Deficient social problem-solving in boys with ODD/CD, with ADHD, and with

both disorders.

Author/s: Walter Matthys

Issue: March, 1999

Children with aggressive behavior are likely to evince poor interpersonal relations
(Kazdin, 1995) as is reflected in high levels of peer rejection (Coie et al., 1990). It has
been demonstrated that aggressive children are deficient in cognitive problem-solving
skills that underlie social interaction (Dodge, 1993; Lochman et al., 1991). Therefore,
intervention strategies have been developed that focus on the child's social problem-
solving skills (Kazdin et al., 1987; Lochman et al., 1987). This training of social problem-
solving skills is one of the promising treatments for persistent aggression in childhood not
only because several controlled outcome studies have shown that this method can effect
change, but also because there is a growing body of basic research in aggressive
children's social problem-solving skills that supports the conceptualization of this
treatment method (Kazdin, 1997).

In problem-solving theories (d'Zurilla, 1986; Goldfried and d'Zurilla, 1969) it is assumed

that persons in everyday situations are faced again and again with problems and that
they are motivated to solve these problems. For this purpose they have at their disposal a
series of cognitive skills such as defining the problem, generating possible solutions, and
deciding which solution will be implemented. Social information-processing models
describe in detail how these cognitive skills can be applied to social problems.

In Dodge's model (1986), it is proposed that when children are faced with a social
situational cue they engage in 4 mental steps before enacting competent social
behaviors: encoding of situational cues, representation and interpretation of these cues,
mental search for possible responses to the situation, and selection of a response. Skillful
processing at each step will lead to competent performance within the situation, whereas
deficient or biased processing will lead to deviant behavior. This model has formed the
basis for a number of empirical studies of aggressive children and adolescents (for
extensive reviews refer to Dodge, 1993; and Lochman et al., 1991).

In the first step, encoding social cues, attention plays an important role. Aggressive boys
have been found to encode fewer cues than nonaggressive boys (Dodge and Newman,
1981); this deficit in encoding has also been found in aggressive/hyperactive boys (Milich
and Dodge, 1984). The second step of processing involves the mental representation and
interpretation of the encoded cues. Studies of aggressive children have demonstrated
biases in attributing hostile intentions to peers (Dodge, 1980; Lochman, 1987). In step 3,
children access one or more possible behavioral responses from long-term memory or
construct a novel response. Both the quantity and the quality of responses generated
have been studied. Richard and Dodge (1982) found that aggressive boys generated
fewer responses than popular boys, but Dodge et al. (1986) found no difference in the
number of responses generated in aggressive and in nonaggressive children. In the latter
study, it was found that aggressive children generated a higher proportion of aggressive
responses than nonaggressive children. In step 4, children evaluate the previously
accessed or constructed responses and select the most positively evaluated response for
enactment. A number of factors are involved in children's evaluations of responses,
including the moral ("good" versus "bad") acceptability of a response and the degree of
confidence children have in their ability to enact each response (self-efficacy). Aggressive
children judge aggression to be less morally "bad" than do other children (Deluty, 1983).
Aggressive children also are more confident in their ability to aggress than are
nonaggressive children, i.e., they are more likely than others to expect that engaging in
aggression will come easily for them (Perry et al., 1986).

In most of the studies reviewed, aggressive children were recruited from regular schools
and defined in terms of the teacher's ratings on aggression. However, it is highly probable

1
that populations of aggressive children consist not only of children with persistent
antisocial behavior but also of children with hyperactivity, impulsivity, and attention
problems. It has indeed been demonstrated that hyperactive children are more likely
than their nonhyperactive classmates to become involved in aggressive interactions
(Abikoff et al., 1980). In one study (Milich and Dodge, 1984), psychiatrically impaired
children with externalizing disorders were involved. However, subjects were not defined
in psychiatric terms (i.e., it was not examined whether they met criteria for one or more
disorders) but they were divided into groups (hyperactive/aggressive, hyperactive,
aggressive, psychiatric control) based on scores on rating scales; also, in that study only
encoding and interpretation were investigated.

The aim of our study was to examine social problem-solving extensively (from encoding
to response selection) in aggressive children who are defined in psychiatric terms, i.e.,
subjects met criteria of one or more of the DSM-III-R (American Psychiatric Association,
1987) categories of disruptive behavior disorders. Five groups of children were involved
in the study: (1) an oppositional defiant and conduct disorder (ODD/CD) group consisting
of children who satisfied criteria for oppositional defiant or conduct disorder (in children
aged 7 to 12 years ODD and CD are similar enough to combine; see Hinshaw et al.,
1993); (2) an attention-deficit hyperactivity disorder (ADHD) group; (3) an ODD/CD group
with comorbid ADHD (ODD/CD+ADHD); (4) a psychiatric control group with internalizing
disorders (INT); and (5) a normal control (NC) group. In this way we were able to find out
(a) whether disruptive behavior disorder groups (ODD/CD, ADHD, ODD/CD+ ADHD) differ
from the normal control group in social problem-solving; (b) whether these differences - if
they exist - are specific for disruptive behavior disorders by comparing these groups with
a psychiatric control group with internalizing disorders; and (c) whether certain features
in social problem-solving are characteristic of each of the 3 disruptive behavior disorder
groups. On the basis of our review of social-cognitive research in children with aggressive
behavior (Matthys and Van Engeland, 1992), we predicted that children with ADHD would
encode fewer situational cues and generate fewer responses and that ODD/CD children
would show hostile biases in interpretation, in the generation of responses, and in the
evaluation of these responses. It was also predicted that the ODD/CD+ADHD group would
show even more deviances because this group is considered to be more seriously
affected than either the ODD/CD or the ADHD group (Moffit, 1993; Taylor et al., 1991;
Walker et al., 1987). In the present study, only boys were included.

The most commonly used method to study children's social problem-solving is the
hypothetical situation interview or questionnaire. According to this method children are
presented with hypothetical social situations and questions designed to elicit responses
that indicate their processing patterns at various steps. The choice of problematic
situations to be represented is of particular relevance because the characteristics of
social information-processing in aggressive children depend highly on the situation
presented (Crick and Dodge, 1994). In the past, researchers have sometimes arbitrarily
determined the problematic situations to be presented to the children. In this study the
identification of the social problem situations was based on the results of factor analysis
of Taxonomy of Problematic Social Situations for Children (TOPS) (Cuperus, 1997; Dodge
et al., 1985). Three prototypical social problem situations thus were identified: (1) Being
Disadvantaged including peer group entry (e.g., when a boy wants to play with a group of
children and they tell him to wait) and response to provocation (e.g., when a peer takes
the boy's turn in a game); (2) Coping With Competition including response to failure and
response to success (e.g., when a boy loses or wins a game against a peer); and (3)
Social Expectations (e.g., when a boy is working on a class project that requires sharing
and cooperation). In most studies of aggressive children's social problem-solving skills,
peer group entry and response to provocations were chosen as problematic situations. To
our knowledge this is the first study to investigate social problem-solving with respect to
coping with success or failure and coping with expectations of peers.

METHOD

Samples

2
The boys with psychiatric disorder were recruited from the outpatient clinic of the
Department of Child and Adolescent Psychiatry, University Hospital Utrecht, and from
special schools. Because treatment could have had an effect on social problem-solving
skills, the boys with psychiatric disorder were currently not undergoing treatment
(psychotherapy, medication). Normal controls were recruited from regular elementary
schools. Subjects could be included in the study only if the following criteria were met: (1)
age between 7 and 12 years; (2) Full Scale IQ at least 80; (3) no psychotherapeutic
treatment in the previous 3 months; and (4) no psychopharmacological treatment
(specifically, no methylphenidate in the previous 3 days).

Boys were placed in the ODD/CD group (n = 48) when they satisfied DSM-III-R criteria for
ODD (n = 33) or CD (n = 15). Boys were placed in the ADHD group (n = 27) when they
satisfied DSM-III-R criteria for ADHD. Boys who met criteria for both ADHD and ODD (n =
25) or CD (n = 4)) were placed in the ODD/CD+ADHD group (n = 29). Neither the ADHD,
the ODD/CD, nor the ODD/CD+ADHD boys met the criteria for any type of anxiety or
mood disorder. The INT group (n = 23) consisted of boys who met criteria for overanxious
disorder (n = 9), dysthymia (n = 10), obsessive-compulsive disorder (n = 3), or
separation anxiety disorder (n = 1). With respect to the NC group (n = 37), boys were
excluded who had received scores on the Child Behavior Checklist for parents (CBCL) and
Teacher's Report Form (TRF) (Achenbach, 1991a,b) in the clinical range for Total
Problems, for the 2 broad-band groups of syndromes (designated as Externalizing and
Internalizing), and for the syndrome scores Attention Problems, Aggressive, Delinquent,
and Anxious/Depressed.

Diagnoses of the subjects from the outpatient clinic (n = 84) were based on extensive
semistructured psychiatric interviews, psychological assessment of the child, interviews
with the parents including discussion of the developmental history, and information from
the child's teacher. On the basis of information from these various informants, consensus
about the diagnosis was reached between the resident child psychiatrist and a board-
certified supervising child psychiatrist during case discussions focused particularly on
DSM-III-R criteria for the various disorders in childhood. Diagnoses of the subjects from
the special schools (n = 43) were based on a semistructured psychiatric interview with
the child, i.e., the Child Assessment Schedule (Hodges et al., 1982) conducted by a
graduate educational psychologist.

To describe subjects from a dimensional point of view, the CBCL and the TRF in the Dutch
translation (Verhulst et al., 1985) were completed by each subject's parents and
teachers. Because of the cognitive nature of the method used, subjects were
administered the WISC-R (Wechsler, 1974). Also, because performance in a social-
cognitive task may be influenced by cognitive stimulation in the family, the educational
level of the parents was checked. This was done by scoring the educational level of the
better-schooled parent according to a Dutch register (Beroepenklapper) (Van Westerlaak
et al., 1976).

Descriptive characteristics of the sample are included in Table 1. Groups differed in age,
the ODD/CD group being older than the ADHD group, the ODD/CD+ADHD group, and the
NC group. Groups differed in Full Scale IQ and in Verbal IQ: Full Scale IQ was higher in the
NC group than in both the ODD/CD group and the ODD/CD+ADHD group, and Verbal IQ
was higher in the NC group than in the ODD/CD group. Results of the CBCL and the TRF
demonstrated expected differences between, on the one hand, the psychiatrically
disordered groups and the NC group, and, on the other hand, among the psychiatrically
disordered groups. Groups also differed in educational level, the NC group having the
highest level and the ODD/CD group having the lowest level: ODD/CD = 1.7 (0.8), ADHD
= 2.0 (0.8), ODD/CD+ADHD = 1.9 (0.9), INT = 2.2 (0.8), and NC = 2.6 (0.6) ([[Chi].sup.2]
= 1.63, p [less than] .001).

The study protocol was approved by the Committee for Research on Human Subjects of
the University Hospital Utrecht, and parents gave written informed consent.

3
Procedure

Preparation of Videotaped Vignettes. To assess social problem-solving, we followed the

procedure developed by Dodge et al. (1986). According to this method children are
presented with videotaped vignettes of problematic social situations and questions
designed to elicit responses that indicate their processing patterns at various steps. The
identification of the social problem situations in which peers were involved was based on
the results of factor analysis of [TABULAR DATA FOR TABLE 1 OMITTED] TOPS (Cuperus,
1997; Dodge et al., 1985). In this way 3 problem domains were identified: Being
Disadvantaged, Coping With Competition, and Social Expectations. The TOPS items with
the highest correlations on these 3 factors were chosen as problematic situations for
elaboration into scripts. Twelve videotaped stimulus tapes were prepared with the
assistance of child actors: 5 in which the social problem consisted of how to cope with
being disadvantaged, 4 in which the problem was how to cope with competition, and 3 in
which the problem was how to deal with social expectations. Each videotaped vignette
consisted of 2 parts. First the social problem was presented, e.g., the protagonist is
building a plane but does not succeed, another boy offers to help him, the result being
that the plane breaks into pieces (Being Disadvantaged). Then, 3 solutions to the problem
are enacted by the protagonist: a prosocial/assertive solution, an antisocial/aggressive
solution, and a passive/submissive one.

Computerized Presentation of Videotaped Vignettes, Questions, and Answers. A

computerized instrument was developed in which for each problem situation an
introductory text, the videotaped vignette, and the questions are shown on the monitor of
a personal computer. In the introductory text for each problem situation the subject is
asked to identify with the protagonist. In the case of multiple-choice questions the subject
himself answered the question on the keyboard of the computer; in the case of open
questions the answer was noted by the research assistant. The 12 videotaped vignettes
were presented to subjects in random order; in each videotaped vignette, the 3
prototypical solutions were also presented in random order.

In the example of the broken plane, the first question assessed the interpretation of cues:
"The plane broke. Did the other boy break it on purpose?" The subject chose 1 of 3
answers: (1) on purpose; (2) it is not clear; (3) accidentally. The second question
assessed children's attention to and encoding of cues: "How do you know [he did it on
purpose, etc.]?" The number of responses was scored as well as the number of responses
in which an interpretation (inference) was made. The third question assessed problem
recognition, which is part of the interpretation: "Would you be upset that your plane was
broken?" The subject chose 1 of 4 answers: (1) very upset; (2) upset; (3) a bit upset; (4)
not upset. The fourth question assessed response generation: "What are you going to do
or say?" The subject was then asked to think of other ways of responding to the situation.
Responses were coded by research assistants for number and for quality as
"pro-social/assertive," "antisocial/aggressive," and "submissive/passive."

