Denham Workman Et Al 2000

See
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/12541331
Prediction of behavior problems from early to

middle childhood: The role of parental
socialization and emotion expression
ARTICLE in DEVELOPMENT AND PSYCHOPATHOLOGY · FEBRUARY 2000

Impact Factor: 4.89 · DOI: 10.1017/S0954579400001024 · Source: PubMed
CITATIONS READS
235 695
6 AUTHORS, INCLUDING:
Susanne Ayers Denham Pamela Cole

George Mason University Pennsylvania State University
107 PUBLICATIONS 4,928 CITATIONS 64 PUBLICATIONS 4,250 CITATIONS
SEE PROFILE SEE PROFILE
Available from: Susanne Ayers Denham

Retrieved on: 21 March 2016
Development and Psychopathology, 12 (2000), 23–45
Copyright  2000 Cambridge University Press
Printed in the United States of America
Prediction of externalizing behavior

problems from early to middle childhood:
The role of parental socialization and
emotion expression
SUSANNE A. DENHAM,a ELIZABETH WORKMAN,a PAMELA M. COLE,b

CAROL WEISSBROD,c KIMBERLY T. KENDZIORA,d AND
CAROLYN ZAHN–WAXLERd
a
George Mason University; bPennsylvania State University; cAmerican University; and
d
National Institutes of Mental Health
Abstract
Parental emotions and behaviors that contribute to continuity and change in preschool children’s externalizing
problems were examined. Mothers and fathers were observed interacting with their children, and child-rearing styles
were reported. Teachers, mothers, and children reported children’s antisocial, oppositional behavior. Externalizing
problems showed strong continuity 2 and 4 years later. Proactive parenting (i.e., supportive presence, clear
instruction, and limit setting) predicted fewer behavior problems over time, after controlling for initial problems; the
converse was true for parental anger. In contrast, the hypothesized ameliorative contribution of parents’ positive
emotion was not found. Parental contributions were most influential for children whose initial problems were in the
clinical range. In particular, parental anger predicted continuation of problems over time. Paternal, as well as
maternal, influences were identified. Examination of parental emotions and inclusion of fathers is important to
research and intervention with young antisocial children.
The goal of the present study was to examine Campbell, 1987, 1994, 1995; Campbell,
socialization experiences that may influence Breaux, Ewing, & Szumowski, 1984; Camp-
the developmental course of behavior prob- bell & Ewing, 1990; Campbell, Ewing,
lems during the transition from preschool to Breaux, & Szumowski, 1986; Campbell,
school. Behavior problems become stable Pierce, March, Ewing, & Szumowski, 1994;
over this period for many children (Bates, Egeland, Kalkoske, Gottesman, & Erickson,
Bayles, Bennett, Ridge, & Brown, 1991; 1990; Fischer, Rolf, Hasazi, & Cummings,
1984; McConaghy, Stanger, & Achenbach,
1992; Pianta & Caldwell, 1990; Wehby,
This study is based on a larger NIMH Intramural Re- Dodge, & Valente, 1993; Zahn–Waxler, Ian-
search Program longitudinal investigation of children notti, Cummings, & Denham, 1990). Early
identified during their preschool years as at risk for be- onset of aggression, in particular, is a precur-
havior disorders. Our thanks go to the families who gave
so much of their time to participate and to Barbara Usher sor of life-course, persistent antisocial behav-
for her invaluable contributions as project manager. ior (Moffitt, 1993). Given the chronicity and
Dr. Susanne Denham, Department of Psychology, severity of many children’s behavior prob-
Mailstop 3F5, George Mason University, 4400 University lems and the costs to the individual, the fam-
Drive, Fairfax VA 22030–4444, E-mail: sdenham@osf1. ily, and society, it is important to identify
gmu.edu, or to Dr. Carolyn Zahn–Waxler, Section on De-
velopmental Psychopathology, National Institute of Men- early conditions that contribute to ameliora-
tal Health, Bldg. 15-K, 9000 Rockville Pike, Bethesda, tion, as well as continuation, of antisocial be-
MD 20892. havior patterns (Haapasalo & Tremblay,
23
24 S. A. Denham et al.
1994; Kendziora & O’Leary, 1993). As interchange of the hour, a mother or father
Campbell et al. (1994) have stated, “Under- may understandably feel frustrated, impatient,
standing the early manifestations as well as furious, or afraid of their inability to teach the
factors influencing the onset and develop- child an important lesson. The parent who can
mental course of behavior problems in young manage these feelings while responding in an
children is among the major challenges in the attuned way to the particular needs and chal-
growing field of developmental psychopathol- lenges of a difficult child is more likely to
ogy” (p. 836). generate creative, flexible solutions in diffi-
Externalizing behaviors—that is, the broad cult situations and teach the child how to
band of symptoms that are distressing to oth- manage his or her own emotions and behav-
ers (Achenbach, 1990)—represent the most ior. Parents who manage their own frustra-
common form of mental health problems in tions when interacting with their children also
children. Externalizing symptoms include may be more able to use negotiation and ra-
hostile defiance, destructive behavior, temper tional guidance in dealing with child misbe-
tantrums, impulsivity, and poor frustration havior, to allow a measure of autonomy, and
tolerance. Notably, many of these symptoms to acknowledge the child’s feelings.
involve emotions, suggesting that difficulties Hence, emotions may be central to the en-
in regulating emotions may be an important hancement of proactive parenting, which is
aspect of early behavior problems (Cole, Mi- associated both with fewer behavior problems
chel, & Teti, 1994; Cole & Zahn–Waxler, (Pettit & Bates, 1989) and a decrease in these
1992). Problems in managing emotion jeopar- problems over time (Zahn–Waxler et al.,
dize children’s successful interpersonal func- 1990). In the latter study, maternal proactive
tioning (Thompson, 1994). parenting practices with aggressive, disruptive
Difficult temperament has also been identi- 2-year-olds predicted fewer problems by kin-
fied as an antecedent of these early behavior dergarten. Proactive approaches included an-
problems (Bates et al., 1991). Low frustration ticipating the child’s point of view; exerting
tolerance, poor impulse control, and observed modulated, respectful control; and providing
high activity levels in 3- and 4-year-olds pre- structure and organization for the child’s play
dict externalizing problems in 5-year-olds activities with peers. Moreover, Russell and
(Zahn–Waxler, Schmitz, Fulker, Robinson, & Russell (1996) have suggested that positive
Emde, 1996). Because only a subset of chil- child behavior is associated with positive
dren with difficult temperament go on to de- emotional states arising from such warm, af-
velop chronic behavior problems, it becomes fectionate, involved parenting (see also Gard-
important to consider, as well, the role of so- ner, 1987).
cialization. In this study we considered both Emotions, particularly anger, are also key
emotional and behavioral aspects of parenting elements in more destructive parenting pat-
as predictors of externalizing behavior prob- terns. Anger can motivate people to overcome
lems—problems commonly characterized by an obstacle. But when that obstacle is a non-
strong negative emotions and difficulty regu- compliant, defiant child, these emotions can
lating emotion. develop to the point that it becomes difficult
The emotional nature of early behavior for parents to generate proactive disciplinary
problems first suggests the need to consider tactics and sustain positive, balanced ex-
the emotional nature of parenting. Feelings changes with the child (Crockenberg & Lit-
are ever present in parents’ child-rearing in- man, 1990; Dumas & LaFreniere, 1993).
teractions; they organize either effective, re- Poorly modulated negative emotion is a con-
sponsive parenting or the converse (Dix, comitant of reactive parenting, often accom-
1991). For example, a parent may feel panying excessive prohibitions, yelling, and
strongly that a preschooler should begin com- physical discipline. Taken together, these pa-
plying with demands for simple, familiar rental emotions and behaviors exacerbate
tasks. If this child has just thrown a toy rather children’s difficulties (Zahn–Waxler et al.,
than put it away, during their 12th negative 1990).
Prediction of behavior problems 25
The presence of parents’ anger and the ab- rental emotion, as well as behaviors and child-
sence of their positive emotions during inter- rearing styles, in the prediction of continuity
action could contribute to behavior problems and change in young children’s behavior
and their stability, whereas presence of plea- problems. Just as high negative affect and low
sure and little anger should be associated with positive affect may function as risk factors,
more positive outcomes. The coercive cycle, the converse pattern of parental emotion (low
in which parent and child each try to termi- negative affect, high positive affect) may pre-
nate each other’s angry, aversive behavior dict improvement in the child’s behavior
with their own angry, aversive behavior, is a problems over time.
well-documented correlate of conduct disor- Generalizations from the extant literature
der (Patterson, Reid, & Dishion, 1992). Moth- about parental influences are based mainly on
ers of externalizing boys are more negative, studies of parents and children where assess-
impatient, and controlling during everyday ments are concurrent. Hence, they do not pro-
child-rearing tasks than comparison mothers vide a good opportunity to examine the role
(e.g., Campbell, March, Pierce, Ewing, & of parenting in the prediction of individual
Szumowski, 1991). Although maternal anger differences in children’s later behavior prob-
is presumed to be a significant part of these lems. It is not possible to disentangle parent
difficult encounters, research has emphasized and child effects in such designs. In this study
the escalation of aversive behaviors, not the we sought to remedy these dual problems by
emotional tone of these interactions. How- focusing on observed patterns of parental
ever, a meta-analytic review of correlates of emotions and behaviors as predictors of child
conduct disorder and delinquency showed that outcomes across time, after controlling for
diminishing involvement and positive af- continuity of behavior problems.
fectional ties were the most powerful predict- We examined both parental emotion and
ors of concurrent and later conduct problems. behavior, from both observational and parent-
Ineffective, harsh parenting was secondary report perspectives. Emotional factors were
(Loeber & Stouthamer–Loeber, 1986; cf. assessed in (a) observed parent–child interac-
Haapasalo & Tremblay, 1994; Pettit, Bates, & tions and (b) the parents’ reports of their
Dodge, 1993). levels of hostility. Behavioral factors were as-
The emotional tone of parenting is not sessed in (a) observed parent–child interac-
completely captured in the immediate context tions and (b) maternal reports of child-rearing
of parent–child interaction. Other, more styles. This second set of factors represented
global contextual factors are also vital to an a broader context of parental socialization ap-
understanding of the development, mainte- proaches that varied in the degree to which
nance, and amelioration of behavior prob- they were proactive and child-oriented versus
lems. Stable tendencies in emotional reactions reactive and harsh. The ways in which pre-
to family events are also an aspect of the school children and their parents interacted
child’s emotional environment. When parents were examined across a series of challenging
enjoy and support each other and their chil- and nonchallenging situations, in order to as-
dren, including the coparenting relationship, sure representativeness of assessments. Chal-
there is usually a more optimal atmosphere in lenging contexts, such as playing a compli-
the family (Kendziora & O’Leary, 1993; cated game that is difficult for a child to win,
Miller, Cowan, Cowan, Hetherington, & allowed examination of interactions in situa-
Clingempeel, 1993). Parental hostility reduces tions that were more likely to involve frustra-
the likelihood that this will occur. Hostility, tion. Nonchallenging situations, such as hav-
even when not directed toward the child in ing a snack and time that is not structured by
a parenting situation, can be felt by children, task demands, allowed examination of inter-
sensitizing them and affecting their emotional actions even when externally induced emo-
security (Cummings & Davies, 1994; Miller tional demands were minimal.
et al., 1993). Taken together, these findings Although previous research has suggested
highlight the need to examine the role of pa- that fathers also play an important role in the
emotional nature of parenting, most studies of In sum, we sought in this study to extend
parent–child interaction focus on mothers and previous research on the etiology of children’s
their children. In one study, mothers’ parent- behavior problems by (a) examining the role
ing warmth (i.e., self-reported supportiveness, of parental emotions (happiness and anger)
observed positive emotion and responsiveness and parenting behaviors (proactive and reac-
during teaching and unstructured play tasks) tive or restrictive) in predicting child out-
was negatively related to children’s behavior comes, based on multiple assessment con-
problems. Fathers’ parenting warmth was texts; (2) including fathers, in addition to
similarly negatively related to behavior prob- mothers, to examine their potentially unique
lems in their children. However, fathers’ en- contributions to adjustment and psychopathol-
dorsement of authoritarian control, and their ogy; (3) using a sample of preschool children
negative limits and demands, also were posi- already known to be on a continuum of risk
tively related to behavior problems in 3-year- for the development of disruptive behavior
olds (Miller et al., 1993). Such research sug- disorders; (4) implementing a longitudinal re-
gests the utility of exploring whether and how search design so that parental emotions and
the positive and negative aspects of both behaviors identified early could be examined
mothers’ and fathers’ emotions and behaviors as predictors of later problems, over and
are associated with continuity and change in above early child problems; (5) searching for
behavior problems, particularly during the parenting predictors specific to ongoing exter-
transition from preschool to early and middle nalizing problems in children initially at high-
childhood. est risk; and (6) incorporating analytic strate-
Finally, research on prediction of antisocial gies that included person-centered as well as
behavior patterns over time has traditionally variable oriented approaches.
emphasized explanation of developmental It was hypothesized that negatively toned
outcomes for the group as a whole, neglecting parental emotions (anger, hostility) and be-
the individual in the process (Richters, 1997). haviors (authoritarian, nonsupportive acts and
Bergman and Magnusson (1997) have pointed parenting styles) would predict a continuation
out that “a standard variable-oriented ap- of children’s externalizing problems. Emo-
proach, focusing on the variable as the main tionally negative parent–child interactions,
theoretical and analytical unit, has limita- family climates of pervasive parental hostil-
tions” (p. 291) in identifying meaningful co- ity, and lack of support and limit setting may
herence and structure in individual growth all exacerbate children’s constitutional predis-
and individual differences in developmental positions. Because the study included young
processes. A person-centered approach allows children who varied in early problems, it was
for the identification of different develop- possible to test the interactions of parental
mental trajectories for different antisocial factors and child risk. Young children whose
children (i.e., improvement, worsening, no problems appear early in development, and
change). It is then possible to examine factors are severe in nature, may be the most ad-
that differentiate developmental outcomes for versely affected by negatively toned parent-
these subtypes of children. Because there is ing. It was also expected that positively toned
little evidence of “growth” of externalizing parental emotions (happiness) and behaviors
problems from middle childhood to early ado- (positive support and structure) would miti-
lescence (Cairns & Cairns, 1994; Haapa- gate continuing difficulties and hence predict
salo & Tremblay, 1994; Loeber & Stou- improvement of behavior problems over time.
thamer–Loeber, 1998; cf. Anderson, 1993), Here, too, we examined whether positive
the more salient distinction may be between emotion and proactive behavior had the stron-
children who do, or do not, continue to have gest constructive influence on the children
problems over time. In the present study we with the most serious problems. It was not the
