JOURNAL OF APPLIED BEHAVIOR ANALYSIS 1991,24,509-522 NUMBER 3 (FALL 199 1)

A BRIEF FUNCTIONAL ANALYSIS OF AGGRESSIVE AND

ALTERNATIVE BEHAVIOR IN AN OUTCLINIC SETTING
JOHN NORTHUP, DAVID WACKER, GARY SASSO, MARK STEEGE,
KARLA CIGRAND, JONI COOK, AND AGNES DERAAD
THE UNIVERSITY OF IOWA

We conducted a brief functional analysis to identify maintaining variables for aggressive behavior
and an alternative replacement response during a 90-min outpatient evaluation of 3 individuals
with severe handicaps. During the initial analogue assessment, which focused on identifying main-
taining contingencies for aggressive behavior, each participant displayed a substantially greater
frequency of aggressive behavior during one condition than during any other. The contingency that
produced the highest percentage of aggressive behavior was then presented for the occurrence of a
specific alternative behavior (a mand). During this contingency reversal phase, each participant
displayed a substantial reduction in aggressive behavior and a substantial increase in alternative
behavior, thus providing a direct analysis of the equivalency of the contingency for maintaining
either behavior.
DESCRIPTORS: functional analysis, contingency reversal, outpatient assessment, aggressive
behavior, severely handicapped

Based on Carr's (1977) conceptual analysis of not with response topography or diagnosis. Similar
idiosyncratic maintaining conditions for aberrant findings have been reported across distinct groups
behavior, Iwata, Dorsey, Slifer, Bauman, and Rich- of children (e.g., Carr & Durand, 1985; Cooper,
man (1982) applied the functional analysis as an Wacker, Sasso, Reimers, & Donn, 1990; Steege,
assessment procedure to identify environmental Wacker, Berg, Cigrand, & Cooper, 1989).
variables affecting self-injurious behavior. Self-in- At least three general condusions can be reached,
jurious behavior was measured across four distinct based on a review of the available literature. First,
conditions during which environmental events were it has been demonstrated that self-injurious and
manipulated by the experimenters, including the aggressive behaviors can be lawful; they have a
presentation of demands, contingent social atten- functional relationship to specific environmental
tion, noncontingent social attention, and denial of events. Second, as proposed by Carr (1977), a
access to toys and other stimulating aspects of an number of investigators have demonstrated that
environment. The results showed that within-sub- these behaviors are multiply determined and there
ject variability was not random; it was consistently is substantial individual variability (e.g., Carr &
associated with a specific maintaining condition but Durand, 1985; Iwata et al., 1982; Steege et al.,
1989). Finally, the substantial individual variabil-
Portions of this artide were used to partially fulfill the ity and the frequently equivocal results of standard
requirements of an Ed.S. degree by the first author. M. Steege behavioral treatments strongly suggest the need for
is now at the University of Southern Maine; J. Cook is now individual assessment to indude a functional anal-
at Mankato State University.
Portions of this investigation were funded by the Iowa ysis of maintaining conditions (Iwata, Pace, Kal-
University Affiliated Facility (UAF). However, the opinions sher, Cowdery, & Cataldo, 1990).
expressed do not necessarily reflect the opinions of the UAF. The majority of previous research, however, has
The authors gratefully acknowledge the outpatient staff in been
the Division of Developmental Disabilities, Department of conducted in highly controlled, long-term in-
Pediatrics, and the participants and their families. In addi- patient settings. A functional analysis of maintain-
tion, we express our appreciation to Phyllis Harper-Bardach ing conditions has typically involved multiple as-
and the Office of Student Services for their support. sessment sessions (e.g., 40 to 60 sessions) over an
Requests for reprints should be sent to John Northup,
343 University Hospital School, The University of Iowa, extended period of time. Although generally rec-
Iowa City, Iowa 52242. ognized as a superior assessment, functional analysis
509
510 JOHN NORTHUP et al.

procedures have been characterized as complex, time level, such a demonstration would enhance the
consuming, and cumbersome (Axelrod, 1987; Doss practitioner's ability to develop an effective inter-
& Reichle, 1989). As a result, less accurate and vention directly derived from an empirical assess-
less reliable assessment procedures may be used ment of individual maintaining contingencies. On
(e.g., descriptive analyses, structured interviews). a more conceptual level, it would be instructive to
To provide further evidence of the utility of func- demonstrate response covariation between an ap-
tional analysis as an assessment procedure for severe propriate and an inappropriate behavior both main-
behavior problems, it is necessary to demonstrate tained by the same contingency.
the generalizability of the assessment procedures Finally, there is a relatively small amount of
and to determine if a more brief version of assess- research concerning persons with severe handicaps
ment is feasible. If an individual functional analysis who are aggressive (Lundervold & Bourland, 1988);
can be conducted in a shorter period of time, then analyses of aggression are urgently needed, because
many more individuals who display aberrant be- most community placements will not accept indi-
havior can be assessed during a period typical of viduals who are aggressive.
psychological evaluations in outclinic settings. Pre- The primary purpose of this investigation, there-
liminary research (Cooper et al., 1990) suggests fore, was to determine the feasibility of conducting
this possibility with children of average intelligence a brief functional analysis of aggressive behavior of
who display conduct problems, but no previous severely handicapped clients in an outpatient setting
studies have evaluated a brief approach to assess- during a 90-min period typical of outpatient psy-
ment of severely handicapped individuals who dis- chological evaluations. We conducted a brief func-
play severe behavior problems. tional analysis consisting of a series of analogue
A second concern relates to selection of treatment conditions lasting 10 min or less implemented dur-
based on the results of a functional analysis. Knowl- ing a 1-day outpatient evaluation.
edge of a maintaining contingency does not nec- This study also extended the procedural appli-
essarily dictate the selection of the most effective cation of functional analyses of maintaining con-
intervention. This concern with treatment selection tingencies to include an evaluation of replacement
has been discussed in terms of identifying appro- behavior. We accomplished this by including a
priate replacement behavior (Carr & Durand, 1985). contingency reversal component within the 90-min
The development of a replacement response is of outpatient assessment protocol. During the contin-
concern for several reasons; however, it is an essen- gency reversal phase, the contingency identified as
tial concern for establishing the long-term effects maintaining aggressive behavior was provided for
of treatment. The development of an appropriate an alternative replacement behavior. This compo-
replacement response is particularly necessary in nent provided an analysis of the equivalency of the
those cases in which the aberrant behavior serves a contingency to maintain an alternative replacement
specific function (e.g., results in attention). Failure behavior as well as to maintain aggressive behavior.
to provide a replacement behavior might increase The results of the contingency reversal also provided
the probability, by default, that another inappro- an initial empirical demonstration of a potentially
priate behavior will emerge to serve the same func- effective treatment.
tion, especially with individuals who have very re-
stricted repertoires of appropriate behavior. METHOD
An important question for the development of
effective treatments is whether the same contingen- Participants
cies maintaining inappropriate behavior can be used Participants were 3 individuals evaluated through
to maintain an alternative, replacement behavior; the Self-Injurious and Aggressive Behavior Service,
it has not been demonstrated that a contingency Department of Pediatrics, The University of Iowa.
identified as maintaining an aberrant behavior will They were the first 3 patients referred to the clinic
also maintain alternative behavior. On a practical for evaluation of aggressive behavior. No other
 FUNCTIONAL ANALYSIS 5 11