The subject was then presented with 3 solutions enacted by the protagonist. Each
solution was followed by questions to assess response evaluation and self-efficacy
evaluation. The first question referred to evaluation based on moral values: "Was this a
good way for the child to respond?" The subject chose 1 of 4 answers: (1) very good; (2)
good; (3) not so good; (4) not good at all. The second question referred to the subject's
confidence that he would be able to behave in this way (self-efficacy): "Would it be
difficult or easy for you to respond in this way?" The subject chose 1 of 4 answers: (1)
very easy; (2) easy; (3) difficult; (4) very difficult.

Finally, after 3 responses had been presented and related questions had been answered,
the 3 videotaped responses were presented again one after the other and the subject
was asked: "Which of the 3 responses would you choose?" Thus response selection was
assessed.

4
Scoring and Reliability of Measures. From these questions, 15 measures were generated
as assessments of boys' social problem-solving. Boys' responses to the individual
questions (described above) were summed across the 5 Being Disadvantaged problem
situations, the 4 Coping With Competition problem situations, and the 3 Social
Expectations problem situations. This was permissible because the correlation
coefficients between the scores of the 15 measures of the various problem situations
within the corresponding problem domain (Being Disadvantaged, Coping With
Competition, Social Expectations) were all significant and most were 0.50 or more.

For the 2 measures requiring a judgment on the part of the interviewer, reliability of
scoring was assessed in 15 subjects. Responses were scored by 2 research assistants
who were blind to the subjects' diagnosis. The [Kappa] for the quality of encoded cues
(whether an inference was made or not) [TABULAR DATA FOR TABLE 2 OMITTED] was
0.72, and the [Kappa] for the quality of responses (prosocial/assertive,
antisocial/aggressive, submissive/passive) was 0.87.

RESULTS

First, we checked whether the data supported the domain specificity of social problem-
solving. Therefore, the social problem-solving variables (45) were subjected to a
multivariate analysis of covariance (MANCOVA) with psychiatric diagnosis (5) and
problem domain (3) as independent variables, and age and Verbal IQ as covariates, since
groups differed in age and IQ (Table 1); we chose Verbal IQ and not Full Scale IQ as the
covariate because of the verbal character of the assessment method. There was an
interaction effect between diagnosis and problem domain ([F.sub.120,514] = 1.4, p [less
than] .01). Further analyses, therefore, were conducted separately for the 3 problem
domains.

Second, in each problem domain social problem-solving variables (15) were subjected to
a MANCOVA with diagnosis (5) as independent variable and age and Verbal IQ as
covariates. In all 3 problem domains there was a multivariate main effect of diagnosis
(Being Disadvantaged: [F.sub.60,566] = 2.24, p [less than] .001; Coping With
Competition: [F.sub.60,566] = 1.71, p [less than] .001; Social Expectations:
[F.sub.60,566] = 1.73, p [less than] .001). There were no interactions between diagnosis
on the one hand and age and Verbal IQ on the other hand in all 3 problem domains; the
effect of age and Verbal IQ on social problem-solving measures thus did not differ among
diagnostic groups. To identify the source of an overall MANCOVA diagnosis effect, pair-
wise comparisons were made between, on the one hand, the ODD/CD group, the ADHD
group, and the ODD/CD+ADHD group, and, on the other hand, the NC group (Table 2). In
addition, to examine whether possible differences between the disruptive behavior
disorder groups and the NC group were specific, pairwise comparisons were made
between the disruptive behavior disorder groups and the INT group (Table 2). In all
problem domains the ODD/CD, the ADHD, and the ODD/CD+ADHD group differed
significantly from the NC group. However, these groups differed less frequently from the
INT group.

Third, when groups differed significantly, follow-up ANCOVAs were conducted (see Tables
3-5 for means and SD of the 15 measures for the 5 groups in the 3 problem domains).
The ADHD group only differed from the NC group in encoding significantly fewer cues in
all problem domains and in generating significantly fewer responses in 2 problem
domains. By contrast, not only did the ODD/CD and ODD/CD+ADHD groups in all problem
domains encode significantly fewer cues and in one problem domain generate
significantly fewer responses than the NC group, but the ODD/CD and ODD/CD+ADHD
groups also often differed from the NC group in self-efficacy evaluation and in response
selection. For example, in all problem domains both ODD/CD and ODD/CD+ADHD boys
felt significantly more confident than NC boys in their ability to enact an aggressive
response. Also, compared with the NC group, the ODD/CD group significantly more often
selected an aggressive response in 2 problem domains and the ODD/CD+ADHD group
significantly more often selected an aggressive response in one problem domain.

5
Because the diagnostic groups differed in educational level, a post hoc analysis was
conducted for each problem domain. Multivariate analyses of variance were conducted
with the social problem-solving measures as dependent variables, with diagnosis and
educational level as independent variables, and with age and Verbal IQ as covariates.
There was no interaction effect between diagnosis and educational level ([F.sub.120,994]
= 0.94, p = .56). The effect of educational level on social problem-solving measures thus
did not differ among diagnostic groups.

DISCUSSION

Studies of aggressive children's social problem-solving have failed to differentiate various

subgroups among these children. Specifically, it is highly probable that the children
involved in those studies consist not only of children with persistent antisocial behavior
but also of children with hyperactivity, impulsivity, and attention problems. In this study,
social problem-solving skills were investigated in ODD/CD, ADHD, and ODD/CD+ADHD
boys, with NC boys and INT boys as control groups. Social problem-solving was studied
with respect to 3 peer-related social problem domains: Being Disadvantaged, Coping With
Competition, and Social Expectations. In accordance with the theory (Crick and Dodge,
1994), the data of our study supported the domain (situation) specificity of social
problem-solving.

First, when the social problem-solving process was taken as a whole, the 3 disruptive
behavior disorder groups differed from the NC group in all 3 problem domains. In
addition, both the ODD/CD group and the ODD/CD+ADHD group differed from the INT
group in the problem domain Being Disadvantaged. Thus, differences between the
ODD/CD group or the ODD/CD+ADHD group and the NC group were specific in the social
problem domain that has been studied the most thoroughly up to now. However, the 3
disruptive behavior disorder groups often did not differ from the INT group in the 2 other
social problem domains.

Second, when the various problem-solving skills were considered separately, the groups
had certain systematic differences in social problem-solving skills. ADHD boys' social
problem-solving was affected only with respect to encoding cues and response
generation. As expected, compared with the NC group the ADHD group encoded fewer
cues in all 3 problem domains. This may have been due to ADHD boys' attention deficits
or to their impulsivity. However, in all 3 problem domains ODD/CD and ODD/CD+ADHD
boys also encoded fewer cues than NC boys. Thus, encoding fewer cues than NC boys
seems not to be a characteristic of ADHD but seems to be a characteristic of disruptive
behavior disorders. Other studies have demonstrated this deficiency in aggressive boys
(Dodge and Newman, 1981) and aggressive/hyperactive boys (Milich and Dodge, 1984).
Furthermore, the ADHD group, as expected, generated fewer responses than the NC
group in 2 problem domains. The fact that ADHD children generate fewer responses than
normal children when they are asked to give as many responses as possible may be a
manifestation of ADHD boys' impulsivity. However, in one problem domain both the
ODD/CD group and the ODD/CD+ADHD group also generated fewer responses. Thus, as
with the case of encoding, generating fewer responses than do normal children seems
not to be specific for ADHD. In all other problem-solving [TABULAR DATA FOR TABLE 3
OMITTED] [TABULAR DATA FOR TABLE 4 OMITTED] [TABULAR DATA FOR TABLE 5
OMITTED] skills the ADHD group did not differ from the NC group. In conclusion, ADHD
boys' social problem-solving was affected only with respect to encoding cues and
response generation, 2 skills that were also affected in the ODD/CD and ODD/CD+ADHD
groups.

By contrast, social problem-solving in the ODD/CD and ODD/CD+ADHD groups was

affected not only with respect to encoding cues and response generation but also with
respect to self-efficacy and response selection. Not only did ODD/CD and ODD/CD+ADHD
boys in all problem domains encode fewer cues and in one problem domain generate
fewer responses than normal controls, but in all problem domains ODD/CD and
ODD/CD+ADHD boys were more confident than normal controls in their ability to enact

6
an aggressive response (self-efficacy). Other studies have shown that aggressive children
feel more confident about behaving aggressively than control children (Erdly and Asher,
1996; Perry et al., 1986). Furthermore, the ODD/CD group when compared with the NC
group more often selected an aggressive response in 2 problem domains, and the
ODD/CD+ADHD group when compared with the NC group more often selected an
aggressive response in one problem domain. Also, in one problem domain, when
compared with the NC group both the ODD/CD group and the ODD/CD+ADHD group less
often selected a prosocial response. Thus, when ODD/CD and ODD/CD+ADHD boys were
given the opportunity to select a response among various responses shown, these boys,
when compared with normal controls, more often selected an aggressive response and
less often selected a prosocial response. To our knowledge, these characteristics in
response selection have never before been demonstrated in aggressive school-age
children. In conclusion, social problem-solving in the ODD/CD and ODD/CD+ADHD groups
was affected throughout the process.

Finally, when the various social problem-solving skills were considered in the 3 problem
domains, the ODD/CD+ADHD group differed more often from the NC group (17 times)
than did the ODD/CD group (12 times) or the ADHD group (6 times). This is in line with
research that demonstrates that the functioning of the comorbid group is more seriously
affected than either the ODD/CD or the ADHD group (Moffit, 1993; Taylor et al., 1991;
Walker et al., 1987).

Contrary to expectations, we did not find differences in interpretation between the

groups. The reason might be that in most studies of hostile attributional biases open
questions are used, whereas in our computerized assessment method the question
regarding interpretation was in a multiple-choice format. In the Milich and Dodge (1984)
study the hyperactive/aggressive group differed from the NC group in interpretation
(assessing hostile attributional biases) when asked in an open-ended manner to give an
explanation for the peer's behavior, but groups did not differ significantly when asked a
similar question in a forced-choice format.

Also, we did not find differences between the groups in recognizing that the situation was
problematic. To our knowledge, problem recognition has not been studied empirically.
Problem recognition needs to be distinguished from problem identification. Problem
recognition precedes problem identification and consists of becoming aware whether
there is a problem, whereas problem identification consists of defining the problem.
Problem recognition seems difficult to investigate. In fact, a direct question that examines
whether the subject recognizes the situation as problematic is inadequate, the issue
being whether the child spontaneously perceives the situation as problematic. Therefore,
we used the more indirect way of investigating problem perception ("Would you be upset
when/that. . . ?") but did not find differences between the groups. Further study of
problem recognition in ODD and CD children is desired because these children tend to
deny the existence of problems, and a sensitivity to problems is an important prerequisite
for effective problem-solving (d'Zurilla, 1986): it sets in motion adequate (reflective)
problem-solving activity. In other words, without problem recognition there can be no
reflective problem-solving.

It should be acknowledged that different diagnostic methods were used for children from
the outpatient clinic and for children from special schools; in future studies we shall use
similar diagnostic procedures. Similarly, no disruptive behavior group with comorbid
anxiety or mood disorders was involved, whereas high rates of comorbidity with these
disorders in, for example, ADHD have been found (Biederman et al., 1998). In future
studies it will be interesting to involve disruptive behavior groups with comorbid anxiety
and mood disorders.

Clinical Implications

7
This study demonstrated that in ADHD boys, social problem-solving was affected only in
encoding and in the generation of responses, whereas in ODD/CD and ODD/CD+ADHD
boys, social problem-solving was also affected with respect to self-efficacy and response
selection; also, social problem-solving was more deficient in the comorbid group than in
either the ODD/CD or the ADHD group. Thus, not all children with disruptive behavior
disorders (aggression) are alike in the social problem-solving deficits they exhibit.
Therefore, in the training of these children's social problem-solving skills (Kazdin et al.,
1987; Lochman et al., 1987; Matthys, 1997) one needs to focus on different skills
depending on the child's psychiatric disorder.

The authors thank Drs. H. t'Hart and P. Westers for their statistical advice, Ms. E. Scholte
for her assistance with statistical analysis, and Dr. J. Lochman for his comments on a draft
of this article.

REFERENCES

Abikoff H, Gittelman R, Klein D (1980), A classroom code for hyperactive children: a

replication of validity. J Consult Clin Psychol 48:555-565

Achenbach TM (1991a), Manual for the Child Behavior Checklist/4-18 and 1991 Profile.
Burlington: University of Vermont Department of Psychiatry

Achenbach TM (1991b), Manual for the Teacher'[s Report Form and 1991 Profile.
Burlington: University of Vermont Department of Psychiatry

American Psychiatric Association (1987), Diagnostic and Statistical Manual of Mental

Disorders, 3rd edition-revised (DSM-III-R). Washington, DC: American Psychiatric
Association

Biederman J, Faraone SV, Taylor A, Sienna M, Williamson S, Fine C (1998), Diagnostic

continuity between child and adolescent ADHD: findings from a longitudinal clinical
sample. J Am Acad Child Adolesc Psychiatry 37:305-313

Coie JD, Dodge KA, Kupersmidt JB (1990), Peer group behavior and social status. In: Peer
Rejection in Childhood, Asher SR, Coie JD, eds. Cambridge, England: Cambridge University
Press, pp 17-59

Crick NR, Dodge KA (1994), A review and reformulation of social information-processing

mechanisms in children's social adjustment. Psychol Bull 115:74-101

Cuperus JM (1997), Sociale probleemoplossing bij kinderen met gedragsstoornissen. PhD

thesis, University Utrecht, The Netherlands

Deluty R (1983), Children's evaluations of aggressive, assertive and

submissive responses. J Clin Child Psychol 12:124-129

Dodge KA (1980), Social cognition and children's aggressive behavior. Child Dev 51:162-
170

Dodge KA (1986), A social information processing model of social competence in children.