examine early parenting behaviors and emo- purpose of this research to compare the rela-
tions that may explain these different develop- tive impact of parental emotions and parent-
mental pathways. ing behaviors on child outcomes. Rather, it
was to begin to incorporate explicitly the role ipating fathers was utilized. Of the 69 chil-
of emotion in the conceptualization and mea- dren, 38 were boys; the mean age of the group
surement of parental processes relevant to de- as a whole was 55 months (range = 48–61
velopmental outcomes in children with early months). Twenty-nine percent were 1–2 SD
behavior problems. above the norm on behavior problems and
34.8% were 2 or more SD above the norm,
based on reports from mother (Child Behavior
Method
Checklist, CBCL/4-16; Achenbach, 1991) or
teacher (Teacher Report Form [TRF], the
Participants
teacher version of the CBCL; Achenbach &
Recruitment and screening procedures. Parti- Edelbrock, 1986). Hence, 36.2% of the sam-
cipants were recruited from a major urban ple was in neither of these risk groups. The
community through newspaper announce- homogeneous SES distribution, using the Hol-
ments and flyers to preschools and day cares lingshead 4-factor formula, indicated the ma-
in the metropolitan area. In order to maximize jority of cases (95.7%) were middle to upper
variability among the children, some an- class with a mean Hollingshead SES of 56.7.
nouncements targeted hard-to-manage (non- The ethnic composition was 95.7% Cauca-
compliant, aggressive, overactive) boys and sian.
girls, and others simply recruited any children Child outcomes were assessed at two later
within the age range. This procedure made it time points. The average age of the 69 return-
possible to identify children who ranged from ing participants at Time 2 was 7.0 years, and
nonproblem to subclinical to clinical levels. the child had completed or was completing
Interested parents who telephoned for infor- first grade (the average time span between
mation received a description of a project dur- Time 1 and Time 2 was 2.4 years, σ = 0.4).2
ing the transition from preschool to school The average age of the 60 returning partici-
age (Cole & Zahn–Waxler, 1988). Children pants (34 boys) at Time 3 was 9.7 years; al-
with any indication of any difficulties beyond most all the children were in third or fourth
the scope of the study (e.g., sensorimotor or grade. Nonreturning participants at Time 3
orthopedic handicaps, autism, mental retarda- did not differ significantly from returning par-
tion) were excluded from participation. Ar- ticipants on any of the earlier criterion vari-
rangements were made for a contact with the ables.
other parent (where appropriate) and for a
first visit. Overview of procedures
Sample. Forty-six boys and 34 girls and their Participation in the larger longitudinal study
mothers and fathers were studied. They came involved four visits to the laboratory at Time
from the original sample of 82 children who 1. During the first three visits, different as-
were on a continuum of risk for development pects of physical, cognitive, social–emotional
of disruptive behavior disorders (Cole, Zahn– development, and mother–child interaction
Waxler, & Smith, 1994).1 Nine of the children were assessed. The fourth visit to the labora-
were from single-parent (mother) families. tory included fathers as well, and families
Fathers in two other families did not partici- were observed interacting in an apartment set-
pate. These 11 children did not differ signifi- ting. The present work focuses on family in-
cantly from those in intact families on any teraction measures obtained during the fourth
Time 1 predictor or criterion variable.
Only data from the 69 families with partic- 2. Several scales loaded on both factors across the nine
principal components analyses of interaction variables
(six each for situations in which mother took part, and
1. Videotapes for two families inadvertently were not three for fathers). However, the pattern for parental
coded. By the time this became known, coding had task orientation and support was clear enough across
been completed and coders, no longer in the area, were the analyses to warrant the creation of the aggregates
unavailable to code the two tapes. described.
visit, as well as maternal report measures of about emotions. In one, mother and child
child-rearing styles and parental reports of were asked to tell the stories implied in two
hostility at that time. wordless picture books (10 min). The two
At Times 1, 2, and 3 mothers completed books were Frog Goes to Dinner (about dis-
the CBCL. The TRF also was mailed to the ruptive behavior in a fancy restaurant; Mayer,
teacher of every enrolled child at Time 1, 1974) and One Frog Too Many (about sibling
once the child had been in preschool or day rivalry; Mayer & Mayer, 1975). The second
care for at least 3 months. Teachers also com- paradigm consisted of an emotion-reminis-
pleted the TRF at Times 2 and 3. Finally, chil- cence task (15 min). The mother asked the
dren completed the Youth Self Report version child to recall and describe a time where she
of the CBCL at Time 3 (YSR; see Thurber & or he had experienced different emotions.
Hollingsworth, 1992). In this study, the exter- Two coding systems were used to observe
nalizing problem scales for each of these mea- aspects of parent–child interaction in these
sures were used to assess severity of these be- observational contexts. One consisted of rat-
havioral and emotional problems. ings of maternal and paternal behavior pat-
Mothers’ and teachers’ reports of external- terns that included supportive presence, limit
izing were correlated significantly, on aver- setting, allowance of autonomy, negative af-
age, across all three time periods, r = .30, p < fect, quality of instructions, and confidence.
.01 (one-tailed). Children’s YSR externalizing The other coding system consisted of ratings
scores were also significantly associated with of maternal and paternal expressions of anger
mothers’ and teachers’ reports at Time 3, on and happiness. Coders for both systems were
average, r = .26, p < .05 (one-tailed). Both uninformed as to the study’s hypotheses and
mothers’ and teachers’ reports were signifi- subject identities.
cantly autocorrelated across time, despite the The variables for the first system were
fact that teachers differed, rmothers = .74, p < coded on 7-point scales, depicted in Egeland,
.001, and rteachers = .41, p < .001 (one-tailed ps). Sroufe, and colleagues’ original system (Den-
ham, Renwick, & Holt, 1991; Egeland,
Sroufe, & Erickson, 1983) based on detailed
Measures: Parenting
scale definitions and rank-ordered descrip-
Observed interaction. Six paradigms, some tions of behavioral episodes fitting each scale
involving challenge and competition, were point. Supportive presence was marked by a
used to observe family interactions in dyadic parent’s positive regard and emotional sup-
and triadic contexts. All were chosen to create port of their child, with allowance of auton-
naturalistic contexts in which both positive omy evidenced by the parent’s recognition of
and negative interactions could easily take the validity of the child’s individual motives
place. Mother, father, and child played two and perspective. Quality of instruction was
competitive games together. One was “Block- scored high when parental directives were
head” (10 min), where individuals take turns clear, and the parent made sure that the child
building a tower with odd shaped blocks. understood task objectives. Parents rated high
When the tower falls, the person whose block in limit setting established and reinforced
made it fall is the Blockhead. The other game their clear expectations for their child’s be-
was “Double Trouble” (20 min), where each havior, and parents rated high in confidence
player tries to get to the finish line first, mov- demonstrated their belief that their interaction
ing two pieces by pushing two dice rollers. with the child would be successful. Negative
Mother, father, and child also had a snack and affect was rated higher when a parent showed
spent free time together (15 min). Interactions anger or discounted or rejected their child. In-
of mother and child were observed in three terrater reliabilities, based on 21 cases, were
situations. In a game called “Make-a-Mouse” assessed by Finn’s r, a statistic for ordinal rat-
(10 min), the players competed to complete ing scales which controls for random agree-
the highest number of small puzzles of mice. ment (Whitehurst, 1984). Finn’s rs were in
The remaining two tasks involved discussions the very good to excellent range: they ranged
from .68 to .99, with a mean of .81 for ratings for mothers’ and fathers’ happiness were .72
of mothers and .89 for ratings of fathers. and .68, respectively; κs for anger were .47
Moreover, 3- to 6-point ranges on the 7-point for father and .60 for mother. Percentage
scales were noted across paradigms, suggest- agreement across all codes was relatively
ing that the contexts chosen varied in terms high, with infrequency of codes largely con-
of challenge. tributing to the lower κ for father anger.
Ratings of maternal and paternal interac- Emotion ratings were then aggregated
tions were factor analyzed separately for each across the six maternal and three paternal ob-
setting (six for mother, three for father), using servational contexts. Internal consistency re-
varimax rotation. Scree plots of eigenvalues liability supported the aggregation of data; re-
indicated the emergence of two clear parental liability for both mother and father happy was
factors, consistently across settings. For moth- .81. For anger, κ was .71 for mother and .66
ers, the two factors accounted for approxi- for father. Data screening procedures (Kolmo-
mately 81% of variance. For fathers, two fac- gorov–Smirnov Goodness of Fit Tests [K-S
tors similar to mothers’, and also accounting test]) indicated that the distributions for par-
for approximately 81% of variance, emerged. ents’ happy emotions were not significantly
Factor loadings for ratings were used to mark different from normal. Distributions for ma-
each factor when they met a .45 criterion.2 ternal and paternal anger had modal values of
Aggregates for each maternal and paternal zero, with significant positive skewness and
factor equaled the mean of scores for compo- kurtosis for paternal anger. Hence, logarith-
nent variables loading on each factor, across mic transformations were performed after
all settings. Thus, both maternal and paternal adding one to each of these emotion scores.
support were marked by supportive presence, Skewness and kurtosis were markedly re-
allowance of autonomy, and nonhostility duced for log-transformed data for maternal
(hostility was weighted negatively; κs = .94 and paternal anger. Happiness and anger rat-
and .92 for mother and father, respectively). ings used in the present analyses represent av-
Both maternal and paternal task orientation erages across all intervals observed.
were marked by limit setting, quality of in-
struction, and confidence in the setting (κs = Parental reports of behavior and emotion.
.92 and .91 for mother and father, respec- Two maternal self-report measures were used
tively). to assess aspects of parenting, both parenting
The definitions of the constructs for the style and negative affective environment. The
second coding system, used to rate mothers’ Child Rearing Practices Report (CRPR) was
and fathers’ expressions of anger and happi- used to create indices of more general parent-
ness, are as follows: (a) anger: sullen, petu- ing attitudes, values, goals, and behaviors
lant; intense outbursts with raised voice; glar- (Block, 1981). Mothers completed this mea-
ing; verbal expressions and vocalizations of sure at home, by sorting 91 cards, each con-
anger, sarcasm, irritation, exasperation; and taining a statement about parenting practices,
(b) happiness: positive disposition; joyful- into seven equal piles order from most to least
ness; enthusiasm; cheerfulness; smiles and endorsed. The procedure’s forced-choice, ip-
laughter. These variables were coded on a 7- sative nature reduces the possibility of obtain-
point scale using detailed definitions and ing socially desirable responses. There is
rank-ordered descriptions of affective epi- considerable evidence for the test–retest
sodes fitting the scale points. Ratings were reliability and validity of the CRPR (Block,
done after each 5-min segment for each task, 1981; Roberts, Block, & Block, 1984; Trick-
then summed over segments within para- ett & Susman, 1988). In this study, the CRPR
digms. Consonant with our emphasis on anger measure was scored using the system devel-
and pleasure, for this study we focused on rat- oped by Rickel and Biasatti (1982). Two
ings of happiness and anger for each parent. scores, for restrictiveness and nurturance,
Kappa coefficients were used to compute were obtained. For the 22-item restrictiveness
rater reliabilities, based on 18 cases. The κs scale, α was .54; for the 18-item nurturance
scale, α was .77. The restrictiveness scale in- levels of supportive presence and task orienta-
cludes a predominance of items referring to tion, and they reported similar levels of hostil-
reactive modes of parenting or negative par- ity. Given the larger number of situations
enting patterns (e.g., “I teach my child that in coded for mothers’ emotions, maternal and
one way or another, punishment will find him paternal emotions were also roughly equiva-
when he is bad”; “I control my child by warn- lent.
ing him about the bad things that can happen Next, bivariate correlations between pre-
to him”; “I let my child know how ashamed dictors and criteria are presented. Multiple re-
and disappointed I am when he misbehaves”). gression equations are then used to examine
In contrast, the nurturance scale includes a the contribution of parenting during the chil-
number of items referring to proactive or pos- dren’s preschool years to later behavior prob-
itive modes of parenting (e.g., “I talk it over lems, controlling for initial level of child
and reason with my child when he misbe- problems. These are followed by analyses of
haves”; “I believe in praising a child when he moderating influences (i.e., whether certain
is good and think it gets better results than aspects of parenting are more important to the
punishing him when he is bad”; “I trust my prediction of later problem behaviors depend-
child to behave as he should, even when I am ing upon the initial severity of child prob-
not with him”; “I usually take into account lems). Last, subgroups of these children ini-
my child’s preferences when making plans for tially at risk, whose behavior problems show
the family”). differing patterns of change over time, are ex-
Both parents completed the Buss–Durkee amined. Do different aspects of parenting pre-
Hostility Inventory (BDHI; Buss & Durkee, dict membership in these groups?
1957), which was used as an index of the fam-
ily’s affective environment. The BDHI in-
Bivariate correlations among
cludes 75 items, such as “I often find myself
parenting variables
disagreeing with people,” “When I look back
on what’s happened to me, I can’t help feeling Correlations among maternal and paternal
mildly resentful,” “When I get mad, I say predictors can be seen in Table 2. The mean
nasty things,” and “When I am mad, I some- magnitude of correlations among parental pre-
times slam doors,” rated as true (1) or false dictors equaled .28 for both mother and fa-
(0). The BDHI subscales include a variety of ther. The most substantial correlations were
types of hostility (e.g., assault, indirect hostil- between support and task orientation for both
ity, irritability, and negativism). Total scores maternal and paternal measures, and between
for mothers’ and fathers’ hostility, summing CRPR restrictiveness and nurturance (i.e., rs
across these eight scales, are utilized here (αs in the .50s and .60s). Because of these sub-
= .74 and .76 for mothers and fathers, respec- stantial intercorrelations, two aggregates each
tively). were created: (a) Proactive Parenting Behav-
iors (Observed), for mother and father, was
the sum of standard scores for supportive
Results
presence and task orientation (αmaternal = .82;
αpaternal = .79); and (b) Proactive Parenting
Analytic plan
Styles (Reported), for mother only, summing
First, descriptive data for all predictors and standard scores for nurturance and restrictive-
criteria are presented in Table 1, for children’s ness, with restrictiveness weighted negatively
problem behaviors over time, parents’ ob- (α = .69).
served behavior and emotions, and parents’
reports of child-rearing styles and hostility.
Bivariate correlations between parenting and
The range of behavior problems included in
child problems
the study can be clearly seen in the descriptive
data for CBCL, TRF, and YSR externalizing Observed parenting predictors. Correlations
scales. Mothers and fathers showed similar of parental predictors with child outcomes are
Table 1. Description of measures utilized in the study
Standard
Mean Deviation Range Possible Range
Time 1 Externalizing problems

