT. A brief glance should not be coded AT ance would be coded TA or possibly DI. These are
when it is an initiation. examples of what might be called future com-
mands.
CM (COMMAND): This category is used when
a direct, reasonable, and clearly stated request or CR (CRY): Use this category whenever a person
command is made to another person. The verbal sobs or cries tears. Actual tears do not have to be
statement must clearly specify the behavior which present.
is expected from the person to whom the com-
mand is directed. The code system requires that ei- DI (DISAPPROVAL): Use this category when-
ther compliance or non-compliance be coded ever a person gives a verbal or gestural criticism of
within 12 seconds. If the command requires com- another person’s behavior or characteristics. In
pliance in the future, code TA. verbal statements, it is essential that the content of

S14
the statement explicitly states criticism or disap- also be humiliating. Playful verbal statements or
proval of the subject’s behaviors or attributes, nicknames are not humiliations. Some people call
looks, clothes, possessions, etc. DI can be coded each other “stupid” more in terms of endearment
simultaneously with CM but never with CN, as than in humiliation.
CN always implies disapproval. Code DI only
when verbal disapproval (i.e., “I do not like you IG (IGNORE): Ignore is an intentional and de-
doing that”) or gestural disapproval is implied by liberate non-response to an initiated behavior.
facial expression, vigor of the gesture, or the criti- There is no doubt that the subject has heard but
cal tone of voice. In addition, a DI can only be has chosen not to respond.
coded if either the subject or the person interacting
with the subject directs the DI at the other member IN (INDULGENCE): Behavior is coded IN
of the dyad. Disapproval of a third person would when, without being asked, a person stops what
be coded TA. he is doing in order to do some behavior for an-
other person which that person is fully capable of
DP (DEPENDENCY): Behavior is coded DP doing for himself. Common kindness, i.e., pour-
when a person is requesting assistance in doing a ing a cup of coffee for another while also pouring
task that he is obviously capable of doing himself. one’s own, handing a nearby dictionary to some-
Everyday requests should not be coded DP—for one who has asked how to spell a word, are not to
example, requests made at dinner would be coded be coded IN. The helping person must stop his
TA unless the statement falls under the rules for own ongoing chain of behavior and perform an
coding CM. To code a behavior DP, it must meet unnecessary service for a capable person. General-
two criteria: the person is capable of doing the act ly, the consequence of IN is RC. Care must be tak-
himself, and it is an imposition on the other per- en to distinguish this category from DP and WK.
son to fulfill the request.
LA (LAUGH): Whenever a person laughs aloud
DS (DESTRUCTIVENESS): This category ap- pleasantly and in an agreeable manner, code LA.
plies to behavior in which a person destroys, dam- Simultaneous talking and laughing, code only LA.
ages, or attempts to damage anything other than a
person; attacks on persons are coded PN. The NC (NON-COMPLIANCE): This code is used
damages need not actually occur, but the potential when a person does not do what is requested of
for damage must exist, e.g., grabbing another’s him in response to a CM, CN or DP within 12 sec-
breakable materials. The value of the object is of onds of the request being made. Non-compliance
no consideration, nor is the actual amount of can be verbal or non-verbal in nature. Care must
damage done. be taken to distinguish DI from NC.

HR (HIGH RATE): This code is used for any NE (NEGATIVISM): This category is used only
very physically active, repetitive behavior not cov- when a person makes a statement in which the ver-
ered by other categories that, if carried on for a bal message is neutral, but which is delivered in a
sufficient period of time, would become aversive. tone of voice that conveys an attitude of “don’t
If the behavior can be coded by other categories, bug me,” or “don’t bother me.” Also included are
i.e., YE, PN, DS, then HR is not to be used. HR defeatist, “I-give-up” statements. This code is nev-
may be intermittently coded with other specific de- er to be used if the verbal meaning of the statement
viant behaviors. The prime goal in coding HR is to is interpreted as disapproving (DI) or humiliating
represent symbolically the observed behavior as (HU).
occurring excessively as measured by its frequency
and/or intensity. High rate behavior is the culmi- NO (NORMATIVE): The normative code is
nation of a series of behaviors which have acceler- used for routine behavior when no other code is
ated until they have reached an intolerable level as applicable.
judged by the observer.
NR (NO RESPONSE): Use this code when a
HU (HUMILIATE): This category is used when behavior does not require a response, or when a
a person makes fun of, shames, or embarrasses an- behavior is directed at another person but the per-
other person. The tone of voice (in terms of nasti- son to whom the behavior is directed fails to per-
ness or derisiveness), as well as the language used, ceive the behavior.
is of prime importance in meeting the criteria for
coding HU. Derisive or inappropriate laughter can PL (PLAY): This category is used when a person

312
is amusing himself, either alone or with other peo- ing activity required in PL or WK. Thus, in a game
ple. Play need not be restricted to games in which where one person says, “It’s your turn,” that is not
clear rules are defined, i.e., monopoly, scrabble, coded TA, but simply as PL. Likewise, in a work
or card games, but is applicable to many activities situation when one member of a dishwashing team
such as amusing oneself alone, with a pet, or play- says, “Here are some more dishes,” the proper
ing with toys. Play can be verbal or non-verbal. code is WK and not TA.

PN (PHYSICAL NEGATIVE): Use whenever a TE (TEASE): Teasing is defined as the act of

subject physically attacks or attempts to attack an- annoying, pestering, mocking, or making fun of
other person. The attack must be of sufficient in- another person. Teasing behavior is directed in
tensity to potentially inflict pain, i.e., biting, kick- such a manner that the other person is likely to
ing, slapping, hitting, spanking, or taking an ob- show displeasure and disapproval. This behavior
ject roughly from another person. The circum- is potentially provocative and disruptive to the
stances surrounding the act need not concern the other person.
observer, only the potential of inflicting pain.
TH (TOUCH): Use of this behavior code indi-
PP (PHYSICAL POSITIVE): This code is used cates non-verbal passing of objects or neutral non-
when a person caresses or communicates with verbal physical contact.
touch to another person in a friendly or affection-
ate manner. WH (WHINE): When a person uses a slurring,
nasal, or high-pitched voice, use this category.
RC (RECEIVE): Use this category when a per- The content of the statement can be of an approv-
son receives an object from another person or is ing, disapproving, or neutral quality; the main ele-
touched physically by a person and is passively ment is the voice quality.
showing no response to the contact. If the person
touched responds in some way, then the specific WK (WORK): Work is a behavior necessary to
response should be coded rather than RC. maintain the smooth functioning of a household;
it is necessary for a child to perform work in order
SS (SELF-STIMULATION): Use of this code is to learn behaviors that will help him to assume an
for a narrow class of behaviors which the individ- adult role. A definite service performed for anoth-
ual does to or for himself and cannot be coded by er person is also coded as WK.
any other codes.
YE (YELL): This category is to be used whenev-
TA (TALK): This code covers the exchange of er a person shouts, yells, or talks loudly. The
conversation between family members. It is used if sound must be intense enough that it is unpleasant
none of the other verbal codes are applicable. Do or potentially aversive if carried on for a sufficient
not use TA in cases when Talk is part of the ongo- length of time.

Sts
 _Appendix 4.1
OSLC Family Crisis List
Date

Case Name

Case Number

Therapist
When you fill this out, just circle whichever of these crises came up in the last seven days. We will ask
you to do this once a month.

Family 11. Pet picked up by dogcatcher.

i; Someone moved in with the family for a day 12. Pet injured—requires veterinarian.
13. Sentimental, useful, or valuable item lost.
or more.
2 Someone that was living with the family for a 14. Automobile accident, no one injured.
month or more left (not a parent). 15. Automobile accident, someone injured.
One of the parents left town temporarily 16. The car needs repairs.
(more than one day). 17. The car broke down or wouldn’t start.
18. Caught in a traffic jam.
One of the children left town overnight or
19. Ran out of gas.
longer.
20. Other
Someone returned from a long trip (over a
day).
Argument with spouse. Economic
Argument with child.
1. Lost some money.
Adult came home from work very upset.
Child came home from play, work, school, 2. Received unexpected bill.
3. Went to apply for welfare or unemployment
etc., very upset.
Conflict with ex-spouse. funds.
. Conflict with local relative. 4. Welfare or unemployment payments began.
. Conflict with out-of-town relative. 5. Welfare or unemployment stopped payment.
. Pleasant long-distance call from relative. 6. Something stolen from the house.
. Received bad news about a family member. 7. Something stolen from family member.
8. Other
. Physical fighting with family member.
. Wife is pregnant.
. Pregnancy suspected. Health
. Birth of a child.
. Other 1. Family member had a routine visit to doctor
or dentist.
2. Family member saw psychiatrist, psycholo-
Household and Transportation gist, counselor or other (list
Paid the bills.
Didn’t have enough money to pay the bills. 3. Someone in the family is ill.
A major repair was necessary for household 4. Sickness lasted more than three days.
or household item. 5. More than one person in the family is ill.
Check bounced. 6. Severe injury to a family member, e.g., bro-
Got a new babysitter. ken leg, pneumonia.
Babysitter quit. 7. Someone in the family learned they have a
Didn’t have any clean clothes. chronic illness, e.g., cancer, TB, muscular
Meal burned or ruined. dystrophy, etc.
Got evicted. 8. Someone in.the family, a relative, died.
Moved. 9. Other

314
School 2. Child had a serious disagreement with a neigh-
1. Child started new school. bor or friend.
ae School called to complain about child’s be- 3. Friend of a family member is having serious
havior. problems—called or came by to talk.
3; Child was sent home from school for behav- 4. Family member had an argument with repair
man, business person, government official,
ior.
4. Child was suspended from school. ete:
SiOrher
bie Child skipped school.
6. School complained about child’s academic
progress (doing poorly).
School called to say child may fail one or
more subjects. Legal
. School called to say child may have to repeat 1. Someone in the family went to see a lawyer.
grade. 2. Someonein the familyhada trafficviolation
. Child’s report card came out today. and gota ticket.
10. Other 3. Family member was arrested.
4. Policeman came to the door.
5. Somebody accused a family member of a
Social interchange crime.
im Adults had a serious disagreement with a 6. Family member appeared in court.
neighbor or friend. 7. Other

SES
 Appendix 5.1
Samples for Comparison Studies

The samples were drawn from a population of that their rates of coercive behavior exceeded .45
problem and nonproblem families studied at the responses per minute.
Oregon Social Learning Center (OSLC) since The normal sample was obtained by advertising
1968. Only those cases were selected falling into in the newspaper for families to be observed hav-
the age group 3 through 8.5 years or 8.6 through ing nonproblem children. A nonproblem family
13.5 years. For the clinical sample, only those was operationally defined as one in which the
cases were included that satisfied the dual require- child had not received psychiatric or counseling
ments for both age and diagnostic label socially treatment during the previous year. The families
aggressive. This is the group for which coercion were paid for participation in the study.
theory was designed. It consisted of boys referred There were 30 children in the clinical sample,
by parents and/or community agents because of and 20 in the nonproblem sample. The following
high rates of coercive behaviors directed toward table summarizes the demographic data for the
other people, e.g., tease, whine, yell, does not four samples including father absence, number of
mind, and fight. Of those referred for such prob- siblings, age of the target child, and occupational
lems, cases were selected for the sample when the level. The findings showed the samples to be com-
observation data collected in the home indicated parable on all variables.

Demographic Data for Two Samples

Mean Percent of
Mean Number Occupational Father-Present
Samples Age of Siblings Level? Families

Socially
aggressive 8.3 Dee, 4.9 68%
(N = 30) (2.8-13.5)t (0-5) (1-7)
Nonproblem 8.7 2.0 4.1 65%
(N = 20) (4.9-13.5) (1-4) (1-7)
‘Parent occupations were classified using Hollingshead and Redlich (1958), where level one was unskilled labor and levels four and
five were clerical and skilled labor, respectively.

tRanges are shown in parentheses.

316
 Appendix 6.1
The Mean Likelihood the Target Child Will Stop After a Single Coercive Response
Given that the Consequence was Neutral or Positive
(Corrected Base Rate)

Agent with Whom Child Interacts

Sample Mother Father Brother Sister

x 0.76 SaObS x 0.81

Abused SD 0.17 SDF071 i SD 0.17

N 19 N 13 N 15

x 0.74 x 0.76 xe O57) FeO.7o)

Normal SD 0.32 SID OSY SD 0.28 SDEO29

N 35 N 26 N 23 N 26

sxanORGY), xe 0872 x 0.70 x 0.82

Social Aggressor SD 0.20 SD 0.26 SD 0.25 SDHOEIS

N 43 N 23 N 23 N 30

tThere were less than 10 subjects in this cell.

317
 _ Appendix 6.2

Three samples were drawn from the total set of the family participated in a minimum of three
all available families studied at OSLC. The groups baseline home observations.
were Stealers, Social Aggressors, and Normals. To
be included in any of these groups the target child The demographic characteristics of the subjects
had to be a male between the ages of 5 and 13, and in the three samples are described below:

Demographic Information

Social Aggressors Stealers Normals

N=37 N=38 N=38

Target child’s
age at intake
ex 8.25 9.45 8.90
SD 1.81 2S 1.98
range 5.75—» 12.58 5.08—» 12.75 §.25—» 12.08

Family size
ax 4.4 a4 4.5
SD 1.41 1672 L227,
range 2+ 8 2+9 2+7

Father absent
in family
51% 32% 26%

Occupational status t
x 4.81 pil 4.1
SD 191 TESS: 1.86
range 1-7 3-7 17

tAs measured by the Hollingshead index.

The criteria for inclusion in one of the three stealing records for two of the three stealing mea-
groups were defined as follows:
sures.
There were 38 subject families with a target boy
who met the criteria to be included in the Stealer
Stealers
sample.
The target child had to be stealing at a high rate.
High-rate stealing was defined by three or more
events. Three methods of recording stealing were Social Aggressors
used: parent report at intake, Parent Daily Report Thirty-seven families were identified in which
(PDR), and court records. At intake, parents were there was a target boy who was identified as a So-
asked whether or not stealing was a problem be- cial Aggressor. The home observation data for
havior for their child. Eighty-two percent of the these boys showed them to be emitting aversive
parents of the boys in this sample responded affir- behaviors at 2.45 per minute. The parental com-
matively. During Parent Daily Report telephone plaints at intake identified the child as a severe
calls, 100% of the parents reported the target conduct problem. However, stealing was not iden-
child to be stealing. Fifty percent of the boys in tified as amajor problem. Criteria for stealing (de-
this sample also had court records for stealing. scribed above) were used, and the following deci-
Ninety-two percent of the boys in this sample had sion rules were applied. If stealing was mentioned

318
by parents at intake, it was required not to have
Normals
occurred more than once (or be defined as a prob-
lem) for a subject to be included in the Social Ag- Thirty-eight subject families were found who
gressor sample. If stealing was reported during the satisfied the criteria to be included in this sample.
PDR calls, it could not occur more than once. If The home observation data for these boys showed
there was a court record of stealing, the event them to be emitting aversive behaviors at <.45
could not have occurred during the baseline or per minute. Stealing was not a problem for these
treatment periods. No subject was included in this youngsters. Three subjects were reported to have
sample for whom there was a record of stealing on stolen once during PDR calls. None of these sub-
more than one measure. jects had a court record for stealing.

319
 _Appendix 8.1
The Pumpkin Experiment
(prepared by M. Forgatch)

This study was designed to test for the effect of $5.00 per observation hour during baseline and
negative reinforcement upon two categories of $7.50 per hour for experimental manipulations.
child behavior. On each of three days the mother The manipulations for this study were carried
and child participated in an ABA reversal design. out with the father absent.
The study took place in the home.
Apparatus
A cuing device was developed to provide imme-
Methods
diate feedback to the mother during the manipula-
tion. The apparatus consisted of a transmitter
Sample small enough to fit in the experimenter’s hand and
The subject family consisted of a Mother, a Fa- a receiver worn on the mother’s waist, with a tiny
ther, and a 3'%-year-old boy, who was the target hearing device in her ear. This enabled the experi-
child, and a 22-year-old sibling. At intake, the menter to speak directly, yet privately, to the
mother described the child as being an extremely mother. This improved communication between
difficult child to manage. On the problem behav- the experimenter and the mother during the ma-
iors checklist she indicated that 20 of the possible nipulation.
26 were problems. For the first baseline observa-
tion probe, the target child’s TAB score was .54. Manipulation
There were 21 baseline observation hours. The fa- The sequence of behaviors selected to be manip-
ther was present for seven baseline observations. ulated was: mother-Command —*child-Provoke
The family lived in a pleasant house in a middle- —» mother-Positive or -Neutral. These behaviors
class residential neighborhood in Eugene. were easy for the mother to distinguish and to per-
form. She did, however, emit many more com-
Observation Procedures mands than requested.
The data was collected by two professional ob- The mother learned the code definitions for the
servers who had experience and demonstrated reli- behavior relevant to the manipulation. She then
ability in using the FICS. As in the earlier study, wrote her commands to Pumpkin for one day.
observer agreement was checked for 20% of the Then the mother and experimenter practiced the
observations. Weekly retraining sessions were sequence of behaviors.
held. The observers were uninformed of the hy- Following each day’s manipulation, the experi-
potheses and manipulations. Most of the sessions menter provided the mother with corrective feed-
were in the late afternoon. The family was paid back. On the first day of the experiment, during

Effects of Manipulating Negative Reinforcement for Pumpkin’s Provoking Behavior

Days p(Pumpkin-Provoke|Mother-Command) Frequency Negative Reinforcement

Ay B Az Ay B A>
One .10 20 .00 0 5 0
Two 16 .40 .00 3) 5 0)
Three Kip ed wm
pours oy pik,2
Mean: 14 ee) .04 Sum: 4 ; 115) 0
As shown there, the mean baseline (A;) was .14. The three sessions of providing negative reinforcement in-
creased the strength of the connection to .29.
The right side of the table showed that the procedures were largely effective in controlling the reinforcing
behavior of the mother.

320
the manipulation (B) phase, the child became ex- structed to talk nicely to him, or to play with him,
tremely out of control. During each of the follow- or to begin doing something else. She was in-
ing three days, the instructions were changed so structed not to be aversive following a Provoke in
that, in addition to the negative reinforcement ar- this sequence.
rangement, the mother was to engage in a struc-
tured interaction with Pumpkin, such as play or
Results
work together.
The base rate for mother-Command was .08,
Design for Pumpkin-Ignore .01, and for Pumpkin-Pro-
The basic design was an A,, B, A; sequence on voke .04. Both of these child target events were
each of four days in the home. During the daily functionally related to mother-Command. For
baseline periods, which lasted 20 minutes each, child-Ignore, given mother-Command, the condi-
the mother was instructed to give Pumpkin ap- tional p value was .10; the comparable value for
proximately four commands. This was her base child-Provoke, given mother-Command, was .09.
rate for commands. These two behaviors constitute a functionally de-
During the 20-minute daily experimental phase, fined response class.
the mother was instructed to increase her com- The table summarizes the relation between
mands fourfold. Whenever Pumpkin responded to Child-Provoke and mother-Command during the
the command with a Provoke, the mother was in- three conditions for each of the three days.

321
 Appendix 11.1
Sample Comparison by Code Categories for Target Children

Social
Stealers Aggressors Normals
Variable (N =37) (N = 34) (N = 36) PZ oo) Duncan

AP .02 .02 04 Solar? N>ST&SA

AT .84 a 1.07 4.83" N>SA
CM 08 .07 .03 OM Ome ST&SA>N
CN 02 01 .00 6.33" ST&SA>N
pals) .20 05 W746i ST#SA#N
.04 .04 00 SP ST&SA>N
01 03 00 4.14* SA>N
01 01 .00 4:65" ST&SA>N
08 ol) 08 LON ie ST #SA#N
.02 10 01 4316 SA>N&ST
.09 .08 .04 Ve bik: ST&SA>N
.03 .04 01 5.44** ST&SA>N
.04 .07 01 EST eo SA>ST&N
.04 a1)i) 01 D930 SA>ST&N
YE 05 .06 01 62m ST&SA>N

Only those categories for which the differences were significant are listed.
*p<.05
"25<01
***5<001

aie
About the Author

Gerald R. Patterson is a research scientist at the Oregon Social

Learning Center. His work has been in the areas of aggression, -
family intervention in the homes of antisocial and delinquent |
boys, field observation procedures, and marital conflict. Dr. &
Patterson is the author of Families and Living with Children,
and is co-author of A Social Learning Approach to Family Inter-
vention, Volume 1: Families with Aggressive Children. He has written many articles for professional
journals and has been a consulting editor for Child Development.
Dr. Patterson is a past president of the Association for the Advancement of Behavior Therapy and
has received a Career Development Award from the National Institute of Mental Health. Much of the
present volume was written while he was a fellow in residence at the Center for Advanced Studies in
the Behavioral Sciences at Stanford University. In 1979 he was one of the recipients of an award for
Outstanding Research on Aggression from the International Society for Research on Aggression. In
1982 he has been selected to receive the Distinguished Scientist Award from the American Psychologi-
cal Association, Division of Clinical Psychology, Section III.
When Dr. Patterson is not working on research problems he can be found fly fishing for trout and
steelhead, and in the Winter searching for powder snow in the trees on the other side of the moun-
tains.

Sek
 References

Achenbach, T.M. The classification of children’s Antonovsky, H.F. A contribution to research in

psychiatric symptoms: A factor analytic study. area of mother/child relationships. Child De-
Psychological Monographs, 1966, 80 (7, velopment, 1959, 30, 37-51.
Whole No. 615). Argyle, M., & Cook, M. Gaze and mutual gaze.
Achenbach, T.M. The child behavior profile, I: London: Cambridge University Press, 1976.
Boys aged 6 through 11. Journal of Consulting Arnold) J:, Levine; “A:, °S Patterson, G:R.
and Clinical Psychology, 1978, 46(3). 478-488. Changes in sibling behavior following family
Achenbach, T.M., & Edelbrock, C.S. The classi- intervention. Journal of Consulting and Clini-
fication of child psychopathology: A review cal Psychology, 1975, 43, 683-688.
and analysis of empirical efforts toward con- Atkinson, J.A. Nonavailability of mother atten-
sulting and clinical psychology, 1978, in press. tion as an antecedent event for coercive de-
Ackerson, L. Children’s behavior problems (Vol. mands in the preschool child. Unpublished
1). University of Chicago Press, 1931. master’s thesis, University of Oregon, 1971.
Adler, A. The practice and theory of individual Attneave, F. Applications of information theory to
psychology. London: Routledge & Kegan Paul psychology: A summary of basic concepts,
Lid...1929. methods, and results. New York: Holt-Dryden,
Alexander, J.F., & Parsons, B.V. Short term be- 1959.
havioral intervention with delinquent families: Averill, J.R. Personal control over aversive stimuli
Impact on family process and recidivism. Jour- and its relationship to stress. Psychological Bul-
nal of Abnormal Psychology, 1973, 81, 219- letin, 1973, 80, 286-303.
DAY Azrin, N.H., Holtz, W. Ulrich, R., & Goldia-
Altmann, S. Sociobiology of the rhesus monkey. mond, I. The control of conversation through
II. Stochastics of social communication. Jour- reinforcement. Journal of Experimental Analy-
nal of Theoretical Biology, 1965, 8, 490-522. sis of Behavior, 1961, 4, 25-30.
Anderson, L.M. Personality characteristics of par- Azrin, N.H., Naster, B.J., & Jones, R. Reciproci-
ents of neurotic, aggressive, and normal pre- ty counseling: A rapid learning based procedure
adolescent boys. Journal of Consulting and for marital counseling. Behavior Research and
Clinical Psychology, 1969, 33(5), 575-581. Therapy, 1973, 11, 1-18.