criteria were used for subject selection. Curtis was gressive behavior, and Heidi was receiving thorazine
a 24-year-old male referred by his parents for eval- and naltrexone. Heidi resided at a large state res-
uation of aggression. Curtis had been diagnosed as idential facility.
functioning in the severe to profound range of men- Genia was a 13-year-old female with cerebral
tal retardation, was nonverbal, and displayed no palsy, referred by her parents for evaluation of
formal means of communication. His aggressive noncompliance and aggressive behavior. Genia had
behavior consisted of attempts to scratch, pinch, been diagnosed as functioning in the moderate to
grab, hit, or pull hair. These behaviors were re- severe range of mental retardation, and her primary
ported to occur a minimum of three times a day means of communication was verbal expression. A
during the past 7 years and had been increasing in review of her records indicated that her receptive
frequency and intensity during the previous 4 or 5 language was sufficient for basic daily needs, with
months. Approximately 1 month prior to evalua- expressive language abilities that allowed her to
tion, Curtis was admitted to a local hospital for 2 complete five- to six-word sentences; her overall
weeks for treatment of aggression. At that time, intelligibility was described as being fair to good.
he was placed on tegretol and haldol, which he Parental report indicated that Genia expressed a
continued to receive at the time of our evaluation. variety of requests, comments, and questions. Ge-
No other formal treatment programs had been re- nia's aggressive behavior consisted of attempts to
ported. Curtis was legally blind, but he presented pinch, bite, and hit. These behaviors were reported
no other medical concerns. At the time of evalu- to occur at least daily for the past 5 to 10 years,
ation, Curtis resided with his parents, because no both at home and in school. A variety of interven-
group home would admit him. He had resided tions had been attempted, induding redirection,
previously at a state institution and at a community time-out, and various punishment procedures. Genia
residential facility. resided at her parents' home.
Heidi was a 21-year-old female referred by the
staff of her residential facility for evaluation of ag- Setting
gressive behavior. Heidi had been diagnosed as The Self-Injurious and Aggressive Behavior Ser-
functioning in the severe to profound range of men- vice is an interdisciplinary service located at a Uni-
tal retardation, was nonverbal, and displayed no versity Affiliated Facility (Wacker, Steege, Nor-
formal means of communication. Her aggressive thup, Reimers et al., 1990). The protocol used in
behavior consisted of attempts to pinch, hit, or bite, this investigation was incorporated into the stan-
and occurred as frequently as 280 times per month. dard evaluation conducted by the Self-Injurious and
Heidi also had a long history (over 10 years) of Aggressive Behavior Service. All conditions were
self-injurious behavior consisting of face slapping conducted in a dassroom at the hospital's inpatient
and self-pinching. The frequency and intensity of unit. The classroom was equipped with a one-way
these behaviors were reported to have been quite mirror to permit unobtrusive observation. A lengthy
variable, occurring as frequently as 1,200 times per questionnaire was requested from the referring agent
month during the previous year. Approximately 2 for each participant prior to the evaluation. The
months prior to our evaluation, Heidi was referred referring agent was also briefly interviewed on the
to a psychiatric facility for inpatient treatment of day of the evaluation. Specific items and materials
her self-injurious behavior, which was considered used with analogue conditions for each participant
to be under control following discharge. Heidi pre- were developed based on this information.
sented no significant medical conditions. Previous
treatments induded differential reinforcement of Response Definitions and Measurements
other behavior, differential reinforcement of alter- Response definitions. Three dasses of responses
native behavior, time-out, physical restraints, and were recorded for each participant: (a) aggressive
medication. At the time of our evaluation, a grad- behavior, (b) appropriate behavior, and (c) an al-
uated time-out procedure was being used for ag- ternative behavior (a mand). The topography of
512 JOHN NORTHUP et al.