CBCL Time 1
Raw scores 16.20 8.48 0–38 0–66
T scores 59.20 10.27 30–77
TRF Time 1
Raw scores 11.00 10.84 0–41 0–68
T scores 55.66 10.43 39–80
Criteria: Externalizing problems
CBCL Time 2
Raw scores 13.26 9.59 0–42 0–66
T scores 55.15 11.28 30–80
CBCL Time 3
Raw scores 13.19 8.75 0–38 0–66
T scores 54.98 11.29 32–77
TRF Time 2
Raw scores 8.76 10.50 0–51 0–68
T scores 53.36 10.04 39–86
TRF Time 3
Raw scores 9.67 9.38 0–43 0–68
T scores 54.63 9.56 39–77
YSR Time 3
Raw scores 8.46 5.71 0–25 0–60
T scores 45.49 10.39 25–66
Time 1 and 2 parental predictors: Observed
Maternal supportive presence 2.78 0.84 0.39–4.00 −1.66–4.33a
Maternal task orientation 4.90 0.92 2.62–6.39 1–7
Maternal happiness 16.67 4.19 5.17–24.50 0–96
Maternal angerb 3.24 2.20 0.00–14.50 0–96
Paternal supportive presence 2.72 0.98 −0.67–4.33 −1.66–4.33a
Paternal task orientation 4.77 1.08 1.11–6.78 1–7
Paternal happiness 8.82 2.35 1.50–13.92 0–45
Paternal angerb 1.90 1.74 0.00–7.83 0–45
Time 1 and 2 predictors: Parent reports
Maternal hostility 23.81 9.55 5–48 0–75
Maternal restrictiveness 59.68 6.86 47–80 22–154
Maternal nurturance 106.33 8.66 81–119 18–126
Paternal hostility 22.75 8.73 7–55 0–75
a
Negative values possible because hostility ratings weighted negatively.
b
Before transformation.
shown in Table 3. Mothers’ observed proac- nalizing problems at Times 2 and 3, approxi-
tive parenting was negatively related to all mately 2 and 4 years later.
outcome measures except the YSR at Time 3. Fathers’ observed proactive parenting was
Moreover, maternal anger was positively re- associated with mothers’ reports of fewer ex-
lated to all outcome measures. Maternal hap- ternalizing problems at Time 3. Father’s anger
piness was modestly negatively related to was associated with teachers’ reports of exter-
teacher and child reports of externalizing at nalizing problems at Time 2, but paternal hap-
Time 3. Thus, mothers rated as more proac- piness was unrelated to child outcomes. These
tive and less emotionally negative at Time 1 significant findings parallel those found for
had children whom they and the children’s observational maternal predictors. In sum,
teachers rated as demonstrating fewer exter- several observed aspects of parenting pre-
Table 2. Correlations among parental predictors
Maternal Predictors 2 3 4 5 6 7
1. Maternal supportive presence .69*** −.40*** .43*** −.27* .18 −.43***