325
Bach, G.R., & Wyden, P. The intimate enemy. Clinical Psychology, 1979. 47, 946-957.
New York: Avon Books, 1968. Baron, R.A. Social reinforcement effects as a func-
Bales, R.F. Interaction process and analysis. Cam- tion of social reinforcement history. Psychologi-
bridge, Massachusetts: Addison-Wesley, 1950. cal Review, 1966, 73, 527-540.
Bales, R.F. The equilibrium problem in small Baron, R.A. Aggression as a function of magni-
groups. In T. Parsons, R. Bales, & E. Shilo tude of victim’s pain cues, level of prior anger
(Eds.), Working papers in the theory of action. arousal, and aggressor-victim similarity. Jour-
New York: Free Press, 1953. nal of Personality and Social Psychology,
Baltes, M., & Reese, H.W. Operant research in 1971a, 18, 48-54.
violation of the operant paradigm. In B.C. Et- Baron, R.A. Magnitude of victim’s pain cues and
zel, J.M. LeBlanc, & D.M. Baer (Eds.), Contri- level of prior anger arousal as determinants of
butions to behavioral research. Hillsdale, New adult aggressive behavior. Journal of Personali-
Jersey: Lawrence Erlbaum Assoc., 1977. ty and Social Psychology, 1971b, 17(3), 236-
Bandura, A. Aggression: A social learning analy- 243.
sis. Englewood Cliffs, New Jersey: Prentice- Barrett, D.E., & Radke-Yarrow, M. Prosocial be-
Hall 1973. havior, social inferential ability, and assertive-
Bandura, A. Behavior theory and the models of ness in children. Child Development, 1977, 48,
man: Presidential address for the American Psy- 475-481.
chological Association. American Psychological Bartlett, M.S. The frequency goodness of fit test
Association Monitor, 1974, 5(11). for probability chains. Proceedings of the Cam-
Bandura, A. Social learning theory. New York: bridge Philosophical Society, 1951, 47, 86-95.
General Learning Press, 1977. Baumrind, D. Effects of authoritative parental
Bandura, A., & Walters, R.H. Adolescent aggres- control on child behavior. Child Development,
sion. New York: Ronald Press Co., 1959. 1966, 37(4), 887-907.

Bandura, A., & Walters, R.H. Social learning and Baumrind, D. Current patterns of parental au-
personality development. New York: Holt, thority. Developmental Psychology, 1971, 4(1),
Rinehart & Winston, 1963. 12.
Barber, T.X. Toward a theory of pain: Relief of Baumrind, D. Parental disciplinary patterns and
chronic pain by pre-frontal leucotomy, opiates, social competence in children. Youth and Socie-
placebos, and hypnosis. Psychological Bulletin, ty,:1978,)9(3 239-272.
1959, 56, 430-460. Baumrind, D. New directions in socialization re-
Barker, R.G. One boy’s day. New York: Harper search. Paper presented at the Society of Re-
and Row, 1951. search in Child Development, San Francisco,
OY).
Barker, R.G. The stream of behavior as an empiri-
cal problem. In R.G. Barker (Ed.), The stream Baumrind, D., & Black, A. Socialization practices
of behavior. New York: Appleton-Century- associated with dimensions of competence in
Crofts, 1963. preschool boys and girls. Child Development,
1967, 38, 291-307.
Barker, R.G. Ecological psychology. Stanford,
Bechtel, R.B. The study of man: Human move-
California: Stanford University Press, 1968.
ment and architecture. Transaction, May 1967,
Barker, R.G., Gump, P.V., Campbell, W.J., Bark- 53-56.
er, L.S., Willems, E., Friesen, W., Le Compte,
W., & Mikesell, E.H. Big school-small school. Becker, W.C. The relationship of factors in paren-
(University cooperative project #594). Kansas:
tal ratings of self and each other to the behavior
Midwest Psychological Field Station, 1962. of kindergarten children as rated by mothers,
fathers, and teachers. Journal of Consulting
Barker, R.G., & Wright, H.F. Midwest and its Psychology, 1960, 24, 507-527.
children. Evanston, Illinois: Row Peterson,
1954.
Becker, W.C., Madsen, C.H., Arnold, C.R., &
Thomas, D.R. The contingent use of teacher
Barnett, L., & Nietzel, M. Relationship of instru- attention and praise in reducing classroom be-
mental and affectional behavior and self esteem havior problems. Journal of Special Education,
to marital satisfaction in distressed and nondis- 1967, 1, 287-307.
tressed couples. Journal of Consulting and

326
Bell, R.Q. A reinterpretation of the direction of ef- Bing, E. Effect of child-rearing practices on devel-
fects in studies of socialization. Psychological opment of differential cognitive abilities. Child
Review, 1968, 75, 81-95. Development, 1963, 34, 631-648.
Bell, R.Q., & Harper, L.V. Child effects on Birchler, G.R., & Webb, L.F. Discriminating in-
adults. Hillsdale, New Jersey: Laurence Erl- teraction patterns in happy and unhappy mar-
baum Associates, 1977. riages. Journal of Consulting and Clinical Psy-
Bem, D. Self perception: An alternative interpreta- chology, 1977, 45(3), 494-495.
tion of cognitive dissonance phenomenon. Psy- Birchler, G.R., Weiss, R.L., & Vincent, J.P. Mul-
chological Review, 1967, 74, 183-200. ti-method analysis of social reinforcement ex-
Benning, J.J., Feldhusen, J.F., & Thurston, J.R. change between maritally distressed and non-
Delinquency prone youth: Longitudinal and distressed spouse and stranger dyads. Journal
preventive research. (Eau Claire County Youth of Personality and Social Psychology, 1975,
Study, Phase III). Wisconsin State Department 31(2), 349-360.
of Health and Social Services, 1968. Black, R.W. Shifts in magnitude of reward and
Berkowitz, L. The contagion of violence: An S-R contrast effects in instrument and selective
mediational analysis of some effects of observed learning. Psychological Review, 1968, 75, 114-
aggression. In W.J. Arnold & M.M. Page 126.
(Eds.), Nebraska Symposium on Motivation. Blau, P.M. Exchange and power in social life.
Lincoln: University of Nebraska Press, 1970. New York: John Wiley and Sons, 1964.
Berkowitz, L. Control of aggression. In B.M. Blood, R.O., & Wolfe, D.M. Husbands and
Caldwell & R. Riecute (Eds.), Review of child wives: The dynamics of married living. New
development research. University of Chicago York: The Free Press, 1960.
Press, 1973a. Blumberg, M. When parents hit out. In S.K. Stein-
Berkowitz, L. Words and symbols as stimuli to ag- metz & M.A. Straus (Eds.), Violence in the
gressive responses. In J. Knutson (Ed.), The family. New York: Dodd, Mead & Company,
control of aggression: Implications from basic 1974,
research. Chicago: Aldine Press, 1973b. Blumenthal, M., Kohn, R.L., Andrews, F.M., &
Berkowitz, L. Some determinants of impulsive ag- Head, K.B. Justifying violence: Attitudes of
gression: Role of mediated association with re- American men. Institute of Juvenile Research,
inforcement for aggression. Psychological Re- University of Michigan, 1972.
view, March 1974, 165-176. Blurton-Jones, N.G. Some aspects of the social be-
Berkowitz, L. Is criminal violence normative be- havior of children in nursery school. In D. Mor-
havior? Journal of Research in Crime and De- ris (Ed.), Primate ethology. London: Weiden-
linquency, 1978, 15(2), 148-161. feld & Nicholson, 1966.

Berlyne, D.E. Arousal and reinforcement. In D. Bobbitt, R., Gourevitch, V., Miller, L., & Jensen,
Levine (Ed.), Nebraska Symposium on Motiva- G. Dynamics of social interaction behavior: A
tion. Lincoln: University of Nebraska Press, computerized procedure for analyzing trends,
1967. patterns, and sequences. Psychological Bulletin,
Bernal, M.E., Delfini, L.F., North, J.A., & 1969, 71, 110-112.
Kreutzer, S.L. Comparison of boys’ behavior in Bogaard, L. Relationship between aggressive be-
homes and classrooms. In E. Mash, L. Handy, havior in children and parent perception of
& L. Hamerlynck (Eds.), Behavior modifica- child behavior. Unpublished doctoral disserta-
tion and families. New York: Brunner/Mazell, tion, University of Oregon, 1976.
1976. Brackbill, Y., & O’Hara, J. The relative effective-
Bernal, M.E., Gibson, D.M., Williams, D.E., & ness of reward and punishment for discrimina-
Pesses, D.I. A device for recording automatic tion learning in children. Journal of Compara-
audio tape recording. Journal of Applied Be- tive and Physiological Psychology, 1958, 51,
havior Analysis, 1971, 4, 151-156. 747-751.
Bernard, J.L., & Eisenman, R. Verbal condition- Bradburn, N.M., & Caplovitz, D. Reports on
ing in sociopaths with social and monetary rein- Happiness. Chicago: Aldine Publishing Co.,
forcement. Journal of Personality and Social 1965.
Psychology, 1967, 6(2), 203-206.

S27,
Brady, J.V., Porter, R.W., Conrad, D.G., & Ma- Burgess, R.L., & Huston, T.L. (Eds.). Social ex-
son, J.W. Avoidance behavior and the develop- change in developing relationships. New York:
ment of gastroduodenal ulcers. Journal of the Academic Press, 1979.
Experimental Analysis of Behavior, 1958, 1,
Burton, R.V. Validity of retrospective reports as-
69-72. sessed by the multitrait-multimethod analysis,
Bronfenbrenner, U. Socialization and social class Developmental Psychology Monographs,
through time and space. In E.E. Maccoby, M. 1970 43(Bie Ze
Newcomb, & E.L. Hartley (Eds.), Readings in Buss, A.H. The psychology of aggression. New
social psychology. New York: Holt, Rinehart York: John Wiley & Sons, 1961.
& Winston, 1958.
Buss, A.H. Instrumentality of aggression feedback
Bronfenbrenner, U. Towards an experimental and frustration as determinants of physical ag-
ecology of human development. American Psy- gression. Journal of Personality and Social Psy-
chologist, 1977, 513. chology, 1966a, 3, 153-162.
Brown, G.W., Bhrolchain, M.N., & Harris, T. Buss, A.H. The effect of harm on subsequent ag-
Social class and psychiatric disturbance among gression. Journal of Experimental Research in
women in an urban population. Sociology, Personality, 1966b, 1, 249-255.
1976, 9;,226-259.
Buss, A.H. Aggression pays. In J.L. Singer (Ed.),
Brown, G.W., Birley, J.L., & Wing, J.K. Influ- The control of aggression and violence. New
ence of family life on the course of schizophre- York: Academic Press, 1971.
nic disorders: A replication. British Journal of
Psychology, 1972, 121, 241-258. Byrne, D., & Rhamey, R. Magnitude of positive
and negative reinforcements as a determinant of
Brown, G.W., & Rutter, M. The measurement of attraction. Journal of Personality and Social
family activities and relationships: A methodo- Psychology, 1965, 2, 884-889.
logical study. Human Relations, 1966, 19,
241-263. Cairns, R.B. Fighting and punishment from a de-
velopmental perspective. Nebraska Symposium
Bryan, J.H., & Kapche, R. Psychopathy and ver- on Motivation. Lincoln: University of Nebras-
bal conditioning. Journal of Abnormal Psychol- ka Press, 1972.
ORV 9675) / 2574.
Cairns, R.B. (Ed.). The analysis of social interac-
Budd, K.S., Green, D.K., & Baer, D.N. An analy- tion: Methods, issues, and illustrations. Hills-
sis of multi misplaced parental social contingen- dale, New Jersey: Lawrence Erlbaum Associ-
cies. Journal of Applied Behavior Analysis, ates, 1979a.
1976, 9, 459-470.
Cairns, R.B. Social development: The origins and
Buehler, R.E., Patterson, G.R., & Furniss, J.M. plasticity of interchanges. San Francisco: W.H.
The reinforcement of behavior in institutional Freeman and Company, 1979b.
settings. Behavior Research and Therapy,
1966, 4, 157-167. Cairns, R.B., & Green, J.A. How to assess per-
sonality and social patterns: Observations or
Burgess, J.M., Kimball, W.H., & Burgess, R.L. ratings? In R.B. Cairns (Ed.), The analysis of
Family interaction as a function of family size. social interaction: Methods, issues, and illus-
Paper presented at the Southeastern Conference trations. Hillsdale, New Jersey: Lawrence Erl-
on Human Development, 1978. baum Associates, 1979.
Burgess, R.L. Child abuse: A behavioral analysis. Cairns, R.B., & Nakelski, J.S. On fighting mice:
In B.B. Lahey & A.E. Kazdin (Eds.), Advances Ontogenetic and experimental determinants.
in child clinical psychology. New York: Plen- Comparative and Physiological Psychology,
um, 1978. 1971, 74, 354-364.
Burgess, R.L., & Bushnell, D. Behavioral sociolo- Cairns, R.B., & Scholz, S. Fighting mice: Dyadic
gy: Experimental analysis of social process. escalation and what is learned. Comparative
New York: Columbia University Press, 1969. and Physiological Psychology, 1973, 85, 540-
Burgess, R.L., & Conger, R.D. Family interaction 550;
patterns related to child abuse and neglect: Some Caldwell, B.M. A new “approach” to behavioral
preliminary findings. Paper presented at the In- ecology. In J.P. Hill (Ed.), Minnesota Symposia
ternational Congress on Child Abuse and Neg- on Child Psychology (Vol. 2). Minneapolis:
lect, Geneva, Switzerland, September 20, 1976. University of Minnesota Press, 1968.

328
Caldwell, B.M. A new approach to behavioral psychoanalysis, and the sociology of gender.
ecology. In J.P. Hill (Ed.), Minnesota Symposia Berkeley, California: University of California
on Child Psychology (Vol. 5). Minneapolis: Press, 1978.
University of Minnesota Press, 1971. Christensen, A. Cost effectiveness in behavior
Call, D.J. Delinquency, frustration, and noncom- family therapy. Unpublished doctoral disserta-
mitment. Unpublished doctoral dissertation, tion, University of Oregon, 1976.
University of Oregon, 1965. Christiansen, K.O. The genesis of aggressive crim-
Camp, B.W. Verbal mediation in young aggressive inality: Implications of a study of crime in a
boys. Journal of Abnormal Psychology, 1977, Danish twin study. In J. DeWit & W.W. Har-
86, 145-153. tup (Eds.), Determinants and origins of aggres-
sive behavior. The Hague, Paris: Mouton,
Camp, B.W. Zimet, S., Doomineck, W., &
Dahlm, N. Verbal abilities in young aggressive 1974.
boys. Journal of Educational Psychology, 1977, Church, R.M. The varied effects of punishment
69 7A29-135. on behavior. Psychological Review, 1963, 70,
369-402.
Campbell, A. The American way of mating. Psy-
chology Today, 1975, 8, 37-43. Church, R.M. Response suppression. In B.A.
Campbell & R.M. Church (Eds.), Punishment
Campbell, B.A., & Church, R.M. (Eds.). Punish-
and aversive behavior. New York: Appleton-
ment and aversive behavior. New York: Apple-
Century-Crofts, 1969.
ton-Century-Crofts, 1969.
Church, R.M. & Getty, D.J. Some consequences
Carlson, W. Hyperactive, socially aggressive, and
of the reaction to an aversive event. Psychologi-
normal children observed in their natural home
cal Bulletin, 1972, 78, 21-27.
environment. Unpublished manuscript, 1979.
(Available from Humboldt State University, Ar- Cicirelli, V.G. Mother-child and sibling-sitting in-
cata, California 95521.) teractions on a problem solving task. Child De-
velopment, 1976, 47, 588-596.
Carlson, W. Factor generated response class scal-
ing. Unpublished manuscript, 1981a. (Availa- Clarke-Stewart, K. Interactions between mothers
ble from Humboldt State University, Arcata, and their young children: Characteristics and
California 95521.) consequences. Monographs of the Society for
Research in Child Development, 1973, 38(6-7,
Carlson, W. Hyperactive, conduct disordered,
Serial No. 153).
and normal children observed in their natural
home environments. Unpublished manuscript, Clarke-Stewart, K.A. And daddy makes three:
1981b. (Available from Humboldt State Uni- The father’s impact on mother and young child.
versity, Arcata, California 95521.) Child Development, 1978, 49, 466-478.
Casey, 1.G. A comparison of the effectiveness of Clement, P.W., & Milne, D.C. Group play thera-
approval and disapproval in modifying chil- py and tangible reinforcers used to modify the
dren’s behavior. Unpublished doctoral disserta- behavior of eight-year-old boys. Behavior Re-
tion, University of Wisconsin, 1967. search and Therapy, 1967, 5, 301-312.
Chamberlain, P. Standardization of a parent re- Coates, J.M. Population and education: How
port measure. Unpublished doctoral disserta- demographic trends will shape the U.S. Futur-
tion, University of Oregon, 1980. ist, February 1978, 35-42.
Chamberlain, P. Parent report data: Methodologi- Cobb, J.A. Survival skills and first grade academic
cal considerations. OSLC manuscript in prepa- achievement. Unpublished manuscript, Center
ration. of Research and Demonstration in the Early
Education of Handicapped Children, Universi-
Charlesworth, R., & Hartup, W.W. An observa-
ty of Oregon, December 1970.
tion study of positive social reinforcement in
the nursery school peer group. Child Develop- Cobb, J.A. The relationship of discrete classroom
ment, 1967, 38, 993-1002. behaviors to fourth-grade academic achieve-
ment. Journal of Educational Psychology,
Charny, I.W. The origins of violence between
1972, 63(1), 74-80.
spouses and siblings. Family Process, 1969, 8,
1-24. Cobb, J.A. & Hops, H. Effects of academic sur-
vival skill training on low achieving first grad-
Chodorow, N. The reproduction of mothering,
ers. Journal of Educational Research, 1973.

322
Cohen, J.A. Coefficient of agreement for nominal lems (Final Report, Vol. 1). American Institute
scales. Educational and Psychological Measure- for Research, Silver Springs, Maryland, 1970.
ment, 1960, 20(1), 37-46. Dawe, H.C. An analysis of two hundred quarrels
Collins, R.C. The treatment of disruptive behav- of preschool children. Child Development,
ior problems by employment of a partial-milieu LIS 45 SylS9 1572
consistency program. Unpublished doctoral Dawes, R.M. Jackknifing Guttman scales and un-
dissertation, University of Oregon, 1966. folded scales. Unpublished manuscript, 1978.
Conger, J.J. & Miller, W.C. Personality, social Delfini, L.F., Bernal, M.E., & Rosen, P.M. Com-
class, and delinquency. New York: John Wiley parison of deviant and normal boys in home
& Sons, 1966. settings. In E.J. Mash, L.A. Hamerlynck, &
Conger, R.D., & Burgess, R. Reciprocity: Equity L.C. Handy (Eds.), Behavior modification and
system stability. Unpublished manuscript, families. New York: Brunner/ Mazel, 1976.
1978. (Available from the University of Geor- de Madariaga, S. Hernan Cortes: Conqueror of
gia, Athens, Georgia.) of Mexico. New York: Doubleday, 1967.
Conger, R.D., & Smith, $.S$.Examining equity in De Master, B., Reid, J.B., & Twentyman, C. The
dyadic and family interactions: Is there any jus- effects of different amounts of feedback on ob-
tice? In E. Filsinger & R. Lewis (Eds.), Marital servers’ reliability. Behavior Therapy, 1976, 8,
observation and behavioral assessment: Recent 317-329.
developments and techniques. Beverly Hills,
California: Sage Press, 1981, in press. Denenberg, V.H. Developmental factors in ag-
gression. In J.F. Knutson (Ed.), The control of
Connolly, K., & Smith, P.K. Reactions of pre- aggression. Aldine Publishing Co., 1973.
school children to a strange observer. In N.
Blurton-Jones (Ed.), Ethological studies of Dengerink, H.A. Personality variables as media-
child behavior. Cambridge: Cambridge Univer- tors of attack instigated aggression. In R. Geen
sity Press, 1972. & E. O’Neal (Eds.), Perspectives in aggression.
New York: Academic Press, 1976, in press.
Coombs, C., Dawes, R., & Tversky, A. Mathe-
matical psychology: An elementary introduc- Dengerink, H.A., Schnedler, R.S., & Covey,
tion. Englewood Cliffs, New Jersey: Prentice- M.K. The role of avoidance in aggressive re-
Hall, Inc., 1970. sponses to attack and no attack. Journal of Per-
sonality and Social Psychology, 1978, 36,
Coopersmith, S. The antecedents of self-esteem. 1044-1053.
San Francisco: W.H. Freeman & Co., 1967.
Deur, J.L., & Parke, R.D. Effects of inconsistent
Corah, N.L., & Boffa, J. Perceived control, self- punishment on aggression in children. Develop-
observation, and response to an aversive stimu- mental Psychology, 1970, 2, 403-411.
lation. Journal of Personality and Social Psy-
chology, 1970, 16, 1-4. Devereaux, E. The role of peer group experience
in moral development. In J. Hill (Ed.), Minne-
Corte, H.E., Wolf, M.M., & Lecke, B.J. A com- sota Symposia on Child Psychology. University
parison of procedures for eliminating self injuri- of Minnesota Press, 1970.
ous behavior of retarded adolescents. Journal
of Applied Behavioral Analysis, 1971, 4, 201- Devine, V.T. The coercion process: A laboratory
analogue. Unpublished doctoral dissertation,
213:
State University of New York at Stony Brook,
Cox, R.D., Gunn, W.B., & Cox, M.J. A film as- 19744
sessment and comparison of the social skillful-
ness of behavior problem and nonproblem male Dodge, K. Social cognition and children’s aggres-
children. Paper presented at the Association for sive behavior. Child Development, 1980, 51,
the Advancement of Behavior Therapy, New 162-170.
York, December 3-5, 1976. Dohrenwend, B.S., & Dohrenwend, B.P. (Eds.).
Stressful life events: Their nature and effects.
Cronbach, L.J., Gleser, G.C., Nanda, H., & Ra-
jaratnam, N. The dependability of behavioral New York: John Wiley & Sons, 1974.
measurements: Theory of generalizability of Dollard, J., Doob, L.W., Miller, N.E., Mowrer,
scores and profiles. New York: John Wiley & O.H., & Sears, R.R. Frustration and aggres-
Sons, 1972. sion. New Haven, Connecticut: Yale University
Press, 1939.
Cureton, L. Early identification of behavior prob-

330
Douglas, J.W. The school progress of nervous and Eriksen, C.W. (Ed.). Behavior and awareness: A
troublesome children. British Journal of Psychi- symposium of research and interpretation. Dur-
atry, 1966, 112, 1115-1116. ham, North Carolina: Duke University Press,
Douglas, J.W., Lawson, A., Cooper, J.E., 1962.
Cooper, E. Family interaction and the activities Erlanger, H.S. Social class and corporal punish-
of young children. Journal of Child Psychology ment in child rearing: A reassessment. Ameri-
and Psychiatry, 1968, 9, 157-171. can Sociological Review, 1974, 39, 68-85.
Dreger, R.M., Lewis, P.M., Rich, T.A., Miller, Eron, L.D., Walder, L.O., Huesmann, L.R., &
K.S., Reid, M.P., Overlade, D., Taffel, C., & Lefkowitz, M.M. The convergence of labora-
Flemming, E. Behavioral classification project. tory and field studies of the development of ag-
Journal of Consulting Psychology, 1964, 28, gression. In J. DeWit & W.W. Hartup (Eds.),
1-13. Determinants and origins of aggressive behav-
Driscoll, R. Incidental and intentional learning ior. The Hague, Paris: Mouton, 1974.
and social competence in high risk children. Pa- Eron, L.D., Walder, L.O., & Lefkowitz, M.M.
per presented at the 87th Annual Convention of Learning of aggression in children. Boston: Lit-
the American Psychological Association, New tle Brown & Co., 1971.
York, September 1979. Estes, W.K. An experimental study of punish-
Dulany, D.E. The place of hypotheses and inten- ment. Psychological Monographs, 1944, 57(3,
tions: An analysis of verbal control in verbal Whole No. 263).
conditioning. In C. Eriksen (Ed.), Behavior and Estes, W.K. Reward in human learning: Theoreti-
awareness: A symposium of research and inter- cal issues and strategic choice points. In R. Gla-
pretation. Duke University Press, 1962. ser (Ed.), The nature of reinforcement. New
Eibl-Eibesfeldt, I. Phylogenetic adaptation as de- York: Academic Press, 1971.
terminants of aggressive behavior in man. In J. Fagan, O.S., Langner, T.S., Gersten, J.C., & Eis-
DeWit & W.W. Hartup (Eds.), Determinants enberg, J. Violent and antisocial behavior: A
and origins of aggressive behavior. The Hague, longitudinal study of urban youth. Interim
Paris: Mouton, 1974. report. New York: Division of Epidemiology,
Elliott, D.S. Experimental study of the labelling Columbia University School of Public Health,
hypothesis. Paper presented at the Annual Con- £977.
ference of the American Society on Criminolo- Fagot, B. Sex related stereotyping of toddlers’ be-
gy, Philadelphia, November 1979. haviors. Developmental Psychology, 1973, 9,
Elliott, D.S., Ageton, S.S., & Canter, R.J. An in- 429.
tegrated theoretical perspective on delinquent Fagot, B. Sex differences in toddlers’ behavior and
behaviors. Journal of Research in Crime and parental reaction. Developmental Psychology,
Delinquency, January 1979, 3-27. 1974, 10(4), 554-558.
Elliott, D.S., & Huizinga, D. Defining patterned Fagot, B. Reinforcing contingencies for sex role
delinquency: A conceptual typology of delin- behavior: Effect of experience with children.
quency offenders. Paper presented at the 32nd Child Development, 1978a, 49, 30-36.
Annual Conference of the American Society on
Fagot, B. The influence of sex of child and paren-
Criminology, San Francisco, November 1980.
tal reactions to toddler children. Child Devel-
Elliott, D.S., & Voss, N.L. Delinquency and- opent, 1978b, 49, 459.
dropout. Lexington, Massachusetts: D.C.
Heath & Co., 1974. Fagot, B., & Patterson, G.R. An in vivo analysis
of reinforcing contingencies for sex-role behav-
Emerson, R. Exchange theory (Part I): A psycho- iors in the preschool child. Developmental Psy-
- logical basis for social exchange. In J. Berger, chology, 1969, 1(5), 563-568.
M. Zelditch Jr., & B. Anderson (Eds.), Socio-
logical theories in progress. New York: Hough- Fawl, C.L. Disturbances experienced by children
ton-Mifflin Co., 1972. in their natural habitats. In R. Barker (Ed.),
The stream of behavior. New York: Appleton-
Erickson, M.R. Effects of initial information and Century-Crofts, 1963.
observational set upon social perception. Un-
published doctoral dissertation, University of Ferguson, L., Partzka, L., & Lester, B. Patterns of
Oregon, 1973. parent perception differentiating clinic from