In: Minnesota Symposium on Child Psychology, Perlmutter M, ed. Hillsdale, NJ: Erlbaum,
pp 77-125

Dodge KA (1993), Social-cognitive mechanisms in the development of conduct disorder

and depression. Annu Rev Psychol 44:559-584

8
Dodge KA, McClaskey C, Feldman E (1985), Situational approach to the assessment of
social competence in children. J Consult Clin Psychol 53:344-353

Dodge KA, Newman J (1981), Biased decision making processes in aggressive boys. J
Abnorm Child Psychol 9:375-379

Dodge KA, Pettit G, McClaskey C, Brown M (1986), Social competence in children. Monogr
Soc Res Child Dev 58:213-251

d'Zurilla TJ (1986), Problem-Solving Therapy: A Social Competence Approach to Clinical

Intervention. New York: Springer

Erdly CA, Asher SR (1996), Children's social goals and self-efficacy perceptions as
influences on their responses to ambiguous provocation. Child Dev 67:1329-1344

Goldfried MR, d'Zurilla TJ (1969), A behavioral-analytic model for assessing competence.

In: Current Topics in Clinical and Community Psychology, Vol 1, Spielberger C, ed. New
York: Academic Press, pp 151-196

Hinshaw SP, Lahey BJ, Hart EL (1993), Issues of taxonomy and comorbidity in the
development of conduct disorder. Dev Psychopathol 5:31-49

Hodges K, Kline J, Stern L, Cytryn L, McKnew D (1982), The development of a child

assessment interview for research and clinical use. J Abnorm Child Psychol 10:173-189

Kazdin A (1995), Conduct Disorders in Childhood and Adolescence, 2nd ed. Thousand
Oaks, CA: Sage

Kazdin A (1997), Practitioner review: psychosocial treatment for conduct disorder in

children. J Child Psychol Psychiatry 38:161-178

Kazdin A, Esveldt-Dawson K, French N, Unis A (1987), Problem-solving skills training and

relationship therapy in the treatment of antisocial child behavior. J Consult Clin Psychol
55:76-85

Lochman J (1987), Self and peer perceptions and attributional biases of aggressive and
nonaggressive boys. J Consult Clin Psychol 55:404-410

Lochman J, Lampron L, Gemner T, Harris S (1987), Anger coping intervention with

aggressive children: a guide to implementation in school settings. In: Innovations in
Clinical Practice: A Source Book, Vol 6, Keller P, Heyman S, eds. Sarasota, FL: Professional
Resource Exchange, pp 339-356

Lochman J, Meyer B, Rabiner D, White K (1991), Parameters influencing social problem-

solving of aggressive children. In: Advances in Behavioral Assessment of Children and
Families, Vol 5, Prinz R, ed. Greenwich, CT: JAI Press, pp 31-63

Matthys W (1997), Residential behavior therapy for children with conduct disorders.
Behav Modif 21:512-532

Matthys W, Van Engeland H (1992), Social-cognitive research in the aggressive child: new
perspectives. Eur Child Adolesc Psychiatry 1:178-185

Milich R, Dodge KA (1984), Social information processing in child psychiatric populations. J

Abnorm Child Psychol 12:471-490

9
Moffit TE (1993), The neuropsychology of conduct disorder. Dev Psychopathol 5:135-151

Perry D, Perry L, Rasmussen P (1986), Cognitive social learning mediators of aggression.

Child Dev 57:700-711

Richard B, Dodge KA (1982), Social maladjustment and problem solving in school-aged

children. J Consult Clin Psychol 50:226-233

Taylor E, Sandberg S, Thorly G, Giles S (1991), The Epidemiology of Childhood

Hyperactivity (Maudsley Monographs No. 33). Oxford, England: Oxford University Press

Van Westerlaak JM, Kropman JA, Collaris JW (1976), Beroepenklapper. Nijmegen, The
Netherlands: Instituut voor Toegepaste Sociologie

Verhulst FC, Berden GFMG, Sanders-Woudstra JAR (1985), Mental health in Dutch
children, II: the prevalence of psychiatric disorder and relationship between measures.
Acta Psychiatr Scand Suppl 324:1-45

Walker JL, Lahey BB, Hynd GW, Frame CL (1987), Comparison of specific patterns of
antisocial behavior in children with conduct disorder with or without coexisting
hyperactivity. J Consult Clin Psychol 55:910-913

Wechsler D (1974), Manual for the Wechsler Intelligence Scale for Children-Revised. New
York: Psychological Corporation

Drs. Matthys and Van Engeland are with the Department of Child and Adolescent
Psychiatry, University Hospital Utrecht, and the Rudolf Magnus Institute of Neurosciences,
Utrecht University, Utrecht, The Netherlands. Dr. Cuperus is with the Institute for Children
With Auditory and Communicative Handicaps, St. Marie, Eindhoven, The Netherlands.

COPYRIGHT 1999 Lippincott/Williams & Wilkins

COPYRIGHT 2000 Gale Group

Cognitive-Behavioral Predictors of Asthma Morbidity in Inner-City Children.

Author/s: Shari L. Wade

Issue: Oct, 2000

ABSTRACT. Asthma is a growing health problem among children in the United States,
particularly in urban, inner-city areas. This article examines the relationship between
cognitive-behavioral aspects of asthma management (caretaker asthma knowledge,
expectations, and problem-solving) and asthma morbidity in a sample of 1376 inner-city
children with physician-diagnosed asthma. In the analyses, baseline symptom severity
served as a covariate, and the average of the 3-, 6-, and 9-month follow-up data served
as the outcome measure. Children of caregivers with ineffective problem-solving
strategies had significantly more days of wheezing over a 14-day period. Ineffective
problem-solving capabilities were also associated with poorer functional status; however,
positive caregiver expectations were associated with better functional status. Of the
cognitive-behavioral factors studied in a high-risk urban population, caregiver problem-
solving skills and expectations emerged as meriting further investigation and possible
intervention. J Dev Behav Pediatr 21:340-346, 2000. Index terms: asthma, problem
solving, expectations.

Asthma is a serious and growing health problem in the United States.[1] It affects 4.8
million children and is the leading cause of school absences.[2] Between 1980 and 1993,
the hospitalization rate for asthma among individuals between the ages of 0 and 24

10
 increased by 28%.[1] Furthermore, African-American children were significantly more
likely to be hospitalized than white children. The National Cooperative Inner-City Asthma
Study (NCICAS) was undertaken in response to this growing health problem.

NCICAS focused on three broad areas--allergens and airway irritants in the home
environment, access to medical care, and psychosocial factors--that were seen as
potential contributors to asthma morbidity in the inner city.[3-5] This article examines
three cognitive-behavioral factors believed to affect the management of childhood
asthma: knowledge of the illness, practical problem-solving skills, and caretaker
expectations regarding asthma management.

Much research on asthma management has focused on the caregiver's knowledge about
asthma as an illness.[6,7] Knowledge of asthma has been shown to have a modest, yet
statistically significant, association with asthma morbidity.[3] Furthermore, the level and
accuracy of knowledge about asthma have been targets for intervention in most pediatric
asthma self-management programs.[8,9] Although these programs have documented
success in improving knowledge of asthma, they have been much less successful in
controlling asthma symptoms or reducing health care utilization for acute episodes.[7-10]
Thus, although a factual understanding of asthma is important, it is not clear that such
information translates into decreased morbidity.

In recent years, psychological research on adherence to health regimens

in chronic diseases has moved beyond knowledge-focused interventions
to include personal capabilities such as personal resourcefulness,
flexibility, and creativity in confronting the day-to-day challenges of
illness management. Within this framework, interventions have
increasingly targeted practical problem-solving skills, including the
ability to anticipate problematic illness management situations, control
one's emotional reactions to these situations, and devise effective
responses. Newer approaches to health behavior change, such as the
Social Action Theory, emphasize that simply altering knowledge about an
illness may not produce long-term change if an intervention fails to build
practical problem-solving skills and positive personal expectations
regarding one's ability to use these skills effectively.[11]

The caretaker's expectations regarding asthma management include beliefs regarding the
efficacy of recommended treatments, confidence in one's ability to carry them out
successfully, and the perceived difficulty of doing so (e.g., expense, inconvenience), all of
which have been predicted to influence adherence to medical recommendations.[12-19] In
the case of childhood asthma, however, the primary adult caregiver (typically a parent)
often shares the management with the child or with other caretakers. The degree to which
management activities axe shared is likely to affect the impact of caretaker knowledge,
beliefs, and problem-solving skills on the child's asthma morbidity.

Of these three factors, practical resourcefulness in solving day-to-day management

problems (i.e., applied problem-solving skills) may be especially important. Preliminary
descriptive analyses of the NCICAS data revealed high levels of asthma knowledge and
relatively positive expectations regarding their ability to prevent and manage asthma
symptoms effectively among the adult caregivers interviewed.[5] However, these inner-city
caregivers were able to generate surprisingly few problem-solving strategies in response to
typical scenarios of asthma problems, suggesting that translating knowledge into effective
management strategies may be an area of difficulty in this sample.[5]

The aim of this study was to build on our previous findings by examining the relationship of
cognitive-behavioral aspects of asthma management (asthma knowledge, problem-solving,

11
and caretaker expectations) to asthma morbidity in the NCICAS sample of inner-city children
with asthma. We hypothesized that high levels of asthma knowledge, effective problem-
solving skills, and positive caretaker expectations would be related to lower levels of asthma
morbidity.

METHODS

Participants

Families of children aged 4 to 9 years with asthma were recruited from emergency
departments or outpatient clinics in the following 8 metropolitan areas in the Northeastern
and Midwestern United States: Baltimore, Maryland; Bronx, New York; Chicago, Illinois;
Cleveland, Ohio; Detroit, Michigan; New York, New York; St. Louis, Missouri; and Washington,
D.C. Recruitment criteria included a diagnosis of asthma and the presence of asthma
symptoms for more than 3 days during the previous 12 months or cough, wheezing, or
shortness of breath lasting for more than 6 weeks during the previous 12 months in the
absence of a diagnosis of asthma. A total of 2385 children were identified as eligible for the
study, and 2143 (90%) consented to participate. Families who consented to participate
reported a greater number of emergency visits for asthma during the previous year (3.5 vs
2.8) and a lower proportion of caregivers with Spanish as a primary language than those
who declined. Of those consenting to participate, 1528 (71%) completed the baseline
interview. Among families who consented to participate, there were no differences in key
illness and demographic factors between those who completed the baseline interview and
those who did not. Ten percent (152 patients) of these children had not been diagnosed
with asthma and therefore were not asked to complete baseline measures pertaining to
asthma management.

The current investigation focuses on participants with diagnosed asthma who completed
the baseline interview and two or more follow-up interviews. Because of missing data on
individual measures, the sample sizes for the analyses reported range from 1057 to 1283.
However, there were no significant differences in baseline demographic, cognitive-
behavioral, and morbidity measures between the 1376 children with asthma who
completed the baseline interview and the 1057 children for whom complete follow-up
data were available.

The demographic characteristics of the sample are presented in Table 1. The children
were 6.18 years of age on average (range, 4-9 yr; SD = 1.69). African-Americans and
Puerto Ricans comprised 67% and 14.9% of the sample, respectively. The caretakers
were predominantly female (96.5%) and unmarried (77.3%). The majority of the sample
(60.9%) had family incomes lower than $15,000 a year. Nearly one third of the children in
the sample had been hospitalized for asthma during the previous year.

12
 Table 1. Sample Characteristics

%(a)

Male child 62.8

Race of child
African-American 67.0
Other black 6.8
Puerto Rican 14.9
Other Hispanic 5.0
Other (mixed, white) 6.3
Income <$15,000 60.5
Female caregiver 96.5
Caregiver not married 77.3
Hospitalized in past year 31.4
2+ ED visits in past year 67.3

Mean SD

Age of child (yr) 6.18 1.69

AIQ 0.85 0.09
PS-I 0.64 0.79
ARI (adult) 2.75 0.32
CE 7.26 O.92

SD, standard deviation; ED, emergency department; AIQ, percentage correct from the
Asthma Information Quiz; PS-I, number of ineffective problem-solving strategies; ARI
(adult), average responsibility for asthma care assumed by the primary adult caregiver;
CE, overall level of caretaker expectation.

(a) Sample sizes ranged from 1213 to 1304 because of incomplete data on individual
questionnaires.