2. Maternal task orientation — −.14 .28* −.09 .20† −.24*
3. Restrictiveness — −.52*** .15 −.24* .10
4. Nurturance — −.39*** .06 −.30*
5. BDHI hostility — .19 .19
6. Happiness — .10
7. Anger —
Paternal Predictors 2 3 4 5
1. Paternal supportive presence .66*** −.26* .24* −.49***

2. Paternal task orientation — −.10 .29* −.30*
3. BDHI hostility — .06 .22†
4. Happiness — −.10
5. Anger —
Note: Mean rmaternal = .28, p < .05. Mean rpaternal = .28, p < .05. With the exception of hostility, all maternal
and paternal predictors were significantly correlated: r = .60 for supportive presence, r = .44 for task orien-
tation, r = .52 for happiness, and r = .50 for anger (all ps < .001).
†p < .10. *p < .05. **p < .01. ***p < .001, two-tailed.
Table 3. Zero-order correlations between maternal predictors and

externalizing criteria
Criteria, Time 2 Criteria, Time 3
Predictors, Time 1 CBCL TRF CBCL TRF YSR
Observed
Maternal proactive −.39* −.26* −.46*** −.29* −.18
Paternal proactive −.15 −.16 −.28* −.08 −.11
Maternal happiness −.08 .06 −.06 −.23† −.24*
Maternal anger .27* .47*** .27* .34** .24*
Paternal happiness .11 .13 .10 −.12 −.06
Paternal anger .06 .39*** .04 .04 .01
Parent Report
Maternal proactive attitudes −.54*** −.32* −.42*** −.23† −.34**
Maternal BDHI .31** .22† .27* −.04 .13
Paternal BDHI .12 .16 .06 .14 .36***
Note: Reporters for CBCL were mothers, TRF were teachers, and YSR were the children them-
selves.
†p < .10. *p < .05. **p < .01. ***p < .001, two-tailed.
dicted behavior problems at both time periods hibiting more externalizing problems. More-
for all informants. over, mothers’ proactive parenting styles were
negatively related to all outcome measures.
Parent-report predictors. Relations among Again, mothers who reported they were less
parental hostility and maternal child-rearing hostile and more proactive at Time 1 had chil-
styles and child outcomes also are shown in dren demonstrating fewer externalizing prob-
Table 3. Maternal hostility was significantly lems at Times 2 and 3, approximately 2 and
correlated with mothers’ reports of externaliz- 4 years later.3
ing problems at Times 2 and 3. Fathers who
had reported more hostility at Time 1 had 3. Differences between parenting–behavior problem cor-
children who saw themselves at Time 3 as ex- relation coefficients for boys and girls were tested us-
Multiple regressions predicting ing could be evaluated. When entered as a

behavior problems block, contributions are adjusted for intercor-
relation among predictors. If there is inter-
Analytic strategy. Multiple regression was
correlation among predictors, this method
used to predict children’s externalizing behav-
permits identification of variables whose con-
ior problems, as reported by their mothers and
tributions are suppressed by their relation to
teachers at Time 2 and by their mothers,
other predictors. The significance of semipar-
teachers, and the children at Time 3. On the
tial correlations for each variable, even more
first step of each regression equation, score(s)
than the significance of the block as a whole,
for the outcome variable at each earlier as-
was of interest.
sessment point(s) was or were entered, as re-
For each criterion at Times 2 and 3—
ported by the mother, the child’s previous
mothers’, teachers’, and children’s reports of
teacher, or for predicting children’s self re-
externalizing problems based on the CBCL,
ports, both mother and teachers’ reports. The
TRF, and YSR—the significance of the total
logic behind entering these variables first was
equation, the amount and significance of
that it allowed examination of the stability of
change in R2, and amount and significance of
behavior problems, and also created a very
each variable’s unique contribution to R2 (i.e.,
conservative test of the unique role of parent-
sr2) is given.4
ing from an earlier period. Any significant
contribution of parenting would be seen over
and above the continuity in behavioral prob- Predicting Time 2 behavior problems. Regres-
lems. sion results predicting Time 2 CBCL criteria
Time 1 parental variables were entered on can be seen in Table 4. Predictors of Time 2
the second step in two separate equations for CBCL scores indicated strong crosstime sta-
each criterion. Step 2 for mothers included bility of behavior problems. Furthermore,
maternal proactive parenting (observed), pro- mothers’ anger Time 1, as well as their re-
active parenting styles (reported), BDHI ported lack of proactive parenting, uniquely
scores (hostility), anger, and happiness. Step predicted their reports of externalizing prob-
2 for fathers included paternal proactive par- lems over and above crosstime stability. No
enting (observed), BDHI scores, anger, and paternal parenting variables added unique pre-
happiness. In both of the separate equations diction to mothers’ Time 2 reports of chil-
(maternal and paternal), parental predictors dren’s externalizing. Regression results pre-
were entered together in a block so that dicting Time 2 TRF scores also can be seen
unique contributions of each aspect of parent- in Table 4. Over and above crosstime stability
of behavior problems, both maternal and pa-
ing Fisher’s z. There were only three significant differ-
ternal anger predicted teachers’ reports of ex-
ences out of 45 pairs. Thus, there appeared to be no ternalizing problems. The contribution to R2
pervasive differences in patterns of association for for maternal and paternal predictors as a block
boys and girls; hence, child sex is not considered in was significant for three of four equations.
subsequent analyses. The correlations which differed
for boys and girls suggested that maternal proactive
attitudes, rboys (36) = −.56, p < .001, rgirls (29) = −.14, ns,
and maternal anger, rboys (36) = .53, p < .001, and rgirls 4. Because a few data were missing, but in differing pat-
(29) = −.02, ns, contributed more to boys’ behavior terns for differing variables (especially for teachers and
problems than to girls’. Two other correlations, only fathers), means were substituted in the regression anal-
one of which was significantly different for boys and yses for missing data. This conservative approach ren-
girls, were suggestive: girls’ self-reported behavior ders associations somewhat smaller than counting data
problems were related to mothers’ self-reported hostil- as truly missing; in a case such as this, it is the pre-
ity, rgirls (29) = .48, p < .001, but boys’ were not, rboys ferred method to retain statistical power. For Time 1
(36) = −.07, ns. In contrast, boys’ self-reported behav- predictors, the number of missing data points ranged
ior problems were related to fathers’ self-reported hos- from 0 to 8, with a median of 0.5. For criteria at Time
tility, rboys (36) = .45, p < .001, but girls’ were not, rgirls 2, the number of missing data points ranged from 0 to
(29) = .22, ns. Due to the paucity of sex differences in 11, with a median of 1.0. For criteria at Time 3, the
zero-order correlational patterns, such differences are number of missing data points ranged from 2 to 11,
not considered further. with a median of 3.0.
Table 4. Predicting externalizing at Time 2
Predictor sr 2 a β R 2ch b
Criterion: CBCL Time 2