aggressive behavior was individually defined for students who had previous observer training ex-
each participant. For Curtis, aggressive behavior perience in the service. Interobserver agreement data
was defined as any attempt to scratch, pinch, hit, were obtained on a minimum of 57% of the sessions
or grab the experimenters. For Heidi and Genia, for every individual. Overall measures of agreement
aggressive behavior was defined as any attempt to were calculated on an exact interval-by-interval ba-
pinch, hit, or bite the experimenters. sis by dividing the total number of agreements by
Appropriate behaviors were defined generally as the total number of agreements plus disagreements
being on task, and specifically as being actively and multiplying by 100 (Kazdin, 1982). Overall
engaged with the experimenter, interacting with interobserver agreement averaged 93% for all ses-
toys or with other materials in an appropriate man- sions and ranged from 71% to 100% across indi-
ner, following directions, and engaging in any ap- viduals.
propriate social interaction (e.g., smiling, eye con-
tact). Appropriate behavior did not include sitting Design
passively, staring, or posturing. Manding behavior We used a multielement design, consisting of
was defined as any recognizable verbalization, man- two rapidly changing reversal designs conducted in
ual sign, or any other clearly recognizable gesture. two phases: an initial analogue assessment and a
This response was recorded whether prompted or contingency reversal. Heidi was observed during an
unprompted. initial analogue assessment consisting of the follow-
Independent variables. There were three cate- ing four conditions: alone, tangible, escape, and
gories of independent variables, each representing social attention. Genia was observed during an ini-
the contingent presentation or withdrawal of con- tial analogue assessment consisting of alone, social
sequences. The three types of consequences were attention, and escape conditions. For Curtis, the
identified as tangibles, attention, and escape. Spe- analogue assessment consisted only of alone and
cific tangibles were individually identified for each escape conditions. The social attention and tangible
participant; examples included edibles, toys, and conditions were not conducted for Curtis, because
preferred items and activities. Social attention was he was observed to be unresponsive to social in-
defined as praise, reprimands, verbal comments, or teraction and tangible reinforcement (he initially
physical contact by the experimenters. Escape was lay on the classroom floor and physically resisted
defined as the termination of a task or activity any attempts to engage him in activities or physical
contingent on any specific behavior of the partici- contact), and this observation was consistent with
pant. referral information.
Data collection. During each session, an ob- Following the initial analogue assessment, all
server recorded the occurrence or nonoccurrence of participants were observed during three additional
each of the three categories of participant responses. conditions, referred to as a contingency reversal. In
All responses were manually recorded using a con- the first contingency reversal condition, the contin-
tinuous 6-s partial interval recording procedure. A gency producing the highest percentage of aggres-
tape recorder signaled the recording interval num- sive behavior during the analogue assessment was
ber at the end of each 6-s interval. All observations again presented, but the consequence was provided
were conducted through a one-way mirror adjoining contingently upon the occurrence of appropriate
the classroom. manding rather than for aggressive behavior. Ag-
Interobserver agreement. Two observers simul- gressive behavior was ignored (Heidi and Genia),
taneously but independently scored all responses or graduated guidance was used to redirect the
during 18 sessions, which constituted 90% of all participant to task (Curtis). This condition was
sessions. The observers consisted of the authors, followed by a control condition, which was either
members of the Self-Injurious and Aggressive Be- a complete reversal in which the condition pro-
havior Service team, and graduate or undergraduate ducing the highest percentage of aggressive behav-
 FUNCTIONAL ANALYSIS 513

ior during the initial analogue assessment was re- of the therapist, the absence of preferred items, the
peated (Heidi and Genia) or the alone condition absence of social attention and interaction, and the
was repeated (Curtis). The control condition was absence of demands were compared with this con-
then followed by a second contingency reversal con- dition.
dition to form a reversal design. For the social attention condition, a therapist
was present in the room and maintained a proximity
Procedure of approximately 1.5 to 3 m to the participant at
Prior to each evaluation, all referral information all times. The therapist interacted and attended to
was reviewed, responsibilities were identified for the participant contingent upon the occurrence of
each team member (i.e., data collection, therapist, aggressive behavior, but otherwise ignored the par-
parental interview), the appropriate assessment pro- ticipant. Typically, the therapist was seated and
tocol was selected, and the order and type of as- appeared to read a magazine or complete paper-
sessment conditions were determined. Preferred work. Although no specific tasks or activities were
items or activities were selected for the tangible provided, the participant could engage freely in
conditions based on referral information, question- activities and move about the room. Contingent
naire data, and interview. Similarly, a task was upon the occurrence of aggressive behavior, the
selected for the escape condition based on actual therapist immediately provided attention to the
tasks and demanding situations required of the participant for 10 to 15 s. Attention consisted of
participant locally. Only tasks considered to be verbal reprimands (e.g., "Please don't do that"),
functional, age appropriate, and challenging for the a light touch on the shoulder, and continued social
individual were selected for this condition. interaction for as long as aggression occurred. All
Analogue assessment. For Heidi and Genia, the other responses, including appropriate and mand-
analogue assessment consisted of alone, tangible, ing behavior, were ignored.
demand, and/or social attention conditions, based During the escape condition, the participant was
on the analogue conditions used by Iwata et al. seated at a table, and the therapist presented the
(1982) and Carr and Durand (1985). During these task of folding and sorting towels and washdoths
conditions, the presentation of any consequence was (the same task was selected for all 3 participants).
always contingent upon the occurrence of aggressive Verbal instructions and modeling of the task were
behavior. All sessions lasted from 5 to 10 min, provided initially, followed by graduated guidance
with a brief( 1 to 2 min) break between each session for incorrect or incomplete task responses. The task
during which the experimenter left the room and was presented continuously at a stable rate through-
briefly reviewed the condition for the next session. out the condition, unless aggressive behavior oc-
All sessions began with the alone condition; the curred. Contingent upon the occurrence of aggres-
subsequent sessions occurred in a counterbalanced sive behavior, the task was immediately removed,
order. For Curtis, the analogue assessment consisted and the therapist turned or moved away from the
of alone, escape, alone, and escape conditions for participant for 15 to 30 s or until the participant
the reasons previously discussed. discontinued his or her display of the behavior, at
During the alone condition, the participant was which time the task was immediately reinstated.
directed to the classroom and was given the in- No verbal praise was provided for correct perfor-
struction to "wait." The therapist then left the mance, and all interactions were limited to provid-
room and had no further interaction or contact with ing task instructions and prompts. Neutral and
the participant. In the classroom, a variety of toys appropriate responses were ignored.
and materials were accessible; however, no specific In the tangible condition, the therapist remained
tasks or activities were directly provided to the in the room and maintained a proximity of 1. 5 to
participant. The alone condition served as a baseline 3 m to the participant. Contingent upon the oc-
for the other three conditions; that is, the presence currence of aggressive behavior, the therapist im-
514 JOHN NORTHUP et al.