331
 nonclinic children. Journal of Abnormal Child the Oregon Social Learning Center, 207 E. Sth,
Psychology, 1974, 2, 169-182. Suite 202, Eugene, Oregon 97401.)
Feshbach, N., & Feshbach, S. The relationship Forster, A. Violence of the fanatical left and right.
between empathy and aggression in two age Annals of American Academy of Political and
groups. Developmental Psychology,. 1969, 1, Social Science, 1966, 364, 141-148.
102-107. Fowles, D.C. The three arousal model: Implica-
Feshbach, S. Aggression. In P.H. Mussen (Ed.), tions of Gray’s two factor learning theory for
Carmichael’s manual of child psychology (Vol. heart rate, electrodermal activity, and psycho-
2). New York: John Wiley & Sons, 1970. pathy. Psychophysiology, 1980, 17(2), 87-104.
Festinger, L. The treatment of qualitative data by Freud, S. Beyond the pleasure principle. London:
scale analysis. Psychological Bulletin, 1947, 44, Psychoanalytic Press, 1922.
149-161. Friedrich, L.K., & Stein, A.H. Aggressive and
Field, M.H., & Field, H.F. Marital violence and prosocial television programs and the natural
the criminal process: Neither justice nor peace. behavior of preschool children. Monographs of
Social Service Review, 1973, 47(2), 221-240. the Society for Research in Child Development,
Fiske, D.W. The limits for the conventional sci- 1973, 38(4).
ence of personality. Journal of Personality, Furman, W., & Masters, J. A Bayesian approach
1974, 42(1), 1-11. to convergent discriminant validation: illus-
Fleischman, M.J. Training and evaluation of ag- trated by validationals between social learning
gressive children. Proposal submitted to NIMH constructs of reinforcement and punishment
Crime and Delinquency section, April 1979a. and their affective consequences. Unpublished
manuscript, 1978. (Available from the Univer-
Fleischman, M.J. Using parenting salaries to con- sity of Minnesota Institute of Child Welfare.)
trol attrition and cooperation in therapy. Be-
havior Therapy, 1979b, 10, 111-116. Gardner, W.P., Mitchell, C., & Hartmann, D.P.
Some problems in the analysis of sequential
Fleischman, M.J. A replication of Patterson’s in- data. Paper presented at the annual meeting of
tervention for boys with conduct problems. the Association for the Advancement of Behav-
Journal of Consulting and Clinical Psychology, ior Therapy, San Francisco, December 1979.
1981, 49, 342-354.
Garmezy, N. The experimental study of children
Fleischman, M.J., & Szykula, S.A. Acommunity vulnerable to psychopathology. In A. Davids
setting repliction of a social learning treatment (Ed.), Child personality and psychopathology
for aggressive children. Behavior Therapy, (Vol. 2). New York: John Wiley & Sons, 1976.
reat 12(1), 115-122,
Garmezy, N., & Devine, V.T. Project compe-
Follick, M.J., & Knutson, J.F. Punishment of irri-
tence: The Minnosota studies of children vul-
table aggression. Aggressive Behavior, 1977, 4,
nerable to psychopathology. In N. Watt, J.
1-17. Rolf, & E.J. Anthony (Eds.), Children at risk
Forehand, R., King, H., Peed, S., & Yoder, P. for schizophrenia. New York: Cambridge Uni-
Mother-child interaction: Comparison of a versity Press, 1982, in press.
noncompliant clinic group and a_ nonclinic
Garmezy, N., & Nuechterlein, K. Invulnerable
group. Behavior Research and Therapy, 1975,
children: Fact and fiction of competence and
13, 79-84.
disadvantage. Paper presented at the meeting of
Forehand, R., Wells, K., & Griest, D. An exami- the American Orthopsychology Association,
nation of the social validity of a parent training Detroit, April 1972.
program. Behavior Therapy, 1980, 11, 488-
Garmezy, N., & Streitman, S. Children at risk:
502. The search for the antecedents of schizophre-
Forehand, R., Wells, K., & Sturgis, E. Predictors nia. Part 1. Conceptual models and research
of child noncompliant behavior in the home. methods. Schizophrenia Bulletin, Spring 1974,
Journal of Consulting and Clinical Psychology, No. 8.
1978, 46, 179. Gelfand, D.M., Gelfand, S., & Dobson, W.R.
Forgatch, M., & Wieder, G. The parent adoles- Program reinforcement of patient behavior in a
cent naturalistic interaction code (PANIC). Un- mental hospital. Behavior Research and Thera-
published manuscript, 1981. (Available from py, 1967, 5, 201-207.

332
Gelles, R.J. The violent home: A study of physical Goodenough, F.L. Anger in young children. Min-
aggression between husbands and wives. Bever- neapolis: University of Minnesota Press, 1931.
ly Hills, California: Sage Publications, 1972. Goodstein, L.D., & Rowley, V.N. A further study
Gelles, R.J. An exploratory study of intrafamily of MMPI differences between parents of dis-
violence. Unpublished doctoral dissertation, turbed and nondisturbed children. Journal of
University of New Hampshire, 1973. Consulting Psychology, 1961, 25(5), 460-464.
Gelles, R.J. Violence towards children in the Goody, E., & Groothues, C.M. Stress in mar-
United States. Paper presented at the meeting of riage: West Atrican couples in London. In V.S.
the American Association for the Advancement Khan (Ed.), Minority families in Britain. Mac-
of Science, 1977. Millan, 1979.
Gerstens |)-C., Langner, T.S., Eisenberg, J-G., Gottman, J.M. Toward a definition of social iso-
Simcha-Fagan, O., & McCarthy, E.D. Stability lation in children. Child Development, 1977,
in change in types of behavioral disturbance of 48, 513-517.
children and adolescents. Journal of Abnormal Gottman, J.M. Marital interaction: Experimental
Child Psychology, 1976, 4, 111-127. investigations. New York: Academic Press,
Gewirtz, J. Deprivation and satiation of social 172.
stimuli as determinants of their efficacy. In J.P. Gottman, J.M., & Bakeman, R. The sequential
Hill (Ed.), Minnesota Symposia on Child Psy- analysis of observation data. In S$.Suomi, M.
chology (Vol. 1). Minneapolis: University of Lamb, & G. Stephenson (Eds.), Social interac-
Minnesota Press, 1967. tion analysis: Methodological issues. Madison:
Gewirtz, J. & Boyd, E. Experiments on mother- University of Wisconsin Press, 1979.
infant interaction underlying mutual attach- Gottman, J.M., Gonso, J., & Rasmussen, B. So-
ment acquisition: The infant conditions the cial interaction, social competence, and friend-
mother. In T. Alloway, P. Pliner, & L. Krames ship in children. Child Development, 1975, 46,
(Eds.), Attachment behavior. New York: Plen- 709-718.
um Publishing Corp., 1977.
Gouldner, A.W. The norm of reciprocity: A pre-
Glairn, J.P., & Annesley, F.R. Reading and arith- liminary statement. American Sociological Re-
metic correlates of conduct and problems and view, 1960, 25, 161-177.
withdrawn children. Journal of Special Educa-
Grant, M. The army of the Caesars. New York:
tion, 1971, 5, 213-218.
Charles Scribner’s Sons, 1974.
Glenn, N. Psychological well being in the post pa-
Graubard, P.S. The relationship between academ-
rental stage: Some evidence from national sur-
ic achievement and behavior dimensions. Ex-
veys. Journal of Marriage and Family, 1975,
ceptional Children, 1971, 37, 755-757.
37, 105-110.
Glueck, S., & Glueck, E. Delinquents and nonde- Graziano, A.M., & Fink, R.S. Second-order ef-
linquents in perspective. Cambridge: Harvard fects in mental health treatments. Journal of
Consulting and Clinical Psychology, 1973, 40,
University Press, 1968.
356-364.
Goldin, P.C. A review of children’s reports of pa-
rent behaviors. Psychological Bulletin, 1969, Green, E.H. Group play and quarreling among
VA 22-236. preschool children. Child Development, 1933,
4, 302-307.
Golding, W. Lord of the Flies, New York: Cow-
ard-McCann Inc., 1962. Green, R.G., Stonner, D., & Shope, G.L. The fa-
cilitation of aggression by aggression: Evidence
Goldstein, M.J., and Judd, L., Rodnick, E.H., against the catharsis hypothesis. Journal of Per-
Alkire, A., & Gould, E.A. A method for study- sonality and Social Psychology, 1975, 31, 721-
ing social influence and coping patterns with 726.
families of disturbed adolescents. Journal of
Nervous and Mental Diseases, 1968, 147(3), Greenwood, C.R., Hops, H., Delquardri, J., &
233-251. Guild, J. Group contingencies for group conse-
quences in classroom management: A further
Goodenough, F.L. Inter-relationships in the be- analysis. Journal of Applied Behavior Analysis,
havior of young children. Child Development, 1974, 7, 413-425.
1930, 1, 29-47.
Griest, D., Wells, K., Forehand, R. An examina-

333
 tion of predictors of maternal perceptions of 1968, 162, 1243-1248.
maladjustment in clinic referred children. Jour- Hare, R.D. Psychopathy, autonomic functioning,
nal of Abnormal Psychology, 1979, 88(3), 277- and the orienting response. Journal of Abnor-
281. mal Psychology Monograph Supplement,
Griffith, W. Behavior difficulties of children as 1968, 73(3), 1-24.
perceived and judged by parents, teachers, and Harris, A.M. Observer effect on family interac-
children themselves. Minneapolis: University of tion. Unpublished doctoral dissertation, Uni-
Minnesota Press, 1952. versity of Oregon, 1969.
Grimm, J.A., Parsons, J.A., & Bijou, S.W. A Harris, A., & Reid, J.B. The consistency of a
technique for minimizing subject-observer class of coercive child behaviors across school
looking interactions in a field setting. Journal of settings for individual subjects. Journal of Ab-
Experimental Child Psychology, 1972, 14, normal Child Psychology, 1981, in prepara-
500-505.
tion.
Gurin, G., Verhoff, J., & Feld, S. Americans view
Hartmann, D.P. Influence of symbolically mod-
their mental health: A nationwide survey. New
eled instrumental aggression and pain cues on
York: Basic Books, 1960.
aggressive behavior. Journal of Personality and
Guthrie, E.R. The competing skeletal response Social Psychology, 1969, 11, 280-288.
hypothesis. In E.E. Boe & R.M. Church
Hartmann, D.P. Considerations in the choice of
(Eds.), Punishment: Issues and experiments.
interobserver reliability estimates. Journal of
New York: Appleton-Century-Crofts, 1978.
Applied Behavior Analysis, 1977, 10, 103-116.
Guttman, L.A. A basis for scaling qualitative
Hartup, W.W. Peer interaction and social organi-
data. American Sociological Review, 1944, 9,
zation. In P. Mussen (Ed.), Carmichael’s manu-
139-150. al of child psychology. New York: John Wiley
Haggard, E. Experimental studies in affective & Sons, 1969.
processes: I. Some aspects of cognitive struc- Hartup, W.W. Violence in development: The
tures and active participation on certain auto- functions of aggression in childhood. Papers
nomic reactions during and following experi- presented at the American Psychological Asso-
mentally induced stress. Journal of Experimen- ciation, Montreal, Quebec, Canada, 1973.
tal Psychology, 1943, 33, 257-284.
Hartup, W.W. Aggression in childhood: Develop-
Haley, J. Problem solving therapy. San Francisco: mental perspectives. American Psychologist,
Jossey-Bass, 1978. 1974, 29, 336-339.
Hall, W.M. Observational and interactional de- Hartup, W.W. Peer interaction and the process of
terminants of aggressive behavior in boys. Un- socialization. In M.J. Guralnick (Ed.), Early in-
published doctoral dissertation, Indiana Uni- tervention and the integration of handicapped
versity, 1973. and non-handicapped children. Baltimore: Uni-
Hall, R.V., Lund, D., & Jackson, D. Effects of versity Park Press, 1977.
teacher attention on study behavior. Journal of Hartup, W.W. Children and their friends. In H.
Applied Behavior Analysis, 1968, 1, 1-12. McGurk (Ed.), Child social development. Lon-
Halverson, C.F., & Waldrop, M.F. Maternal be- don: Methuen, 1978, in press.
havior toward own and other preschool chil- Hartup, W.W. Levels of analysis in the study of
dren: The problem of “ownness.” Child Devel- social interaction: An historical perspective. In
opment, 1970, 41, 839-845. M.E. Lamb, S.J. Suomi, & G.R. Stephenson
Hamburg, D., & vanLawick-Goodall, J. Factors (Eds.), Social interaction analysis. Wisconsin:
facilitating development of aggressive behavior Univeristy of Wisconsin Press, 1979.
in chimpanzees and humans. In J. DeWit & Hatfield, E., Utne, M., & Traupman, J. Equity
W.W. Hartup (Eds.), Determinants and origins
theory and intimate relationships. In R. Burgess
of aggressive behavior. The Hague, Paris: Mou- & T. Huston (Eds.), Social exchange in devel-
ton, 1974.
oping relationships. New York: Academic
Harbin, H.T., & Madden, D.J. Battered parents: Press, 1979.
A new syndrome. American Journal of Psychia- Hedlund, C.S. Relationship of positive social rein-
try, 1979, 136(10), 1288-1291. forcement to delinquency and sub types of de-
Hardin, G. The tragedy of the commons. Science, linquent behavior: A text of Patterson’s etiolog-

334
 ical model. Unpublished master’s thesis, Uni- and coping in divorce: Afocus on women. Un-
versity of West Virginia, 1971. published manuscript, 1980.
’ Hefferline, R.F. Learning theory and clinical psy- Heyns, R.W., & Lippitt, R. Systematic observa-
chology: An eventual symbiosis. In A.J. Bach- tional techniques. In G. Lindsey (Ed.), Hand-
rach (Ed.), Experimental foundations in clini- book ofsocial psychology (Vol. 1). Cambridge,
cal psychology. New York: Basic Books, 1963. Massachusetts: Addison-Wesley, 1954.
Hefferline, R.F., Keenan, B., & Harford, R.A. Hibbs, D.A. Problems of statistical estimation and
Escape and avoidance conditioning in human causal influence in time-series regression mod-
subjects without their observation of the re- els. In H.C. Costner (Ed.), Sociological metho-
sponse. Science, 1959, 130, 1338-1339. dology. San Francisco: Jossey-Bass, 1973-74.
Helfer, R.E., & Kempe, C.H. (Eds.). The bat- Hicks, M.W., & Platt, M. Marital happiness and
tered child. Chicago: University of Chicago stability: A review of the research in the sixties.
Press, 1968. In C. Broderick (Ed.), A decade of family re-
Henry, M.M., & Sharpe, D.F. Some influential search. National Council on Family Relations,
factors in the determination of aggressive be- 1970,
havior in preschool children. Child Develop- Hinde, R.A. The study of aggression: Determi-
ment, 1947, 18, 11-28. nants, consequences, goals, and functions. In J.
DeWit & W.W. Hartup (Eds.), Determinants
Herbert, E. Parent programs—bringing it all back
and origins of aggressive behavior. The Hague,
home. Paper presented at the meeting of the
American Psychological Association, Miami, Paris: Mouton, 1974.
1970. Hineline, P. Negative reinforcement and avoid-
ance. In W. Honig & J.E. Staddon (Eds.),
Herbert, E.W., Pinkston, E.M., Hayden, M.L.,
Handbook of operant behavior. New Jersey:
Sajwaj, T.E., Pinkston, S., Cordua, G., &
Prentice-Hall, 1977.
Jackson, C. Adverse effects of differential pa-
rental attention. Journal of Applied Behavior Hines, P.A. How adults perceive children: The ef-
Analysis, 1973, 6, 15-30. fect of behavior tracking and expected deviance
Herrnstein, R.J. Relative and absolute strength of on teachers’ impressions of a child. Unpub-
response as a function of frequency of rein- lished doctoral dissertation, University of Ore-
forcement. Journal of Experimental Analysis of gon, 1974.
Behavior, 1961, 4, 267-272. Hirschi, T. Causes of delinquency. Berkeley: Uni-
Herrnstein, R.J. Formal properties of the match- versity of California Press, 1969.
ing law. Journal of the Experimental Analysis Hobbs, S.A., & Forehand, R. Important parame-
of Behavior, 1974, 21, 159-164. ters in the use of timeout with children: A re-ex-
amination. Journal of Behavioral Therapy and
Hetherington, E.M. Effects of father absence on
Experimental Psychiatry, 1977, 8, 365-370.
personality development in adolescent daugh-
ters. Developmental Psychology, 1972, 7(3), Hoffman, W.H. Statistical models for the study of
313-326. change in the single case. In C. Harris (Ed.),
Problems in measuring change. Madison Wis-
Hetherington, Cox, & Cox, 1976—see Hether-
consin: University of Wisconsin Press, 1967.
ington, Cox, & Cox, 1978.
Hetherington, E.M., Cox, M., & Cox, R. After-
Hokanson, J.E. Vascular and psychogalvanic ef-
math of divorce. In J.H. Stevens, Jr. & M.
fects on experimentally aroused anger. Journal
Matthews (Eds.), Mother-child, father-child re- of Personality, 1961, 29, 30-39.
lations. Washington, D.C.: National Associa- Hokanson, J.E., & Edelman, R. Effects of three
tion for the Education of Young Children, social responses on vascular processes. Journal
1978. of Personality and Social Psychology, 1966, 3,
Hetherington, E.M., Cox, M., & Cox, R. Family 442-447,
interaction and the social, emotional, and cog- Hokanson, J.E., Willers, K.R., & Koropsak, E.
nitive development of children following di- The modification of autonomic responses dur-
vorce. In V. Vaughn & T. Brazelton (Eds.), ing aggressive interchange. Journal of Personal-
The family: Setting priorities. New York: Sci- ity, 1968, 36, 386-404.
ence and Medicine, 1979. Holleran, P.R. Noise and aggressiveness as deter-
Hetherington, E.M., Cox, M., & Cox, R. Stress minants of aggressive behavior. Unpublished

335
 doctoral dissertation, University of Notre Huston, T.L., & Burgess, R.L. Social exchange in
Dame, 1977. developing relationships: An overview. In R.
Holleran, P. Parent perception and delinquency. Burgess & T. Huston (Eds.), Social exchange in
OSLC manuscript in preparation. . developing relationships. New York: Academic
Press, 1979, in press.
Hollingshead, A.B., & Redlich, F.C. Social class
and mental illness. New York: John Wiley & Jacobson, N.S., & Margolin, G. Marital therapy.
Sons, 1958. New York: Brunner/ Mazel, 1979.
Holmberg, M.C. The development ofsocial inter- Jenkins, R.L., & Glickman, S. Common syn-
change patterns from 12 to 42 months’ cross- dromes in child psychiatry. American Journal
sectional and short-term longitudinal analyses. of Orthopsychiatry, 1946, 16, 244-253.
Unpublished doctoral dissertation, University Jenkins, R.L., & Hewitt, L. Types of personality
of North Carolina, 1977. structure encountered in child guidance clinics.
Holmes, T.H., & Masuda, M. Life changes and American Journal of Orthopsychiatry, 1944,
illness susceptibility. In B.S. Dohrenwend & 14(1), 84-94.
B.P. Dohrenwend (Eds.), Stressful life events: Jennings, H.H. Leadership and isolation (2nd
Their nature and effects. New York: John Wil- Ed.). London: Longmans, Green & Co., 1950.
ey & Sons, 1974. Johns, J.H., & Quay, H.C. The effect of social re-
Homans, G.C. Social behavior: Its elementary ward on verbal conditioning in psychopathic
forms. New York: Harcourt, Brace & World, and neurotic military offenders. Journal of
ines. 19618 Consulting Psychology, 1962, 26, 217-220.
Honig, A.S., Tannenbaum, J., & Caldwell, B. Johnson, $.M., & Bolstad, O.D. Methodological
Maternal behavior in verbal report and in la- issues in naturalistic observations: Some prob-
boratory observations. Paper presented at the lems and solutions for field research. In L.A.
meeting of the American Psychological Associa- Hamerlynck, L.C. Handy, & E.J. Mash (Eds.),
tion, San Francisco, 1968. Behavior change: Methodology, concepts, and
Hood, R., & Sparks, R. Key issues in criminolo- practice. Champaign, Illinois: Research Press,
gy. London, Weidenfeld & Nicholson, 1970. 1973.
Hops, H., & Cobb, J. Initial investigations into Johnson, S.M., & Bolstad, O.D. Reactivity to
academic survival-skill training: Direct instruc- home observation: A comparison of audio re-
tion and first-grade achievement. Journal of corded behavior with observers present or ab-
Educational Psychology, 1974, 66(4), 548-553. sent. Journal of Applied Behavior Analysis,
OMS feintestKea
lets.
Hops & Cobb, 1977—see Hops & Cobb, 1974.
Johnson, S.M., Bolstad, O.D., & Lobitz, G.K.
Hops, H., Walker, H., & Greenwood, C.R. Peers Generalization and contrast phenomena in be-
—a program for remediating social withdrawal havior modification with children. In L.A.
in the school setting: Aspects of a research de- Hamerlynck, L.C. Handy, & E.J. Mash (Eds.),
velopment process. CORBEH Report #33, Oc- Behavior modification with families. I. Theory
tober 1977. and research. II. Applications and develop-
Horne, A. Aggressive behavior in referred and ments. New York: Brunner/ Mazel, 1976.
nonreferred one- and two-parent families. Pre- Johnson, S.M., & Christensen, A. Multiple criter-
sented as part of a Symposium on Advances in ia follow-up of behavior modification with
Behavioral Treatment of One-Parent Families. families. Journal of Abnormal Child Psycholo-
Paper presented at the 88th Annual Convention gy, 1975, 3, 135-154.
of the American Psychological Association,
Montreal, Canada, Sept., 1980. Johnson, S.M., & Lobitz, G.K. Parental manipu-
lations of child behavior in home observations.
Hotchkiss, J.M. The modification of maladaptive Journal of Applied Behavior Analysis, 1974a,
behavior of a class of educationally handi- jjogp28265\
capped children: Operant conditioning tech-
niques. Unpublished doctoral dissertation, Uni- Johnson, $.M., & Lotitz, G.K. The personal and
versity of Southern California, 1966. marital adjustment of parents as related to ob-
served child deviance and parenting behaviors.
Hull, C.L. Principles of behavior. New York: Ap- Journal of Abnormal Child Psychology, 1974b,
pleton-Century-Crofts, 1943. pd0) Pi pe24 UF