1. CBCL Time 1 .366*** .606 .367***
2. Maternal anger .044* .169 .151**
Maternal happiness .004 −.050
MBDHI .000 −.001
Maternal proactive style .095** −.269
Maternal proactive behavior .012 −.091
2. Paternal anger .000 −.011 .015
Paternal happiness .009 .081
FBDHI .005 .059
Paternal proactive behavior .010 −.095
Criterion: TRF Time 2
1. TRF Time 1 .213*** .462 .462***
2. Maternal anger .132* .353 .136*
Maternal happiness .001 .040
MBDHI .005 .065
Maternal proactive style .001 −.040
Maternal proactive behavior .000 .000
2. Paternal anger .085* .285 .126*
Paternal happiness .051† .198
FBDHI .004 .054
Paternal proactive behavior .003 −.049
Note: Reporters for CBCL were mothers, TRF were teachers, and
YSR were the children themselves.
a
Squared semipartial correlations, equivalent to the portion of R2
uniquely attributable to each variable.
b
Significance of R2change.
†p < .10. *p < .05. **p < .01. ***p < .001, two-tailed.
Predicting Time 3 behavior problems. Regres- Moderation analyses. Next we examined

sion results predicting Time 3 CBCL and TRF whether the power of parental predictors dif-
criteria, as well as YSR child self-reports of fered, depending on children’s initial level of
externalizing problems, are shown in Table 5. behavior problems. Following Baron and
Usually only one earlier behavior problem Kenny’s (1986) methodology, we used hierar-
score predicted Time 3 behavior criteria be- chical regression equations to assess these
cause of the substantial intercorrelation differences in prediction patterns: For all
among predictors. Over and above this behav- possible combinations, parenting and prior
ioral stability, both fathers’ lack of observed externalizing predictors were entered on the
proactive parenting and lack of maternal hos- first step; their interaction term was entered
tility predicted mothers’ reports of children’s on the second step. Significant interaction
externalizing at Time 3. As for teacher reports terms were found for 12 of the 90 equations,
of Time 3 externalizing, no parental predictor over twice the number expected by chance.
approached significance, although the mater- (An additional 8 interactions, p < .10, were
nal variables as a group added significantly to identified. Although they were not graphed, it
R2. Finally, paternal hostility was a significant is worth noting that the same patterns held;
predictor of children’s reports of their own i.e., both positive and negative parenting fac-
externalizing at Time 3. The contribution to tors predicted subsequent externalizing prob-
R2 for maternal and paternal predictors as a lems in predicted directions for those children
block was significant for two of six equations. who initially showed more of these prob-
Table 5. Predicting externalizing at Time 3
Predictor sr 2 a β R 2ch b
Criterion: CBCL Time 3

1. CBCL Time 1 .033 .133 .684***
CBCL Time 2 .518*** .739
2. Maternal anger .006 .049 .045
Maternal happiness .017 .076
MBDHI .073* −.179
Maternal proactive style .016 .084
Maternal proactive behavior .066† −.167
2. Paternal anger .016 −.073 .027
Paternal happiness .001 .013
FBDHI .000 .000
Paternal proactive behavior .079* −.175
Criterion: TRF Time 3
1. TRF Time 1 .013 .123 .100*
TRF Time 2 .047† .239
2. Maternal anger .044 .228 .152*
Maternal happiness .031 −.165
MBDHI .037 −.195
Maternal proactive behavior .012 −.116
2. Paternal anger .001 −.032 .030
Paternal happiness .020 −.139
FBDHI .013 .115
Paternal proactive behavior .001 .038
Criterion: YSR Time 3
1. CBCL Time 1 .000 −.016 .191*
CBCL Time 2 .033 .224
TRF Time 1 .083* .314
TRF Time 2 .000 .010
2. Maternal anger .027 .175 .119
Maternal happiness .054† −.224
MBDHI .000 .005
Maternal proactive behavior .016 .133
2. Paternal anger .038 −.190 .150*
Paternal happiness .002 −.037
FBDHI .171** .393
Paternal proactive behavior .001 .026
Note: Reporters for CBCL were mothers, TRF were teachers, and
YSR were the children themselves.
a
Squared semipartial correlations, equivalent to the portion of R2
uniquely attributable to each variable.
b
Significance of R2change.
†p < .10. *p < .05. **p < .01. ***p < .001, two-tailed.
lems.) For the 12 significant cases, separate plots for regression equations for children
regression equations were analyzed for chil- high and low on earlier externalizing, with β
dren low and high on the earlier externalizing weights for the significant slope. Differential
predictor (split at the median). Finally, the prediction of Time 2 and Time 3 TRF scores
findings of these separate regression equa- is shown in Figure 1, while plots for differen-
tions were evaluated, and interactive effects tial prediction of Time 3 CBCL91-M and
were plotted following the guidelines of Ai- YSR scores are shown in Figures 2 and 3, re-
ken and West (1991). spectively. No moderation effects were found
The results of these series of steps are de- for CBCL91-M Time 2 scores.
picted in Figures 1–3. Each figure represents The findings depicted in Figures 1 through
Figure 1. Differential prediction of Time 2 and Time 3 TRF scores.
Figure 2. Differential prediction of Time 3 CBCL91-M scores.
3 can be grouped according to several recur- children high in earlier behavior problems. In
ring themes; the important findings are four- fact, parental anger was the most prominent
fold. First, in all but one case, prediction of predictor in these moderation analyses, across
later externalizing problems from parenting reporters (i.e., mother or teacher) and time pe-
variables was found for children with higher riods used in prediction. Third, positive par-
previous externalizing scores. Second, nega- enting variables, such as proactive parenting
tive parenting variables, such as anger or (observed) and maternal proactive parenting
hostility, were especially strong positive pre- styles (reported), strongly predicted fewer
dictors of later externalizing problems for later behavior problems for children high in
Figure 3. Differential prediction of Time 3 YSR scores.
earlier behavior problems. Proactive behavior above the norm on mothers’ or teachers’ re-
was the second most frequent predictor in ports of externalizing problems at the begin-
these moderation analyses, across Time 1 ning of the study were the focus of this analy-
reporters. Mothers’ reported proactive parent- sis. A cluster analysis was used to typify
ing styles were significant predictors in mod- change in these children’s maternal and
erating children’s reports of their own behav- teacher externalizing ratings from Time 1 to
ior problems at Time 3. Maternal happiness Time 2 to Time 3.
was a significant predictor only once— This analysis created two distinct groups
with lack of maternal happiness at Time 1 be- based on profile similarity, by means of
ing the only predictor more important for squared Euclidean distances from initial clus-
children low, rather than high, in Time 1 be- ter centers. ANOVA confirmed the existence
havior problems. Fourth, maternal predictors of two distinct cluster groups of young antiso-
figured in all but two of the moderation analy- cial children by comparing the between the
ses. In sum, parenting factors, especially between-and within-group profiles for each
mothers’ anger, proactive parenting, and pro- cluster group. Examination of the profiles
active parenting styles, were important espe- showed patterns of change from equivalent
cially in predicting behavior problems in Time 1 scores. The first profile group con-
those children showing more difficulty ini- sisted of 28 children with externalizing prob-
tially. lems who improved, and the second consisted
of 10 whose behavior problems were stable
across the period (see Table 6).
Change in externalizing over time and its
Next, we were interested in determining
association with parental predictors
whether parental predictors differed singly
We used a person-centered analytic approach and in combination for these two groups of
to examine parental predictors of different de- initially at-risk children who showed different
velopmental trajectories for young antisocial developmental trajectories. Separate direct
children—that is, different patterns of change discriminant function analyses were per-
in externalizing across the three data collec- formed using maternal or paternal variables as
tion periods. Thus, the 38 children who ini- predictors of membership in the two groups.
tially scored 1 or more standard deviations Maternal predictors were anger, happiness,
Table 6. Clusters of children at moderate or high risk at Time 1: Change in externalizing

over time final cluster centers
CBCL TRF
Type of Child Time 1 Time 2 Time 3 Time 1 Time 2 Time 3 n
Improving 20.86 15.54 14.69 10.68 5.11 6.64 28

Stable 19.00 21.40 20.50 28.10 26.58 17.74 10
Between group F (1, 36) 0.45 2.85† 2.86† 25.76*** 60.01*** 11.99***
Note: “Improving,” or “stable,” scores in either absolute terms, or in terms of T scores.

†p < .10. ***p < .001.
Table 7. Prediction of improving or stable status by grouped parenting variables:

Discriminant function analyses
Correlations of Pooled Within-Group

Predictors With Univariate Correlations With
Predictors Discriminant Function F (1, 36) Discriminant Function
Maternal
1. Maternal proactive style .49 9.36*** .66
2. Maternal hostility −.16 .57 −.12
3. Maternal proactive (observed) .28 9.28** .66
4. Maternal anger −.61 14.85*** −.83
5. Maternal happiness −.18 0.00 .00
Canonical R .61
Eigenvalue .59 Correctly classified 81.6% of cases
Paternal
1. Paternal hostility .08 6.54** .80
2. Paternal proactive (observed) .03 1.17 −.34
3. Paternal anger .97 6.63** .80
4. Paternal happiness .21 0.04 −.06
Canonical R .28
Eigenvalue .47 Correctly classified 81.6% of cases
**p < .01, ***p < .001.
hostility, and observed and reported proactive changes in externalizing problems. Canonical
parenting. Paternal predictors were anger, correlations for each group of predictors
happiness, hostility, and observed proactive showed that, as group, both maternal and pa-
parenting. Groups were children who im- ternal predictors were related to membership
proved or did not improve (based on the clus- in either the improving or the stable or wors-
ter analysis). ening group. Univariate analyses of variance,
For maternal predictors, as shown in Table as well as correlations of single predictor vari-
7, one discriminant function was calculated, ables with each discriminant function (also
with a χ2 (5) = 13.42, p < .01. This discrimi- shown in Table 7) indicate that the mothers
nant function accounted for 100% of the be- of children who improved were more proac-
tween-group variability in changes in exter- tive in both measurement contexts and
nalizing problems. For paternal predictors, showed less anger. A similar pattern was pres-
one marginally significant discriminant func- ent for fathers: Improvement was associated
tions was calculated, with a χ2 (4) = 8.54, p < with proactive behavior (observed) and lack
.10. This discriminant function accounted for of hostility and anger. Finally, mere knowl-
100% of the between-group variability in edge of these parenting dimensions allows
correct cluster grouping (i.e., improving vs. antecedent in time to child outcomes, sugges-
stable) of between 76 and 82% of all 38 chil- tive of parental influence, although the direc-
dren originally at risk. tion of effect in correlational designs cannot
be definitive. Therefore, an analytic approach
was used that allowed for prediction of paren-
Discussion
tal contributions (self-reported as well as ob-
The goals of this research were to identify served), over and above young children’s ini-
some aspects of parenting that characterize fa- tial levels of problems. Such an approach
milial, environmental conditions contributing necessarily makes it difficult to identify influ-
to continuity and change over time in serious ential parental factors, because of the known
behavior problems of young children. The re- strong continuity of child problems over time,
sults of this study are consistent with previous especially for disruptive behaviors that have
research on parenting practices linked to chil- emerged early in development.
dren’s behavior problems. However, few of Findings from the present study were con-
these earlier studies, even longitudinal ones, sistent with previous research in identifying
include multiple time points for assessment of marked continuity in problems over time. De-
child problems and comprehensive assess- spite this strong stability, parenting factors
ments of both maternal and paternal parent- were found to predict developmental trajecto-
ing. Inclusion of these design elements here ries, in addition to the child contributions, in
allowed us to evaluate the unique contribu- over half the regression equations tested. With
tions of parents over and above stability in regard to emotion, positive emotion emerged
behavior problems. Another way in which the as a predictor of fewer behavior problems
present work was intended to extend previous over time in only one instance.
research was in its explicit focus on emotional Observed parental anger predicted chil-
elements of parenting in predicting child out- dren’s later aggressive, antisocial behavior.
comes. Viewing parenting through the lens of Moreover, parental anger was most influential
affective organization (Dix, 1991), and behav- as a disorganizer of the behavior of those al-
ior problems as dysregulation (Cole et al., ready at risk, indicating the interaction of pa-
1994), was found to provide a fruitful avenue rental anger and early vulnerability. Parents’
for investigation. negative emotions are punishing and dysregu-
Both mothers and fathers were observed lating, a disorganizer of young children’s so-
interacting with their children in a variety of cial–emotional development (Cummings, Ian-
representative contexts. Children who had notti, & Zahn–Waxler, 1985; Cummings,
been selected for study varied substantially in Zahn–Waxler, & Radke–Yarrow, 1981; Da-
their initial levels of problems, ranging from vies & Cummings, 1995; Denham, 1989,
nonproblem to the top 2% of disruptive chil- 1993).
dren based on norms from standardized tests. In addition, pervasive negative emotional
Correlations between parenting dimensions patterns, reflected in self-reported parental
when children were 4–5 years old and later hostility, were important in predicting later
child problems revealed many significant as- problems. Parental hostility is one form of
sociations. Mothers’ observed proactive par- chronic distress that has been identified as
enting practices consistently predicted fewer contributing to parents’ inability to attend and
externalizing problems in children over time, respond effectively to children’s behaviors
as did mothers’ reported nurturant, nonrestric- and needs (Campbell, 1994, 1995; Dix, 1991;
tive orientation. Parental anger was consis- Dumas & Wekerle, 1995). An affective envi-
tently related to greater externalizing prob- ronment such as this may exacerbate aversive
lems over time, and parental hostility also was child behaviors.5
associated with externalizing problems in
children. Paternal proactive parenting was 5. There was one unexpected finding. Specifically, Time
sometimes linked to fewer child problems. 1 maternal hostility was associated with a more opti-
These socialization factors measured were mal child outcome at Time 3. Because significant di-
In this study, proactive parenting was also tual and analytic value of a person-centered
a predictor of child outcomes. Its effects were as well as a variable-oriented approach. These
strongest in analyses of moderating variables. parental contributions underscore the need for
That is, the role of constructive, proactive par- very early identification and intervention in
enting in decreasing children’s disruptive an- children’s disruptive behaviors (Haapasalo &
tisocial behavior patterns was strongest for Tremblay, 1994; Shaw, Owens, Vondra, &
children who initially had many problems. Keenan, 1996; Shaw, Winslow, Owens, Von-
This facilitative role of proactive parenting, dra, Cohn, & Bell, 1998).
seen especially for the most troubled children, Most studies of the early course of chil-
was found for both observed parental behav- dren’s behavior problems focus on the role of
iors and self-reported styles. Observed pro- the mother as an agent of change. In coercion
active behaviors consisted of supportive theory, Patterson (1980) has characterized the
presence and provision of a structured envi- roles of mothers and fathers of preschool chil-
ronment. Reported styles consisted of a nur- dren in the following manner. In normal fami-
turant, child-centered approach and a deem- lies, fathers function as social facilitators and
phasis on harsh (restrictive, reactive) methods resident “guests.” Mothers serve as caretak-
of child rearing. These findings confirm the ers, with both parents sharing in child man-
results of the Zahn–Waxler et al. (1990; see agement problems. These same roles occur in
also Wakschlag & Hans, 1999) longitudinal families with an aggressive, disruptive child;
study of even younger children, and are con- however, in such families the role of the
sistent with cross-sectional research on restric- mother is expanded to include that of crisis
tive, nonnurturing child-rearing styles that are manager. Mothers are more likely than fathers
associated with children’s behavior problems (a) to be exposed to high densities of aversive
(e.g., Dumas & LaFreniere, 1993; Dumas, La- events even from normal preschool children
Freniere, Beaudin, & Verlaan, 1992; Gardner, and (b) experience the attack behaviors of
1989; Pettit & Bates, 1989). problem children. Patterson refers to mothers
Because of the high levels of continuity of as “unacknowledged victims.”
externalizing problems over time, parenting Given these different parental roles and
factors accounted for a smaller portion of the functions, it would not be surprising if find-
variance in regression analyses. It is notewor- ings here for maternal and paternal predictors
thy, however, that despite consistency over differed in magnitude, type, and areas of in-
time in the relative rankings of problems, a fluence. When the sample as a whole was
large number of children showed improve- considered in regression analyses, maternal
ment. This improvement was consistently and paternal predictors did not differ. How-
linked with parenting behaviors that provided ever, almost all of the moderation analyses,
positive support, structure, and an affective where the interaction effects of parenting and
environment characterized by a lack of anger child risk were identified, were based on the
and hostility. One of the strengths of these beneficial effects of mothers’ positive parent-
findings is in demonstrating that, although re- ing approaches and the deleterious effects of
lated, both parental emotions and parental be- mothers’ negative emotionality. Mothers were
haviors contribute to risk and resiliency in observed interacting with their children in
children initially at risk for behavior prob- more settings than fathers were, and provided
lems. The identification of two different de- reports of child-rearing styles. Although this
velopmental trajectories for children with more closely reflects daily family life, it also
early onset aggression, based on different par- makes comparative statements somewhat ten-
enting environments, illustrates the concep- uous. Evidence of paternal influence on early
childhood trajectories for antisocial behaviors,
however, does affirm the importance of in-
rectional patterns were consistent for all regressions
save one, and for all moderator analyses, we would cluding fathers in research designs (DeKlyen,
not wish to attach importance to this lone finding by Biernbaum, Speltz, & Greenberg, 1998).
attempting to interpret it. Moreover, intervention efforts to reduce be-
havior problems and develop behavioral com- tively small sample size, which places restric-
petence would benefit from further inclusion tions on the number and types of models that
of the fathers in the process of remediation can be tested. The number of tests conducted
and change (Kendziora & O’Leary, 1993). here may have resulted in some chance find-
The parenting dimensions studied here are ings. Fourth, even though our design featured
by no means exhaustive of children’s experi- analyses that highlighted the unique contribu-
ences in their family environments that may tions of parental emotion and behavior, future
create risk for their development of antisocial research could benefit from coding systems
behavior patterns. Parental psychopathology that even more clearly differentiate them.
and personality also need to be considered, The conceptual focus on the influence of
along with additional child-rearing practices parental emotion on child behavior problems
and parental styles. Moreover, measures of lends itself to the testing of alternative models
parenting behaviors and emotions like those (Dix, 1991), assuming a sufficiently large
used here are in need of further refinement. In sample size. Here we tested a “direct effect”
this study, for example, operational defini- model, whereby parental emotions are pre-
tions of anger and positive emotion were non- sumed to influence child outcomes directly.
overlapping. This makes it possible to esti- Such effects were clearly in evidence, particu-
mate the influence of each in an independent larly for the deleterious role of parental anger
manner. Positive and negative parenting be- in maintaining child problems over time. One
haviors, however, were not measured on sepa- might also have hypothesized causal path-
rate scales, which would influence interpreta- ways where parental emotions played an “in-
tion of findings to some degree. The limited direct” role.
support for the hypothesis that parental posi- That is, parental emotions could influence
tive affect would play an ameliorative role in- parental behaviors that, in turn, impact on
dicates the need to reconsider how we have child outcomes. Dix’s (1991) model frames
conceptualized and measured this quality of three indirect effects of parents’ emotion ex-
parenting. Timing, contingencies, appropriate- pression and regulation. First, anger can affect
ness, intensity of expression, and more—all parent motivation—it is easy to imagine a
may play a role in how the child experiences parent who repeatedly feels angry with a child
the pleasure of a parent. to become overintrusive with a child, or, con-
Several potential limitations must be noted. versely, disengaged. Intense or chronic paren-
First, parenting factors were not assessed at tal anger may be accompanied by inability to
Times 2 and 3. Inclusion of such information monitor and control one’s own behavior or
would have made it possible to examine the the disinclination to control the child’s behav-
relative contribution of both early and con- ior. Second, anger can affect parent percep-
temporaneous parenting environments. Sec- tions and cognitions—increasing parents’
ond, the sample of participants was almost ex- propensity to view the child negatively, dis-
clusively upper-middle SES and Caucasian. rupting the ability to reason clearly about
Although a more diverse sample of partici- child-rearing conflicts and engage in problem
pants might, in theory, provide a greater range solving, or disrupting parental monitoring and
of behavior problems and parenting behavior, attention. As Kendziora and O’Leary (1993)
there was marked variability in the severity of note, such narrowing of attention and moni-
children’s externalizing problems at the onset toring can lead to restriction in response rep-
of the study (see also Haapasalo & Tremblay, ertoires and problematic parenting. There may
1994). Nonetheless, generalization must be be both metaphorical and veridical meaning
limited to the population of antisocial children to the phrase “so angry she couldn’t think
sampled. It is noteworthy, however, that rela- straight.” Finally, anger can be directly com-
tively affluent nonminority children have municated via expressive behaviors, which
been heretofore understudied regarding risk children find easily imitated but also difficult
for later aggressive, antisocial behavior prob- to cope with when predominant in the affec-
lems. A third limitation pertains to the relative environment. Unfortunately our design,
paradigms, and small sample size did not change in the child, and on change in aspects
allow us to completely disentangle these po- of parenting that are not overtly emotional
tential mechanisms at play in the prediction (e.g., Conduct Problems Prevention Research
of behavior problems over time. This places Group, 1992; Reid, 1993; but cf. Greenberg,
constraints on our understanding of how pa- Speltz, & DeKlyen, 1993). Even interventions
rental emotions are implicated in child out- that include components on helping parents
comes. Most likely, the pathways from paren- deal with the stress in their own lives, an as-
tal anger to child externalizing are both pect of training which at least hints at parental
multiply determined and bidirectional. emotions, remain almost solely cognitive be-
Although transactional models are com- havioral (Kazdin, 1993; Reid, 1993). The re-
monly adhered to, their translation into research here, along with a growing body of lit-
search designs dealing with the interactions of erature on parenting factors associated with