mediately presented the tangible item for approx- For Curtis, the consequence during the contin-
imately 15 to 30 s. All other responses were ignored, gency reversal condition was escape. For Heidi, the
and the therapist engaged in no other interaction consequence during the contingency reversal was a
with the participant. preferred tangible item, and for Genia, the con-
Contingency reversal. Three additional condi- sequence was social attention. For Curtis, the same
tions immediately followed the completion of the task, folding towels and washdoths, was presented
analogue assessment phase. The contingency re- in the same manner as in the analogue assessment.
versal phase began with the condition that produced However, after approximately 30 s, he was prompt-
the highest percentage of aggressive behavior dur- ed to use the "please" sign. Prompting following
ing the analogue assessment. However, rather than a least-to-most restrictive prompt sequence, begin-
being presented for aggressive behavior, the con- ning with a verbal instruction and ending with
tingency was now presented for the occurrence of complete physical guidance. Prompts continued to
a specific manding response, which was modeled be provided approximately every 30 s. Contingent
several times for the participant at the beginning upon each occurrence of the "please" sign (prompt-
of the condition. For Curtis and Heidi, the alter- ed or independent), the task was immediately re-
native behavior was the display of the "please" moved, and the therapist turned or moved away
sign. This sign was modeled and physically prompt- from the participant for 15 to 30 s. All occurrences
ed approximately every 30 s. The consequence was of aggressive behavior resulted in graduated guid-
delivered whenever the sign was emitted indepen- ance to redirect Curtis to the task. For Heidi, the
dently or whenever the participant did not resist same procedures were followed; the tangible item
the physical prompt. For Genia, the alternative was presented contingent upon occurrence of the
behavior was verbally stating, "Come here, please." "please" sign, and all aggressive behavior was ig-
At the beginning of each contingency reversal con- nored. For Genia, social attention was provided for
dition, Genia was given the verbal instruction, "If 15 to 30 s contingent upon her stating, "Come
you want to talk to us, just say, 'Come here, here, please," and aggressive behavior was ignored.
please.'" No other prompts were given. The contingency reversal sessions provided a di-
Following this condition, a reversal was achieved rect analysis of the contingency for appropriate and
by repeating the condition from the analogue as- inappropriate behavior. If the frequency of use of
sessment that produced the highest percentage of the alternative, replacement behavior (use of the
aggressive behavior (Heidi and Genia) or the alone "please" sign or stating, "Come here, please") in-
condition (Curtis). For Heidi and Genia, the con- creased and the frequency of aggressive behavior
sequence was again provided contingently upon the decreased during these conditions, a potentially ef-
occurrence of aggressive behavior. As before, all fective treatment intervention was available based
other appropriate or neutral behaviors (induding directly on the assessment data.
use of the alternative, replacement behavior) were
ignored. This condition, conducted for 5 min, pro- RESULTS
vided both a replication of the analogue assessment
condition and a reversal within the contingency The results of the analogue assessment and con-
reversal phase. Following this reversal, the contin- tingency reversal phases for each participant are
gency reversal condition was repeated. For Curtis, displayed in Figures 1, 2, and 3, respectively. Dur-
the alone condition was repeated because of the ing the initial analogue assessments, each of the
intensity of his aggressive behavior displayed during participants displayed a greater percentage of ag-
the analogue assessment (one of the therapists had gressive behavior during one maintaining condition
been bruised). The alone condition provided an than during any other.
adequate control for his display of inappropriate Curtis displayed aggressive behavior during the
behavior and the "please" sign, and thus was con- escape conditions only. Because the tangible and
sidered sufficient for the evaluation. social attention conditions were not conducted, Cur-
 FUNCTIONAL ANALYSIS 515

OANALOGUE ASSESSMET CONTINGENCY REVERSAL ANALOGLE ASSESSMENT ITNGCY REVERSAL

Contingency Conti ngehnc:y
Contilnoenc: Contingencgy Alone Tangible Esape Attention Reversal Tangible Reversal 11
00 Alone Esca pe Al one Escape ReversallI Aloe Reversal 100- 0
90
.Appropriate

70
60
o

~40
~5O 0 U
40
20~~~~~~~~~
30 .0

20
Sign I0
20 -Aggression *
10
- 0L

0 00 0 10 Sign
F~~~~~~~~~~~~
0 0E * m
1 2 3 4 5 6 7
CONSECUTIVE SESSIONS 1 2 3 4 5 6 7

Figure 1. Performance of Curtis across conditions during CONSECUTIVE SESSIONS

the analogue and contingency reversal phases of assessment. Figure 2. Performance of Heidi across conditions during
the analogue and contingency phases of assessment.