336
Johnson, $.M., Wahl, G., Martin, S., & Johans- ships in small groups. In P.H. Leiderman & D.
son, S. How deviant is the normal child: A be- Shapiro (Eds.), Psychological approaches to so-
havioral analysis of the preschool child and his cial behavior. Stanford University Press, 1964.
family. In R.D. Rubin, J.P. Brady, & J.D. Hen- Karpowitz, D. Stimulus control in family interac-
derson (Eds.), Advances in behavior therapy tion sequences as observed in the naturalistic
(Vol. 4). New York: Academic Press, 1973. setting of the home. Unpublished doctoral dis-
Johnston, M. Responsiveness of delinquents and sertation, University of Oregon, 1972.
nondelinquents to social reinforcement. British Kass, R.E., & O’Leary, K.D. The effects of ob-
Journal of Social and Clinical Psychology, server bias in field-experimental settings. Paper
1976, 15, 41-49. presented at a symposium, “Behavior analysis
Jones, R.R. Behavioral observation and frequency in education,” University of Kansas, Lawrence,
data: Problems in scoring, analysis, and inter- Kansas, April 9, 1970.
pretation. In L.A. Hamerlynck, L.C. Handy, & Kelly, R., & Stephens, M.W. Comparison of dif-
E.J. Mash (Eds.), Behavior change: Methodol- ferent patterns of social reinforcement in chil-
ogy, concepts, and practice. Champaign, Illi- dren’s operant learning. Journal of Compara-
nois: Research Press, 1973. tive Physiological Psychology, 1964, 57, 294-
Jones, R.R., Reid, J.B., & Patterson, G.R. Natu- 296.
ralistic observations in clinical assessment. In P. Kendon, A. Some functions of gaze direction in
McReynolds (Ed.), Advances in psychological social interactions. Acta Psychologica, 1967,
assessment (Vol. 3). San Francisco: Jossey- 26, 22-63.
Bass, 1975.
Kent, oRoNsO Leary, «KD Diaments) Conk
Kaffman, M., & Elizur, E. Infants who become Dietz, A. Expectation biases in observational
enuretic: A longitudinal study of 161 Kibbutz evaluation of therapeutic change. Journal of
children. Monographs, 1977, 42(2), 1-61. Consulting and Clinical Psychology, 1974,
Kagan, J. The child’s perception of parent punish- 42(6), 774-780.
ment. Journal of Abnormal and Social Psychol- Kirkpatrick, R. Deviant behavior of children in
Ogy,71956, 53, 257-258. multiple settings. Unpublished master’s thesis,
Kagan, J., & Moss, H. Birth to maturity. New University of Oregon, 1978.
York: John Wiley & Sons, 1962. Knutson, J.F. Aggression as manipulable behav-
Kahneman, D., & Tversky, A. On the psychology ior. In J. Knutson (Ed.), The control of aggres-
of prediction. Psychological Review, 1973, sion: Implications from basic research. Chi-
80(4), 237-251. cago: Aldine Publishing Co., 1973.
Kanfer, F.H. Self-regulation: Research issues and Knutson, J.F., & Hyman, M. Predatory aggres-
speculations. In C. Neuringer & J.L. Michael sion and irritable aggression: Shock-induced
(Eds.), Behavior modification in clinical psy- fighting in mouse-killing rats. Physiology and
chology. New York: Appleton-Century-Crofts, Behavior, 1973, 11, 113-115.
1970. Kohn, M.L. Social class and schizophrenia: A crit-
Kanter, F.11., Cox, L., Greiner, J., & Karoly, P. ical review and reformulation. Schizophrenia
Contracts, demand characteristics, and self- Bulletin, 1973, 7, 60-79.
control. Journal of Personality and Social Psy- Kolb, T.M., & Straus, M.A. Marital power and
chology, 1974, 30, 605-619. marital happiness in relation to problem solving
Kanfer, F.H., & Phillips, J. The learning founda- ability. Journal of Marriage and the Family,
tions of behavior therapy. New York: John Wil- 1974, 36, 756-766.
ey & Sons, 1970. Kopfstein, D. The effects of accelerating and de-
Kanfer, F.H., & Seidner, M. Self-control: Factors celerating consequences on the social behavior
enhancing tolerance of noxious stimulus. Jour- of trainable retarded children. Child Develop-
nal of Personality and Social Psychology, 1973, ment, 1972, 43, 800-809.
25, 381-389. Kuenstler, W.H. Differential effects of positive
Kantor, J.R. Interbehavioral psychology. Gran- and negative social reinforcement on juvenile
ville, Ohio: The Principia Press, Inc., 1959. delinquents and Sunday school students. Un-
Kaplan, H.B., Burch, N.R., & Bloom, S. Physio- published doctoral dissertation, University of
logical covariation and sociometric relation- Houston, 1970.

337
Kuo, Z.Y. The genesis of the cat’s response to the 1973, 28, 417-425.
rat. Journal of Comparative Psychology, 1930, Leiter, M.P. A study of reciprocity in preschool
1151-35, play groups. Child Development, 1977, 48,
Kuo, Z.Y. Further study on the behavior of the cat 1288-1295.
toward the rat. Journal of Comparative Psy- Lennard, H., & Revenstein, A. Patterns in human
chology, 1938, 25, 1-8. interaction. San Francisco: Jossey-Bass, 1969.
Lacey, J.I. Somatic response patterning and stress: Lentz, W.P. Rural urban differentials and juvenile
Some revisions of activation theory. In M. delinquency. Journal of Criminal Law, Crimi-
Appley & R. Trumball (Eds.), Psychological nology, and Police Science, 1956, 47, 331-339.
stress: Issues in research. New York: Appleton-
Century-Crofts, 1967. Lepper, M.R., & Dafoe, J.L. Incentives, con-
straints, and motivation in the classroom: An
Lagerspetz, K. Studies on the aggressiveness in attributional analysis. In I. Frieze, D. Bar-Tal,
mice. Helsinki: Suomalainen Tiedeakatemia, & J. Carroll (Eds.), Attribution theory: Appli-
1964. cations to social problems. San Francisco: Jos-
Lagerspetz, K. Modification of aggressiveness in sey-Bass, 1980, in press.
mice. In S. Feshbach & A. Fraczek (Eds.), Ag- Levin, G.R., & Simmons, J. Response to praise by
gression and behavior change. New York: Prae- emotionally disturbed boys. Psychological Re-
ger, 1980. DontslI62ay TiO:
Laivgueur, H., Peterson, R.F., Sheese, J.G., & Pe- Levin, G.R., & Simmons, J. Response to food and
terson, L. Behavioral treatment in the home: praise by emotionally disturbed boys. Psycho-
Effects on an untreated sibling and long term logical Reports, 1962b, 11, 539-546.
follow-up. Behavior Therapy, 1973, 4, 431-
441. Levinger, G. Sources of marital dissatisfaction
among applicants for divorce. American Jour-
Lamb, M.E. Fathers: Forgotten contributors to nal of Orthopsychiatry, 1966, 36, 803-807.
child development. Human Development,
1975, 18, 245-266. Levinger, G. A social exchange view on the disso-
lution of pair relationships. In R. Burgess & T.
Lamb, M.E. The role of the father: An overview. Huston (Eds.), Social exchange in developing
In M.E. Lamb (Ed.), The role of the father in relationships. New York: Academic Press,
child development. New York: John Wiley & 1979.
Sons, 1976.
Levitt, E.E. Research on psychotherapy with chil-
Lambert, W.W. Promise and problems of cross dren: An evaluation. Journal of Consulting Psy-
cultural exploration of children’s aggressive chology, 1957, 21, 189-196.
strategies. In J. DeWit & W.W. Hartup (Eds.),
Determinants and origins of aggressive behav- Levitt, E.E. Research on psychotherapy with chil-
ior. The Hague, Paris: Mouton, 1974. dren. In A.E. Bergin & S.L. Garfield (Eds.),
Handbook of psychotherapy and behavior
Lang, P.J. The application of psychophysiological change. New York: John Wiley & Sons, 1971.
methods to the study of psychotherapy and be-
havior modification. In A.E. Bergin & S.L. Lewis, O. La vida. Panther Modern Society, Seck-
Garfield (Eds.), Handbook of psychotherapy er, Warbberg, and Panther, 1968.
and behavior change: An empirical analysis. Liberman, R.P., Wallace, C.J., Vaughn, C.E.,
New York: John Wiley & Sons, 1971. Snyder, K.S., & Rust, C. Social and family fac-
Lapouse, R., & Monk, M.A. An epidemiologic tors in the course of schizophrenia: Towards an
study of behavior characteristics in children. interpersonal problem-solving therapy for
American Journal of Public Health, 1958, schizophrenics and their families. Paper pre-
sented at the Conference on Psychotherapy of
48(9), 1134-1144.
Schizophrenia: Current Status and New Direc-
Lazarus, R.S., & Lanier, R. Stress related trans- tion, Yale University School of Medicine, New
actions between person and environment. In Haven, Connecticut, April 1979.
L.A. Pervin & M. Lewis (Eds.), Perspectives in
interactional psychology. Plenum Publishing Lichtenstein, E. How to quit smoking. Psychology
Co., 1978. Today, 1971, 4, 42-44.
Lefcourt, H.M. The function of the illusion of Lieven, E.V.M. Turn-taking and pragmatics: Two
control and freedom. American Psychologist, issues in early child language. In B.N. Campbell

338
 & P.T. Smith (Eds.), Recent advances in the ing of child behavior. Unpublished master’s the-
psychology of language. New York: Plenum sis, University of Oregon, 1978.
Press, 1976. Lorber, R. Parental tracking of childhood behav-
Lindner, R. The jet propelled couch. In R. Lind- ior as a function of family stress. Unpublished
ner, The fifty-minute hour. New York: Rhine- doctoral dissertation, University of Oregon,
hart & Co., 1955. 1981.
Lindskold, S. Trust development, the GRIT pro- Lorenz, K. On aggression. New York: Harcourt,
posal, and the effects of conciliatory acts in Brace & World, Inc., 1966.
conflict and cooperation. Psychological Bulle- Lott, A.J., & Lott, B.E. Group cohesiveness as in-
tin, 1978, 85, 772-793. terpersonal attractions: A review of relation-
Littman, D., Freund, R., & Schmaling, K. Paren- ships with antecedent and consequent varia-
tal categorization of child behaviors. OSLC bles. Psychological Bulletin, 1965, 64,
manuscript in preparation. 259-309.
Littman, D., & Patterson, G.R. Unpredictable ag- Lovaas, O.I. Effect of exposure to symbolic ag-
gression: A common dilemma. OSLC manu- gression on aggressive behavior. Child Devel-
script in preparation, 1980. opment, 1961a, 32, 37-44.
Littman, D., Patterson, G.R., & Forgatch, M.S. Lovaas, O.]. Interaction between verbal and non-
Maternal attribution. . . . OSLC manuscript in verbal behavior. Child Development, 1961b,
preparation, 1980. 32, 329-336.
Liverant, S. MMPI differences between parents of Lykken, D.J. A study of anxiety in the sociopathic
disturbed and non-disturbed children. Journal personality. Journal of Abnormal Psychology,
of Clinical Psychology, 1959, 23, 256-260. 1957-.55.46-10:
Lobitz, W.C., & Johnson, S.M. Parental manip- Lytton, H. Observation studies of parent-child in-
ulation of the behavior of normal and deviant teraction: A methodological review. Child De-
children. Child Development, 1975, 46, 719- velopment, 1971, 42, 651-684.
TXSe Maccoby, E. (Ed.) The development of sex differ-
Lobitz, G.K., & Johnson, $.M. Normal versus ences. Stanford University Press, 1966.
deviant children: A multi method comparison. Maccoby, E., & Jacklin, C. The psychology of
Journal of Abnormal Child Psychology, 1976, sex differences. Stanford University Press,
3(4), 353-373. 1974.
Loeber, R. Optimal prediction of family interac- MacFarlane, J.W., Allen, L., & Honzik, M. A de-
tions on the basis ofseries of antecedents. Paper velopmental study of the behavior problems of
presented at the meeting of the Association for normal children between 21 months and 14
the Advancement of Behavior Therapy, San years. Berkeley, California: University of Cali-
Francisco, December 1979. fornia Press, 1962.
Loeber, R. Child precursors of assaultive behavior Madsen, C., Becker, W., & Thomas, D. Rules,
in males. Research proposal submitted to the praise, and ignoring: Elements of elementary
National Institute of Mental Health, Crime and classroom control. Journal of Applied Behavior
Delinquency Section, October 1980. Analysis, 1968, 1, 139-150.
Loeber, R., & Janda, W. Mother intervention in Magee, R.D. Correlates of aggressive defiant
sibling conflict. OSLC manuscript in prepara- classroom behavior in elementary school boys:
tion. A factor analytic study. Dissertation Abstracts,
Loeber, R., Janda, W., & Reid, J.B. Family in- 1964, 25(2), 1340-1341.
teractions of assaultive adolescents, stealers, Maier, N.R. Frustration in the study of behavior
and nondelinquents. OSLC manuscript in prep- without a goal. Ann Arbor, Michigan: Univer-
aration. sity of Michigan Press, 1961.
Loeber, R., Patterson, G.R., & Dishion, T., Mul- Marcus, L.M. Studies of attention in children vul-
tiple gating: A multi-stage assessment proce- nerable to psychopathology. Unpublished doc-
dure for identifying youth at risk for delinquen- toral dissertation, University of Minnesota,
cy. OSLC working paper, July 1981. 1972.
Lorber, R. The effects of set induction upon track- Margolin, G. A sequential analysis of dyadic com-

$39
 munication, Paper presented at the meeting, of and the validity of interviews for research in
the Association for the Advancement of Behav- child development. Child Development, 1961,
ior Therapy, Atlanta, Georgia, December 32, 171-185,
1977, Mednick, $., & Christiansen, K,O. Biosocial ba-
Margolin, G, Joint marital therapy to erihance an- ses of criminal behavior. New York: Gardner
ger management and reduce spouse abuse, Press, 1977,
American Journal of Family Therapy, 1979, in Mednick, S.A., & Hutchings, B. Criminality in
press. adoptees and their adoptive and biological par-
Martens, E.H,, & Russ, H,Adjustment of behav- ents: A pilot study. In S.A. Mednick & K.O.
ior problems of school children; A description Christiansen (Eds.), Biosocial bases of criminal
and evaluation of the clinical program in Berke- behavior. New York: Gardner Press, Inc.,
ley, California. U.S, Office of Education Bulle WL
yar
tin, 1932, 78(18), Megargee, E.1. Matricide, patricide, and the dy-
Martin, B. Reward and punishment associated namics of aggression, Paper presented at the
with the same goal response: A factor in the Annual Convention of the American Psycholog-
learning of motives. Psychological Bulletin, ical Association, Washington, D.C., 1967.
1963, 60, 441-451, Meichenbaum, D. Cognitive-behavioral modifica-
Martin, J.A., Maccoby, E.E., Baron, K.W., & tion: Future directions. In P, Sjoden, S. Bates,
Jacklin, C.N, The sequential analysis of moth- & W.S. Dockens (Eds.), Trends in behavior
er-child interaction at 18 months: A compari- therapy. New York: Academic Press, 1979.
son of micro analytic methods, Developmental Menzel, H,A new coefficient for scalogram analy-
Psychology, 1980, in press. ses. Public Opinion Quarterly, Summer 1953,
Mash, EJ. Behavior modification and methodolo- 268-280,
gy: A developmental perspective. The Journal Mercatoris, M., & Craighead, W.E. The effect of
of Educational Thought, 1976, 10, 5-21. nonparticipant observation on teacher and pu-
Masling, J., & Stern, G. Effect of the observer in pil classroom behavior. Journal of Educational
the classroom, Journal of Educational Psychol Psychology, 1973.
ogy, 1969, 60, 351-354, Meyer, W.J., & Seidman, S.B. Relative effective-
McAdoo, W., & Connally, EF.MMPI’s of parents ness ofdifferent reinforcement combinations on
in dysfunctional families. Journal of Consulting concept learning of children at two develop-
and Clinical Pseyhology, 1975, 43, 270. mental levels. Child Development, 1961, 32,
117-127,
McCaffrey, J., & Cummings, J. Behavior patterns
associated with persistent emotional distur- Miller, N.E. Theory and experiment relating psy-
bances of children in regular classes of elemen- choanalytic displacement to stimulus reponse
tary grades, Onodaya County, New York: generalization, Journal of Abnormal and Social
Mental Health Research Unit, New York State Psychology, 1948, 43, 155-178.
Dept. of Mental Hygiene, 1967, Miller, N.E., & Dollard, J. Social learning and
MceClearn, G.E., & DeFries, J.C, Genetics and imitation. New Haven, Connecticut: Yale Uni-
mouse aggression, In J.F. Knutson (Ed.), The versity Press, 1941,
control of aggression: Implications from basic Miller, $.M., & Grant, R. The blunting hypothe-
research, Chicago: Aldine Publishing Co., sis: A view of predictability and human stress.
1973. In P. Sjoden, S. Bates, & W.S,. Dockens (Eds.),
McCord, J. A life history approach to criminal be- Trends in behavior therapy. New York: Aca-
havior, Paper presented at the American Society demic Press, 1979,
on Criminology, Atlanta, Georgia, November Minton, C., Kagan, J., & Levine, J. Maternal
1977, control and obedience in the two-year-old.
McCord, J. Antecedents and correlates of vulnera- Child Development, 1971, 42, 1873-1894,
bility and resistance to psychopathology. In R. Minuchin, $., Montalvo, B., Guerney, B., Ros-
Zucker & A. Rabin (Eds.), Further explora- man, B., & Schumer, F. Families of the slums.
tions in personality. New York: John Wiley & New York: Basic Books, 1967,
Sons, L98O,
Mischel, W. A social-learning view of sex differ-
McCord, J., & McCord, W. Cultural stereotypes ences in behavior, In E. Maccoby (Ed.), The

340
 development of sex differences. Stanford, Cali- 1979, 14(54), 313-326.
fornia: Stanford University Press, 1966. Nisbett, R.E., & Wilson, T.D. Telling more than
Mischel, W. Personality and assessment. New we can know: Verbal reports on mental proc-
York: John Wiley & Sons, 1968. esses. Psychological Review, 1977, 84(3), 231-
Mischel, W. On the interface of cognition and per- DOR
sonality: Beyond the person-situation debate. Nord, W. B. Social exchange theory: An integra-
American Psychologist, 1979, 9, 740-755. tive approach to social conformity. Psychologi-
Moore, D. Determinants of deviancy: A behavior- cal Bulletin, 1969, 71, 174-208.
al comparison of normal and deviant children Novaco, R.W. Anger control. Lexington, Massa-
in multiple settings. Unpublished manuscript, chusetts: Lexington Books/D.C. Heath & Co.,
University of Tennessee, 1975. 1975;
Moore, D., Chamberlain, P., & Mukai, L. Chil- Novick, J., Rosenfeld, E., Block, D.A. Situational
dren at risk for delinquency: A follow-up com- variations in the behavior of children. Paper
parison of aggressive children and children who presented at a lecture, “Research on child devel-
steal. Journal of Abnormal Child Psychology, opment,” Minneapolis, Minnesota, 1965.
1979,°7(3); 345-355. O’Conner, R.D. Modification of social withdrawl
Moore, D., Forgatch, M., Mukai, L., & Toobert, through symbolic modeling. Journal of Applied
D. Interactional coding system. Unpublished Behavior Analysis, 1969, 2, 15-22.
manuscript, Oregon Social Learning Center, O’Leary, K.D., Kent, R.N., & Karpowitz, J.
1977. Shaping data collection congruent with experi-
Moore, S. Correlates of peer acceptance in nursery mental hypotheses. Journal of Applied Behav-
school children. Young Children, 1967, 22, ior Analysis, 1975, 8, 43-51.
281-297. Oltmanns, T., Broderick, J., & O’Leary, K. Mari-
Moos, R.H. Behavioral effects of being observed: tal adjustment and the efficacy of behavior ther-
Reactions to a wireless radio transmitter. Jour- apy with children. Journal of Consulting and
nal of Consulting and Clinical Psychology, Clinical Psychology, 1977, 45(5), 724-729.
1968, 32(4), 383-388. Olweus, D. Personality factors and aggression
Moustakas, C.E., Sigel, I., & Schalock, H. An with special reference to violence within the
objective method for the measurement and ana- peer group. In J. DeWit & W.W. Hartup
lysis of child/adult interaction. Child Develop- (Eds.), Determinants and origins of aggressive
ment, 1956, 27, 109-134. behavior. The Hague, Paris: Mouton Press,
Murphy, L.B. Social behavior and child personali- 1974.
ty. New York: Columbia University Press, Olweus, D. Development of multi-faceted aggres-
1937. sion inventory for boys. Reports from the Insti-
Myers, J., Lindenthal, J., & Pepper, M. Social tute of Psychology, 1975, 6.
class, life events, and psychiatric symptoms. In Olweus, D. Longitudinal studies of aggressive re-
B.S. Dohrenwend & B.P. Dohrenwend (Eds.), action patterns: A review. Paper presented at
Stressful life events. New York: John Wiley & the 21st International Congress of Psychology,
Sons, 1974. Paris, July 1976.
Nelson, C., & Knutson, J.F. Sex, strain, and- Olweus, D. Aggression in the schools: Bullies and
housing: Variables influencing the effects of pri- whipping boys. New York: John Wiley & Sons,
or shock exposure on shock-induced aggres- 1973:
sion. Aggressive Behavior, 1978, 4, 237-252. Olweus, D. Stability of aggressive reaction pat-
Nelson, R.O., Lipenski, D.P., & Black, J.L. The terns in males: A review. Psychological Bulle-
relative reactivity of external observations and tin, 1979, 86(4), 852-875.
self monitoring. Behavior Therapy, 1976, 7, Olweus, D. Familial determinants of aggressive
314-321. behavior in adolescent boys: A causal analysis.
Nemeth, C. Bargaining and reciprocity. Psycho- Developmental Psychology, 1981, in press.
logical Bulletin, 1970, 74, 297-308. Omark, D.R., Omark, M., & Edelman, M. Dom-
Nielsen, A., & Gerber, D. Psychosocial aspects of inance hierarchies in young children. Paper pre-
truancy in early adolescence. Adolescence, sented at the Congress of Anthropological and

341
 Ethnological Science, Chicago, 1973. presented at the 4th Biennial Meeting of the
Orzech, M. The effect of verbal reward and verbal Southeastern Conference on Human Develop-
punishment on delinquent and nondelinquent ment, April 1976.
adolescent boys. Unpublished doctoral disser- Patterson, G.R. An empirical approach to the
tation, Wayne State University, 1962. classification of disturbed children. Journal of
Paris, $S.G., & Cairns, R.B. An experimental and Clinical Psychology, 1964, 20, 326-337.
ethological analysis of social reinforcement Patterson, G.R. An application of conditioning
with retarded children. Child Development, techniques to the control of a hyperactive child.
1972, 43;-717-729. In L. Ullmann & L. Krasner (Eds.), Research in
Parke, R.D. Effectiveness of punishment as an in- behavior modification. New York: Holt, Rine-
teraction of intensity, timing, agent, nurtur- hart & Winston, 1965a.
ance, and cognitive structure. Child Devel- Patterson, G.R. Responsiveness to social stimuli.
opment, 1969, 40, 213-235. In L. Ullmann & L. Krashner (Eds.), Research
Parke, R.D. The role of punishment in the sociali- in behavior modification. New York: Holt,
zation process. In Early experiences and the Rinehart & Winston, 1965b.
process of socialization. New York: Academic Patterson, G.R. Behavioral techniques based upon
Press, 1970. social learning: An additional base for develop-
Parke, R.D. Rules, roles, and resistance to devia- ing behavior modification technologies. In C.
tion: Recent advances in punishment, disci- Franks (Ed.), Behavior therapy: Appraisal and
pline, and self control. In Minnesota Symposia status. New York: McGraw-Hill, 1969.
on Child Psychology.' Minneapolis, Minnesota: Patterson, G.R. Changes in status of family mem-
University of Minnesota Press, 1975. bers as controlling stimuli: A basis for describ-
Parke, R.D. Interactional designs. In R.B. Cairns ing treatment process. In L.A. Hamerlynck,
(Ed.), The analysis of social interaction. Hills- L.C. Handy, & E.J. Mash (Eds.), Behavior
dale, New Jersey: Lawrence Erlbaum Associ- change: Methodology, concepts, and practice.
atess 11979. Champaign, Illinois: Research Press, 1973.