nature and nurture is uncommon. This study, child outcomes, suggests the need to incorpo-
for example, would have benefited from the rate explicit efforts to aid parents in their reg-
inclusion of measures of child temperament ulation of emotion as well. Could young chil-
and emotion regulation to assess child as well dren’s behavior problems be further reduced
as parent effects. Also, longitudinal studies by systematic efforts to help mothers and fa-
like this one are not well suited to the study thers to engage in proactive behaviors and
of moment-to-moment changes in the nature regulate their anger?
of the transactions that might provide better There seems to be some reluctance to pur-
understanding of processes implicated in sue these questions, given the reaction that
long-term developmental outcomes. More- followed early theories in which mothers were
over, approaches are needed that combine the seen as causing their children’s problems.
best features of both genetic and nongenetic This reaction and subsequent research appro-
research designs. The present study of the role priately highlighted the primacy of child ef-
of parental socialization is like that of most fects in many disorders, such as schizophrenia
others in that there is no control for biological and autism. However, efforts to reduce par-
relatedness. Such research may yield “envi- ents’ feelings of responsibility have general-
ronmental” influences that are genetically me- ized to a much broader range of problems,
diated, to some degree. Statistical control for some of which do result in part from parent-
child problems helps, but does not fully re- ing styles and behaviors. “No fault” theories
solve the issue. can contribute to a reluctance to pursue in-
Hostility and proneness to anger, in partic- quiry into parental influence on child out-
ular, may be part of an irritable disposition comes. Both research and clinical intervention
with biological underpinnings. A genetic pre- with young children with behavior problems
disposition, passed on from parent to child, might benefit from a sympathetic reappraisal
could be reflected in child defiance and nega- of this position. Children with early onset
tivity that is not the sole result of exposure to aggression are at very high risk for many
an angry parent or imitation of an aggressive types of emotional and behavioral problems
model. Seen as traits or dispositions, these throughout their lives. It is worth pursuing
emotions may be difficult to modify, com- further the ways that parents might learn more
pared to the alteration of specific behaviors. about changes that they could make to alter
It is not surprising, then, that behavior these maladaptive developmental trajecto-
modification research and interventions tend ries for their children (Wakschlag & Hans,
to focus heavily on creating behavioral 1999).
References
Achenbach, T. M. (1990). Conceptualization of develop- Achenbach, T. M. (1991). Manual for the Child Behavior
mental psychopathology. In M. Lewis & S. M. Miller Checklist/4–16 and 1991 Profile. Burlington, VT:
(Eds.), Handbook of developmental psychopathology University Associates in Psychiatry.
(pp. 3–13). New York: Plenum Press. Achenbach, T. M., & Edelbrock, C. (1986). Manual for
the Teacher’s Report Form and teacher version of the Cole, P. M., Michel, M. K., & Teti, L. O’D. (1994). The
Child Behavior Profile. Burlington, VT: University development of emotion regulation and dysregulation:
Associates in Psychiatry. A clinical perspective. Monographs of the Society for
Aiken, L. S., & West, S. G. (1991). Multiple regression: Research in Child Development, 59(2/3, Serial No.
Testing and interpreting interactions. Newbury Park, 240), 73–102.
CA: Sage. Cole, P. M., & Zahn–Waxler, C. (1988). Prediction
Anderson, E. R. (1993). Analyzing change in short-term of conduct problems during the transition from
longitudinal research using cohort-sequential designs. preschool to school age (Protocol No. 88-M-0217).
Journal of Consulting and Clinical Psychology, 61, Bethesda, MD: National Institutes of Mental
929–940. Health.
Baron, R. M., & Kenny, D. A. (1986). The moderator– Cole, P. M., & Zahn–Waxler, C. (1992). Emotional dys-
mediator variable distinction in social psychological regulation and disruptive behavior disorders. In D.
research: Conceptual, strategic, and statistical consid- Cicchetti & S. L. Toth (Eds.), Developmental per-
erations. Journal of Personality and Social Psychol- spectives on depression (pp. 173–209). Rochester,
ogy, 51, 1173–1182. NY: University of Rochester Press.
Bates, J. E., Bayles, K., Bennett, D. S., Ridge, B., & Cole, P. M., Zahn–Waxler, C., & Smith, K. D. (1994).
Brown, M. M. (1991). Origins of externalizing behav- Expressive control during a disappointment: Varia-
ior at eight years of age. In D. J. Pepler & K. H. Rubin tions related to preschoolers’ behavior problems. De-
(Eds.), The development and treatment of childhood velopmental Psychology, 30, 835–846.
aggression (pp. 93–120). Hillsdale, NJ: Erlbaum. Conduct Problems Prevention Research Group. (1992). A
Bergman, L. R., & Magnusson, D. (1997). A person-ori- developmental and clinical model for the prevention
ented approach in research on developmental psycho- of conduct disorder: The FAST TRACK program.
pathology. Development and Psychopathology, 9, Development and Psychopathology, 4, 509–527.
291–319. Crockenberg, S., & Litman, C. (1990). Autonomy as
Block, J. H. (1981). The child-rearing practices report competence in 2-year-olds: Maternal correlates of
(CRPR): A set of Q items for the description of paren- child defiance, compliance, and self-assertion. Devel-
tal socialization and values. Berkeley, CA: Institute opmental Psychology, 26, 961–967.
of Human Development, University of California, Cummings, E. M., & Davies, P. (1994). Marital conflict.
Berkeley. New York: Guilford Press.
Buss, A. H., & Durkee, A. (1957). An inventory for as- Cummings, E. M., Iannotti, R. J., & Zahn–Waxler, C.
sessing different kinds of hostility. Journal of Con- (1985). Influence of conflict between adults on the
sulting Psychology, 21, 343–349. emotions and aggression of young children. Develop-
Cairns, R. B., & Cairns, B. D. (1994). Lifelines and risks: mental Psychology, 21, 495–507.
Pathways of youth in our time. New York: Cambridge Cummings, E. M., Zahn–Waxler, C., & Radke–Yarrow,
University Press. M. (1981). Young children’s responses to expressions
Campbell, S. B. (1987). Parent-referred problem three- of anger and affection by others in the family. Child
year-olds: Developmental changes in symptoms. Development, 52, 1274–1282.
Journal of Child Psychology and Psychiatry, 28, Davies, P. T., & Cummings, E. M. (1995). Children’s
835–845. emotions as organizers of their reactions to interadult
Campbell, S. B. (1994). Hard-to-manage preschool boys: anger: A functionalist perspective. Developmental
Externalizing behavior, social competence, and fam- Psychology, 31, 677–684.
ily context at two-year followup. Journal of Abnor- Denham, S. A. (1989). Maternal affect and toddlers’ so-
mal Child Psychology, 22, 147–166. cial–emotional competence. American Journal of Or-
Campbell, S. B. (1995). Behavior problems in preschool thopsychiatry, 59, 368–376.
children: A review of recent research. Journal of Denham, S. A. (1993). Maternal emotional responsive-
Child Psychology and Psychiatry, 36, 113–149. ness and toddlers’ social–emotional functioning.
Campbell, S. B., Breaux, A. M., Ewing, L. J., & Szumow- Journal of Child Psychology and Psychiatry, 34,
ski, E. K. (1984). A one-year follow-up of parent-iden- 725–728.
tified behavior problem toddlers. Journal of the Ameri- Denham, S. A., Renwick, S., & Holt, R. (1991). Working
can Academy of Child Psychiatry, 23, 243–249. and playing together: Prediction of preschool social–
Campbell, S. B., & Ewing, L. J. (1990). Hard-to-manage emotional competence from mother-child interaction.
preschoolers: Adjustment at age nine and predictors Child Development, 62, 242–249.
of continuing symptoms. Journal of Child Psychology DeKlyen, M., Biernbaum, M. A., Speltz, M. L., &
and Psychiatry, 31, 871–889. Greenberg, M. T. (1998). Fathers and preschool be-
Campbell, S. B., Ewing, L. J., Breaux, A. M., & havior problems. Developmental Psychology, 34,
Szumowski, E. K. (1986). Parent-identified be- 264–275.
havior problem toddlers: Follow-up at school entry. Dix, T. (1991). The affective organization of parenting:
Journal of Child Psychology and Psychiatry, 27, Adaptive and maladaptative processes. Psychological
473–488. Bulletin, 110, 3–25.
Campbell, S. B., March, C., Pierce, E., Ewing, E. J., & Dumas, J., & LaFreniere, P. J. (1993). Mother–child rela-
Szumowski, E. K. (1991). Hard-to-manage preschool tionships as sources of support or stress: A compari-
boys: Family context and stability of externalizing be- son of competent, average, aggressive, and anxious
havior. Journal of Abnormal Child Psychology, 19, dyads. Child Development, 64, 1732–1754.
301–318. Dumas, J., LaFreniere, P. J., Beaudin, & Verlaan, P.
Campbell, S. B., Pierce, E. W., March, C. L., Ewing, (1992). Mother–child interactions in competent and
L. J., & Szumowski, E. K. (1994). Hard-to-manage aggressive dyads: Implications of relationship stress
preschool boys: Symptomatic behavior across con- for behaviour therapy with families. New Zealand
texts and time. Child Development, 65, 836–851. Journal of Psychology, 21, 3–13.
Dumas, J. E., & Wekerle, C. (1995). Maternal reports Miller, N. B., Cowan, P. A., Cowan, C. P., Hetherington,
of child behavior problems and personal distress as E. M., & Clingempeel, W. G. (1993). Externalizing
predictors of dysfunctional parenting. Developmental in preschoolers and early adolescents: A cross-study
and Psychopathology, 7, 465–479. replication of a family model. Developmental Psy-
Egeland, B., Kalkoske, M., Gottesman, N., & Erickson, chology, 29, 3–18.
M. F. (1990). Preschool behavior problems: Stability Moffitt, T. E. Adolescent-limited and life-course persis-
and factors accounting for change. Journal of Child tent antisocial behavior: A developmental taxonomy.
Psychology and Psychiatry, 31, 891–909. Psychological Review, 100, 674–701.
Egeland, B., Sroufe, L. A., & Erickson, M. (1983). The Patterson, G. R. (1980). Mothers: The unacknowledged
developmental consequences of different patterns of victims. Monographs of the Society for Research in
maltreatment. International Journal of Child Abuse Child Development, 5, (5, Serial No. 186), 1–64.
and Neglect, 7, 459–469. Patterson, G. R., Reid, J. B., & Dishion, T. J. (1992).
Fischer, M., Rolf, J. E., Hasazi, J. E., & Cummings, L. Antisocial boys. Eugene, OR: Castalia.
(1984). Follow-up of a preschool epidemiological sam- Pettit, G. S., & Bates, J. E. (1989). Family interaction
ple: Cross-age continuities and predictions of later ad- patterns and children’s behavior problems from in-
justment with internalizing and externalizing dimen- fancy to four years. Developmental Psychology, 25,
sions of behavior. Child Development, 55, 137–150. 413–420.
Gardner, F. E. M. (1987). Positive interaction between Pettit, G. S., Bates, J. E., & Dodge, K. A. (1993). Family
mothers and conduct-problem children: Is there train- interaction patterns and children’s conduct problems
ing for harmony as well as fighting? Journal of Ab- at home and school: A longitudinal perspective.
normal Child Psychology, 15, 283–293. School Psychology Review, 22, 403–420.
Gardner, F. E. M. (1989). Inconsistent parenting: Is there Pianta, R. C., & Caldwell, C. B. (1990). Stability of ex-
evidence for a link with children’s conduct problems? ternalizing symptoms from kindergarten to first grade
Journal of Abnormal Child Psychology, 17, 223–233. and factors related to instability. Development and
Greenberg, M. T., Speltz, M. L., & DeKlyen, M. (1993). Psychopathology, 2, 247–258.
The role of attachment in the early development of Reid, J. B. (1993). Prevention of conduct disorder before
disruptive behavior problems. Development and Psy- and after school entry: Relating interventions to de-
chopathology, 5, 191–214. velopmental findings. Development and Psychopath-
Haapasalo, J., & Tremblay, R. E. (1994). Physically ag- ology, 5, 263–276.
gressive boys from ages 6 to 12: Family background, Rickel, A. U., & Biasatti, L. I. Modification of the Block
parenting behavior, and prediction of delinquency. Child Rearing Practices Report. Journal of Clinical
Journal of Consulting and Clinical Psychology, 62, Psychology, 38, 129–134.
1044–1052. Roberts, G. C., Block, J. H., & Block, J. (1984). Continu-
Kazdin, A. E. (1993). Treatment of conduct disorder: ity and change in parents’ child-rearing practices.
Progress and directions in psychotherapy research. Child Development, 55, 586–597.
Development and Psychopathology, 5, 277–310. Russell, A., & Russell, G. (1996). Positive parenting and
Kendziora, K. T., & O’Leary, S. G. (1993). Dysfunctional boys’ and girls’ misbehaviour during a home observa-
parenting as focus for prevention and treatment of child tion. International Journal of Behavioral Develop-
behavior problems. In T. H. Ollendick & R. J. Prinz ment, 19, 291–307.
(Eds.), Advances in Clinical Child Psychology (Vol. Shaw, D. S., Owens, E. B., Vondra, J. I., & Keenan, K.
15, pp. 175–206). New York: Plenum Press. (1996). Early risk factors and pathways in the devel-
Loeber, R., & Stouthamer–Loeber, M. (1986). Family opment of early disruptive behavior problems. Devel-
factors as correlates and predictors of juvenile con- opment and Psychopathology, 8, 679–699.
duct problems and delinquency. In M. Tonry & N. Shaw, D. S., Winslow, E. B., Owens, E. B., Vondra,
Morris (Eds.), Crime and justice: An annual review J. I., Cohn, J. F., & Bell, R. Q. (1998). The develop-
of research (Vol. 7, pp. 29–149). Chicago: University ment of early externalizing problems among children
of Chicago Press. from low-income families: A transformational per-
Loeber, R., & Stouthamer–Loeber, M. (1998). Develop- spective. Journal of Abnormal Child Psychology, 26,
ment of juvenile aggression and violence: Some com- 95–107.
mon misperceptions and controversies. American Thompson, R. (1994). Emotion regulation: A theme in
Psychologist, 53, 242–259. search of definition. In N. A. Fox (Ed.), The develop-
Maccoby, E., & Martin, J. (1983). Socialization in the ment of emotion regulation: Biological and behav-
context of the family: Parent–child interaction. In ioral considerations. Monographs of the Society for
P. H. Mussen (Series Ed.), E. M. Hetherington (Vol. Research in Child Development, 59(2/3, Serial No.
Ed.), Handbook of child psychology: Vol. 4. Social- 240), 25–52.
ization, personality, and social development (pp. 1– Thurber, S., & Hollingsworth, D. K. (1992). Validity of
102). New York: Wiley. the Achenbach and Edelbrock Youth Self-Report with
Mayer, M. (1974). Frog goes to dinner. New York: Dial hospitalized adolescents. Journal of Clinical Child
Books for Young Readers. Psychology, 21, 249–254.
Mayer, M., & Mayer, M. (1975). One frog too many. Trickett, P. K., & Susman, E. (1988). Parental percep-
New York: Dial Books for Young Readers. tions of child-rearing practices in physically abusive
McConaghy, S. H., Stanger, C., & Achenbach, T. M. and nonabusive families. Developmental Psychology,
(1992). Three-year course of behavioral/emotional 24, 270–276.
problems in a national sample of 4- to 16-year-olds: Wakschlag, L. S., & Hans, S. L. (1999). Relation of ma-
I. Agreement among informants. Journal of the Amer- ternal responsiveness during infancy to the develop-
ican Academy of Child and Adolescent Psychiatry, ment of behavior problems in high-risk youths. De-
31, 932–940. velopmental Psychology, 35.
Wehby, J. H., Dodge, K. A., & Valente, E. (1993). Denham, S. A. (1990). Antecedents of problems be-
School behavior of first grade children identified as haviors in children of depressed mothers. Develop-
at-risk for development of conduct problems. Behav- ment and Psychopathology, 2, 271–291.
ioral Disorders, 19, 67–78. Zahn–Waxler, C., Schmitz, S., Fulker, D. Robinson, J., &
Whitehurst, G. J. (1984). Interrater agreement for jour- Emde, R. (1996). Behavior problems in five-year-old
nal manuscript reviews. American Psychologist, 39, MZ and DZ twins: Genetic and environmental influ-
22–28. ences, patterns of regulation and internalization of con-
Zahn–Waxler, C., Iannotti, R. J., Cummings, E. M., & trol. Development and Psychopathology, 8, 103–122.