tis was observed during additional alone and escape

conditions. The results of these two additional con-havior was of low intensity and frequently consisted
ditions provided both a partial replication of the of only light taps on the therapist's back.
results of the first escape condition and a reversal Curtis' use of the "please" sign increased sub-
in performance across conditions. stantially, from 0% during the analogue assessment
Curtis displayed no aggressive, appropriate, or to 20% and 32%, respectively, during the contin-
manding behavior during either alone condition. gency reversal conditions. Curtis continued to dis-
Instead, Curtis sat or stood passively with little play no appropriate behavior other than signing;
movement, although he occasionally engaged in that is, at no time did he actively participate or
some posturing behavior. During the escape con- cooperate with the completion of the task. How-
ditions, physical guidance was used to attempt to ever, it was anecdotally noted that Curtis became
engage Curtis in the task of towel folding. Most increasingly less resistant during the contingency
reversal conditions. A replication of the alone con-
interaction consisted of providing physical guidance
to attempt to position him at the work setting; he dition resulted in an immediate return to 0% oc-
never participated in completing the task. Althoughcurrence of all target behaviors.
prompted and unprompted signing were not re- Heidi displayed aggressive behavior during the
corded separately, anecdotally it was noted that all
tangible and escape conditions only. In the tangible
signs for Curtis were prompted, but there was some condition, she displayed aggressive behavior during
increased participation as the sessions progressed.17% of the intervals, whereas in the escape con-
In the first escape condition, Curtis displayed ag-dition, she displayed aggressive behavior during
gressive behavior during 23% of the recorded in- 13% of the intervals. Although the percentage of
tervals. In the second escape condition, Curtis dis-
occurrence was highest during the tangible condi-
played aggressive behavior during 12% of the tion, the relatively high percentage of occurrence
intervals. However, the intensity of his aggressionduring the escape condition suggested that Heidi's
during both conditions was sufficient to cause bruis-
aggressive behavior may have served multiple func-
ing to an expenmenter. tions; that is, it functioned to obtain access to pre-
For Curtis, the consequence of the contingency ferred items in some situations and to escape from
reversal conditions was escape from task demands. undesirable tasks or activities in other situations.
The contingency reversal conditions substantially Heidi displayed some appropriate behavior dur-
reduced the frequency of occurrence of Curtis' ag- ing the analogue assessment. In the escape condi-
gressive behavior, which decreased to 3% and 8%, tion, Heidi participated in completing the task dur-
respectively. Of equal importance, aggressive be- ing 26% of the intervals. During the alone and
516 JOHN NORTHUP et al.
ANALOGUE ASSESSMENT CONTINGENCY REVERSAL tensity of Heidi's aggressive behavior decreased in
Conti ngencut Contingency,
100
Alonc Attention Escape Reversal
0
Attention Reversal
0
III( the contingency reversal conditions.
90
Heidi appeared to discriminate quickly the change
.Appropriate
in contingencies. During the reversal condition, her
4 70- use of the "please" sign decreased to 0%, but
60
0

following an initial prompt during the second con-

tingency reversal condition, she again independently
used the "please" sign to request her preferred item.
30
Genia displayed aggressive behavior during the
20 -Aggression
escape and attention conditions. In the escape con-
10- Sign
dition, Genia displayed aggressive behavior during
only 6% of the intervals; however, the aggressive
1 2 3 4 5 6 behavior was of sufficient intensity that it was nec-
CONSECUTIVE SESSIONS
essary to discontinue the assessment condition for
Figure 3. Performance of Genia across conditions during about 4 min. Following this break, Genia was
the analogue and contingency phases of assessment.
instructed to return to the towel-folding task, and
her behavior was appropriate for the remainder of
attention conditions, Heidi sat quietly or was ac- the observation period. Genia's aggressive behavior
tively engaged with items throughout those con- occurred with substantially greater frequency (24%
ditions. Heidi displayed no manding responses dur- of the recorded intervals) during the attention con-
ing the analogue assessment. dition.
Heidi displayed no aggressive behavior during Genia displayed some appropriate behavior dur-
the contingency reversal conditions, compared to ing each initial assessment condition. In the alone
17% during the analogue assessment. In addition, condition, she appropriately engaged in play activ-
when the contingencies were again provided for ities throughout the observation period. In the es-
aggressive behavior, her aggressive behavior in- cape condition, she appropriately participated in
creased to 20%. task completion during 81% of the intervals. Ap-
Heidi's use of the "please" sign also increased propriate behavior was lowest during the attention
substantially, from 0% during the analogue as- condition (65% of the intervals). Genia displayed
sessment to 46% and 50% during the contingency no manding behavior during the initial assessment
reversal conditions. Anecdotally, it was noted that conditions.
Heidi began to sign independently after only two Genia also displayed no aggressive behavior dur-
prompts. Appropriate behavior increased from 0% ing either contingency reversal condition. Genia
during the tangible condition of the analogue as- quickly followed the instructions; she stated, "Come
sessment to 54% and 50% when the same contin- here, please," during the first 6-s interval of both
gency was provided for use of the "please" sign contingency reversal conditions, and she continued
and aggressive behavior was ignored. to request attention appropriately during 18% and
For Heidi, the tangible condition of the analogue 20% of the intervals, respectively. It was noted that
assessment was repeated as the control condition. all of Genia's verbalizations were independent fol-
The results provided further replication of the re- lowing the initial instruction.
sults of the analogue assessment and provided a For Genia, the attention condition of the ana-
reversal within the contingency reversal phase. Dur- logue assessment was repeated. Attention was again
ing the replication of the tangible condition, ag- provided only upon the occurrence of aggressive
gressive behavior increased to 20%, but returned behavior, and all other behavior (including the
to 0% when the contingency reversal condition was statement, "Come here, please") was ignored. Ge-
reinstated. Anecdotally, it was noted that the in- nia appropriately requested attention during the
 FUNCTIONAL ANALYSIS 517

o 50
LU
CONTINGENCY REVERSAL I ALONE CONTI INGENCY REVERSAL II

8 45
" 40
or
w 35
Z 30
X'0 25
o 20 0000
m 15
E 10
>
5 _ _

-J
DO
C) ICURTISI
1 2 3 4 5 6 7 8 10 1 2 3 4 5 1 2 3 4 5 6 7
Consecutive Minutes
Figure 4. Minute-by-minute analysis of signing and aggression for Curtis during the contingency reversal phase of
assessment.