Parke, R.D., Deur, J.L., & Sawin, D.B. The in- Patterson, G.R. Interventions for boys with con-
termittent punishment effect in humans: Condi- duct problems: Mutliple settings, treatments,
tioning or adaptation? Psychonomic Science, and criteria. Journal of Consulting and Clinical
1970, 18(4), 193-194. Psychology, 1974a, 42(4), 471-481.
Parke, R.D., & Levy, J.L. Schedule of punishment Patterson, G.R. Retraining of aggressive boys by
and inhibition of aggression in children. Devel- their parents: Review of recent literature and
opmental Psychology, 1972, 7, 266-269. follow-up evaluation. Canadian Psychiatric As-
sociation Journal, 1974b, 19, 142-161.
Parke, R.D., & Walters, R.H. Some factors deter-
mining the efficacy of punishment for inducing Patterson, G.R. A basis for identifying stimuli
response inhibition. Monographs of the Society which control behavior in natural settings.
for Research in Child Development, 1967, Child Development, 1974c, 45, 900-911.
32(109). Patterson, G.R. Stimulus control in natural set-
Parmillee, A. European neurological studies for tings. In J. DeWit & W.W. Hartup (Eds.), De-
terminants and origins of aggressive behavior.
the newborn. Child Development, 1962, 33,
The Hague, Paris: Mouton Press, 1974d.
169-180.
Parton, D.A., & Ross, A.O. Social reinforcement Patterson, G.R. Families. Champaign, Illinois:
of children’s motor behavior: A review. Psycho- Research Press, 1975a.
logical Bulletin, 1965S. Patterson, G.R. Multiple evaluations of a parent-
Parton, D.A., & Ross; A.O. A reply to “The use training program. In T. Thompson & W.S.
of rate as a measure of response in studies of so- Dockens (Eds.), Applications of behavior mod-
cial reinforcement.” Psychological Bulletin, ification. New York: Academic Press, 1975b.
1967, 67, 323-325. Patterson, G.R. The aggressive child: Victim and
Patterson, D.S. Social ecology and social behav- architect of a coercive system. In L.A. Hamer-
ior: The development of the differential usage lynck, L.C. Handy, & E.J. Mash (Eds.), Be-
of play materials in preschool children. Paper havior modification and families: Theory and

342
 research (Vol. 1). New York: Brunner/ Mazel, for classes of noxious behaviors. In J.F. Knut-
1976. son (Ed.), The control of aggression: Implica-
Patterson, G.R. Accelerating stimuli for two tions from basic research. Chicago: Aldine Pub-
classes of coercive behaviors. Journal of Abnor- lishing Co., 1973.
mal Child Psychology, 1977a, 5(4), 334-350. Patterson, G.R., Cobb, J.A., & Ray, R.S. Direct
intervention in the classroom: A set of proce-
Patterson, G.R. A three-stage functional analysis
dures for the aggressive child. In F. Clark, D.
for children’s coercive behaviors: A tactic for
Evans, & L. Hamerlynck (Eds.), Implementing
developing a performance theory. In D. Baer,
behavioral programs for schools and clinics.
B.C. Etzel, & J.M. LeBlanc (Eds.), New devel-
opments in behavioral research: Theories, Champaign, Illinois: Research Press, 1972.
methods, and applications. In honor of Sidney Patterson, G.R., Cobb, J.A., & Ray, R.S. A so-
W. Bijou. Hillsdale, New Jersey: Lawrence Erl- cial engineering technology for retraining the
baum Associaties, 1977b. families of aggressive boys. In H.E. Adams &
Patterson, G.R. Siblings: Fellow travelers in coer- I.P. Unikel (Eds.), Issues and trends in behavior
cive family processes. In R.J. Blanchard (Ed.), therapy. Springfield, Illinois: C.C. Thomas,
Advances in the study of aggression. New 1973.
York: Academic Press, 1979a, in press. Patterson, G.R., & Dawes, R.M. A Guttman
Patterson, G.R. A performance theory for coer- scale of children’s coercive behaviors. Journal
cive family interaction. In R. Cairns (Ed.), So- of Clinical and Consulting Psychology, 1975,
cial interaction: Methods, analysis, and illus- 43(4), 594.
trations. Hillsdale, New Jersey: Lawrence Erl- Patterson, G.R., & Fleischman, M.J. Mainte-
baum Associates, 1979b. nance of treatment effects: Some considerations
Patterson, G.R. Treatment for children with con- concerning family systems and follow-up data.
duct problems: A review of outcome studies. In Behavior Therapy, 1979, 10, 168-185.
S. Feshbach & A. Fraczek (Eds.), Aggression Patterson, G.R., & Gullion, M.E. Living with
and behavior change: Biological and social children: New methods for parents and teach-
processes. New York: Praeger, 1979c. ers. Champaign, Illinois: Research Press, 1968.
Patterson, G.R. Mothers: The unacknowledged Patterson, G.R., & Hinsey, W. Investigations of
victims. Monographs of the Society for Re- some assumptions and characteristics of a pro-
search in Child Development, 1980, 45(5, Seri- cedure for instrumental conditioning in chil-
al No. 186), 1-64. dren. Journal of Experimental Child Psychol-
Patterson, 1980a—see Patterson, 1980. ogy, 1964, 1, 111-122.
Patterson, 1980b—-see Patterson, 1979c. Patterson, G.R., Hops, H., & Weiss, R.L. Inter-
personal skills training for couples in early
Patterson, G.R. Stress: Agent of change for family stages of conflict. Journal of Marriage and the
process. In M. Rutter & N. Garmezy (Eds.), Family, May 1975, 295-303.
Coping and stress: A developmental view. New
York: McGraw-Hill, 1981a, in press. Patterson, G.R., Littman, R.A., & Bricker, W.
Assertive behavior in children: A step toward a
Patterson, G.R. A bilateral trait for aggression: theory of aggression. Monographs of the So-
An interactional analysis of mother-child dy- ciety for Research in Child Development, 1967,
ads. Paper presented at the Conference on 32(5), 1-43.
Boundary Areas in Psychology: Developmental
and Social, Vanderbilt University, Nashville, Patterson, G.R., Littman, R.A., & Hinsey, W.C.
Tennessee, June 17, 1981b. Parental effectiveness as reinforcers in the la-
boratory and its relation to child rearing prac-
Patterson, G.R., & Anderson, D. Peers as social tices and child adjustment in the classroom.
reinforcers. Child Development, 1964, 35,
Journal of Personality, 1964, 32(2), 180-199.
951-960.
Patterson, G.R., & Loeber, R. Family manage-
Patterson, G.R., & Cobb, J.A. A dyadic analysis ment skills and delinquency. Paper presented at
of “aggressive” behaviors. In J.P. Hill (Ed.), the meeting of The International Society for the
Minnesota Symposia on Child Psychology (Vol.
Study of Behavioral Development, Toronto,
5). Minneapolis: University of Minnesota
August 1981.
Press, 1971.
Patterson, G.R., McNeal, $.A., Hawkins, N., &
Patterson, G.R., & Cobb, J.A. Stimulus control

343
 Phelps, R. Reprogramming the social environ- Loucks. Contemporary Psychology, 1978, 23,
ment. Journal of Child Psychology and mien 642-644.
try, 1967, 8, 181-195. Paul, J.S. Observer influence on the interactive be-
Patterson, G.R., & Moore, D. Interactive pat- havior of a mother and a single child in the
terns as units of behavior. In S.J. Suomi, M.E. home. Unpublished master’s thesis, Oregon
Lamb, & G.R. Stephenson (Eds.), Social inter- State University, 1963.
action analysis: Methodological issues. Madi- Paykel, E.S. Life stress and psychiatric disorder.
son, Wisconsin: University of Wisconsin Press, In B.S. Dohrenwend & B.P. Dohrenwend
1979. (Eds.), Stressful life events. New York: John
Patterson, G.R., Ray, R.S., & Shaw, D.A. Direct Wiley & Sons, 1974.
intervention in families of deviant children. Peine, H. Behavioral recording by parents and its
Oregon Research Institute Research Bulletin, resultant consequences. Unpublished master’s
1968, 8(9). thesis, University of Utah, 1970.
Patterson, G.R., Ray, R.S., Shaw, D.A., & Perry, D., & Bussey, K. Self reinforcement in high
Cobb, J.A. Manual for coding of family inter- and low aggressive boys following acts of ag-
actions. New York: Microfiche Publications, gression. Child Development, 1977, 48, 653-
1969 (revised). 657.
Patterson, G.R., & Reid, J.B. Reciprocity and co- Perry, D.G., & Perry, L.C. Denial of suffering in
ercion: Two facets of social systems. In C. the victim as a stimulus to violence in aggressive
Neuringer & J.L. Michael (Eds.), Behavior boys. Child Development, 1974, 45, 55-62.
modification in clinical psychology. New York:
Appleton-Century-Crofts, 1970. Petrinovich, L. Probabilistic functionalism: A con-
ception of research method. American Psychol-
Patterson, G.R., & Reid, J.B. Intervention for ogist, 1979, 34(5), 373-390.
families of aggressive boys: A replication study.
Phillips, L.D. Bayesian statistics for social scien-
Behavior Research and Therapy, 1973, 11,
tists. New York: Thomas Y. Crowell Co.,
383-394.
1973.
Patterson, G.R., Reid, J.B., & Chamberlain, P.
Family typology and clinical resistance. OSLC Phipps-Yonas, S. Reaction time, peer assessment,
and achievement in vulnerable children. Paper
manuscript in preparation.
presented at the 87th Convention of the Ameri-
Patterson, G.R., Reid, J.B., Jones, R.R., & Cong- can Psychological Association, New York, Sep-
er, R.E. A social learning approach to family tember 1979.
intervention. Vol. 1. Families with aggressive
Platt, J. Social traps. American Psychologist,
children. Eugene, Oregon: Castalia Publishing
1973, 28, 641-651.
Co.F 19751
Polk, K. Teenage delinquency in small town
Patterson, G.R., Schwartz, R., & Van der Wart,
E. The integration of group and individual America. Research Report No. 5. Washington,
D.C.: National Institute of Mental Health,
therapy. The American Journal of Orthopsychi-
1978.
atry, 1956, 26(3), 618-629.
Patterson, G.R., & Stouthamer-Loeber, M. Pa- Pollack, S., Vincent, J., & Williams, B. Demand
rental discipline and children’s antisocial behav- characteristics in the classroom observation of
hyperactive children: Reactivity to naturalistic
ior. OSLC manuscript in preparation.
observation. Unpublished manuscript, 1977.
Patterson, G.R., Stouthamer-Loeber, M., & Loe-
ber, R. Parental monitoring and antisocial child Pritchard, M., & Graham, P. An investigation of
behavior. OSLC manuscript in preparation. a group of patients who have attended both the
child and adult departments of the same psychi-
Patterson, G.R., Weiss, R.L., & Hops, H. Train- atric hospital. British Journal of Psychiatry,
ing of marital skills: Some problems and con- 1966, 112, 603-613.
cepts. In H. Leitenberg (Ed.), Handbook of be-
havior modification and behavior therapy. En- Purcell, K., & Brady, K. Adaptation to the inva-
glewood Cliffs, New Jersey: Prentice-Hall, Inc., sion of privacy: Monitoring behavior with a
miniature radio transmitter. Merrill Palmer
1976.
Quarterly of Behavior and Development, 1966,
Paul, G.L., & Lentz, R.J. Psychosocial treatment 12, 242-254.
of chronic mental patients: A review by Sandra

344
Quay, H.C. Personality and delinquency. In H.C. Personality and Social Psychology, 1965, 2(4),
Quay (Ed.), Juvenile delinquency research and 487-499.
theory. New York: Van Nostrand Reinhold
Redl, F., &Wineman, D. Children who hate.
Gorl965. New York: The Free Press, 1951.
Quay, H.C., & Hunt, W.A. Psychopathy, neur- Redl, F., & Wineman, D. The aggressive child.
oticism, and verbal conditining: A replication New York: The Free Press, 1957.
and extension. Journal of Consulting Psycholo-
gy, 1965, 29, 283. Reid, J.B. Reciprocity in family interaction. Un-
published doctoral dissertation, University of
Quinsey, V.L. Some applications of adaptation Oregon, 1967.
level theory to aversive behavior. Psychological
Bulletin, 1970, 73, 441-450. Reid, J.B. Reliability assessment of observation
data: A possible methodological problem.
Quinton, D. Family life in the inner city: Myth Child Development, 1970, 41, 1143-1150.
and reality. In W. Marland (Ed.), Education
for the inner city. London: Heineman, 1980. Reid, J.B. (Ed.). A social learning approach to
family intervention. Vol. II. Observation in
Quinton, D., & Rutter, M. Early hospital admis- home settings. Eugene, Oregon: Castalia Pub-
sions and later disturbances of behavior: An at- lishing Co., 1978.
tempted replication of Douglas’ findings. De-
velopmental Medicine and Child Neurology, Reid, J.B., & Chamberlain, P. Treatment of
1976, 18, 447-459. chronic delinquents: An outcome evaluation.
OSLC manuscript in preparation.
Rachlin, H., & Herrnstein, R.J. Hedonism revis-
ited: On the negative law of effect. In B.A. Reid, J.B., & Hendricks, A.F.C.J. A preliminary
Campbell & R.M. Church (Eds.), Punishment analysis of the effectiveness of direct home in-
and aversive behavior. New York: Applieton- tervention for treatment of predelinquent boys
Century-Crofts, 1969. who steal. In L. Hamerlynck, L. Handy, & E.
Mash (Eds.), Behavior therapy: Methodology,
Radke-Yarrow, M., Campbell, J., & Burton, concepts, and practice. Champaign, Illinois:
R.V. Reliability of maternal retrospection: A Research Press, 1973.
preliminary report. Family Process, 1964, 3,
207-218. Reid, J.B., Hinojosa-Rivero, G., & Lorber, R. A
social learning approach to the outpatient treat-
Radke-Yarrow, M., Campbell, J., & Burton, ment of children who steal. Unpublished manu-
R.V. Child rearing: An inquiry into research script, 1980. (Available from Oregon Social
and methods. San Francisco: Jossey-Bass,
Learning Center, 207 E. Sth, Eugene, Oregon.)
1968.
Reid, J.B., & Patterson, G.R. A social learning
Radke-Yarrow, M., Campbell, J., & Burton, approach to family therapy. 16mm film. Cham-
R.V. Recollections of childhood: A study of the paign, Illinois: Research Press, 1974.
retrospective method. Monographs of the Soci-
ety for Research in Child Development, 1970, Reid, J.B., & Patterson, G.R. The modification of
35(5, Serial No. 138). aggression and stealing behavior of boys in the
home setting. In A. Bandura & E. Ribes (Eds.),
Radke-Yarrow, M., & Waxler, C.Z. Dimensions Behavior modification: Experimental analyses
and correlates of prosocial behavior in young of aggression and delinquency. Hillsdale, New
children. Child Development, 1976, 47, 118- Jersey: Lawrence Erlbaum Associates, 1976.
125.
Rem, |. b., Patterson, Gn. & Poeber, RO The
Radke-Yarrow, M., & Waxler, C.Z. Observing abused child: Victim, instigator, or innocent
interaction: A confrontation with methodolo- bystander? Paper presented at the Nebraska
gy. In R.B. Cairns (Ed.), The analysis of social Symposium on Motivation, Lincoln, Nebraska,
interactions. Hillsdale, New Jersey: Lawrence March 1981.
Erlbaum Associates, 1979.
Reid, J.B., Taplin, P.S., & Lorber, R. A social in-
Randall, T.M. An analysis of observer influence teractional approach to the treatment of abu-
on sex and social class differences in mother in- sive families. In R. Stuart (Ed.), Violent behav-
fant interaction. Paper presented at the Society ior: Social learning approaches to prediction,
for Research in Child Development, Denver, management, and treatment. New York: Brun-
1975. ner/ Mazel, 1981.
Raush, H.L. Interaction sequences. Journal of Reynolds, G.S., Catania, A.C., & Skinner, B.F.

345
 Conditioned and unconditioned aggression in Roff, M., Robins, L.N., & Pollack, M. (Eds.).
pigeons. Journal of the Experimental Analysis Life history research in psychopathy (Vol 2).
of Behavior, 1963, 6, 73-74. Minneapolis: University of Minnesota Press,
Reynolds, M.M. Negativism of preschool chil- 1972.
dren: An observational and experimental study. Rogers, M.E., Lilienfeld, A.M., and Pasamancek,
Contributions to education (No. 228). Bureau B. Prenatal and paranatal factors in the devel-
of Publications, New York: Teachers College, opment of child behavior disorders. Johns Hop-
Columbia University, 1928. kins University Press, Baltimore, 1954.
Ricketts, A.F. A study of the behavior of young Rollins, B.C., & Feldman, H. Marital satisfaction
children in anger. Unpublished master’s thesis, over the family life cycle. Journal of Marriage
University of Iowa, 1931. and the Family, 1970, 32(1), 20-28.
Riddle, M., & Roberts, A. Delinquency, delayed Romanczyk, R.G., Kent, R.N., Diament, C., &
gratification, recidivism, and the Porteus image O’Leary, K.D. Methodological problems in
test. Psychological Bulletin, 1977, 84(3), 417- naturalistic observation. Paper presented at the
425. 2nd Annual Symposium on Behavior Analysis,
Roach, J.L. Some social-psychological character- Lawrence, Kansas, May 1971.
istics of a child guidance clinic caseload. Jour- Rosenfeld, H.M., & Baer, D.M. Unnoticed ver-
nal of Consulting Psychology, 1958, 22, 183- bal conditioning of an aware experimenter by a
186. more aware subject: The double agent effect.
Robins, L.N. The accuracy of parental recall of Psychological Review, 1969, 76, 425-432.
aspects of child development and child-rearing Rosenthal, R. Experimental effects in behavioral
practices. Journal of Abnormal Social Psychol- research. New York: Appleton-Century-Crofts,
ogy, 1963, 55S,261-270. 1966.
Robins, L.N. Deviant children grown up: A socio- Rosenthal, R., & Fode, K.L. Three experiments
logical and psychiatric study of sociopathic per- in experimenter bias. Psychological Reports,
sonality. Baltimore: Williams & Wilkins, 1966. 1963, 12, 491-511.
Robins, L.N., & Ratcliff, K.S. Risk factors in the Rosenthal, R., & Lawson, R. A longitudinal
continuation of childhood antisocial behavior study of the effects of experimenter bias on the
into adulthood. International Journal of Men- operant learning of laboratory rats. Journal of
tal Health, 1978, 7(3-4), 96-116. Psychiatric Research, 1964, 2, 61-72.
Robinson, B.E. Sex typed attitudes, sex typed Rothchild, J., & Wolf, S$. The children of the
contingency behavior, and personality charac- counter culture. New York: Doubleday, 1976.
teristics of male care givers. Unpublished doc- Rule, B.G., & Nesdale, A.R. Emotional arousal
toral dissertation, University of North Caro- and aggressive behavior. Psychological Bulle-
lina, 1976. tin, 1976, 83(5), 851-863.
Robinson, J.P. Toward a more appropriate use of Ruppenthal, G.C., Arling, G.L., Harlow, H.F.,
Guttman scaling. The Public Opinion Quarter- Sackett, G.P., & Suomi, S.J. A ten year per-
ly, 1973, 37, 260-267. spective of motherless-mother monkey behav-
Rodnick, E.H., & Garmezy, N. An experimental ior. Journal of Abnormal Psychology, 1976,
approach to the study of motivation in schizo- 85(4), 341-349.
phrenia. In M.R. Jones (Ed.), Nebraska Sym- Russell, C.S. Transition to parenthood: Problems
posium on Motivation. Lincoln, Nebraska: and gratifications. Journal of Marriage and the
University of Nebraska Press, 1957. Family, May 1974, 294-301.
Roff, M. Childhood social interactions and young Rutter, M. Children of sick parents: An environ-
adult bad conduct. Journal of Abnormal and mental and psychiatric study. London: Oxford
Social Psychology, 1961, 63, 333-337. Univeristy Press, 1966.
Roff, M. A two factor approach to juvenile delin- Rutter, M. Epidemiological strategies and psychi-
quency and the later histories of juvenile delin- atric concepts in research on the vulnerable
quency. In M. Roff, L.N. Robins, & M. Pol- child. In E. Anthony & C. Koupernick (Eds.),
lack (Eds.), Life history research in psycho- The child in his family (Vol. 3). New York:
pathy. (Vol. 2).. Minneapolis: University of John Wiley & Sons, 1974.
Minnesota Press, 1972.

346
Rutter, M. Protective factors in children’s re- Schaefer, E.S. Development of hierarchical config-
sponses to stress and disadvantage. In M. Kent urant tonal models for parent behavior and
& J.E. Rolf (Eds.), Primary prevention of psy- child behavior. In J.P. Hill (Ed.), Minnesota
chopathology. Vol. 3. Social competence in Symposia on Child Psychology (Vol. 5). Min-
children. Hanover, New Hampshire: Universi- neapolis, Minnesota: University of Minnesota
ty Press of New England, 1979. Bressemlovis
Rutter, M., Tizard, J., & Whitmore, R. Educa- Schaffer, H.R. The growth of sociability. Balti-
tion, health, and behavior. New York: John more, Maryland: Penguin Books, Inc., 1977.
Wiley & Sons, 1970. Schelle, J. A brief report on invalidity of parent
Ryan, W. Blaming the victim. New York: Vintage evaluations of behavior change. Journal of Ap-
Books, Inc., 1976. plied Behavior Analysis, 1974, 7, 341-343.
Sackett, G.P. A nonparametric lag sequential ana- Schlesinger, S.E. The prediction of dangerousness
lysis for studying dependency among responses in juveniles. Crime and Delinquency, January
in observational scoring systems. Unpublished 1978, 40-48.
manuscript, Regional Primate Research Center, Schmauk, F.J. Punishment, arousal, and avoid-
University of Washington, 1976. ance learning in sociopaths. Journal of Abnor-
Sackett, G.P. The lag sequential analysis of contin- mal Psychology, 1970, 76, 325-335.
gency and cyclicity in behavioral interaction re- Schoenfeld, W.N. “Reinforcement” in behavior
search. In J. Osofsky (Ed.), Handbook of in- theory. Pavlovian Journal, 1978, 13, 135-144.
fant development. New York: John Wiley &
Sons, 1977. Schoggen, M., Barker, L., & Barker, R. Structure
of the behavior of American and English chil-
Sallows, G. Comparative responsiveness of nor- dren. In R.G. Barker (Eds.), The stream of be-
mal and deviant children to naturally occurring havior. New York: Appleton-Century-Crofts,
consequences. Unpublished doctoral disserta- 1963.
tion, University of Oregon, 1972.
Schoggen, P. Environmental forces in the everyday
Salzinger, K., Portnoy, S., Zlotogura, P., & Keis- lives of children. In R.G. Barker (Ed.), The
ner, R. The effect of reinforcement of continu- stream of behavior. New York: Appleton-Cen-
ous speech and on plural nouns in grammatical tury-Crofts, 1963.
context. Journal of Verbal Learning and Verbal
Behavior, 1963, 6(1), 477-485. Schoggen, P. Mechanical aids for making speci-
men records of behavior. Child Development,
Sarbin, T., Allen, V., & Rutherford, E.E. Social 1964, 35, 985-989.
reinforcement, socialization, and chronic delin-
quency. British Journal of Social and Clinical Schooler, C. A note of extreme caution on the use
Psychology, 1965, 4, 179-184. of Guttman scales. American Journal of Sociol-
ogy, 1968, 74, 296-301.
Saxe, R.M., & Stollak, G.E. Curiosity and the
parent-child relationship. Child Development, Schuck, G.R. The use of causal models in aggres-
1971, 42, 373-384. sion research. In J. DeWit & W.W. Hartup
(Eds.), Determinants and origins of aggressive
Scarpitti, F.R. Can teachers predict delinquency?
behavior. The Hague, Paris: Mouton, 1974.
The Elementary School Journal, 1964, 65(3),
130-136. Sears, R.R. Comparison of interviews with ques-
tionnaires for measuring mother attitudes to-
Scarr, S. The origins of individual differences in
ward sex and aggression. Journal of Personality
adjective checklist scores. Journal of Consulting
and Social Psychology, 1965, 2, 37-44.
Psychology, 1966, 30, 354-357.
Sears, R.R., Maccoby, E., & Levin, H. Patterns
Schachter, S. The interaction of cognitive and
of child rearing. New York: Harper & Row,
physiological determinants of emotional state.
19574
In L. Berkowitz (Ed.), Advances in experimen-
tal social psychology (Vol. 1). New York: Aca- Sears, R.R., Whiting, J.W., Nowlis, V., & Sears,
demic Press, 1964. P.S. Some child rearing antecedents of aggres-
sion and dependency in young children. Genet
Schachter, S., & Singer, J.E. Cognitive, social,
Psychological Monograph, 1953, 57, 135-234.
and physiological determinants of emotional
state. Psychological Review, 1962, 69, 379- Segal, J., & Yahraes, H. A child’s journey: Forces
399. that shape the lives of our young. New York:

347
 McGraw-Hill, 1978. Simard, K. From here to delinquency: An investi-
Seligman, M.E. Helplessness on depression, de- gation of achievement and home-quality char-
velopment, and death. San Francisco: W.H. acteristics in delinquent and socially aggressive
Freeman & Co., 1975. preadolescent males. Unpublished master’s the-
sis, University of Oregon, 1981.
Sells, S.B., & Roff, M. Peer acceptance-rejection
and personal development. Final Report OE-5- Singer, D.L. The control of aggression and vio-
0417 and OE2-10-051, 1967. lence. New York: Academic Press, 1971.