Procedures

The study was approved by the institutional review boards of the eight
academic centers. Informed consent was obtained during recruitment,
and the baseline interview was scheduled to occur 2 to 4 weeks later.
This 2.5-hour interview focused on medical history, asthma symptoms,
health care utilization, adherence, access to health care, home
environment, and demographics, in addition to the psychosocial factors.
Interviewees were screened regarding their literacy and, based on this,
were either read the entire interview or allowed the option of self-
administering the psychosocial measures that had been developed for a
self-report format. Hispanic families were given the option of having the
interview and questionnaires administered in either Spanish or English
by a bilingual interviewer. Each family was contacted subsequently by
phone at 3, 6, and 9 months after baseline to assess asthma symptoms
and utilization. The study methodology is described in detail by Mitchell
et al.[4]

Measures

13
Asthma Information Quiz. The Asthma Information Quiz (AIQ) is a questionnaire with 23 tree-
false items assessing knowledge of asthma symptoms, triggers, medication management,
and prevention. Items were selected from existing measures of asthma knowledge (Air
Power, Open Airways), with additional items created by the National Cooperative Inner City
Asthma Study (NCICAS) staff. A preliminary list of items was reviewed by five staff
physicians at each site with the purpose of eliminating items that were ambiguous or of
questionable relevance.[5] "Coughing is often a symptom of asthma" is an example of an
item assessing knowledge of asthma symptoms. The AIQ score is the proportion of correct
responses on the questionnaire. Scores ranged from .32 to 1.00 (mean = .85), with higher
scores indicating greater knowledge about asthma.

Caretaker's Expectations Regarding Asthma Management. The Caretaker's Expectations

Regarding Asthma Management (CE) measure is a 15-item scale developed by NCICAS to
assess the caretaker's expectations across several domains of asthma management
behaviors. We have reported the details of the development and initial psychometric testing
of the CE elsewhere.[20] In brief, asthma experts were surveyed to identify key asthma
management behaviors. Through this process, five asthma management behaviors were
identified: (1) keeping medical appointments for preventive asthma care; (2) protecting the
child from exposure to asthma triggers; (3) following ongoing treatment procedures
precisely; (4) recognizing impending attacks early; and (5) following treatment procedures
precisely during an attack. For each domain, an item was developed to assess outcome
expectations, response difficulty, and self-efficacy, for a total of 15 items.[20] For example,
after a description of a typical medical appointment for asthma, the item "how much do you
think it will help [child]'s asthma to bring him/her to these medical appointments?" was used
to assess outcome expectations in the domain of appointment keeping. Responses to the CE
items were rated on a 9-point scale (1 = not at all; 5 = somewhat; 9 = extremely). Subscale
scores were created using the mean item rating for the 5 items in the subscale. Sequential
pilot tests included initial pretests, concurrent and retrospective think-aloud interviews,
[21,22] and final pretests, with revisions at each step. In analyses based on the NCICAS
sample, we obtained a Cronbach's alpha of .75 for the total score. Findings supporting a
series of hypothesized relationships provided evidence of construct validity. Preliminary
analyses indicated that effect sizes for the three subscale scores and the total score were
similar. Therefore, to reduce the number of predictors in the model, the CE total score
(mean across all 15 items) was used in the analyses below. Scores ranged from 4.2 to 9.0,
with higher scores indicating more positive expectations.

Problem Solving in Situations Involving Asthma Management. The NCICAS developed an

interviewer-administered measure of caregiver problem-solving regarding the child's
asthma management based on a methodology previously developed by one of the
investigators (Problem Solving in Situations Involving Asthma Management [PS]).[23]
Participants were presented with five scenarios about a parent and child facing a problem
with the child's asthma that was ultimately resolved successfully. The respondent was then
asked to fill in the middle portion of the story by describing all of the actions taken by the
parent to handle the situation. Thus, it was possible for respondents to generate multiple
responses for each scenario. The hypothetical problem situations in the stories were
adapted from "Twenty Problems" from the "Living with Asthma" education program.[24]
Problem-solving scenarios were piloted in an initial sample of 93 caregivers of children with
asthma. Each action reported in response to the problem-solving scenarios was reviewed by
33 experienced asthma clinicians who sorted responses into three classes: "effective,"
"ineffective," and "detrimental." These categories were further refined to include only
responses that could be categorized with [is greater than or equal to] 80% agreement.
Raters were trained to a criterion of [is greater than or equal to] 90% agreement, which was
maintained through regular monitoring and retraining as data collection proceeded. The
correlations among summary scores for effective, ineffective, and detrimental responses
were as follows: effective with ineffective = -.02; effective with detrimental = -.37;
ineffective with detrimental = -.03, with only the correlation between effective and
detrimental problem-solving achieving statistical significance. Spearman correlation
analyses indicated that only ineffective problem-solving responses were correlated with

14
asthma morbidity at the p [is less than] .05 level. Thus, the current analyses used the total
number of ineffective problem-solving responses in the multivariate models. Scores ranged
from 0 to 4, with higher scores indicating less effective problem-solving.

Asthma Responsibility Interviews for Adult Caretakers. NCICAS developed the Asthma
Responsibility Interviews for Adult Caretakers (ARIA) to examine the division of
responsibility for asthma management tasks between the parent and child and among
multiple adult caretakers.[25] Pilot questions pertaining to asthma responsibility were
developed by asthma experts on the basis of asthma self-management and caretaker
competencies identified previously.[26,27] Item selection and wording were refined after
pilot interviews with 40 adults and 40 children. The final ARIA consisted of an initial
question asking the adult caregiver to identify all of the individuals who helped care for
the child's asthma. This item was followed by nine questions pertaining to the level of
responsibility assumed by each caregiver for management of the child's asthma in
varying domains. The nine-item ARIA had a Cronbach's alpha of .69 for the primary
caregiver and .79 for the child as caregiver.[25] Findings supporting a series of
hypothesized relationships provided preliminary evidence of construct validity. In the
current investigation, the average responsibility of the primary caregiver and the total
number of caregivers served as covariates in the multivariate analyses to control for the
degree of involvement of the primary adult caregiver in the child's asthma management.

Measures of Morbidity. Two measures of asthma symptoms were considered as outcomes

in the current investigation: days of wheezing and functional status. To assess wheezing,
caregivers were asked, at each assessment, to recall the number of days during the past
2 weeks that the child experienced wheezing, coughing, or shortness of breath.[3] The 2-
week time frame was chosen on the basis of a review of literature regarding the stability
of recall over various lengths of time. A calendar depicting the 2-week period in question
was also used to facilitate accurate responding. This measure of symptoms correlated .9
with an aggregate measure of visible symptoms, disruptions in sleep, and disruptions in
daily activities because of asthma. Functional status was assessed using a modified
version of the Functional Status II (Revised) [FSII(R)], 14-item version.[28] The FSII(R) was
developed to provide a measure of health status in children and was normed on a sample
of 732 healthy and ill children. Items on the FSII(R) assess sleeping, eating, energy level,
and mood over a 2-week time frame. For example, one item asked "did the child seem
lively and energetic?" The FSII(R) was shown to have adequate internal consistency
(Cronbach's [Alpha] [is greater than] .80), and evidence was supportive of construct,
content, and discriminant validity. In the current study, the FSII(R) was modified such that
the caregiver was asked to rate each of the 14 behaviors on a 5-point scale (0 = never, 1
= rarely, 2 = some of the time, 3 = almost always, 4 = always) instead of the 3-point
scale used in the standardized version. This modification was made to facilitate
documentation of the economic impact of pediatric asthma in the inner city. As in the
original FSII(R), responses that indicated a potential problem were further queried to
determine whether they were due fully, partly, or not at all to the child's asthma. For both
the FSII(R) and the days with wheezing outcome measures, we used an average of the 3-,
6-, and 9-month follow-up assessments to control for the effects of seasonal
exacerbations in asthma symptoms, thereby providing a more stable and reliable
measure of morbidity

Analyses

Regression analyses were used to examine the relationships between the three cognitive-
behavioral factors and two outcome measures. To characterize the relationship between
the cognitive-behavioral factors and the asthma outcomes, we dichotomized the
cognitive-behavioral factor scores at their medians into low and high groups (AIQ = .87;
PS-I = 1; CE = 7.3). The decision to dichotomize the predictor variables rather than to
examine them as continuous measures was made to facilitate the interpretability of
findings and had virtually no impact on the proportion of variance explained by the
predictors. Each of the three cognitive-behavioral factors was first entered separately into

15
the regression models to assess its individual effect on asthma morbidity after controlling
for baseline levels of asthma morbidity. The level of symptoms (or functional status) at
the baseline assessment was included as a measure of asthma severity. In this way, we
sought to examine the relationship between cognitive-behavioral factors and asthma
morbidity after taking into account the severity of symptoms at the baseline assessment.

In the second set of regression analyses, the three cognitive-behavioral factors were
entered together with standard demographic variables (child's age, gender, ethnicity),
recruitment site, number of caretakers, number of medications, and average
responsibility level of the primary caregiver to assess their relationship to asthma
morbidity after accounting for the effects of the other variables. Baseline levels of the
outcome variable of interest were included as covariates in these analyses as well to
control for the effects of the severity of symptoms at baseline. The number of
medications also served as a proxy measure for asthma severity in these multivariate
analyses. Square-root transformations were used to normalize the distributions of
wheezing and functional status scores. To provide a readily interpretable measure of the
effects of the cognitive-behavioral factors on asthma morbidity, results are presented as
difference scores in wheezing and functional status between caregivers who scored high
and low on the cognitive-behavioral predictor of interest (e.g., problem-solving). These
difference scores were created by converting the adjusted scores (adjusting for baseline
asthma severity and other potential covariates in the regression model) back into the
untransformed scale using the low-bias estimates approach.[29]

RESULTS

Correlations Among the Cognitive-Behavioral Predictors

The correlations among the cognitive-behavioral predictors were as follows: asthma

knowledge and ineffective problem solving, .02; asthma knowledge and caregiver
expectations,. 11; and ineffective problem-solving and caretaker's expectations, .05. Only
the relationship between knowledge and expectations was significant at the p [is less
than] .05 level. These low intercorrelations suggest that the cognitive-behavioral factors
assessed in this investigation are independent

Regression Models of Individual Cognitive-Behavioral Factors on Asthma Morbidity

Table 2 presents the results of the regression analyses examining the individual effect of
each of the cognitive-behavioral predictors on wheezing and functional status, after
controlling for baseline levels of the severity of asthma symptoms. Ineffective problem-
solving strategies were most consistently related to higher levels of asthma morbidity,
with small associations with the number of days of wheezing (p = .0049) and functional
status (p = .0510). Caretakers reporting more ineffective problem-solving strategies had
children with .4 more days of wheezing on average over a 2-week period and lower
functional status. Caretaker expectations were related to the child's functional status,
with better functional status reported by caretakers having more positive expectations
regarding the child's asthma, although the clinical magnitude of the effect was small.
Caregiver knowledge of asthma was not related to either of the morbidity measures.

Table 2. Regression Analyses of the Cognitive-Behavioral Factors on Wheezing and

Functional Status Adjusting for Baseline Asthma Severity

Wheezing (days/2 wk) Functional Status

AIQ(a)
Difference(b) 0.25 -0.22
CI (-0.05, 0.54) (-1.65, 1.20)
Total [R.sup.2] (%) 7.4 23.3
PS-I(a)
Difference 0.40(**) -1.36(c)

16
 CI (0.12, 0.68) (-2.73, 0.01)
Total [R.sup.2] (%) 8.5 23.3
CE(a)
Difference 0.07 1.69(*)
CI (-0.22, 0.35) (0.28, 3.10)
Total [R.sup.2] (%) 7.8 23.5

CI, 95% asymptotic confidence interval of the difference in the original scale; AIQ, total
percentage correct on the Asthma Information Quiz; PS-I, Number of ineffective problem-
solving strategies; CE, overall level of caretaker expectations.

(a) Scores were dichotomized at the median.

(b) Difference in morbidity in original scale between high and low scores after adjusting
for baseline morbidity.

(c) Statistical significance of p = .051.

(*) Statistical significance of p < .05 testing for the difference in morbidity between high
and low cognitive-behavioral scores using the partial F-test in the regression model.

(**) Statistical significance of p < .01 testing for the difference in morbidity between high
and low cognitive-behavioral scores using the partial F-test in the regression model.

Regression of the Cognitive-Behavioral Factors on Morbidity Adjusting for Severity and

Other Covariates

In the next set of regression analyses, we examined the relationship of each of the
cognitive-behavioral predictors to the two morbidity measures (wheezing and functional
status) after controlling for baseline asthma severity, social and demographic
characteristics, the other cognitive-behavioral predictors, as well as the responsibility of
the primary caregiver for asthma management. As presented in Table 3, ineffective
problem-solving was associated with .37 more days of wheezing on average per 2-week
period (p = .022) and lower levels of functional status (p = .0083). Positive care taker
expectations were associated with better functional status (p = .0003). Although the
effect sizes are relatively small, these findings suggest that ineffective problem-solving
and caretaker's expectations make independent contributions to asthma morbidity even
after accounting for baseline symptom severity, sociodemographic factors, and the other
cognitive-behavioral predictors.