A steep, upward increase in rate occurred im-

first five intervals of the control condition by stating
"Come here, please" three times. However, she mediately for the appropriate manding response
then became aggressive toward the therapist and that continued throughout the contingency reversal
continued to display aggressive behavior through- conditions for all participants. However, once the
out the observation period, during 54% of the contingency was removed during the control con-
intervals overall. In addition, her overall behaviordition, the manding response remained at 0% for
appeared to change, and she displayed a variety of Curtis and Heidi and returned to 0% after the first
inappropriate postural or gestural movements, minute for Genia. Once the contingency was re-
laughed inappropriately, and became increasingly instated during the second contingency reversal con-
active. These behaviors were immediately reduced dition, the upward trend occurred again.
when she again was provided attention for an ap- For Curtis, aggressive behavior slowly stabilized
propriate request. during the contingency reversal conditions, and all
To determine whether aggressive behavior and target behaviors returned to 0% during the control
the use of an appropriate manding response varied condition. For Heidi and Genia, aggressive behav-
within as well as across conditions, a minute-by- ior remained at 0% during the contingency reversal
minute analysis was conducted for the conditions conditions, but steadily increased during the control
in the contingency reversal phase. The number of condition. Thus, immediate control over behavior
intervals in which aggression or an appropriate was established in all conditions for all participants,
manding response occurred were plotted cumula- and all trends occurred in the predicted direction.
tively against time (minutes of the session). Of
interest were the rate of response and change of
rate occurring within conditions for each of the
DISCUSSION
target behaviors. The results are displayed in Fig- The results of this investigation replicated and
ures 4, 5, and 6. extended existing research in three ways. First, the
518 JOHN NORTHUP et al.

50 -
CONTINGENCY REVERSAL T ANG IBLE CONTINGENCY REVERSAL 11
w

0
45 -

Co
-J
40 -

w 35 -
z
30 -
UJ)

L.. 25 -
0

m 20 -

15 -

I-
-AJ
Sign
10 -
4fI

5- 7g Aggression
_r - - _ _ _
0-
I Heidi |
I I

1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 1 2 3 4 5
CONSECUTIYE MINUTES
Figure 5. Minute-by-minute analysis of signing and aggression for Heidi during the contingency reversal phase of
assessment.

results provided further replication of the feasibility desired direction of behavior change. Rather, an
of conducting a brief functional analysis of behavior individual assessment of the function of behavior
in a setting and time frame (90 min) typical of is necessary prior to the implementation of treat-
psychological evaluations. The results of a series of ment. In each case, the results of assessment were
brief (5 to 10 min) analogue conditions, conducted not predictable based on structural or demographic
during a 90-min outpatient assessment, indicated analyses of the participants, and distinct treatments
that the aggressive behavior of each participant was were recommended based on the results of the
associated with a specific maintaining contingency. functional analysis procedures.
These results, and the results of Cooper et al. (1990), Third, and most importantly from a conceptual
suggest that functional analysis procedures used standpoint, the results of this investigation dem-
during assessment are generalizable across settings, onstrated that the contingencies identified as main-
response topographies, demographic characteristics, taining aggressive behavior also served to reinforce
and maintaining conditions. an altemative, replacement behavior. These results
Second, these results provide additional support are important because they provide direct evidence
to previous investigations (Carr & Durand, 1985; of the treatment utility of functional analysis as an
Iwata et al., 1982; Steege et al., 1989), which assessment procedure (Hayes, Nelson, & Jarrett,
suggested that (a) severe behavior disorders should 1987).
not be considered only as motoric responses that A distinctive feature of this study was the very
can be reduced or suppressed and (b) treatment rapid effects obtained during all conditions of the
should not be selected solely on the basis of the study and especially for the contingency reversal
 FUNCTIONAL ANALYSIS 519

a 50- ATTENT ION CONTINGENCY REVERSAL 11

LUI CONTINGENCY REVERSAL I
0
)
(A
45-
40-
LU
I--
35-
z
c.,LU 30-
C,)
% 25-
0
W- 20-
m
D
z
15-
LUI
Sign
> 10 -

I-
: 5- gere3ion
L) 0- I Genia I

I I I I I I I I I I

1 2 3 4 5 6 7 1 2 3 4 5 1 2 3 4 5 6 7 8 9 10
CONSECUTIVE MINUTES
Figure 6. Minute-by-minute analysis of signing and aggression for Genia during the contingency reversal phase of
assessment.