Selye, H. The stress of life. McGraw-Hill, 1976 Skindrud, K.D. An evaluation of observer bias in
(revised). experimental field studies of social interaction.
Unpublished doctoral dissertation, University
Shaw, G. Meat on the hoof: The hidden world of of Oregon, 1972.
Texas football. New York: Dell Publishing Co.,
Skindrud, K.D. Field evaluation of observer bias
1972.
under overt and covert monitoring. In L. Ham-
Shea, M.J. A follow-up study into adulthood of erlynck, L. Handy, & E. Mash (Eds.), Behav-
adolescent psychiatric patients in relation to in- ior change: Methodology, concepts, and prac-
ternalizing and externalizing symptoms, MMPI tice. Champaign, Illinois: Research Press,
configurations, social competence, and life his- 1973.
tory variables. Unpublished doctoral disserta-
Skinner, B.F. Walden two. New York: Macmil-
tion, University of Minnesota, 1972.
lan Co., 1948.
Shelly, M.W. Isolated and restricted environ-
Skinner, B.F. Science and human behavior. New
ments. (Final Report for ONR Contract N 14-
York: Macmillan Co., 1953.
68-C-V286 and NR 177-920.) Lawrence, Kan-
sas: University of Kansas, 1970. Skinner, B.F. Contingencies of reinforcement.
New York: Appleton-Century-Crofts, 1969.
Shepherd, M., Oppenheim, A.N., & Mitchell, S.
Childhood behavior disorders and the child- Skinner, B.F. Negative conditioning and periodic
guidance clinic: An epidemiological study. reinforcement. In E.E. Boe & R.M. Church
Journal of Child Psychology and Psychiatry, (Eds.), Punishment issues and experiments.
1966, 7, 39-52. New York: Appleton-Century-Crofts, 1978.
Sherman, H., & Farina, A. Social adequacy of Slovic, P., Fischoff, B., & Lichtenstein, $. Behav-
parents and children. Journal of Abnormal Psy- ioral decision theory. Annual Review of Psy-
chology, 1974, 83(3), 327-330. chology, 1977, 28, 40.
Shields, M.M. Some communicational skills of Smith, H.T. A comparison of interview and obser-
young children: A study of dialogue in nursery vation measures of mother behavior. Journal of
school. In R.N. Campbell & P.T. Smith (Eds.), Abnormal Social Psychology, 1958, 57, 278-
Recent advances in the psychology of language. 282.
New York: Plenum, 1976. Smith, M. Concerning the magnitude of the be-
Shirley, M. A behavior syndrome characterizing havioral sample for the study of behavioral
prematurely born children. Child Develop- traits in children. Journal of Applied Psycholo-
ment, 1939, 115-129. gy, 1931, 15, 480-485.
Siddle, D.A. Electrodernal activity and psycho- Smith, P.K., & Green, M. Aggressive behavior in
pathy. In S.A. Mednick & K.O. Christiansen English nurseries and playgroups: Sex differ-
(Eds.), Biosocial bases of criminal behavior. ences and response of adults. Child Develop-
New York: Gardner Press, 1977. ment, 1975, 46, 211-214.
Sigal, J.J. Enduring disturbances in behavior fol- Snyder, J.J. A reinforcement analysis of interac-
lowing acute illness in early childhood: Consis- tion in problem and nonproblem families. Jour-
tencies in four independent follow-up studies. nal of Abnormal Psychology, 1977, 86(5), 528-
In E.J. Anthony & C. Koupernik (Eds.), The yOoe
child in his family. New York: John Wiley & Solomon, R.L. Punishment. American Psycholo-
Sons, 1974. gist, 1964, 19, 239-254.
Silverman, M.I. The relationship between self es- Solomon, R.L., & Brush, E.S. Experimentally de-
teem and aggression in two social classes. Dis- rived concepts of anxiety and aversion. In M.R.
sertation Abstracts, 1964, 25(4), 2616. Jones (Ed.), Nebraska Symposium on Motiva-

348
 tion. Lincoln, Nebraska: University of Nebras- olence in the Americanfamily. New York:
ka Press, 1956. Doubleday/Anchor,1979, in press.
Solomon, R.L., & Corbit, J.D. An opponent Strayer, J. Social conflict and peer group status.
process theory of motivation: Temporal dy- Paper presented at the Society for Research in
namics of affect (Part 1). Psychological Review, Child Development Conference, New Orleans,
1974, 81, 119-145. 19778
Solomon, R.L., & Wynne, L.C. Traumatic avoid- Stuart, R.B. Behavioral remedies for marital ills: A
ance learning: The principles of anxiety conser- guide to the use of operant interpersonal tech-
vation and partial irreversibility. Psychological niques. In T. Thompson & W. Dockens (Eds.),
Review, 1954, 61, 353-384. Applications of behavior modification. New
Spielberger, C. The role of awareness in verbal York: Academic Press, 1975.
conditioning. In C. Eriksen (Ed.), Behavior and Stunkard, A.J., & Rush, J. A critical review of re-
awareness: A symposium of research and inter- ports of untoward responses during weight re-
pretation. Duke University Press, 1962. duction for obesity. Annual Review of Internal
Spivack, G., Platt, J.J., & Shure, M.B. The prob- Medicine, 1974, 81, 526-533.
lem solving approach to adjustment. San Fran- Suomi, S.J. Levels of analysis for interaction data
cisco: Jossey-Bass, 1976. collected on workers living in complex social
Spivack, G., & Shure, M.B. Social adjustment of groups. In S. Suomi, M. Lamb, & G. Stephen-
young children: A cognitive approach to solv- son (Eds.). Social interaction analysis: Metho-
ing real life problems. San Francisco: Jossey- dological issues. Madison: University of Wis-
Bass, 1974. consin Press, 1979.

Staats, A.W., & Staats, C.K. Attitudes established Suomi, S., Lamb, M., & Stephenson, G. (Eds.),
by classical conditioning. Journal of Abnormal Social interaction analysis: Methodological is-
and Social Psychology, 1958, 57, 37-40. sues. Madison: University of Wisconsin Press,
17s
Staats, A.W., Staats, C.K., & Crawford, H.L.
Surrott, P.R., Ulrich, R.E., & Hawkins, R.P. An
First order conditioning of meaning and the
elementary student as a behavioral engineer.
parallel conditioning of G.S.R. Journal of Gen-
Journal of Applied Behavior Analysis, 1969, 2,
eral Psychology, 1962, 67, 159-167.
85-92.
Stark, R., & McEvoy, J. Middle class violence.
Swart, C., & Berkowitz, L. The effects of stimu-
Psychology Today, 1970, 4(6), 107-112.
lus associated with a victim’s pain on later ag-
Steinmetz, S.K., & Straus, M.A. The family as a gression. Journal of Personality and Social Psy-
cradle of violence. Society, 1973, 10(6), 50- chology, 1976, 33, 623-631.
56.
Sykes, R.E., & Brent, E.E. Strategies of “taking
Steinmetz, S.K., & Straus, M.A. (Eds.). Violence charge” in police-civilian interaction. In J. Kin-
in the family. New York: Dodd, Mead & Co., ton (Ed.), Police roles in the seventies, 1978, in
1974. press.
Stone, L. The family, sex, and marriage in Eng- Szpiler, J.A., & Epstein, S. Availability of an
land 1500-1800. San Francisco: Harper & avoidance response as related to autonomic
Row, 1977. arousal. Journal of Abnormal Psychology,
Straus, M.A. Some social antecedents of physical 1976, 85, 73-82.
punishment: A linkage theory interpretation. Szykula, S$. The Helena Family Center. Report
Journal of Marriage and the Family, 1971, presented at the Conference of the Association
33(4), 658-663. for Advancement in Behavior Therapy, Chica-
Straus, M.A. A general systems theory approach go, November 1979.
to a theory of violence between family mem- Taplin, P.S. Changes in parental consequation as a
bers. Social Science Information, 1973, 12(3), function of intervention. Unpublished doctoral
105-125. dissertation, University of Wisconsin, 1974.
Straus, M.A. Wife beating: How common and Taplin, P.S., & Reid, J.B. Effects of instructional
why? Victimology: An International Journal, set and experimenter influence on observer reli-
1978, 2(3-4), 443-458. ability. Child Development, 1973, 44, 547-
Straus, M.A., Gelles, R.J., & Steinmetz, S.K. Vi- 554.

349
Taplin, P.S., & Reid, J.B. Changes in parental New York: Teachers College, 1932.
consequation as a function of family interven- Tinbergen, N. The study of instinct. New York:
tion. Journal of Consulting and Clinical Psy- Oxford University Press, 1951.
chology, 1977, 4, 973-981.
Toch, H. Violent men. New York: Aldine Publish-
Taquiri, R. Relational analysis: An extension of ing Co., 1969.
sociometric methods with emphasis upon social
perception. Sociometry, 1952, 15, 91-104. Tonge, W.L., James, D.S., & Hillam, $.M. Fami-
lies without hope: A controlled study of 33
Tarrier, N., Vaughn, C., Lader, M.H., & Leff, problem families. Royal College of Psychia-
J.P. Bodily reactions to people and events in trists, Ashford, Kent: Headley Bros., Ltd.,
schizophrenics. Archives of General Psychiatry, 1975.
1979; 362-311-315;
Toobert, D. Nonviolent punishments that work.
Tavormina, J.B., Boll, T.J., Dunn, N.J., Lus- In G.R. Patterson & J.B. Reid (Eds.), Systema-
comb, R.L., & Taylor, J.R. Psychosocial ef- tic common sense, 1980, in preparation.
fects of raising a physically handicapped child
on parents. Paper presented at the Annual Con- Toobert, D., Patterson, G.R., & Moore, D.
vention of the American Psychological Associa- MOSAIC. In G.R. Patterson & R. Loeber
(Eds.), Assessment of family interaction, 1981,
tion, San Francisco, 1975.
in preparation.
Terdal, L., Jackson, R.H., Garner, A.M. Mother-
child interactions: A comparison between nor- Tuchman, B. A distant mirror: The calamitous
mal and developmentally delayed groups. In 14th century. New York: Alfred A. Knopf,
L.A. Hamerlynck, L.C. Handy, & E.J. Mash 1978.
(Eds.), Behavior modification and families.
New York: Brunner/ Mazel, 1976. Ullmann, L., & Krasner, L. Case studies in behav-
ior modification. New York: Holt, Rinehart &
Terrell, G., & Kennedy, W. Discrimination learn- Winston, 1965.
ing and transposition in children as a function
of the nature of the reward. Journal of Experi- Ulrich, R.E., Dulaney, S., Arnett, M., & Mueller,
mental Psychology, 1957, 53, 257-260. K. An experimental analysis of nonhuman and
human aggression. In J. Knutson (Ed.), The
Theorell, T. Life events before and after the onset control of aggression: Implications from basic
of a premature myocardial infarction. In B.S. research. Chicago: Aldine Press, 1973.
Dohrenwend & B.P. Dohrenwend (Eds.),
Stressful life events: Their nature and effects. Ulrich, R.E., Johnson, M., Richardson, J., &
New York: John Wiley & Sons, 1974. Wolff, P.C. The operant conditioning of fight-
ing behavior in rats. Psychological Record,
Thomas, A., Becker, W., & Armstrong, M. Pro- 1963, 13, 465-470.
duction and elimination of disruptive classroom
Verna, G.B. The effects of four-hour delay of pun-
behavior by systematically varying teachers’ be-
ishment under two conditions of verbal instruc-
havior. Journal of Applied Behavioral Analysis,
tion. Child Development, 1977, 48, 621-624.
1968, 1, 35-45.
Thomas, A., Chess, J., Birch, H., Hertzog, M. A Vincent, J.P., Friedman, L.C., Nugent, J., Mes-
longitudinal study of primary reaction patterns
serly, L. Demand characteristics in observation
of marital interaction. Journal of Consulting
in children. Comparative Psychiatry, 1960, 1,
and Clinical Psychology, 1979, in press.
103-112.
‘Vreeland, R.G., & Waller, M.B. Social interac-
Thomas, D., Loomis, A., & Arrington, R. Obser-
vational studies of social behavior. Vol. 1. So- tions in families of firesetting children. Unpub-
cial behavior patterns. New Haven, Connecti- lished manuscript, 1980.
cut: Yale University Press, 1933. Wadsworth, M. Roots of delinquency: Infancy,
adolescence, and crime. New York: Barnes &
Thomas, E.A., & Malone, T.W. On the dynamics
of two-person interactions. Psychological Re- Noble, Inc., 1979.
view, 1979, 86(4), 331-360. Wahl, G. Operant analysis of family interactions.
Thomas, E.A., & Martin, J.A. Analysis of par-
Unpublished doctoral dissertation, University
ent-infant interaction. Psychological Review, of Oregon, 1971.
1976, 83(2), 141-156. Wahler, R.G. Behavior therapy for oppositional
Thorndike, E.L. The fundamentals of learning. children: Love is not enough. Paper presented

350
 at the Eastern Psychological Association, 1968. tiation: An outcome of frequency or ambiguity?
Wahler, R.G. The insular mother: Her problems Journal of Experimental Child Psychology,
_ in parent child treatment. Unpublished manu- 1972,.13, 249-260,
script, 1979. Watson, J.S. Perception of contingency as a deter-
Wahler, R.G., Leske, G., & Rogers, E. The insu- minant of social responsiveness. In E. Thomas
lar family: A deviance support system for oppo- (Ed.), Origins of the infant’s social responsive-
sitional children. Paper presented at the Banff ness (Vol. 1). New York: Halsted Press, 1979.
International Conference on Behavior Modifi- Weick, K.E. Systematic observational methods. In
cation, 1977. G. Lindsey & A. Aronsen (Eds.), The hand-
Wahler, R.G., & Moore, D. School-home behav- book of social psychology (Vol. 2, 2nd Ed.).
ior change procedures in a high risk communi- Reeding, Massachusetts: Addison-Wesley,
ty. Paper presented to the Association of the 1968.
Advancement of Behavior Therapy, San Fran- Weinrott, M., Bauske, B., & Patterson, G.R. Sys-
cisco, 1975. tematic replication of a social learning ap-
Waksman, S. An empirical investigation of Camp- proach. In P.O. Sjoden, S. Bates, & W.S. Dock-
bell and Fiske’s multistate-multimetrics using ens (Eds.), Trends in behavior therapy. New
social learning measures. Journal of Abnormal York: Academic Press, 1979.
Psychology, 1977. Weinrott, M.R., Garrett, B. & Todd, N. The in-
Walker, H.M., & Buckley, N.K. Programming fluence of observer presence on classroom be-
generalization and maintenance of treatment ef- havior. Unpublished manuscript, ORI, 1976.
fects across time and across settings. Journal of Weintraub, S.A. Self control as a correlate of an
Applied Behavior Analysis, 1972, 5, 209-224. internalizing-externalizing symptom dimen-
Walker, H.M., & Buckley, N.K. Teacher atten- sion. Journal of Abnormal Child Psychology,
tion to appropriate and inappropriate class- 1973} 1(3);'292-307.
room behavior: An individual case study. Focus Weiss, J.M. Effects of coping behavior in different
on Exceptional Children, May 1973, 5, 5-11. warning signal conditions on stress pathology
Walker, H.M., Hops, H., & Johnson, S.M. Gen- in rats. Journal of Comparative and Physiologi-
eralization and maintenance of classroom treat- cal Psychology, 1971, 1, 1-14.
ment effects. Behavior Therapy, 1975, 6, 188- Weiss, R.L. Marital bittersweets: Conceptual and
200. empirical samples. Paper presented at the West-
Wallerstein, J.S., & Kelly, J.B. The effects of pa- ern Psychologist’s Association, San Francisco,
rental divorce: Experiences of the child in later 1978.
latency. American Journal of Orthopsychiatry, Weissman, M.M., & Paykel, E.S. The depressed
1976, 46(2), 256-269. woman: A study of social relationships. Chica-
Wallerstein, J.S., & Kelly, J.B. Surviving the go Press, 1974. i
breakup: How children and parents cope with Welsh, R.S. Severe parental punishment and delin-
divorce. New York: Basic Books, 1980. quency: A developmental theory. Journal of
Walter, H.I., & Gilmore, $.K. Placebo versus so- Clinical Child Psychology, 1976, 5, 17-21.
cial learning effects in parent training proce- Werner, E., & Smith, R. Kauai’s children come of
dures designed to alter the behaviors of aggres- age. Honolulu: University of Hawaii Press,
sive boys. Behavior Therapy, 1973, 4, 361- 1977.
37. Wessman, A.E., & Ricks, D.F. Mood and person-
Walters, R.H., & Brown, M. Studies of reinforce- ality. New York: Holt, Rinehart & Winston,
ment of aggression: III. Transfer of responses to 1966.
an interpersonal situation. Child Development,
West, D.J., & Farrington, D.P. Who becomes de-
1963, 34, 563-571.
linquent? New York: Crane, Russak & Co.,
Warden, G.J., & Aylesworth, M. The relative val- 1973;
ue of reward and punishment in the formula-
White, G.D. The effects of observer presence on
tion of a visual discrimination habit in the white family interactions. Unpublished doctoral dis-
rat. Journal of Comparative Physiological Psy-
sertation, University of Oregon, 1972.
chology, 1927, 7, 117-127.
White, G.D., Nielsen, G., & Johnson, S.M. Time
Warren, V., & Cairns, R. Social reinforcement sa-

351
 out duration and the suppression of deviant be- appropriate aggressive behavior of boys. Be-
havior in children. Journal of Applied Behavior havior Therapy, 1974, 5, 215-221.
Analysis, 1972, 5, 111-120. : Wolff, S. Behavioral characteristics of primary
White, M.N. Natural rates of teacher approval school children referred to a psychiatric depart-
and disapproval in the classroom. Journal of ment. British Journal of Psychiatry, 1967, 113,
Applied Behavior Analysis, 1975, 8, 367-372. 885-893.
Whiting, B.B., & Whiting, J.M. Children of six Wolfgang, M.E. From boy to man—from delin-
cultures. Cambridge, Massachusetts: Harvard quency to crime. Paper presented at the Nation-
University Press, 1975. al Symposium on Serious Juvenile Offenders,
Wiggins, J.S. (Ed.). Personality and prediction: Department of Corrections, Minneapolis, Min-
Principles of personality assessment. Reeding, nesota, September 1977.
Massachusetts: Addison-Wesley, 1973. Wolfgang, M.E., Figlio, R., & Sellin, T. Delin-
Willems, E.P., & Raush, H.L. Naturalistic view- quency in a birth cohort. Chicago: University
points in psychological research. New York: of Chicago Press, 1972.
Holt, Rinehart & Winston, 1969. Wolking, W.D., Dunteman, G.H., & Bailey, J.P.
Willerman, L. Activity level and hyperactivity in Multivariate analyses of parents’ MMPI’s based
twins. Child Development, 1973, 44, 288-293. on psychiatric diagnosis of their children. Jour-
nal of Consulting Psychology, 1967, 31, 521-
Wills, T.A. The contribution of instrumental and 524.
affective events to perceived pleasure and dis-
pleasure in marital relationships. Unpublished Woo, D. Experimental studies of the reinforce-
master’s thesis, University of Oregon, 1971. ment trap. Unpublished master’s thesis, Univer-
sity of Oregon, 1978.
Wills, T.A., Weiss, R.L., & Patterson, G.R. A be-
havioral analysis of the determinannts of mari- Wright, H.F. Observational child study. In P.
tal satisfaction. Journal of Consulting and Clin- Mussen (Ed.), Handbook of research methods
ical Psychology, 1974, 42(6), 802-811. in child development. New York: John Wiley
& Sons, 1960.
Wilson, C.C., Robertson, S.J., Herlong, L.H., &
Haynes, S.N. Vicarious effects of time out in Wright, Q. A study of war. Chicago: University of
the modification of aggression in the classroom. Chicago Press, 1972.
Behavior Modification, 1979, 3, 97-111. Zajonc, R.B. Family configuration and intelli-
Wilson, T.D., & Nisbett, R.E. The accuracy of gence. Science, 1976, 192, 227-236.
verbal reports about the effects of stimuli on Zegiob, L., Arnold, S., & Forehand, R. An exam-
evaluations and behavior. Social Psychology, ination of observer effects in parent-child inter-
1978, 41(2), 118-131. action. Child Development, 1975, 46, 509-
Sa
Wilson, W. Correlates of avowed happiness. Psy-
chological Bulletin, 1967, 67, 294-306. Zylstra, J.L. Assessing delinquent juveniles from
Wiltz, N.A., & Patterson, G.R. An evaluation of
their responses to verbal conditioning. Unpub-
parent training procedures designed to alter in- lished doctoral dissertation, University of
Washington, 1966.