Table 3. Regression Analyses of the Cognitive-Behavioral Factors on Wheezing and

Functional Status, Adjusting for Baseline Severity, Social and Demographic
Characteristics, and Other Cognitive-Behavioral Predictors

Wheezing (days/2 wk) Functional Status

AIQ(a)
Difference(b) 0.18 -.07
CI (-0.14, 0.50) (-1.66, 1.51)
PS-I(a)
Difference 0.37(*) -2.06(**)
CI (0.05, 0.69) (-3.63, -0.50)
CE(a)
Difference -0.08 3.03(**)
CI (-0.40, 0.24) (1.42, 4.64)
Total [R.sup.2] (%) 11.0 28.8

17
Differences were adjusted for baseline morbidity, child's age, gender, ethnicity,
recruitment site, number of caretakers, average responsibility of the adult caregiver,
number of medications, and the other cognitive-behavioral factors (AIQ, PS-I, CE). CI, 95%
asymptotic confidence interval of the difference in the original scale; AIQ, total
percentage correct on the Asthma Information Quiz; PS-I, number of ineffective problem-
solving strategies; CE, overall level of caretaker expectations.

(a) Scores were dichotomized at the median.

(b) Difference in morbidity in original scale between high and low scores after adjusting
for baseline morbidity, child's age, gender, ethnicity, recruitment site, number of
caretakers, average responsibility of the adult caregiver, number of medications, and the
other cognitive-behavioral factors (AIQ, PS-I, CE).

(*) Statistical significance of p < .05 testing for the difference in morbidity between high
and low cognitive-behavioral scores using partial F-test in the regression model.

(**) Statistical significance of p < .01 testing for the difference in morbidity between high
and low cognitive-behavioral scores using partial F-test in the regression model.

DISCUSSION

The goal of the psychosocial battery administered fit the beginning of the National
Cooperative Inner City Asthma Study (NCICAS) was to identify psychosocial factors that
were related to asthma morbidity among inner-city children, providing the basis for an
empirically derived intervention. Our hypotheses regarding cognitive-behavioral
components of asthma management were partially supported in both sets of regression
analyses. Although the relationships were small in magnitude, some have potentially
important clinical relevance. Children of caretakers who provided one or more ineffective
problem-solving responses spent .37 days more wheezing during a 2-week period than
those of caretakers with superior problem-solving skills. The magnitude of this effect can
be contrasted to the NCICAS intervention that resulted in a .5-day reduction in wheezing
per 2-week period.[30] As noted, the measure of wheezing was an average across three
2-week periods distributed throughout 9 months of the year, providing a more
representative measure of morbidity than one based on a single report. Ineffective
problem-solving was also associated with lower levels of functional status. The
caretaker's expectations were a consistent predictor of subsequent caretaker reports of
the child's functional status. Considering that interventions to enhance problem-solving
and alter health expectancies have proven effective in a variety of health promotion
contexts,[11,31] efforts to improve these cognitive-behavioral aspects of asthma
management may offer hope of reducing symptoms and improving functional status in a
substantial number of children.

Of interest, subsequent asthma morbidity was predicted by a tendency to generate many

ineffective solutions to problems, rather than by an inability to generate effective
solutions. Because effective and ineffective problem-solving responses were unrelated to
one another, many individuals provided both effective and ineffective solutions. For such
individuals, their tendency to generate ineffective strategies may dilute or negate the
impact of their effective responses on preventing symptoms. If this is the case, clinical
interventions may need to target and modify ineffective responses to problems, as well
as equip caregivers with new (and more effective) problem-solving strategies. Similarly,
health care providers may need to understand all of the strategies the caregiver may try
when managing the child's symptoms, not merely whether he or she knows the right
response.

The caretaker's expectations regarding his or her ability to manage the child's asthma
were predictive of the child's functional status, suggesting that attitudes as well as
Problem-solving skills play a role in determining child health outcomes in this population.

18
Parents of children with frequent and seemingly unmanageable asthma symptoms over
time may come to have more negative expectations regarding the management of their
child's asthma. Although it is possible that the relationship between expectations and
functional status may merely be an artifact of the relationship between disease severity
and asthma morbidity, we postulate a reciprocal relationship between asthma symptoms
and caretaker expectations (see the article by Bandura[19] on triadic reciprocal
causation). Specifically, severe symptoms may lead to negative expectations regarding
the likely outcome of asthma management activities, with these expectations
contributing, in turn, to subsequent nonadherence and increased symptom activity.
Because we controlled for baseline levels of symptom severity (e.g., functional status),
our data suggest that the relationship between expectations and functional status is not
solely attributable to illness severity. That is, caregiver expectations are predictive of
functional status independent of illness severity. Thus, positive caregiver expectations
may serve as a protective factor for certain children, whereas unfavorable expectations
may place other children at increased risk for asthma morbidity.

The caregiver's knowledge of asthma was unrelated to either measure of asthma

symptoms, suggesting that knowledge, above a certain threshold, is less important than
practical problem-solving skills and expectations regarding their success.[11] These
findings support the notion that asthma interventions must target the attitudes and
practical skills that determine day-to-day asthma management practices rather than
factual knowledge if they are to be successful in reducing morbidity.

The relatively small magnitude of the findings may be attributable to aspects of the study
design as well as the nature of asthma itself. This study focused on the caretaker's
knowledge, expectations, and problem-solving abilities in relation to the child's asthma
morbidity. It is possible that stronger relationships would have emerged, particularly
among the older children, if we had examined the child's own expectations and skills,
rather than the caregiver's.[23] Our data indicate that responsibility for asthma care is
increasingly shared between parent and child as the child gets older Children also spend
more time out of the home as they age with a corresponding increase in their reliance on
self, rather than adult, management. Additionally, children with asthma constitute a
heterogeneous group with respect to the age of onset and putative cause, allergic
sensitivity, access to health care, and exposure to environmental irritants (e.g., cigarette
smoke), among other things. Thus, it is unlikely that any single psychosocial or medical
variable accounts for a large portion of the variance in morbidity. Instead, cognitive-
behavioral asthma management factors may be important predictors for some children,
but they may play a lesser role in the illness expression of others. For example, factors
such as knowledge, expectations, and problem-solving abilities may have limited impact
on the morbidity of children with rare or seasonal symptoms.

Several important caveats should be noted. First, all of the data in this study were
provided via caretaker self-report. Thus, the morbidity measures may have been subject
to distortions in caregiver recall. However, reliance on a 2-week time frame and the use
of multiple morbidity measurements over the course of the year serve to reduce this bias.
Our measures may also have been subject to social desirability biases (i.e., the
caretaker's desire to present herself and the family in a favorable light). Evidence
suggests that Hispanic respondents may be particularly likely to provide socially desirable
and/or extreme responses.[32-34] However, in the current sample, we did not find
differences between the African-American and Hispanic participants on the Caretaker's
Expectations Regarding Asthma Management measure, which required self-appraisal.
Furthermore, the high level of caretaker psychological symptoms reported in the sample
as a whole suggests that participants felt comfortable in acknowledging problems.[5]
Within the context of the current study, we were unable to classify children according to
the National Asthma Education Program guidelines for asthma severity.[35] Instead,
baseline asthma symptoms and the number of asthma medications were used as proxy
measures. Clearly, measures of pulmonary function, such as spirometry data, would have
provided a potentially more objective measure of asthma severity. Another limitation,
given the scope of the current study, was our inability to assess other potentially

19
important cognitive predictors such as the child's problem-solving ability and
expectations regarding asthma management, perceived control over symptoms, or level
of psychological distress. Clearly, large epidemiological studies such as this one must be
balanced by smaller ones that focus on such issues in greater depth. Finally, because this
was a nonrandom sample that was designed to oversample children with severe
symptoms, the representativeness of the sample with respect to families of inner-city
children with asthma is uncertain.[22]

In sum, these findings provide additional new information about a growing health
problem affecting children in the inner city and have potentially useful implications for
clinical practice. Health care providers may improve asthma outcomes for children in the
inner city by identifying and changing ineffective problem-solving strategies among
caregivers. Furthermore, by talking with families about their expectations regarding
managing asthma, it may be possible to identify families at increased risk for
nonadherence as a result of less than optimal expectations. Addressing problem-solving
skills and attitudes represents a departure from traditional, information-focused
approaches. Even though the explanatory power of these variables is modest, these
factors have been shown to be amenable to behavioral interventions. Thus, improving
caretaker problem-solving capabilities and expectations could reduce symptoms and
enhance functional status in a substantial number of children. Future research may serve
to identify which segment of children with asthma is most greatly influenced by cognitive-
behavioral factors.

Acknowledgment. Preparation of the manuscript was supported in part by funds

supported by grants UO1 A1-30751, A1-30752, A1-30756, A1-30772, A1-30773-01, A1-
30777, A1-30779, A1-30780, and N01 A1-15105 from the National Institute of Allergy and
Infectious Diseases (National Institutes of Health, Bethesda, MD).

REFERENCES

[1.] CDC: Asthma--United States: 1989-1992. MMWR Morb Mortal Wkly Rep 43:-955, 1996

[2.] Healthy People: National Health Promotion and Disease Prevention Objectives. United
States Department of Health and Human Services DHS Publication No. (PHS) 91-50212.
Washington, DC, U.S. Government Printing Office, 1991

[3.] Kattan M, Mitchell H, Eggleston P, et al: Characteristics of inner-city children with

asthma: The National Cooperative Inner-City Asthma Study. Pediatr Pulmonol 24:253-262,
1997

[4.] Mitchell H, Senturia Y, Gergen P, et al: Design and methods of the National
Cooperative Inner-City Asthma Study. Pediatr Pulmonol 24:237-252, 1997

[5.] Wade S, Well C, Holden G, et al: Psychosocial characteristics of inner-city children

with asthma: A description of the NCICAS psychosocial protocol. Pediatr Pulmonol 24:263-
276, 1997

[6.] Rubin D, Bauman L, Lauby D: The relationship between knowledge and reported
behavior in childhood asthma. J Dev Behav Pediatr 10:307-312, 1989

[7.] Wilson-Paesano SR, McNabb WL: The role of patient education in the management of
childhood asthma. Prev Med 75:670-687, 1985

[8.] Howland J, Bauchner H, Adair R: The impact of pediatric asthma education on

morbidity. Chest 94:964-969, 1988

20
[9.] Rachelefsky GA: Review of asthma self-management programs. J Allergy Clin
Immunol 80:506-511, 1987

[10.] Clark NM, Feldman CH, Evans D, Levison MJ, Wasilewski Y, Mellins RB: The impact of
health education on frequency and cost of health care use by low income children with
asthma. J Allergy Clin Immunol 78:108-115, 1986

[11.] Ewart CK: Social action theory for a public health psychology. Am Psychol 46:931-
946, 1991

[12.] Holden G: The relationship of self-efficacy appraisals to subsequent health related

outcomes: A meta-analysis. Psychol Reports 66:1044-1046, 1991

[13.] Janz NK, Becker MH: The health belief model: A decade later. Health Educ Q 11:1-47,
1984

[14.] Strecher VJ, DeVellis BM, Becker MH, Rosenstock IM: The role of self-efficacy in
achieving health behavior change. Health Educ Q 13:73-91, 1986

[15.] Maddux JE, Brawley L, Boykin A: Self-efficacy and healthy behavior: Prevention,
promotion and detection, in Maddux JE (ed): Self-Efficacy, Adaptation, and Adjustment:
Theory, Research, and Application. New York, NY, Plenum Press, 1995, pp 173-202

[16.] Bandura A: Self-efficacy: Toward a unifying theory of behavior change. Psychol Rev
84:191-215, 1977

[17.] Bandura A: Social foundations of thought and action: A social cognitive theory.
Englewood Cliffs, NJ, Prentice-Hall, 1986

[18.] Bandura A: Self-efficacy mechanism in psychobiologic functioning, in Schwarzer R

(ed): Self-Efficacy: Thought Control of Action. Washington, Hemisphere Publishing
Corporation, 1992, pp 355-394

[19.] Bandura A: Self-efficacy: The Exercise of Control. New York, NY, W.H. Freeman, 1997

[20.] Holden G, Wade SL, Mitchell H, Ewart C, Islam S: Caretaker expectations and the
management of pediatric asthma in the inner-city: A scale development study. Soc Work
Res 22:51-59, 1998

[21.] Royston PN: Using intensive interviews to evaluate questions, in Fowler FJ (ed):
Health Survey Research Methods. U.S. Department of Health and Human Services
Publication No. 89-3447. Rockville, MD, U.S. Department of Health and Human Services,
1989, pp 3-7

[22.] Willis GB, Royston P, Bercini D: The use of verbal report methods in the
development and testing of survey questionnaires. Appl Cogn Psychol 5:251-267, 1991

[23.] Hanna KJ, Ewart CK, Kwiterovich PO: Child problem solving competence, behavioral
adjustment, and adherence to lipid-lowering diet. Patient Educ Couns 16:119-131, 1990

[24.] U.S. Department of Health and Human Services: Living with Asthma: Manual for
Teaching Parents Self-Management of Childhood Asthma. NIH Publication No. 86-23-64.
Bethesda, MD, National Institutes of Health, 1989

21
[25.] Wade SL, Islam S, Holden G, Kruszon-Moran D, Mitchell H: Division of responsibility
for asthma management tasks between caregivers and children in the inner-city. J Dev
Behav Pediatr 20:93-98, 1999

[26.] McNabb WL, Wilson-Pessano SR, Jacobs AM: Critical self-management competencies
for child with asthma. J Pediatr Psychol 11:103-117, 1986