conditions; rapid effects were demonstrated for both of response generalization. An experimental dem-
a decrease in aggressive behavior and an increase onstration of response generalization would require
in an alternative replacement response (a mand). a contingency to be presented for only one dass of
There are at least four interrelated plausible expla- behaviors (R1), with changes noted in another dass
nations for these rapid results. of behaviors (R2). Such a demonstration assumes
The first explanation involves Carr's (1988) sup- that a contingency is not presented simultaneously
position that communicative responding and some for both dasses of behavior; that is, R2 is not
behavior problems
severe may be functionally directly manipulated.
equivalent. This explanation is supported by the In the present investigation, guided compliance
response covariation that occurred during the con- was used contingently for aggressive behavior dur-
tingency reversal conditions; as manding increased, ing the contingency reversal condition for Curtis,
aggressive behavior decreased. If both responses and the aggressive behavior of Heidi and Genia
(aggression and manding) result in the same out- was placed on extinction. Thus, the contingency
come (i.e., are functionally equivalent), strength- reversal conditions actually induded at least two
ening one response should weaken the other (Carr, active treatment components: differential reinforce-
1988). In this investigation, Carr's supposition is ment of appropriate (manding) behavior and guid-
supported not only by the observed response co- ed compliance or extinction. Although an analysis
variation, but also because the same contingency of the separate effects of each component was not
maintained both responses. conducted, it is possible that the indusion of both
However, the inverse relationship between the components was necessary for the very rapid results
mand and aggressive behavior observed in the pres- obtained.
ent investigation cannot be evaluated solely in terms Wacker, Steege, Northup, Sasso et al. (1990)
520 JOHN NORTHUP et al.
conducted a component analysis of a functional an alternative response should result in both an
communication treatment package in which the increase in that response and a decrease in the target
separate effects of a communicative (manding or behavior (Carr, 1988). In the Wacker, Steege, Nor-
activating a prerecorded message) response, guided thup, Sasso et al. (1990) study, the density of
compliance or time-out, and density of reinforce- reinforcement was controlled for and the same over-
ment were examined. The results indicated that all amount of reinforcement was provided on a
guided compliance or time-out were necessary for differential-reinforcement-of-other-behavior sched-
maximal control of self-injury or aggression with 2 ule during communication training, without regard
children. These authors further suggested that if to the mand. The results indicated that the density
the maintaining contingency for a target behavior of reinforcement was an active component contrib-
was not accurately identified and actively disrupted, uting to overall treatment effects. The continuous
the response remained in the individual's repertoire schedule of reinforcement used in this investigation
along with the new alternative response and was for the mand may also have contributed to the
likely to be displayed at least occasionally. How- rapid effects observed during the contingency re-
ever, the trained mand was also demonstrated to versal conditions. Because of the limited practicality
have a needed effect attributed to the self-sched- of CRF schedules in most situations, future inves-
uling of reinforcement made possible by the mand. tigators need to determine the therapeutic validity
Thus, both components were necessary for maximal of these procedures in applied settings.
control and may have been necessary in the present In summary, the rapid results observed during
investigation as well. the contingency reversal conditions may be artrib-
The third explanation is that the use of contin- uted to the cumulative effects of establishing func-
gencies for inappropriate behavior are not neces- tional equivalence between the two responses, a
sarily punishing but facilitate responding through disruption of the maintaining contingency for self-
extinction. Given that the variables maintaining a injurious or aggressive behavior, a dense schedule
behavior are known, their removal would be ex- of reinforcement for a manding response, and the
pected to weaken a behavior without need for pun- self-control of reinforcement provided by the mand.
ishment or suppression of behavior with aversive On a practical level, the two most important
stimuli (Ferster, 1961). Thus, even if guided com- findings were that (a) functional analysis techniques
pliance was not aversive, its use to prevent previous appeared to generalize to outclinic settings (as dem-
escape responding constitutes an extinction proce- onstrated by the control established over each par-
dure for aggressive behavior (Iwata et al., 1990). ticipant's behavior), and (b) the results of the as-
Because an additional contingency was in effect for sessment were useful for prescribing specific
each participant's aggressive behavior in the present treatments. The demonstration of the effect of the
study, it is reasonable to assume that the effects contingency reversal provided the practitioner with
observed during the contingency reversal condi- an empirically based rationale for initiating inter-
tions, at least in part, may be attributed to extinc- vention. However, some concerns with the present
tion or escape extinction. study should be considered when interpreting the
The fourth explanation is the dense schedule of results.
reinforcement provided for the manding response It is of substantial concern that constructing an-
during the contingency reversal conditions. The alogue conditions in which a severe behavior might
manding response was consistently responded to on occur could place both the dient and the therapist
a continuous schedule of reinforcement (CRF). A in a position of some risk. For both therapeutic and
continuous schedule ofreinforcement usually results ethical reasons, this may not be tolerable for some
in a higher overall density of reinforcement; this clients with severe behavior problems. In such in-
might have contributed to the treatment effects. stances, it may be necessary to rely on more de-
Thus, increasing the density of reinforcement for scriptive analyses conducted in the natural envi-
 FUNCTIONAL ANALYSIS 521

ronment to suggest possible functional relationships subsequent experience, this type of brief functional
(Bijou, Petersen, Harris, Allen, & Johnston, 1969). analysis at times could not be conducted simply
Potential functional relationships could then be because the client did not display any problem
subjected to a functional analysis in which an al- behavior during the evaluation.
ternative, replacement behavior being considered One question, then, is whether our clinical as-
for use in an intervention is designated as the de- sessment should be referred to as a functional anal-
pendent variable. ysis per se; perhaps it is better characterized as
A second limitation of the procedures used in simply being a direct or functional assessment
this investigation may be the use of a generic (Wacker & Steege, in press). We deliberately avoid-
"please" sign. Although the use of a generic sign ed the term functional assessment because of its
is practical and may be essential during initial train- current widespread use in education to refer to the
ing, long-term results of the use of such a sign are content of curricular or vocational programs (i.e.,
unknown, and consideration should be given to a functional curriculum). The term behavioral as-
implications for an overall communication program sessment might be adequately descriptive; however,
(Sigafoos, Doss, & Reichle, 1989). this term has become so heuristic that it refers to
We suggest that future investigators separately a variety of direct and indirect procedures. Al-
record prompted and unprompted mands and that though it can be argued that the term functional
their relationship to a decrease in problem behaviors analysis should only be used with experimental
receive further scrutiny. In this investigation, as well designs involving repeated measurements, we do
as in others (e.g., Wacker, Steege, Northup, Sasso not believe that such an experimental design is
et al., 1990), a substantial initial decrease in prob- explicitly required by the term. If only by default,
lem behavior occurred even when the mands were we retained the term functional analysis but, be-
fully prompted. However, independent manding cause of our use of limited data points, we refer to
may be essential for zero rates of occurrence or for our analysis as being a brief functional analysis.
long-term maintenance. Although the feasibility of conducting a func-
In most previous studies, the functional analysis tional analysis within a very short time span (90
procedures involved repeated assessment within min) has been suggested by both Cooper et al.
conditions to identify maintaining contingencies. (1990) and the present investigators, the length of
The results of our brief functional analysis in a observation conditions and the number of obser-
clinical setting may not always identify existing vations necessary to obtain convincing results may
maintaining contingencies because we often rely on be of some concern. The minute-by-minute analysis
only one data point per condition. This is, obvi- of Cooper et al. (1990) revealed that at least 10
ously, not ideal. However, at the very least, if the min per condition are required for determining re-
effects of a particular consequence can empirically liable effects. Future research in outclinic settings
reduce a problem behavior and reinforce an alter- appears justified to determine further the needed
native behavior, a beginning for treatment is sug- length of observation conditions. Just how many
gested. Alternatively, one or more follow-up ap- observations are necessary can, perhaps, never be
pointments could be requested in which our brief definitively determined, because the number and
version of functional analysis is repeated. Such fol- duration of observations needed to obtain a reliable
low-ups could further verify the initial results and picture of actual environmental events are simply
provide an opportunity for any necessary adjust- unknown (Johnston & Pennypacker, 1980); or,
ments to the initial recommendations as well as alternatively, as Bijou et al. (1969) stated, "It de-
document the initial success (or lack thereof) of the pends on the data" (p. 202). Baer, Wolf, and
initial intervention. Similarly, it may be fruitful to Risley (1987) suggested that "It might prove valu-
repeat periodically the brief functional analysis in able to the field to recall its original designs and
other settings. It should also be noted that, in our their logic-a good design is one that answers the
522 JOHN NORTHUP et al.