Sod
 Author index

Editor’s Note: In cases where “et al.” has been Arrington, R., 42, 59
used to abbreviate references in the text, all of the Atkinson, J.A., 144, 177, 178
authors (not just the first author) have been in- Attneave, F., 175
cluded in the Author Index. Averill }.R24 71
Aylesworth, M., 117
Ayllon, N., 2
A Azrin, N.H., 2, 52, 209
Achenbach, T.M., 14, 27, 30, 31, 33, 39, 243,
246, 250
Ackerson, L., 21 B
Adler, A., 96, 266 Bach, G.R., 163
Ageton, S.S., 37, 265 Baer, D.M., 92, 244, 266
Alexander, J.F., 209 Bailey, J.P., 281, 282
AllenL., 20; 25, 39, 43, 260, 261 Bakeman, R., 6, 172, 188, 197
Allen, V., 244 Bales, R.F., 58, 203, 230
Altmann, S. 170, 171 Baltes, M., 196
Anderson, D., 45 Bandura Acs2 6, 12. 18h 68=7 Orn2607/ oa Oi oles
Anderson, J.E., 33 93, 102, 104-106, 121, 137, 146, 147, 150,
Anderson, L.M., 25, 38, 282, 289, 290 172
Andrews, F.M., 114 Barber, T.X., 112
Annesly, F.R., 34 Barker, L.S., 28, 96
Antonovsky, H.F., 43 Barker, R.G., 3, 28, 42, 45, 54, 96, 186, 196,
Argyle, M., 201 199
Acting, G.L., 271 Barnett hs /5.°233
Armstrong, M., 52 Baron, K.W., 172, 187, 189, 204
Aronfreed, J., 125 Baron, R.A., 106, 108
Arnett, M., 104 Barrett, D.E-; 35
ArmoldeG.R-, 113, 222 Bartlett, M.S., 186
Arnold, J., 121 Baumrind, D., 44, 269, 272, 273, 284
Arnold, S., 54 Bauske, B., 62; 306, 307

$53
Bechtel, R.B., 54 Bussey, K., 107
Becker, W.C., 44, 52, 98, 113, 222 Byrne, D., 211
Beil R.O., 10225135, 196,277
Bem, D., 93
Benning, J.J., 34, 37 GC
Berkowitz,/L., 12) 13,°69;:70;-72) 73; 82) 106, Cairns, R.B., 1, 3, 6, 14, 16, 17, 108 asaya
1375°057-159, 177 160-162, 192, 199, 244, 245, 310
Berlyne, D.E., 86 Caldwell, B., 42, 44, 58
Bernal, M.E., 16, 29, 57, 228 Call al 265)
Bernard, J.L., 244 Camp, B.W., 34, 243
Bhrolchain, M.N., 78, 79 Campbell, A., 77
Bijou, S.W., 54 Campbell, B.A., 116
Bing, E., 44 Campbell, J., 43, 122
Birch, H., 244 Campbell, W.J., 28, 96
Birchler, G.R., 212, 229, 232-234 Canter, RJ. 37,265
Birley; J.L., 67,70 Caplovitz, D., 75
Black, A., 44, 272 Carlson, W., 62, 254, 255, 267
Black, J.L., 54 Meaceys1.G., 118, 122
Black, R.W., 108 Catania, A.C., 104
Blau, P.M., 210 Chamberlain, P., 28, 31, 32, 61, 62, 250, 265,
Block, D.A., 44 266, 283, 295, 296, 306, 307
Blood, R.O., 233 Charlesworth, R., 206
Bloom, S., 70 Charny, I.W., 286
Blumberg, M., 114 Chess, J., 244
Blumenthal, M., 114 Chodorow, N., 284
Blurton-Jones, N.G., 2 Christensen, A., 44, 45, 61
Bobbitt, R., 42, 58 Christiansen, K.O., 135, 136, 244, 262
Boffa, J., 71 Church, R.M., 112, 116-118, 143
Bogaard, L., 235 Circirelli, V.G., 24
Bollgelsf: 289 Clarke-Stewart, K.A., 76, 286
Bolstad, O.D., 29, 42, 50, 54, 57-59, 77, 78, Clement, P.W., 44
DS ale G e241 Coates, J.M., 290
Boyds bn392 Cobb, J.A., 7, 12, 28, 29, 34, 36, 45, 46, 54, 59,
Brackbill, Y., 118 100, 103, 104, 109, 111, 121, 128, 129, 144,
Bradburn, N.M., 75 150, 151, 164, 181, 182, 184, 220, 255, 296,
Brady, Je¥ cast 297 305
Brady, K., 57 Cohen, J.A., 50
Brent, E.E., 158 Collins, R.C., 44
Bricker, W., 7, 25, 45, 89, 98, 103, 105, 108, Conger, J.J., 34
109, 135, 136, 141 Conger, R.D., 61, 208, 212, 286
Broderick, J., 231, 286, 291 Conger, R.E., 31, 42, 45, 118, 137, 147, 221,
Bronfenbrenner, U., 65, 116 tal, 2284 235; 285, 3042305
Brown, G.W., 67, 70, 78, 79 Connally, F., 281
Brown, M., 105 Connolly, K., 54
Brush, E.S., 116 Conrad, D.G., 71
Bryan, J.H., 244 Cook, M., 201
Buckley, N.K., 98, 244, 266, 269 Coombs, C., 5, 190
Budd, K.S., 244, 266 Cooper, E., 44
Buehler, R.E., 45, 98 Cooper, J.E., 44
Burch, N.R., 70 Coopersmith, S., 37
Burgess, J.M., 23 Corah, N.L..-71
Burgess, R.L., 23, 61, 157,209) 210928230725 Corbit, J.D., 226
286 Cordua, G., 244
Burton, R.V., 43, 44, 122 Corte, H.E., 244, 266
Bushnell, D., 209 Covey, M.K., 145
Buss, AF, 12, 13, 73, 206:451 Comin 2
Cox ., 24,25, 73, 217,.231, 240, 246;:281, Ericksen, C.W., 91
285, 286, 290 Erickson, M.R., 224
Cox, R.. 24, 25, 73, 217, 231, 240, 246; 281, Exianger, 11.S., 116
285, 286, 290 BronbeD., 22, 275 34376280 2264277,
Craighead, W.E., 54, 59 234
Crawford, H.L., 177 Estes, W.K., 86, 87, 108, 116
Cronbach, L.J., 51
Cummings, J., 38
Cureton, L., 34 F
Fagan, O.S., 34, 234
Fagot, B., 226; 2/1; 285
D Farina, A., 273
Dafoe, J.L., 138, 235 Farrington, D.P., 27, 34, 36, 37, 234, 259
Dahlm, N., 34, 243 Renal, (CollogU5 WE
Dawe, H.C., 16, 20 Feld, S075
Dawes n., 5,75, 155, 190, 255 Feldhusen, J.F., 34, 37
DeFries, J.C., 134, 135 Feldman, H., 81
Delfini, L.F., 16, 29, 228 Ferguson, L., 33
Delqaurdri, J., 113 Feshbach, N., 35, 246
DeMaster, B., 52 Feshbachs, $r41.2.520 0225,35,375 12751572246
Denenberg, V.H., 135 Festinger, L., 266
Dengerink, H.A., 145, 157, 162, 179 Field, H.F., 156
Deur,,J.L., 124, 125 Field, M.H., 156
Devereaux, E., 21 Figlio, R., 264, 265
Devine, V.T., 144, 153, 154, 178, 259 Fink, R.S., 65
Diament, C., 51, 52 Fischhoff, B., 90
Dietz, A., 52 Fiske, D.W., 2, 7, 169
Dishion, T., 258, 259 Fleischman, M.J., 228, 262, 284, 297, 305-307
Dobson, W.R., 99 Flemming, E., 250
Dodge, K., 72, 73, 82, 234 Fode, K.L., 52
Dohrenwend, B.P., 74 Follick, M.J., 13
Dohrenwend, B.S., 74 Forehand, R., 32, 54, 77, 81, 222, 240, 241,
Dollard, J., 2, 104 281, 283
Doob, L.W., 104 Forgatch, M., 64, 180, 194, 229, 231, 232, 258,
Doomineck, W., 34, 243 27S 27 308s 320
Douglas, J.W., 23, 34, 44 Forster, A., 156
Dreger, R.M., 250 Fowles, D.C., 142, 146
Driscoll, R., 243 Freud, S., 92
Dulany, D.E., 244 Freund, R., 278
Dulaney, S., 104 Friedman, L.C., 55, 230
Dunn, N.J., 289 Friedrich, L.K., 35, 93
Dunteman, G.H., 281, 282 Friesen, W., 28, 96
Furman, W., 129
Furniss, J.M., 45, 98
E
Ebner, M., 45
Edelbrock, C.S., 14, 246 G
Edelman, M., 270 Gardner, W.P., 172
Edelman, R., 146 Garmezy, Nz, 12; 27,37, 2163 243,259
Eibl-Eibesfeldt, I., 12, 93, 134 Garner, A.M., 241
Eisenberg, J.G., 27, 34, 234, 265 Garrett, B., 54
Eisenman, R., 244 Gelfand, D.M., 99
Elizur, E., 34 Gelfand, S., 99
Elliott, D.S., 32-34, 37, 234, 260, 264, 265 Gelles, R.J., 113, 114, 156, 157, 274
Emerson, R., 210 Gerber, D., 266
Epstein, S., 70, 146 Gersten, J.C., 27, 34, 234, 265

355
Getty, D.J., 143 Halverson, C.F., 228, 230
Gewirtz, J., 92, 134 Hamburg, D., 134
Gibson, D.M., 57 Harbine Hee 2225
Gilmore, S.K., 44, 65, 307, 308 j Hardin, G., 241
Glairn, J.P., 34 ElaressRe Din 156
Glenn, N., 81 Harford, R.A., 91
Gleser, G.C., 51 Harlow, H.F., 271
Glickman, S., 30, 31 Harper, L.V., 102, 196
Glueck, E., 34 Harris, Age295139 545, 97-9 95)LoS
Glueck, S., 34 InkeieGS,Wagdey WY
Goldiamond, I., 2, 52 Hartmann, D.P., 5, 6, 50, 106, 1079872
Goldin, P.-C .°227 Hartup, W.W., 1, 6, 13, 16, 20, 35, 38, 64, 73,
Golding, W., 68 106, 150, 185, 186, 206, 212, 246, 260
Goldstein, M.J., 27 Hatfield, E., 209, 210
Gonso, J., 35 Hawkins, N., 45
Goodenough, F.L., 17, 21, 42, 47, 115, 127 Hawkins, R.P., 54
Goodstein, L.D., 281, 282, 289, 290 Hayden, M.L., 244
Goody, E., 285 Haynes, S.N., 121
Gottman, J:M:, 1; 6,35, 102,108, U723ersG- Head, K.B., 114
188, 192, 193, 200, 201, 204, 205, 207, 242, Hebb, D., 134
246 Hedlund, C.S., 133, 244
Gouldner, A.W., 209 Hefferline, R.F., 91
Gourevitch, V., 42, 58 Helfer, R.E., 157
Graham, P., 28 Hendricks,..A.F.C.J., 31, 251, 23252oe
Grant, M., 94 Henry, M:M., 16; 20; 215.35
Granth Oneal Herbert, E., 45, 244
Graubard, P.S., 34 Herlong, L.H., 121
Graziano, A.M., 65 Herrnstein, R.J., 101, 116, 117
Green, D.K., 244, 266 Hertzog, M., 244
Green, E.H., 16, 38 Hetherington, E.M., 24, 25, 73, 112, 217, 231,
Green, [sAx, 3 240, 281, 285, 286, 290, 291
Green, M., 20 Hewitt, L., 30, 31, 34, 36, 250
Green, R.G., 146 Heyns, R.W., 42
Greenwood, C.R., 113, 137, 244 Hibbs, D.A., 82
Gromer. )., 72 Hicks, M.W., 229
Griest, D., 32, 81, 240, 281, 283 Hillam, S.M., 78, 80, 217, 233
Griffith, W., 20 Hinde, R.A., 107, 134, 170, 196
Grimm, J.A., 54 Hineline, P., 142, 147, 164
Groothues, C.M., 285 Hines, P.A., 224
Guerney, B., 296 Hinojosa-Rivero, G., 306
Guild, J., 113 Hinsey, W.C., 118, 187, 244
Gullion, M.E., 244, 304 Hirschi, T., 34, 36, 121} 262; 297
Gump, P.V., 28, 96 Hobbs, S.A., 222
Gunn, W.B., 246 Hoffman, W.H., 82
Gurin, G., 75 Hokanson, J.E., 146
Guthrie, E.R., 2, 6, 116, 143 Holleran, P.R., 69, 235, 248, 258
Guttman, L., 250, 251 Hollingshead, A.B., 63, 316, 318
Holmberg, M.C., 16, 17, 21
Holmes, T.H., 74, 217
Holtz, W., 52
H Homans,G.C., 209, 210
Hafner, A.J., 27 Honig, A.S., 44
Haggard, E., 71 Honzik, M., 20, 25, 39, 43, 260, 261
Haley, J., 221 Hood, R., 21, 22, 260
Hall, R.V., 108 Hops, H., 29, 33-36, 46, 113; 137, 18292675
Hall, W.M., 162 210, 220, 233, 244, 304

356
Horne, A., 24, 25, 291 King, H., 77, 240, 241
Hotchkiss, J.M., 108 Kirkpatrick, R., 29
Huesmann, L.R., 27, 80, 122, 137 Knutson, J.F., 13, 69, 106, 147, 160, 162, 165,
Huizinga, D., 264 166
Hult, C.L., 25-6, 86, 88 Kohn, M.L., 79
Hunt, W.A., 243 Kohn, R.L., 114
Huston, T.L., 209, 210 Kolb, T.M., 285
Hutchings, B., 135, 136 Kopfstein, D., 128
Hyman, M., 13 Koropsak, E., 146
Krasner, L., 2
Krentzern 9).Waal
G29
Kuenstler, W.H., 133
Jaen, CN., 20, 21, 150, 172,. 187, 189, 204, Kio 74Y 47135
270
Jackson, C., 244
Jackson, D., 108 L
Jackson, R.H., 241 Eacey,J.E; 70
Jacobson, N.S., 87, 163, 211, 230 Lader, M.H., 70
JaimessDro.. 7.85 80;217, 233 Lagerspetz, K., 134, 135, 160
Janay W., 257, 259 Laivgueur, H., 244, 266
Jenkins, RL., 30, 31, 34, 36, 250 FambyMGr.. 191, 192, 286
Jennings, H.H., 34 Lambert, W.W., 13, 16, 17, 106, 127
Jensen, G., 42, 58 WangerP|i 70
Johansson, S., 96, 99, 101, 108, 240 Langner, T.S., 27, 34, 234, 265
Johns, J.H., 243 Kaniens Re, 28
Johnson, S.M., 16, 29, 42, 44, 45, 50, 54-59, 77, Lapouse, R., 43, 250
Geo 99-101.,,.108, 115, 1195120, 126, 173, Lawson, A., 44
#7 on 876, 182, 228, 231; 232, 240, 241, 283, Lawson, R., 52
286 Lazarus, R.S., 28
Johnston, M., 104, 133, 160, 243 Lecke, B.J., 244, 266
Vomes, RGR:, 12, 31, 42, 45, 51, 55, 58-60, 77, Le Compte, W., 28, 96
932404, 118):137, 147, 161, 209, 221, 227, Lefcourt, H.M., 71
228, 235, 283, 304, 305 Leff}: P. 470
Lefkowitz, M.M., 22, 27, 34, 37, 80, 122, 127,
137, 234
K Leiter, M.P., 206
Kaffman, M., 34 Lennard, H., 232
Kagan, J., 26, 34, 77, 162, 285 Lentz> Rij.7 121
Kahneman, D., 90, 226, 227 Lentz, W.P., 265
Kantem ©11., 6, 72, 91, 93 Lepper, M.R., 138, 235
Kantor, J.R., 170 Leske, G., 80, 234
Kapche, R., 244 Mester BesS
Kaplan, H.B., 70 Levin, G.R., 243, 244
Karoly, P., 72 ibe, 1Bl4 13s Ail, WA, A 2, 2S
Karpowitz, D., 170, 171 Levine, A., 121
Karpowitz, J., 53 Kevine,J., 34, 77, 162
Kasse RE.) 52 Levinger, G., 156, 210
Keenan, B., 91 Levitt, E.E., 1, 43, 44
Keisner, R., 187 Bevy.) 2.57125:
Kelyejeo., 24, 25, 156, 231,291, 296 Lewis, O., 78, 217
Kelly, R., 118 KewisbaMes 250
Kemp, ©.H., 157 Liberman, R.P., 230
Kendon, A., 201 Lichtenstein, E., 87
Kennedy, W., 118 Lichtenstein, S., 90
Kent, R.N., 51-53 Lieven, E.V.M., 200
Kimball, W.H., 23 Lilienfeld, A.M., 39

B57
Lindenthal, J., 74 McEvoy,J., 113, 114, 156
Lindner, R., 68 McNeal, S.A., 45
Lindskold, S., 162 Mednick, S., 135, 136, 244, 262
Lindsley, O., 2 Megargee, E.I., 165
Lipenski, D.P., 54 Meichenbaum, D., 242
Lippitt, R., 42 Menzel, H., 266
Littman, D., 70, 73, 234, 248, 278 Mercatoris, M., 54, 59
Littman, R.A 7, 25,45; 89, 989 103, 105, 108% Messerly, L., 55, 230
109, 118, 135, 136, 141 Meyer, W.J., 117
Liverant, S.,°282, 289, 290 Mikesell, E.H., 28, 96
Lobitz, G.K.,.16; 29,, 55"826n 2386723220041- Miller, K.S., 250
283, 286 Miller, L., 42, 58
Lobitz, W.C., 55, 56, 228, 240 Miller, N.E., 2, 104
Loeber, R., 60; 61, 68,155, 159; 170; 475,477, Miller, S.M., 70, 71
186, 224, 228, 257-259, 263, 264, 274, 275, Miller, W.C., 34
279.250 Milne, D.C., 44
Loomis, A., 42, 59 Minton, C., 34, 77, 162
Lorber, R., 56, 62, 67, 68, 82; 112, Daiets6s Minuchin, S., 296
157, 223-225, 227: 232, 235; 249) 246200 Mischel, W., 20, 25, 28, 29, 181
Lorenz, K., 94 Mitchell, C., 172
|e oa al epee) Mitchell, S., 39
Lott. bE. oo Monk, M.A., 43, 250
ovaass:| Of LOS Montalvo, B., 296
Lund, D., 108 Moore, D., 28, 29, 34, 50, 64, 190, 191, 205,
Luscomb, R.L., 289 258, 265, 266, 306
Lykken, D.J., 136 Moore, S., 34, 36
Lytton, H., 58 Moos, R.H., 54
Moss, H., 26
Moustakas, C.E., 58
M Mowrer, O.H., 104
Maccoby, Ez, 20,721, 340° its, 7121.17 252 Mueller, K., 104
150, 172, 187, 189,204, 27022723 273 Mukai, L., 28, 64, 265, 266, 306
MacFarlane, J.W., 20, 25, 39, 43, 260, 261 Murphy, L.B., 35
Madden D.J7,) 2215223 Myers, J., 74
Madsen, C., 98, 113, 222
Magee, R.D., 37
Maier, N.R., 71, 142, 147 N
Malone, T.W., 187, 189, 192 Nakelski, J.S., 160
Marcus, L.M., 243 Nanda, H., 51
Margolin, G., 87, 156, 163, 202, 204, 208, 211, Naster, BJ... 209
230 Nelson, C., 147
Martens, E.H., 39 Nelson, R.O., 54
Martin, B., 124 Nemeth, C., 212
Martin, S., 96, 99, 101, 108, 172, 187, 189, 192, . Nesdale, A.R., 13, 69, 73, 106
204, 240 Nielsen, A., 266
Mash, E.J., 58 Nielsen, G., 119, 120
Masling, J., 59 Nietzel, M., 75, 233
Mason, J.W., 71 Nisbett, R.E., 90, 91
Masters, J., 129 Nord, W.B., 209
Masuda, M., 74, 217 North, J.Av,16, 29
McAdoo, W., 281 Novaco, R.W., 283
McCaffrey, J., 38 Novick, J., 44
McCarthy, E.D., 27, 265 Nowlis, V., 121
McClearn, G.E., 134, 135 Nuechterlein, K., 37, 216
McCord, J., 34, 37, 259, 262, 299 Nugent, J., 55, 230
McCord, W., 299

358
 Pollack, M., 28
O Pollack, S., 53-55
O’Connor, R.D., 52 Porter, R.W., 71
O’Hara, J., 118 Portnoy, S., 187
OTVeary, K:D., 51-53, 231, 286, 291 Pritchard, M., 28
Oltmanns, T., 231, 286, 291 Purcell, K., 57
Olweus, D., 14, 22, 26, 34, 35, 37, 273, 304
Omark, D.R., 270
Omark, M., 270
Oppenheim, A.N., 39 Quast, W., 27
Orzech, M., 133, 244 Quay, H.C., 31, 242, 243
Overlade, D., 250 Quinsey, V.L., 134
Quinton, D., 216, 227, 229

P
Paris,$.G.,244,245 R
Parkes R.D:, 6, 7, 121-125, 133 Rachlin, H., 116, 117
Parmillee, A., 76 Radke-Yarrow, M., 3, 35, 43, 58, 107, 122, 246
Parsons, B.V., 209 Rajaratnam, N., 51
Parsons, J.A., 54 Randall, T.M., 54
Parton, D.A., 89, 244 Rasmussen, B., 35
Partzka, L., 33 Ratcliff KeS:. 221.223
Pasamancek, B., 39 Raushhels,.3;, 4273, 150,1707171, 186, 190"
Patterson, D.S., 21 192, 193, 203, 204, 293
Patterson, G.R., 1, 3, 7, 12, 14, 16, 18, 23, 25, Raye i295, 28,293 40, 46, 120, 296,297,305
28, 29, 31, 34, 38, 42-46, 50, 51, 54, 58-62, Redl, F., 293, 294
6465. 70-73, 7/5, 77, 81, 89, 93, 98, 100, Redlich, F.C., 63, 316
1O9703-105, 108, 109,111, 112,:118,.421, Reese, H.W., 196
128-130, 132, 134-137, 139, 141, 144, 145, Reid, J.B., 12, 14, 21, 23, 29, 31, 41-46, 49-52,
279150, 155, 159, 160, 161, 164, 166, 173, 56, 58, 59, 60-62, 64, 65, 67-69, 77, 93, 96,
175, 176, 178, 181-191, 194, 204-212, 221, M21 8e30: 1300137; 1394147 5455-159
223, 224, 226-228, 229, 231, 233-235, 244, 161572062209, 21252217, 228% 2262723.232,
250, 255-259, 262-264, 274-276, 279, 282- PBS 2AI-2 5 20 259, 262, 267, 2742276, 2/93
284, 286, 289, 294-297, 304-307 283, 286, 291, 295, 296, 304-307, 311
PaulyG.L., 121 Reid, M.P., 250
Pauley.o.; 53, 94, 57, 59 Revenstein, A., 232
Pavlov, I.P., 88, 108 Reynolds, G.S., 104
Paykel, E.S., 74, 283 Reynolds, M.M., 76
Peed, S., 77, 240, 241 Rhamey, R., 211
Peine, H., 45, 224 Rich, T.A., 250
Pepper, M., 74 Richardson, J., 104, 160
Perry, D., 106, 107 Ricketts, A.F., 16
Peery, &-C., 106 RicksaiDFe. 75. 78
Pesses, D.I., 57 Riddle, M., 242
Peterson, L., 244, 266 Roach, J.L5 39
Peterson, R.F., 244, 266 Roberts, A., 242
Petrinovitch, L., 7 Roberts, E., 234
Phelps, R., 45 Robertson, S.J., 121
Phillips, J., 6, 93 Robins) L:N., 28; 34, 43, 221, 223, 259,262
Phillips, L.D., 5, 164 Robinson, B.E., 226
Phipps-Yonas, S., 243 Robinson, J.P., 250, 251, 266
Pinkston, E.M., 244 Rodnick, E.H., 27
Pinkston, S., 244 Roff, M., 28, 34-36
Platt, J., 230, 241 Rogers, E., 80
Plate, M1...229 Rogers, M.E., 39
Polk, K., 105 Rollins, B.C., 81