Denham Workman Et Al 2000

Uploaded by

Copyright:

Available Formats

Denham Workman Et Al 2000

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Denham Workman Et Al 2000

Uploaded by

Copyright:

Available Formats

See

Prediction of behavior problems from early to

ARTICLE in DEVELOPMENT AND PSYCHOPATHOLOGY · FEBRUARY 2000

Susanne Ayers Denham Pamela Cole

SEE PROFILE SEE PROFILE

Available from: Susanne Ayers Denham

Prediction of externalizing behavior

SUSANNE A. DENHAM,a ELIZABETH WORKMAN,a PAMELA M. COLE,b

Table 1. Description of measures utilized in the study

Time 1 Externalizing problems

Table 2. Correlations among parental predictors

1. Maternal supportive presence .69*** −.40*** .43*** −.27* .18 −.43***

1. Paternal supportive presence .66*** −.26* .24* −.49***

Table 3. Zero-order correlations between maternal predictors and

Criteria, Time 2 Criteria, Time 3

Predictors, Time 1 CBCL TRF CBCL TRF YSR

Multiple regressions predicting ing could be evaluated. When entered as a

Table 4. Predicting externalizing at Time 2

Criterion: CBCL Time 2

Predicting Time 3 behavior problems. Regres- Moderation analyses. Next we examined

Table 5. Predicting externalizing at Time 3

Criterion: CBCL Time 3

Figure 1. Differential prediction of Time 2 and Time 3 TRF scores.

Figure 2. Differential prediction of Time 3 CBCL91-M scores.

Figure 3. Differential prediction of Time 3 YSR scores.

Table 6. Clusters of children at moderate or high risk at Time 1: Change in externalizing

Type of Child Time 1 Time 2 Time 3 Time 1 Time 2 Time 3 n

Improving 20.86 15.54 14.69 10.68 5.11 6.64 28

Note: “Improving,” or “stable,” scores in either absolute terms, or in terms of T scores.

Table 7. Prediction of improving or stable status by grouped parenting variables:

Correlations of Pooled Within-Group

**p < .01, ***p < .001.

You might also like

1. Maternal supportive presence .69* −.40* .43*** −.27* .18 −.43***

p < .01, *p < .001.