[27.] Wilson SR, Mitchell JH, Rolnick S, Fish L: Effective and ineffective management
behaviors of parents of infants and young children with asthma. J Pediatr Psychol 18:63-
81, 1993

[28.] Stein PEK, Jessop DJ: Functional status II(R): A measure of child health status. Med
Care 28:1041-1055, 1990

[29.] Miller DM: Reducing transformation bias in curve fitting. Am Statistician 38:124-126,
1984

[30.] Evans R, Gergen P, Mitchell H, et al: A randomized clinical trial to reduce asthma
morbidity among inner-city children: Results of the National Cooperative Inner-City
Asthma Study (NCICAS). J Pediatr 135:332-338, 1999

[31.] Ewart CK: Self-efficacy and recovery from heart attack: Implications for a social
cognitive analysis of exercise and emotion, in Maddux JE (ed): Self-Efficacy, Adaptation,
and Adjustment: Theory, Research, and Application. New York, NY, Plenum Press, 1995,
pp 203-226

[32.] Hui CH, Triandis HC: Effects of culture and response format on extreme response
style. J Cross Cult Psychol 20:296-309, 1989

[33.] Marin G, Gamba R J, Matin BV: Acquiescence and extreme response sets among
Hispanics: The role of acculturation and education. J Cross Cult Psychol 23:498-509, 1992

[34.] Ross CE, Mirowsky J: Socially-desirable response and acquiescence

in a cross-cultural survey of mental health. J Health Soc Behav 25:189-
197, 19

[35.] National Heart, Lung and Blood Institute: Guidelines for the Diagnosis and
Management of Asthma. NIH Publication No. 91-3042. Bethesda, MD, National Institutes
of Health, 1991

SHARI L. WADE, PH.D. Department of Pediatric Rehabilitation, Children's Hospital Medical

Center, Cincinnati, Ohio

GARY HOLDEN, D.S.W. New York University Ehrenkranz School of Social Work, New York,
New York

HENRY LYNN, PH.D. HERMAN MITCHELL, PH.D. Rho Inc., Chapel Hill, North Carolina

CRAIG EWART, PH.D. Psychology Department, Syracuse University, Syracuse, New York

Address for reprints: Shari Wade, Ph.D., Children's Hospital Medical Center, Room 2310,
Pavilion 3333 Burnet Ave., Cincinnati, OH 45229-3039.

COPYRIGHT 2000 Lippincott/Williams & Wilkins

COPYRIGHT 2001 Gale Group

22
Regulation of Negative Affect During Mother-Child Problem-Solving
Interactions: Adolescent Depressive Status and Family Processes.

Author/s: Lisa Sheeber

Issue: Oct, 2000

Lisa Sheeber [1,4]

Nicholas Allen [2]

Betsy Davis [1]

Erik Sorensen [3]

Despite recent suggestions that depression can be conceptualized as a disorder of affect

regulation, relatively little research has focused on affect regulation skills in depressed
individuals. This paper investigated whether depressed adolescents (N = 25) differ from
nondepressed adolescents (N = 25) on two indices of affect regulation (i.e., duration of
negative affective states and reciprocity of maternal negative affect) as well as whether
these indices are related to microsocial family interactional processes. Analyses revealed
that depressed teens differed from their nondepressed peers with regard to duration of
negative affective states but not in their likelihood of reciprocating negative affect.
Additionally, indices of adolescent affect regulation were related to family interactional
processes. Duration of depressive affect was positively associated with maternal display
of facilitative behavior contingent on adolescent depressive behavior. Duration of
aggressive behavior was inversely related to ma ternal problem-solving responses to
aggressive behavior. Finally, adolescent reciprocity of maternal depressive and
aggressive behaviors was strongly associated with mothers' reciprocity of adolescents'
negative affective behavior.

KEY WORDS: adolescent; depression; affect regulation; microsocial observations.

INTRODUCTION

Although unipolar affective disorders are comprised of several classes of symptoms (e.g.,
somatic, cognitive, and affective), depression is primarily an affective disorder, whose
essential feature, according to the DSM-IV (American Psychological Association, 1994), is
sad or depressed affect. A number of researchers, including Gray (1994), Fowles (1988),
Depue and Iacono (1989), and Clark and Watson (1991), have proposed that depression
is associated with heightened activity and sensitivity of aversive emotional systems, and
(more specifically to depression) lowered activity and sensitivity of appetitive emotional
systems. The presence of heightened aversive and reduced appetitive functioning has
been supported in psychometric (Clark & Watson, 1991), psychophysiological (e.g.,
Gatchel, McKinney, & Koebernick, 1977; Lewinsohn, Lobitz, & Wilson, 1973; G. E.
Schwartz, Fair, Salt, Mandel, & Klerman, 1976), and information processing (Henriques,
Glowacki, & Davidson, 1994) studies of depression.

In an attempt to identify the key mechanisms responsible for the abnormal emotional
responding in depressed states, it has been suggested that depression can be
conceptualized as a failure to regulate emotion (Cole & Kaslow, 1988; Gross & Munoz,
1995; Tomarken & Keener, 1998). Emotional regulation has been defined as "an
individual's reflexive reactions and effortful intrapersonal and interpersonal strategies for
altering or maintaining an affective state" (Thompson, 1990, pp. 367-467). Gross and
Munoz (1995) have further noted that emotion regulation occurs in "the context of an
ongoing stream of emotional stimulation and behavioral responding" (p. 153), where
emotional responses, and efforts to regulate them, become antecedents for new, or
escalating, emotional responses. Additionally, though emotion regulation is often
conceptualized as a within-person variable, there is a significant interpersonal component

23
(Gottman, Guralnick, Wilson, Swanson, & Murray, 1997). Dyads or larger groups develop
strategies f or managing negative affect and maintaining constructive engagement within
stressful interpersonal exchanges (Lindahl & Markman, 1990; McDonough, Carlson, &
Cooper, 1994). As such, interactional processes that reflect on the interactants' ability to
avoid negative affective behaviors in the face of others' aversive behavior, may well
serve as important indices of the capacity to regulate negative affects within
relationships.

Researchers working within the emotion regulation model are beginning to examine
whether depressed individuals demonstrate deficits in their ability to manage negative
affect. Based on evidence that depression may involve prefrontal brain dysfunction
related to approach and withdrawal emotional behaviors, Tomarken and Keener (1998)
have predicted that depressed individuals should display longer maintenance (i.e.,
"greater temporal continuity") of negative emotional responses and poorer ability to
maintain temporal continuity of positive emotions. Consistent with predictions based on
evidence that the prefrontal cortex is critical for the regulation of goal-directed behavior
overtime, a few studies using self-report of depressive states have indicated that
depressed adults demonstrate greater temporal continuity of negative affective
responses to naturalistic stressors (e.g., Goplerud & Depue, 1985) as well as to
laboratory-induced depressed mood (Gilboa & Gotlib, 1997). Tomarken and Keener
(1998) note, howev er, that relatively few studies have examined the time course of
affective responses in depressed and nondepressed individuals.

There is preliminary evidence that depressed youth may also have difficulty regulating
negative affect. In particular, it appears that they may have a more limited repertoire of
strategies for regulating affect, use less effective strategies, or fail to use strategies
within their repertoire. Garber and colleagues (Garber, Braafladt, & Weiss, 1995; Garber,
Braafladt, & Zeman, 1991) found that depressed youth reported less frequent use of
adaptive interpersonal and intrapersonal strategies for reducing negative affect (e.g.,
problem-solving, cognitive restructuring), as well as more frequent use of avoidance
responses than did their nondepressed peers. Moreover, they were less likely to believe
that their strategies would be effective. Similarly, adolescents' depressive
symptomatology has been shown to be related to the quality of their responses to
depressed mood, with ruminative responses predicting greater depressive affect and
distracting responses predicting less (J. A. Schwartz & Koenig, 1996).

It is notable, however, that in each of these studies, strategies for managing and
responding to negative affect were examined as factors associated with elevated levels of
depressive symptomatology (e.g., as indicated by self-report measures). Though
pertinent, these studies do not directly address the question of whether depressed youth
have specific deficits in altering and maintaining emotional states. Though many
investigators appear to assume the connection between depression and emotion
regulation deficits, treating depressive symptomatology as evidence for difficulties in
emotion regulation (see for example, Breen & Weinberger, 1995; Cooper, Shaver, &
Collins, 1998), it is unclear that the association has been demonstrated empirically. As
noted by Rutter (1991), assuming this connection and thereby conceptualizing depression
as a problem of affect regulation, confounds the mechanism with the behavior. To the
extent that depressed youth experience deficits in emotional regulation, they should
display greater difficulty transitioning out of negative affective states once evoked, as
well as greater vulnerability to such states in the face of aversive events or interactions.
Whether this is indeed the case is a significant question in that evidence of such deficits
would indicate the relevance of the literature on the development of emotion regulation
skills for understanding the etiology and maintenance of depressive disorders.

Interpersonal and Familial Influences on the Development of Emotion Regulation Skills

It has been hypothesized that children may develop their skills for regulating negative
affective arousal in the course of parent-child interactions (Cole & Kaslow, 1988; Garber
et al., 1995). As noted by McDonough et al. (1994), "nurturing [and we would add

24
socializing] children requires that adults both accept and limit the child's expression of
affect [and] minimize their expressions of negative affect toward the child" (p. 67).
Parent-child interactions may thus comprise an ongoing process of teaching children
(through modeling, instruction, and social contingencies) to maintain, alter, and modulate
their emotional experiences and expression. As will be discussed now, several lines of
research, both developmental and clinical, suggest that family processes do in fact play a
role in children's development of emotion regulation skills. The work of Gottman and
colleagues (e.g., Gottman, Katz, & Hooven, 1996, 1997) on parents' meta-emotion
philosophies stands out in this regard. Based on a longitudinal, biop sychosocial
assessment of parenting behavior, children's emotional regulation, and children's
developmental outcomes, they reported that parents' validation of children's negative
emotions, and their engagement in coaching their child in recognizing and coping with
these emotions, were related to children's emotional regulation abilities as well as to their
peer, academic, and health functioning over time.

It is notable, however, that nearly all research in this area has been based on infants and
very young children, and focused on the role of attachment relationships in the
development of children's affective control (e.g., Cole & Kaslow, 1988; Fox, 1994). Only
recently, has attention been directed to understanding the role that family and parenting
processes may play in the development of older childrens' and adolescents' ability to
regulate distressed affect. Perhaps this is because one of the recognized shifts that occur
as children mature is that they take on increasingly more of the responsibility for
managing their own affective arousal. That is, their parents and other adults do less of
the work for them as they develop the necessary motor and cognitive skills, and over
time, develop increasingly sophisticated approaches for managing their own affect (see
Cole & Kaslow, 1988; Gross & Munoz, 1995). Nonetheless, the fact that children and
adolescents need to acquire and maintain a developmentally appropri ate set of skills for
regulating negative affective arousal (Forgatch, 1989) leaves open the question of what
factors influence their ability to do so. Attention to proximal variables that may facilitate
or impede their accomplishment of this ongoing developmental challenge will most likely
offer insights into normal emotional development and inform the development of
interventions for psychopathology associated with impairments in this area.

Preliminary evidence suggests that older children and early adolescents may also learn to
regulate their arousal through parents' responses to their emotions. Eisenberg, Fabes,
and Murphy (1996) reported that school-aged children's negative emotionality was
inversely related to maternal problem-focused reactions and positively related to
mothers' minimizing responses and fathers' punitive or distress responses. Children
whose mothers responded to negative affect with minimizing or punitive responses were
more likely to use avoidant strategies and less likely to use constructive ones. Similar
results have also emerged from studies of school-aged children with depressed mothers.
In particular, depressed mothers have been found to respond to children's negative affect
with more directive, less supportive, and less problem-solving behavior (Garber et al.,
1991). In turn, both they and their children generated fewer and poorer strategies for
responding to negative affect and had lower expectations that their strat egies would be
effective when compared to families of healthy women.

It also appears that children in families characterized by difficulty in de-escalating

negative emotions may not learn effective skills for managing their affect (Lindahl &
Markman, 1990; Lindahl, Clements, & Markman, 1998). In particular, mothers' use of
negative conflict-resolution strategies has been shown to relate strongly to children's
negative affective escalation in interactions with peers (Lindahl et al., 1998). Additionally,
parents' reciprocity of negative affect and difficulty returning to a neutral or positive
affective state during conflict interactions are predictive of children's emotional and
behavioral adjustment (Carson & Parke, 1996).

Research on child and adolescent depression may also be relevant here--particularly to

the extent that our supposition that such disorder represents, at least in part, a failure of
affective regulation systems is supported. With the growing recognition that depressive

25
behavior exists in an interpersonal context, there has been increasing interest in the role
of family relationships and interactional processes as factors relevant to understanding
depression in youth. Initial studies have yielded clear evidence that family functioning is
disrupted in families of adolescents with elevated depressive symptomatology (Kaslow,
Deering, & Racusin, 1994). Of particular relevance, is evidence that families of depressed
persons may respond to negative affective behavior differently than do families of non-
depressed persons, and that social contingencies operating within the family may thus be
relevant for understanding the display of such behavior. In our own work, we discovered
that mothers of adolescents reporting ele vated levels of depressive symptomatology
were more likely than mothers of nondepressed adolescents to increase facilitative
behavior in response to adolescent depressive behavior. Additionally, fathers of
depressed adolescents were more likely than their counterparts in families of
nondepressed adolescents to decrease aggressive behavior subsequent to adolescent
depressive behavior (Sheeber, Hops, Andrews, Alpert, & Davis, 1998). We interpreted
these findings to suggest that parents of depressed adolescents may inadvertently be
reinforcing depressive behavior. Analogous results have been reported in families of
depressed women (Biglan et al., 1985; Dumas & Gibson, 1990; Hops et al., 1987). Of
particular relevance here is evidence that affective behavior appears to be responsive to
contingencies operating within the family environment. It thus provides another layer of
evidence that parental responses to children's affective behavior may be relevant for the
development of skills necessary for self-management.