359
Romanczyk, R.G., 51 Sellin, T., 264, 265
Rosen, P.M., 228 Sells, S.B., 35
Rosenfeld, E., 44 Selye, H., 81, 217
Rosenfeld, H.M., 92 Sharpe, D.F,,°163220, 21, 35
Rosenthal, R., 52 Shaw, D.A., 3, 46
Rosman, B., 296 Shaw, G., 94
Ross, A.O., 89, 244 Shea, M.J., 27, 34, 259
Rothchild, J., 272 Sheese, J.G., 244, 266
Rowley, V.N., 281, 282, 289, 290 Shelly, M.W., 75
Rule, B.G., 13, 69, 73, 106 Shepherd, M., 39
Ruppenthal, G.C., 271 Sherman, H., 273
Rush, J., 87 Shields, M.M., 200
Russ, H., 39 Shirley, M., 76
Russell, C.S., 81 Shope, G.L., 146
Rust, C., 230 Shure, M.B., 230, 242
Rutherford, E.E., 244 Siddle, D.A., 136
Rutter, M., 13, 20-25, 29, 34, 37-39, 64, 67, 70, Sigal, J.J., 273
216, 218, 225, 234, 239, 246, 260, 281, 286, Sigel, I., 58
290 Silverman, M.I., 37
Ryan, W., 246 Simard, K., 253
Simcha-Fagan, O., 27, 265
Simmons, J., 243, 244
2 Singer,D.L., 157
Sackett, G.P., 6, 104, 108, 172, 186-188, 197, Singer, J:E.., 72
271 Skindrud, K.D., 46, 50-53
Sajwaj, T.E., 244 Skinner, B.F., 2, 6, 86-88, 104, 106-108, 116
Sallows, G., 228, 243 Slovic, P., 90
Salzinger, K., 187 Smith, H.T., 44
Sarbin, T., 244 Smith, M., 59
Sawin, D.B., 125 Smith, P.K., 20, 54
Saxe, R.M., 34 Smith, Ro; 34, 35737, 271
Scarpitti, F.R., 34 Smith, S.S., 208, 212
Seat. 5),4035. Snyder, J.J., 16, 61, 98, 100, 128, 130, 197, 204
Schachter, S., 72 Snyder, K.S., 230
Schaefer, E.S., 302, 303 Solomon, R.L., 112, 116-118, 124, 142, 147,
Schaffer, H.R., 209 226
Schalock, H., 58 Sparks, R,, 21, 22,260
Schelle, J., 43, 44 Spielberger, C., 91, 244
Schlesinger, S.E., 34 Spivak, G., 230, 242
Schmaling, K., 278 Staats, A.W., 177
Schmauk, F.J., 136 Staats; GK 77
Schnedler, R.S., 145 Stark, R., 113, 114, 156
Schoenfeld, W.N., 86, 88, 108 Stein, A.H., 35, 93
Schoggen, P., 3, 45, 77 Steinmetz, S.K., 113, 156, 157
Scholz, S., 160 Stephens, M.W., 118
Schooler, C., 266 Stephenson, G., 191, 192
Schuck, G.R., 122, 273 Stern, G., 59
Schumer, F., 296 Stevenson, H.W., 134
Schwartz, R., 1 Stollak, G.E., 34
Sears, .P:S.,(121 Stone, L., 114
Sears, R.R., 44, 104, 113, 121, 122, 127, 272, Stonner, D., 146
mo Stouthamer-Loeber, M., 224, 229, 259
Segal, J., 272 Straughan, J., 45
Seidman, S.B., 117 Straus, M.A., 113, 156-159, 163, 274, 285
Seidner, M., 72 Strayer, J., 35
Seligman, M.E., 71, 81, 147, 148 Streitman, S., 37, 216

360
 Stuart, R.B., 209 Vreeland, R., 266
Stunkard, A.J., 87
' Sturgis, E., 240
Sies.)., 175, 191, 192, 271 W
Surrott, P.R., 54 Wadsworth, M., 34, 221, 223
Swart, C., 106 Wahl, G., 96, 98, 99, 101, 108, 240
Sykes, R.E., 158 Wahler, R.G., 3, 34, 58, 80, 82, 83, 204, 218,
Szpiler, J.A., 70, 146 231, 234, 244, 266
Szykula, S., 132 Waksman, S., 62
Walder, b.O:, 22, 27, 34434800122, 4275137
234
T Waldrop, M.F., 228, 230
Taffel, C., 250 Walker, H.M., 29, 33, 98, 137, 182, 244, 266,
Tannenbaum, J., 44 269
Taplin, P.S., 42, 43, 51, 56, 60, 62, 67, 68, 101, Wallace, C.J., 230
io eaz, 134, 156, 157, 227, 232, 249 Waller, M.B., 266
Taquiri, R., 35 Wallerstein, J.S., 24, 25, 156, 231, 291, 296
Tarrier, N., 70 Walter, H.I., 44, 65, 307, 308
Tavormina, J.B., 289 Waltersa ithe, 915 95.105. 121 1225.
Taylor, J.R., 289 Warden, G.J., 117
Terdal, L., 241 Warren, V., 244, 245
Terrell, G., 118 Watson, J.S., 102, 107, 108, 133, 244
Theorell, T., 81 Waxler: C:Z.03,355 58,107, 246
Thomas, A., 52, 244 Webb, L.F., 229, 233
Thomas, D., 42, 59, Weick, K.E., 58
Thomas, D.R., 98, 113, 222 Weinrott, M., 54, 55, 62, 64, 306, 307
Dhomiasst.A., 6, 1/2, 187, 189, 190, 192 Weintraub, S.A., 242
Thorndike, E.L., 2, 88, 116, 117 Weiss, J.M., 71
Thurston, J.R., 34, 37 Weiss Relea /o5- 7659207, 2 LO-2112)229)5 232-234
Tinbergen, N., 170 Weissman, M.M., 283
Tizard, J., 20-25, 34, 37, 39, 64, 234, 239, 246, Wells, K., 32, 81, 240, 281, 283
260, 286, 290 Welsh, R.S., 137, 138
Toch, H., 72, 106, 157-159 Werner, E., 34, 35, 37, 271
Todd, N., 54 Wessman, A.E., 75
Tonge W.L., 78, 80, 217, 233 West Di Jise27
5134556510
7502345.209
Toobert, D., 64, 118, 258, 308 Whalen, K., 179
Traupman, J., 209, 210 White, G.D., 57-59, 119, 120
Tuchman, B., 94 White, M.N., 96 i
Tuppin, J., 32 Whiting, B.B., 20, 245, 266, 270
Tver A, 5, 90, 190, 226, 227 Whiting, J.M., 20, 245, 266, 270
Twentyman, C., 52 Whiting, J.W., 121
Whitmore, R., 20-25, 34, 37, 39, 64, 234, 239,
246, 260, 286, 290
U Wieder, G., 232, 258, 279, 308
Ullman, L., 2 Wiggins, J.S., 3, 42, 58
Ulrich, R., 52, 54, 104, 160 Willems, E., 28, 42, 96
Utne, M., 209, 210 Willerman, L., 76, 135
Willers, K.R., 146
Williams, B., 53-55
Vv Williams, D.E., 57
vanLawick-Goodall, J., 134 Wills, T.A., 75, 208, 210-212
Vaughn, C., 70, 230 Wilson, C.C., 121
Verhoff, J., 75 Wilson, T.D., 90, 91
VermarG.b:, 125 Wilson, W., 75
Vincent, J., 53-55, 212, 229, 230, 232-234 Wiltz, N.A., 65, 307
Voss, N.L., 34, 260 Wineman, D., 293, 294

361
Wing, J.K., 67, 70
Wolf,M.M., 2, 244, 266 Y
WolteS2272 ; Yahraes, H., 272
Wolfe, D.M., 233 Yoder, P., 77, 240, 241
Wolff, S., 21
Wolff, P.C., 104, 160
Wolfgang,
M.E.,157,264,265 Z
Wolking, W.D., 281, 282 Zajonc, R.B., 23
Woo, D., 145, 153, 154, 164 Zegiob, L.E., 54
Wright, H.F., 41, 42, 65 Zimet, S., 34, 243
Wright, Q., 94 Zlotogura, P., 187
Wyden, P., 163 Zyistra, J-Lis, 243
Wynne, L.C., 112, 116, 142, 147

362
 Subject index

Editor’s Note: When three or more consecutive see also Aversive events—as determinants of ag-
pages relate to the same topic, the symbol ff. has gression
been used to indicate “folio” or major discussion, see also Escalation
for example, “157 ff.” A page reference to a foot- Ai > R;
note is followed by the symbol n., for example, glossary of symbols, 84
“1S 7H? punishment effects, 85, 116, 117, 126 ff., 133
reinforcement effects, 85, 88 ff., 102 ff.
terms introduced, 85 ff.
Aggression (aggressive behavior) Anarchy progression, 11, 232 ff., 274, 275
angry (irritable) vs. nonangry (instrumental), Antisocial children, 11 ff.
ie, 47, 73,106 arrested socialization hypothesis, 18 ff., 38,
cultural programming for, 21, 81, 93 ff., 114, 241, 242
156, 157, 182, 246 characteristics shared by, 239 ff.
defined, 12, 13 defined, 14, 29 ff.
extinction, resistance to, 124, 136, 137, 147, described, 293 ff.
148, 163, 244 disrupted families and, 232 ff.
genetic determinants for, 21, 93, 134 ff., 244, hyporesponsiveness to social stimuli, 33, 107,
270 108, 133, 134, 136, 242 ff., 266n.
irritability and, 192 ff., 255 ff., 277 ff. marginal
adjustment
asadults,
11,12,27ff.,
learning of, 93 ff. 32ff.,259ff.
media and, 69, 70, 93 noncompliance of, 31, 32, 239 ff.
and negative attribution, 72, 73 peer relations, 33 ff., 245, 246
pain elicited, 106, 153, 160 self-perception, 37
punishment for, 111 ff. skills deficits, 32 ff., 220, 241 ff.
reinforcement for, 85 ff., 95 ff. treatment outcomes, 304 ff.
stability across settings, 28, 29, 181, 182, 241, Arousal, attribution, and anger, 68 ff.
269 anger
asrantrait. 9. 11,25 ff...195 and aggression, 13, 68, 69, 73, 106, 107
vicarious learning of, 6, 69, 70, 93, 121, 137, as a motivational state, 68, 73
162 arousal, 69 ff.

363
 (arousal, continued) and SES, 22
control of aversives and, 71, 72 and settings, 28, 29, 94, 181, 182
physiological measures of, 70, 146 and sex of child, 20 ff., 270
reduction hypothesis, 146, 147 traditional treatment failures, 1, 2
attribution and anger, 72, 73, 157 ff., 234 Coercive process
Aversive events defined, 13
a priori classification of, 12, 16, 77, 78, 82n., first stages in development of described, 271
129 see also Performance theory
as accelerating consequences, 128 ff. Coercion theory
assessment of, 14 ff. defined, 6, 13
and attrition of happiness, 74 ff. development of, 1-8
as determinants of aggression, 67 ff., 148 ff., key assumptions, 13
We iefs see also Performance theory
in family life, 13, 73 ff. Crises
and negative reinforcement, 148 ff. checklist, 78, 79, Appendix 4.1
as punishment, 127 ff. as stressors, 78 ff.
synchronicity of, 203, 204
Data collection (assessment)
Caretakers, 269 ff. development of general techniques, 2 ff., 41 ff.,
early training for role of, 270, 271 186, 187, 199
fathers’ irritability, 287 ff. interactive approach, 42
fathers’ mood, 289 judgments/ratings vs. observation, 3, 43 ff.,
fathers’ role, empirical definition of, 286, 287 Tis
fathers’ role in support system, 284 ff. observer effects on data, 49 ff.
mothers’ irritability, 192 ff., 228, 232, 255 ff., power index defined, 175, 176
277 the probability, conditional vs. base-rate, 5, 88,
mothers’ insularity (support systems), 78 ff., 89, 104, 172 ff., 186
DIS 2195-930, 234330234 sequential format, importance of, 3, 5, 7, 14,
mothers’ mood, 71 ff., 217 ff., 232 ff., 280 ff. 46, 188, 189
mothers as storm centers, 71, 72, 75, 76, 81, serial dependencies, 78, 82n., 172, 187 ff.,
35s 2/5 fk. 208, see also Thematic effects
normal children as stressors, 16 ff., 76 ff., Family Interaction Coding System (FICS)
239 ff., 270, 284 aversive categories (TAB) defined, 14 ff.
parental neglect, 272, 296 ff. code definitions, Appendix 3.1, 46
parental permissiveness, 272 ff., 296, 298, development of, 14, 41, 42, 45 ff., 64
3017302 limitations, 64, 65, 149, 152, 153, 173
single-parent families, 24, 25, 289 ff. methodology, 45 ff.
Catharsis theory, 147, 157 observer effects on data, 49 ff.
Child abuse (violent families) psychometrics, 16, 58 ff.
anarchy progression and, 274, 275, 299 MOSAIC, 64, 65, 106, 227, 258, 267n.
effect of observer presence on, 56, 68 PANIC, 232, 258, 267n.
escalation and violence, 95, 107, 155 ff., 162 Parent Daily Report (PDR)
as a function of family size, 23 development of, 41, 44, 45
and “Mixed” sample, 249, 274 psychometrics of, 58, 61, 62
and negative attribution, 68, 69, 72, 158, 159, TAB, correlation with, 16, 62
302 parent report data
normative data see Normals—physical punish- biases in, 8n., 43 ff., 122, 224, 249, 250
ment and violence TIRO (Telephone Interview Report Observa-
parental use of punishment, 112, 114, 115, tion) e255
(PXS5 As AOL Sib MS Tey DEY ae. Delinquency and crime
299 at risk factors, 27, 28, 30, 31, 34 ff., 157, 159,
Child aggression 165n., 218, 221, 223 ff), 22925 zeae
and age, 1316 ff., 25 ff., 38n., 76, 7/, 99, 04 249, 259 ff., 264 ff.
and broken homes, 24, 25, 289 ff. continuity hypothesis, 27, 28, 259 ff., 264 ff.
and family size, 22 ff., 64 defined, 259, 260
and ordinal position in the family, 25 family management variables and, 263, 264

364
 genetic determinants, 135, 136 overwhelmed parents, 298, 299
hyporesponsiveness hypothesis, 136, 243, 244 parent behavior and child pathology, 302 ff.
juvenile delinquency parental misattribution, 302
belt theory of, 137 the parent-sibling, 296, 297
chronic offenders, 136, 260, 262 ff., 299, “perfect” parents, 301
301, 306 sadomasochistic arabesque, 299 ff.
family management variables and, 157, 221, the unattached parent, 296 ff.
P25 tt., 229,.232, 252, 2545260,.262:ff.
feelings of efficacy and, 37
Learning
vs. performance, 6, 87, 88, 93
impulsiveness, 242
vicarious learning of aggression, 6, 93, 121,
labeling causes delinquency hypothesis, 32,
NSIC,WS2
33, 265
and “Mixed” sample, 249, 254, 259, 266 Marital satisfaction
and sex (male/female), 21 and antisocial child behavior, 156, 217, 231,
and SES, 22, 36 232, 285,3286;,290,.291
treatment outcomes, 306 child-rearing and, 81
see also Stealers for men vs. women, 75, 76, 81
pain as a reinforcer for violence, 106, 158, 160 misattribution by spouses, 200
perception of others (negative attribution), 37, mutuality of positive and negative events and,
38, 68, 69, 82n., 157 ff. 208, 209
self-perception, 33, 37, 159, 265 physical violence among spouses, 113, 114,
violence within families, 156 ff., 274, 275 PS63057 54163
Deviancy power differential between spouses and, 158,
drift, 32 ff. 285
operational definition of, 130 and problem-solving skills, 163, 209, 229 ff.
ratio of positive to negative events and, 75,
Escalation 2105211
and arousal-attribution process, 68, 69, 72, synchronicity of aversive events, 204, 205
158, 159, 302 synchronicity of positive events, 202 ff.
defined, 13, 157 Microsocial analyses of structure and process,
and negative reinforcement, 95, 141, 142, 169 She255-44.
155 f8.5-161 ff. behavioral events, prediction of, 170, 171
trained fighters and, 157, 160, 161 bilateral emphasis, implications of, 5, 192,
Extinction 195g1960255;5,296
and negative reinforcement, 146 ff. functional relations
resistance of aggressive behavior to, 124, 136, experimental manipulations of, 177 ff.
137, 147, 148, 163, 244 networks of, 173 ff., 181 ff.
utility of, 174 ff.
Family management variables, 215 ff., 258, 259 intersubject component, 172 ff.
assessment of, 232, 258 defined, 170
defined, 215, 216 inter- and intrasubject components together,
@naedeunguency, 157, 221, 223 ff., 229, 232, 172, 189 ff.
2526254)260, 262 ff. intrasubject component, 186 ff.
family management practices, crisis disruption defined, 169
of, 216 ff. irritable exchanges, 192 ff., 277 ff., 287 ff.
as mediators, 216, 232 and Markov chain, 190 ff.
parental monitoring, 222 ff. power index defined, 175, 176
habituation-exposure hypothesis, 224 settings, analysis of, 181, 182
parental sanctions, 224 ff. structure, defined, 169
problem-solving skills, 229 ff. see also Turn-taking, synchronicity, and reci-
defined, 230 procity
role in treatment, 230, 235 “Mixed” sample (Social Aggressors and Stealers)
rule setting, 126, 221, 222 and child abuse, 249, 274
traditional approaches to, 220 and delinquency, 249, 254, 259, 266
Family typologies, 31, 295 ff. parenting practices, 247, 249, 253, 254, 279
this child is special, 298 sample criteria, 249, 252, 253

365
Nattering microsocial analyses, 255 ff.
and antisocial child behavior, 69, 112, 118, see also Learning—vs. performance
123,130 ff. ,136, 138,228 Probability theory
defined, 69 comparison of conditional to base-rate, 88, 89,
Normals 104, 172 ff., 186, 250
aggressive child behavior, 16 ff., 39n., 93, 94, methodologists and theorists, 6
230, 241 power index, 175 ff.
anarchy progression and, 232, 274, 275 source material suggested, 5
children as stressors, 16 ff., 76 ff., 239 ff., Psychopathology
270, 284 child psychopathology and broken homes, 24,
criterion threshold for deviancy labeling, 248, 228 OEk.
249 hyporesponsiveness hypothesis, 136, 243, 244
escalation and, 159 parent behavior and child pathology, 302 ff.
guilt induction, 138, 163, 301, 302 at risk factors, 20, 27, 28, 31, 36, 37, 38n., 74,
noncompliance, 39n., 240, 241 216 ff., 230, 259, 260
physical punishment and violence, 113 ff., 157, stress and, 74, 217, 218
25/7, LIA Punishment
positive consequences for coercive behavior, for aggression, 111 ff., 137, 138
98 ff. aversive events as accelerating consequences,
positive consequences for prosocial behavior, 123 tf, 183. £6
96, 222,226, 227, 245 back-up,
111ff.,118,132,138,244
ratio of positive to negative events, 96, 232 cognitive factors (lectures), 113, 122, 123,
sample criteria, 63, 97, 252, Appendices 5.1 125, 126
and 6.2 contingency hypothesis, 133
stealing, 39n., 260 ff. cultural attitudes towards, 111 ff.
effectiveness and attachment of parent to child,
Pain control 120122, 10301 3c;
defined, 13 effectiveness (vs. reinforcement), 112, 113,
Parents of antisocial children 116 ff., 124
and child noncompliance, 32, 239 ff. effects on A;—»R; connections, 85, 116, 117,
nattering and, 112, 118, 123, 130 ff., 136, 126 ff., 133
138, 228 guilt induction, 138, 163, 301, 302
problem-solving skills of, 229 ff. implications, 136 ff.
selectivity of, 32, 123, 223, 224, 241, 247 ff., intensity and timing, 125, 126
YS) is ineffective (The Legend), 116 ff.
setting levels of aggressive behavior, 28, 29, long-term vs. short-term effects, 117, 128,
Ped, 132, 133, 138n.
as uncooperative, 49, 234, 297, 298 mixed schedules of punishment and reinforce-
use of punishment and reinforcement, 126 ff., ment, 117, 124
224 ff., 302 ff. modeling-frustration hypothesis, 137
Performance theory and negative reinforcement as members of the
contribution of family management variables, same family, 143
2VSs 21822227 2244229023225 Sao, noncontingent
147,148 (“learned
h elplessness”),
71,
contribution of functional relations, 174 ff.
contribution of interactional variables, 251 ff. parental monitoring and, 222 ff.
contribution of inter- and intrasubject compo- time out, 118 ff., 127, 133, 137 ff., 162, 163,
nents, 186 181, 222, 244, 266n.
contribution of irritability measures, 192 ff., traditional approaches, 112 ff., 121, 122, 127,
255 tf... 2/735 277 aie sete! 128, 137, 138
contribution of negative reinforcement, 155 variance accounted for, 89, 127, 132, 139n.
contribution of reinforcement, 89, 101 ff. see also Nattering
contribution of parent commands, 240
contribution of punishment, 89, 127, 132, Reinforcement, negative
139m. arousal-reduction and rigidity, 146, 147
development of, 6, 7 aversive antecedents, 148 ff.
goal of, 6, 62, 89, 186 in the classroom, 141

366
 complexity in social interaction, 142 ff. escalation and, 159, 161
defined, 89, 141, 143 family management variables and, 224, 258,
effectiveness (vs. positive reinforcement), 142, 259
144 microsocial analyses, 255 ff.
effects on Aj—» R; connections, 85, 142, 143, parental enmeshment, 247, 249, 257, 303, 304
ios, 185 parents’ criterion threshold for classifying devi-
and escalation, 95, 141 ff., 155 ff., 161, 162, ant behavior, 223 ff., 247 ff.
165, 166 prognosis for adjustment as adults, 259
experimental manipulations in situ, 153 ff., response classes, 182 ff., 254, 255
161, 162, 185 sample criteria, 18, 31, 97, 252, Appendices
and extinction, 147, 148 5.1 and 6.2
implications, 162 ff. symptom progressions for, 249 ff., 267n.
likelihood of counterattack and, 150 ff. treatment outcomes, 305 ff.
long-term vs. short-term effects, 142 ff., 162 Social learning theory
mixed schedules of negative reinforcement and development of, 2-8
positive reinforcement, 147, 148, 155 reinforcement concepts and, 2, 6, 7, 91
and punishment as members of the same family, see also Performance theory
143 Stealers, 259 ff.
reinforcement trap, 144 ff., 279 anarchy progression and, 274, 275
traditional approaches, 142, 143, 147, 162 continuity hypothesis, 259 ff.
utility of, 144, 151 ff. chronic offenders, families of, 262, 263
variance accounted for, 155 drop-out rates for, 262
Reinforcement, positive differences between Social Aggressors and, 31,
for aggression, 85 ff., 95 ff 246 ff.
and awareness, 90 ff., 145 frequency labeling hypothesis, 251, 260,
in the classroom, 96, 202, 226, 245, 271 263 ff.
contingency hypothesis, 102, 107, 108, 244, “normal” stealing, 39n., 260 ff.
245 parents’ criterion threshold for classifying steal-
defined, 86 ff. ing behavior, 223, 247 ff.
effects, learning vs. performance, 87, 88 parents as unattached (distant), 247, 249, 253,
effects on A,;—»R; connections, 85, 88 ff., INP, ZEN, PU) EDD, PX 5 BOB ie
102 ff. sample criteria, 18, 31, 252, Appendix 6.2
effectiveness (vs. negative reinforcement), 142, symptom progression for, 39n., 249 ff., 260,
144 262, 263
effectiveness (vs. punishment), 112, 113, TAB score comparison, 256
116 ff., 124 treatment outcomes, 305 ff.
mixed schedules of positive reinforcement and Symptom progressions, 249 ff.
negative reinforcement, 147, 148, 155 Stress
mixed schedules of punishment and reinforce- children as stressors, 76 ff.
ment, 117, 124 control of aversive events and, 71, 72, 81
noncontingent (“learned laziness”), 71 crises as stressors, 78 ff.
pain as a reinforcer, 106, 107, 157, 158, 160 disruption of family management practices and,
parental rejection of the concept, 113, 164, ORDA Ssci
225, 226 operational definition, 71
synchronicity and, 202 ff. physiological measures of, 70, 71, 81, 146
tests for status as, 103 ff.
traditional approaches, 86 ff., 101, 102, 104 ff. Thematic effects, 89, 100, 103, 104, 173, 186 ff.
variance accounted for, 89, 101 ff. see also Microsocial analyses of structure and
process—intersubject and intrasubject com-
Sample characteristics, 18, 62 ff., 97, 249, 252, ponents
253, Appendices 5.1 and 6.2 Treatment outcomes, 304 ff.
see also Delinquency and crime, “Mixed,” Nor- and drop-out rates, 65 n., 262, 305
mals, Social Aggressors, and Stealers Turn-taking, synchronicity, and reciprocity,
Social Aggressors, 254 ff. AST.
anarchy progression and, 274, 275 mutuality, 207 ff.
differences between Stealers and, 31, 246 ff. reciprocity, 209 ff.

367
(Turn-taking, synchronicity, and reciprocity, con- Variables
tinued) criterion for selection of, 2, 5 ff., 64, 89, 186,
shopping, 206, 207 245
synchronicity, 202 ff. interactional, use of to differentiate samples,
defined, 199 251 ff.
and ripple effects, 204 ff.
turn-taking, 200 ff., 208

368
 i
1 a ye
Wear

a aed
at
ye

te f eaCy zSent
yaBeh
Wade
\ a eaTAS, aa
a oe

Pavey
. a
Ay fh
 Paes yi
t
‘es
=ry rae
3. 3@

eR Na +h, ae
ARP GRE
on Beis
y)
SNAa
en'\

\
RJ 500 .$65 v.3

A Social learning arrroach

to family intervention /