question convincingly, and needs to be constructed Hayes, C., Nelson, R., &Jarrett, R. (1987). The treatment
in reaction to the question and then tested through utility of assessment: A functional approach to evaluating
assessment quality. American Psychologist, 42, 963-
argument in that context, rather than imitated from 974.
a textbook.... Perhaps the important point is that Iwata, B., Dorsey, M., Slifer, K., Bauman, K., & Richman,
convincing designs should be more important than G. (1982). Toward a functional analysis of self-injury.
'proper' designs" (p. 319). We believe the design Analysis and Intervention in Developmental Disabil-
ities, 2, 3-20.
used in this investigation fulfills the intent of Baer Iwata, B., Pace, G., Kalsher, M., Cowdery, G., & Cataldo,
et al.'s suggestion. The rapid reversals obtained in M. (1990). Experimental analysis and extinction of
response to a series of changing conditions provide self-injurious escape behavior. Journal of Applied Be-
havior Analysis, 23, 11-27.
what we believe is both a convincing demonstration Johnston, J., & Pennypacker, H. (1980). Strategies and
of the effect of the changed contingencies and a tactics of human behavioral research. Hillsdale, NJ:
practical, cost-efficient application of functional Eribaum.
Kazdin, A. (1982). Single-case research designs: Methods
analysis procedures to an outpatient setting. for clinical and applied settings. New York: Oxford
University Press.
Lundervold, D., & Bourland, G. (1988). Quantitative
REFERENCES analysis of treatment of aggression, self-injury, and prop-
erty destruction. Behavior Modification, 4, 590-617.
Axelrod, S. (1987). Functional and structural analyses of Sigafoos, J., Doss, S., & Reichle, J. (1989). Developing
behavior: Approaches leading to reduced use of punish- mand and tact repertoires in persons with severe devel-
ment procedures? Research in Developmental Disabil- opmental disabilities using graphic symbols. Research in
ities, 8, 55-70. Developmental Disabilities, 10, 183-200.
Baer, D., Wolf, M., & Risley, T. (1987). Some still-current Steege, M., Wacker, D., Berg, W., Cigrand, K., & Cooper,
dimensions of applied behavior analysis. Journal of Ap- L. (1989). The use of behavioral assessment to prescribe
plied Behavior Analysis, 20, 313-327. and evaluate treatments for severely handicapped chil-
Bijou, S., Petersen, R., Harris, F., Allen, K., & Johnston, dren. Journal of Applied Behavior Analysis, 22, 23-
M. (1969). Methodology for experimental studies of 33.
young children in natural settings..Psychological Record, Wacker, D., & Steege, M. (in press). Providing outclinic
19, 177-210. services: Evaluating treatment and social validity. In S.
Carr, E. (1977). The motivation of self-injurious behavior: Axelrod & R. Van Houten (Eds.), Effective behavioral
A review of some hypotheses. Psychological Bulletin, treatment: Issues and implementation.
84, 800-816. Wacker, D., Steege, M., Northup, J., Reimers, T., Berg,
Carr, E. (1988). Functional equivalence as a mechanism W., & Sasso, G. (1990). Use of functional analysis
of response generalization. In R. Homer, R. Koegel, & and acceptability measures to assess and treat severe be-
G. Dunlap (Eds.), Generalization and maintenance: havior problems: An outpatient clinic model. In A. Repp
Life-style changes in applied settings (pp. 194-219). & N. Singh (Eds.), Aversive and non-aversive treat-
Baltimore: Paul H. Brookes. ment: The great debate in developmental disabilities
Carr, E., & Durand, V. M. (1985). Reducing behavior (pp. 349-359). Sycamore, IL: Sycamore.
problems through functional communication training. Wacker, D., Steege, M., Northup, J., Sasso, G., Berg, W.,
Journal of Applied Behavior Analysis, 18, 111-126. Reimers, T., Cooper, L., Cigrand, K., & Donn, L. (1990).
Cooper, L., Wacker, D., Sasso, G., Reimers, T., & Donn, A component analysis of functional communication train-
L. (1990). Using parents as therapists to assess the ing across three topographies of severe behavior problems.
appropriate behavior of their children: Application to a Journal of Applied Behavior Analysis, 23, 417-429.
tertiary diagnostic clinic. Journal of Applied Behavior
Analysis, 23, 285-296.
Doss, S., & Reichle,J. (1989). Establishing communicative
alternatives to the emission of socially motivated excess
behavior: A review. Journal of the Association for Per- Received February 2, 1990
sons with Severe Handicaps, 14, 101-112. Initial editorial decision July 26, 1990
Ferster, C. B. (1961). Positive reinforcement and behav- Revision received September 3, 1990
ioral deficits of autistic children. Child Development, Final acceptance April 16, 1991
32, 437-456. Action Editor, Terry J. Page