It is important to note that we expect the association between family interactional

processes and children's emotional development to continue into adolescence. We base
this expectation on evidence that the family continues to play a significant role in
determining the emotional health of the children in

adolescence (Barrera & Garrison-Jones, 1992; McDonough et al., 1994; Sheeber, Hops,
Alpert, Davis, & Andrews, 1997). Additionally, adolescence may be a particularly
important point to examine the role of family interactions in emotional development
because of both the increased risk for depressive disorders (Lewinsohn, Hops, Roberts,
Seeley, & Andrews, 1993) and increased family conflict (Collins, 1990; Montemayor,
1983).

In this investigation, we aimed to conduct a preliminary examination of emotion

regulation in a verbal problem-solving interaction between adolescents and their
mothers. Our guiding hypotheses were (a) depression is associated with poorer ability to
regulate negative emotions and (b) family interactional processes are related to
adolescents' ability to regulate negative emotion. In particular, our first hypothesis was
that depressed adolescents would be less able to regulate dysphoric and
aggressive/irritable affect. Secondly, we hypothesized that adolescents' ability to regulate
dysphoric and aggressive affect would be related to maternal responses to the
adolescents' affective behavior.

METHOD

Participants

Inclusion Criteria and Recruitment Procedures

Participants were 50 adolescent--mother dyads. Half of the adolescents met criteria for a
unipolar affective disorder and the remainder were healthy controls. The depressed
adolescents met DSM III-R criteria for a current episode of either Major Depressive
Disorder (MDD; n = 22) or Dysthymia (n = 3) and were currently participating in mental
health treatment. Approximately half of the depressed adolescents also met criteria for a
current nonaffective comorbid disorder (i.e., anxiety disorders, disruptive behavior
disorders, substance use disorders, and attention deficit hyperactivity disorder).
Depressed adolescents were recruited from public and private outpatient facilities.

26
Therapists provided the parents and adolescents with a brief description of the project
and obtained their consent to release the family's name and phone number to research
staff. One of the investigators telephoned the family, described the research project, and
conducted a brief screening for depressive symptomatology.

The healthy adolescents did not meet diagnostic criteria for any current psychiatric
disorder and had no history of mental health treatment. They were matched with
depressed adolescents on race, gender, age, and neighborhood of residence. Most of the
healthy adolescents (n = 21) were recruited via a canvassing technique in which research
staff went door-to-door in the neighborhoods of depressed participants. At each home,
research staff introduced themselves, briefly explained the purpose of the study and the
need for comparison families, and indicated that because there was a depressed girl/boy
in their neighborhood, who was participating in our project, we were looking for a
comparison family with the same age girl/boy, who was not depressed, to take part in the
research. The remaining healthy participants were either referred by the depressed
participants and their mothers (n = 3) or self-referred in response to an ad in a
community newspaper (n = 1). Our objective in doing neighborhood-based recruiting wa
s to make our two groups as similar as possible on demographic and social--cultural
characteristics.

Demographic Data and Equivalence of Groups

As just reported, the depressed and healthy groups were matched on sex, race, and age
(within 1 year). In each group, 17 of the adolescents were female and 9 were male. One
dyad in each group was African-American and the remainder were Caucasian. The
adolescents were between 12 and 19 years old, and there was no mean age difference
between groups (M = 15 years, 6 months; SD = 2 years). Families of depressed
adolescents were more likely to be headed by single parents (58% vs. 15%), [[chi].sup.2]
(1, N = 52) = 10.04, p [less than] .01, and they had a correspondingly lower median
family income, Wilcoxon rank sum Z = -2.11, p [less than] .05; annual family income
ranged from $5000 to $146,000 in the depressed group and from $15,000 to $180,000 in
the nondepressed group. There was no between group difference on mothers' level of
education (M = 13 years). Additional information on participants and recruitment
procedures is presented in Sheeber and Sorensen (1998).

Measures

Schedule of Affective Disorders and Schizophrenia-Children's Version (K-SADS)

The K-SADS (Orvaschel & Puig-Antich, 1986) interview was conducted with both
depressed and healthy adolescents to confirm appropriateness of the group assignment
as well as to determine whether depressed adolescents met criteria for comorbid
disorders; due to practical considerations, the interviewers were not blind to diagnostic
status. Mothers did not participate in the diagnostic assessment both because of the
already lengthy nature of our assessment and because, as noted by others (Lewinsohn et
al., 1993), the reliability of adolescent report increases and the agreement between
parent and adolescent decreases with age. Additionally, our primary focus was on
depressive symptomatology and we expected that adolescents would have more direct
access to information regarding their depressive moods and behaviors. Though this
procedure departs from the standard administration, it has been used successfully by
other research teams (e.g., Lewinsohn et al., 1993).

Interviews were conducted by two of the investigators as well as by graduate and

undergraduate students, who participated in a rigorous training program. The student
interviewers demonstrated 100% agreement with the investigators on the
presence/absence of depressive diagnoses as well as a minimum of 80% agreement on
nonaffective disorders on at least two interviews before doing independent interviews.
Interview-derived diagnoses were confirmed by one of the investigators, who reviewed

27
both the items endorsed and the interviewer's notes. Questions regarding the accuracy of
diagnoses were resolved based upon discussion with the individual interviewer and
review of interview audiotapes as needed. Weekly supervision sessions were held among
the first author and the interviewers. Reliability ratings were obtained by the
investigators on a random 20% of the interviews. With agreement based on specific
diagnoses, the average interrater reliabilities for research interviews were K = 1.00 for
depressive disorders, K = .56 for anxiety disorders, K = .62 for behavior disorders, and K
= 1.00 for substance use disorders. There was insufficient variability in the data to obtain
kappa scores for eating disorders or psychotic disorders. However, interviewers agreed
100% on the absence of these disorders.

Living in Family Environments Coding System (LIFE)

The LIFE coding system (Hops, Biglan, Tolman, Arthur, & Longoria, 1995) was developed
to assess both behaviors characteristic of depressed individuals and the functional
relations between the behavior of the depressed person and that of their family
members. Observers record the affect and verbal content of interactions in real time. The
LIFE is an event-based coding system in which new codes are entered each time the
affect or verbal content of the interactants changes. The time codes enable us to assess
how long particular verbal and nonverbal behaviors are maintained.

Four composite variables, derived from the individual affect and content codes, were
used in the present study. Depressive behavior consists of all codes with dysphoric,
anxious, whining, or pain affect as well as complaints and self-derogatory comments with
neutral affect. Aggressive behavior includes all codes with irritable affect as well as
disapproving, threatening, or argumentative statements with neutral affect. Facilitative
behavior consists of statements made with happy or caring affect as well as approving or
affirming statements and statements that serve to maintain the conversation. The
Problem-Solving construct includes statements that identify problems or propose
solutions as long as the accompanying affect is positive or neutral. Observers were blind
to hypotheses and group. Approximately 30% of the tapes were coded by a second
observer for reliability purposes. Kappa coefficients for the composite codes were as
follows: Aggressive K .80, Depressive K = .70, Problem-Solving K = .65, and Facili tative K
= .83. Validity of these constructs has been established in a series of studies examining
marital and parent-child interactions (e.g., Hops et al., 1987; Hops, Sherman, & Biglan,
1990). More detailed information regarding the development and psychometric
characteristics of LIFE coding system is presented in Hops, Davis, and Longoria (1995).

Two observational indices of the adolescents' ability to regulate emotional arousal were
created from the LIFE data. One index was the average length of time the negative
affective behavior was maintained each time the adolescent displayed it. This variable
was created by dividing the duration of adolescent negative affect over the course of the
interaction by the number of distinct times the negative affect was displayed. This index
was developed to capture the adolescents' ability to shift out of negative affective states.
Separate variables were constructed to assess the average duration of depressive and
aggressive emotional displays.

The second index of the adolescents' ability to regulate their emotional behavior was
reciprocity by the adolescent of maternal negative affect. In creating this variable, we
regarded the mothers' behavior to be a provocative affective stimulus. The extent to
which the adolescents reciprocated maternal negative affect was, therefore, considered
to be an index of their ability to regulate the affect generated by aversive parental
behaviors. This is consistent with the manner in which difficulties in affect regulation have
been assessed in the context of conflictual marital exchanges (e.g., Lindahl & Markman,
1990).

In order to measure adolescent reciprocity of maternal negative affect, Allison-Liker z-

scores (Allison & Liker, 1982) were calculated for interactive sequences initiated by

28
maternal displays of negative emotion. Z-score for sequential relations reflect the extent
to which a specified sequence of behavior, in this case adolescent reciprocity of negative
affect (e.g., adolescent depressive behavior after maternal depressive behavior), occurs
more or less often than would be expected as a function of the base rate of each
behavior. The z-scores were computed by comparing the unconditional probability of
adolescent depressive or aggressive behaviors with the conditional probability of these
behaviors when they were preceded by the same behavior on the part of their mothers.
Thus the scores indicate whether negative maternal affective behaviors increased the
likelihood that the adolescent would display the same class of behavior. Such a relation
would be reflected in a positive z-score. As with the duration vari ables, separate
variables were created for reciprocity of maternal depressive and aggressive behaviors.

In order to examine the hypothesis that parents' responses to adolescents' emotional

behaviors would be related to the adolescents' ability to regulate their affective states,
Allison-Liker z-scores were also created to measure maternal responses to adolescent
emotional behaviors. The z-scores were computed by comparing the unconditional
probability of each parent behavior (i.e., depressive, aggressive, problem solving, and
facilitative) with the conditional probability of that behavior when it was preceded by
either adolescent depressive or aggressive behavior. Hence, a total of four z-scores, each
reflecting the probability of a given parental behavior in response to adolescent affective
behavior were created for adolescent depressive and adolescent aggressive behavior,
respectively. Each score indicated whether adolescent negative affect elicited (positive z-
score) or suppressed (negative z-scores) the relevant parental behavior. The means and
standard deviations for each variable examined in the parent-- adolescent interactions
are presented in Table I.

Assessment Procedures

Home Assessment

The initial assessments were conducted in the families' homes. Mothers and adolescents
read and signed the informed consent forms, and mothers completed the demographic
questionnaire. The affective and psychosis sections of the K-SADS interview were
conducted with the adolescents; as described later, the remainder of the K-SADS was
administered during the laboratory assessment. The adolescents were paid $10.00 for
their participation.

29
Laboratory Assessment

The lab assessments took place an average of 13 days after the home visits. At this
meeting, the diagnostic interview was completed, and the mothers and adolescents
participated in two consecutive 15-minute problem-solving interactions (PSIs), which were
videotaped for later coding. Topics for the interactions were identified based on mothers'
and adolescents' responses on the Issues Checklist (IC; Prinz, Foster, Kent, & O'Leary,
1979), a list of 44 topics about which adolescents and parents may disagree. Examples of
topics discussed by families include "adolescent lying to parents" and "adolescent talking
back to parents." One topic was discussed during each interaction. Participants identified
which of the topics they had discussed during the last month, provided an estimate of
how frequently the topic arose, and rated the anger intensity of the discussions on a 5-
point scale. The items having the greatest conflict ratings (Frequency x Intensity) on the
adolescents' and mothers' IC were chosen. Topics disc ussed by the depressed
adolescents and their mothers had overall greater conflict ratings than did those
discussed by the comparison dyads (mother topics, t[49] = 2.12, p [less than] .05, Ms =
67 and 30, for mothers of depressed and comparison adolescents, respectively;
adolescent topics, t[49] = 2.01, p [less than] .05, Ms = 81 and 34, for depressed and
comparison adolescents, respectively). Adolescents and mothers were paid $20.00 each
for their participation.

RESULTS

Hierarchical multiple regression analyses with backward elimination were used to

examine the relations between adolescent depressive status (i.e., Depressed or Healthy),
the microsocial parent-child interaction variables, and each of the four indices of
adolescent emotion regulation. Hypotheses regarding depressive status and microsocial
family processes were tested in the same analyses. In each analysis, adolescent
depressive status was entered first, followed by simultaneous entry of each of the z-
scores reflecting maternal responses to the adolescents' aggressive or depressive
behaviors, which were then followed by a series of variables representing the interaction
between depressive status and each of the z-scores. Though we did not hypothesize that
the relations between the family process variables and adolescent emotion regulation
would differ as a function of adolescent depressive status, the interaction terms were
included because the relatively limited amount of research in this area did not allow u s
to exclude the possibility of differential relations. Backward elimination procedures were
used such that the interaction terms were removed prior to the elimination of the main
effect variables. The inclusion of the main effect components in evaluating the interaction
effects is necessary to eliminate the possibility that the latter are due to main effects not
included in the equation.

Average Duration of Depressive and Aggressive Behavior per Display

Table II presents the significant results from the regression analyses predicting the
average duration of depressive and aggressive behaviors. Depressive status accounted
for 24% of the variance in average length of the adolescents' depressive displays, with
depressed adolescents maintaining depressive affective states longer than did healthy
adolescents; average durations of depressive behavior for depressed and healthy
adolescents are presented at the bottom of Table I. Additionally, one significant main
effect, the z-score reflecting "maternal facilitative behavior given adolescent depressive
behavior" explained an additional 8% of the variance. No interactions between depressive
status and maternal responses to adolescent depressive behavior were predictive of the
duration of depressive affective states.

The results of the regression predicting duration of aggressive bouts were less strong
than those predicting duration of depressive bouts. Depressive status and maternal
problem-solving in response to adolescent aggression together predicted 10% of the

30
variance. As expected, depressive status was positively associated with longer duration
of aggressive behavior, accounting for 6% of the variance; average durations of
aggressive behavior for depressed and healthy adolescents are presented at the bottom
of Table I. The z-score reflecting maternal problem-solving responses to adolescent
aggression indicated an inverse relation, predicting an additional 4% of the variance; that
is, maternal problem-solving responses were associated with shorter duration of
aggressive bouts. As presented in Table II, the p-values for these findings are marginal
and so these results should be interpreted cautiously. As before, no significant
interactions between depressive status and maternal responses to adolescent aggressive
beh avior emerged in the prediction of the duration of aggressive affective states.

Adolescent Reciprocity of Maternal Depressive and Aggressive Behaviors

The significant results from the regression analyses predicting adolescent reciprocity of
maternal depressive and aggressive behaviors are presented in Table III. Adolescent
reciprocity of maternal negative affect was predicted in each instance by the mothers'
reciprocity of negative affect when it was displayed by the adolescent. The z-score
reflecting maternal reciprocity of adolescent depressive behavior accounted for 49% of
the variance in adolescents' reciprocity of depressive behavior displayed by their
mothers. [5] The parallel z-score for reciprocity of aggressive behavior explained 77% of
the variance. It is of note that neither the main effect of adolescent depressive status nor
its interaction with the relevant z-scores predicted the adolescents' likelihood of
responding in kind to aversive maternal affect.

DISCUSSION

The results provide support for the hypothesis that depressed adolescents are less adept
at regulating negative emotions. Depressive and, to a lesser extent, aggressive behaviors
were maintained for longer durations by depressed than by healthy adolescents. That the
between group differences were more pronounced for depressive than for aggressive
behaviors is consistent with depression being primarily characterized by dysphoric rather
than by irritable affects, and suggests that difficulties in regulating affect may be
somewhat limited to this particular class of emotional behavior. It is notable in this regard
that though families of depressed adolescents report greater levels of conflictual behavior
than do those of nondepressed peers, the evidence that depressed adolescents actually
display greater levels of aggressive or angry behaviors during family interactions is
actually relatively weak (see Sheeber & Sorensen, 1998).

The finding that depressed adolescents have difficulty shifting out of depressive states is
consistent with the preliminary findings in adult samples (Gilboa & Gotlib, 1997; Goplerud
& Depue, 1985). Difficulty in transitioning from bouts of depressive and aggressive affect
may be a significant risk for ongoing depression because of the adverse impact of
negative affects on one's ability to constructively engage in conflict resolution or
interpersonal problem-solving (e.g., Capaldi, Forgatch, & Crosby, 1994; Gotlib &
Hammen, 1992; Nolen-Hoeksema, 1998). In turn, poorer conflict resolution and problem-
solving may lead to continued interpersonal difficulties, resulting in an ongoing risk for
depressive disorder (Gotlib & Hammen, 1992). Our data are different than those obtained
in past research in that because previous findings were based on self-report of affective
states, it was unclear whether an individual's difficulty shifting out of depressive states
would be discernable to others. Although our data cannot tell us whether mothers were
aware of their adolescents' difficulties, they indicate that difficulties are observable and
hence, may be influencing the interactive behavior of other family members. For
example, our earlier finding that mothers of depressed adolescents were more likely than
mothers of nondepressed adolescents to increase facilitative behavior in response to
adolescent depressive behavior (Sheeber et al., 1998) could reflect mothers' awareness
that their children need help resolving aversive affective states. The effect that
adolescents' difficulty in altering negative affective states has on their own and others'
behavior, as well as on the outcome of family problem-solving discussions, is a topic that

31
we plan to pursue further in our ongoing research on family processes related to
adolescent depression.

On the other hand, we were surprised to find no evidence that depressed adolescents
were more likely to reciprocate negative affect. Given our small sample size, it is
worthwhile to note that the Betas associated with these findings were extremely small,
indicating that the absence of significant associations was not a function of our relatively
low power. These results suggest that the emotional control difficulties experienced by
depressed adolescents are fairly specific. That is, though they have a more difficult time
than healthy adolescents have in resolving negative affective states, they are not more
vulnerable to being pulled into negative interactional sequences by others' displays of
aversive affective behaviors--or at least by such behavior as displayed by their mothers.

The results also provide preliminary support for the hypothesis that adolescents' ability to
regulate their affective states is related to how their mothers respond to their affective
behavior. As regards the duration of depressive bouts, it appears that maternal
facilitative behavior that is contingent on the adolescents' display of depressive behavior,
adds unique variance beyond that predicted by the adolescents' clinical status. Although,
on initial reading, some people find this result to be counterintuitive, we consider it to be
suggestive of a reinforcement mechanism in which positive maternal responses to
depressive behavior may maintain it. In a previous study, we have found that mothers of
depressed adolescents were more likely to provide contingent positive responses to
depressive behavior than were mothers of nondepressed adolescents (Sheeber, et al.,
1998). It is important to keep in mind, in this regard, that we are discussing facilitative
behavior that is contingent on adolescents' displays of depressive behavior, not rates of
maternal facilitation.

The duration of adolescent aggressive bouts was also related to maternal behavior. In
particular, maternal problem-solving responses to aggressive behavior were inversely
related to the duration of the aggressive behavior. It seems logical to hypothesize that
maternal problem-solving responses help to defuse adolescent irritability. It is also likely
that adolescents whose mothers problem-solve when confronted with aggressive
behavior, have had the opportunity to develop their own set of problem-solving skills to
use in response to aversive experiences. Although the marginal p-value associated with
this finding argues for a cautious interpretation pending replication, we think that it is
nonetheless worth noting because of its consistency with the previous research findings.
In particular, it is congruent with evidence that children's negative emotionality is
inversely related to maternal problem-focused reactions (Eisenberg et al., 1996). It is
moreover consistent with evidence that depressed mothers respon d to children's
negative affect with less problem-solving behavior, and that both they and their children
generate fewer and poorer strategies for responding to negative affect (Garber et al.,
1991).

Finally, reciprocity of maternal depressive and aggressive behaviors was strongly

associated with mothers' reciprocity of adolescents' negative affective behavior. That is,
adolescents whose mothers responded to aggressive or depressive behavior with like
behavior, were most likely to get similarly pulled into conflictual exchanges. This high
degree of concordance has multiple possible explanations that we are not in a position to
tease apart. First, it may reflect family norms regarding how to respond to aversive
exchanges. The work of Gottman, Katz, et al. (1997) has alerted us to the influence of
parents' meta-emotion philosophies in influencing children's affective behavior.
Additionally, it may be that parents who display this behavioral response style are those
who don't have the skills to refrain from aversive interchanges and therefore don't
provide their children with models of more adaptive responses. This is consistent with
Gross and Munoz's assertion that children's ability to regulate emotional arousal may well
depend on the models provided to them by their parents (Gross & Munoz, 1995). It should
also be noted that the strength of these associations may also indicate that each lag (i.e.,
adolescent aggressive given mother aggressive; mother aggressive given adolescent
aggressive) is a component of a longer conflict sequence, with families differing on the

32
length of conflict chains in which they engage; i.e., in their ability to disengage from such
interchanges.

It is perhaps notable that there were no interactions between maternal responses to

negative affective behavior and the adolescents' clinical status in predicting adolescent
emotional regulation. Hence, it appears that the family interactional processes that relate
to adolescents' ability to manage distressed and angry affect are similar in families of
healthy and depressed children. So as not to make too much of this result, we think it is
important to note here that our sample size provided limited power for testing
interactions, and hence, "nonfindings" need to be approached gingerly. Were these
findings to be replicated, one could make at least two hypotheses to explain them. First,
we would suggest the likelihood that emotion regulation abilities may mediate relations
between microsocial family processes and depression. That is, family processes
associated with emotion regulation difficulties are similar across healthy and depressed
groups, but the groups differ in the extent to which these processes ar e experienced,
and this difference in level accounts in part for their clinical status. In this manuscript we
established only one step relevant to testing such a hypothesis (i.e., that both family
microsocial processes and depression are associated with indices of emotion regulation
skills-the purported mediator). It will be important in future research to examine whether
these same processes are associated with depression and whether that relationship is
"explained" to a significant degree, by emotion regulation skills.

Second, it may well be that there are a number of processes that independently influence
an individual's ability to regulate negative affect. This would raise the question as to what
processes, aside from family interactions, might contribute to the relations between
clinical status and affect regulation. Literature reviewed in the introduction asserts that
the underlying psychobiology of affect regulation is dysfunctional in depression
(Tomarken & Keener, 1998), suggesting one mechanism by which affect regulation might
become dysfunctional independent of (or in interaction with) interpersonal family
processes. For instance, relative hypoactivation of the left frontal cortex has been noted
in the infant offspring of depressed mothers (Dawson, Klinger, Panagiotides, Hill, &
Spieker, 1992; Field, Fox, Pickens, & Nawrocki, 1995). As the left prefrontal cortex has
been associated with the persistence of goal directed appetitive behaviors over time,
these findings suggest that the biological basis for a dysfuncti on in emotion regulation
may be present in some at-risk individuals from the earliest stages of life, possibly
implicating genetic mechanisms. Of course such biologically based temperaments could
also interact with family interaction processes over the child's lifetime, potentially
exacerbating or ameliorating their risk for case-level disorders later in life. The possibility
that indices of depression-related psychobiological dysfunction will predict poor
regulation of negative affect is an important topic for future research.

One limitation of the present study was its sole focus on regulation of negative affect. As
noted earlier, emotion regulation refers to an individual's ability to "maintain" as well as
to "alter" affective states (Thompson, 1990, pp. 367-467). In this regard, Tomarken and
Keener (1998) have predicted that depressed individuals should display poorer ability to
maintain temporal continuity of positive emotions. It will thus be important in future
research to examine whether depressed adolescents evidence greater difficulty
maintaining positive affect, as well as whether their ability to do so is related to
microsocial family processes.

Several caveats are important in interpreting our findings. First, the families of depressed
adolescents were recruited from treatment settings; as only 20% of depressed youth
receive treatment (Keller, Lavori, Beardslee, Wunder, & Ryan, 1991; Rohde, Lewinsohn, &
Seeley, 1995), caution is warranted in generalizing to nontreatment-seeking populations.
This is particularly the case as regards differences in duration of negative affective bouts.
In particular, it remains to be seen whether similar differences would emerge in
nontreatment samples, where preliminary data suggest that participants may be less
likely to evidence comorbid disorders and adverse family environments (John, Offord,
Boyle, & Racine, 1995; Lewinsohn, Clarke, Rohde, Hops, & Seeley, 1996; Lewinsohn,

33
Gotlib, & Seeley, 1995). On the other hand, as the family processes related to the
adolescents' ability to regulate negative affect were equivalent across the depressed and
non-depressed groups, it is unlikely that using a treatment sample influ enced these
results.

Second, we accepted potential threats to the internal validity of the study so as to not to
sacrifice its external validity. For example, though we considered matching participants
on family structure (e.g., maternal marital status), we chose not to do so because such
structural characteristics appear to distinguish the family environments of depressed and
healthy adolescents, and may actually contribute to the processes that influence
adolescent risk for emotion regulation deficits or depressive disorder or both. For
example, both single parenthood and the lower income levels associated with it may
place stressors on mothers that influence their parenting and hence the adolescents' risk
for emotional dysregulation and disorder. Similarly, though we considered excluding
adolescents with comorbid conditions, we were concerned that this would artificially
constrain the sample and pose serious questions regarding the generalizability of
findings. Of course, the inherent cost of these decisions is that we cannot rule out the
possibility that our results may be attributable, at least in part, to characteristics other
than depressive status. This is a particular limitation in that the small sample size did not
permit us to either include covariates in our analyses or test for mediational effects such
as those just hypothesized. These analyses are clearly necessary steps for future
research.

Relatedly, because this was a small and preliminary investigation, we did not include a
psychiatric control group. We are, therefore, unable to determine whether the deficits
observed are specific to depressive disorders or characteristic of psychopathology more
broadly. However, the greater difficulty observed in the regulation of depressive than
aggressive behaviors, suggests that the deficits may well have specificity to depressive
disorders.

Finally, given the cross-sectional and correlational nature of our data, it is unclear
whether the family processes play a causal role in accounting for the adolescents' ability
to regulate negative affect. However, our recent finding of a prospective relation between
family environment and depressive symptoms in a community sample (Sheeber et al.,
1997), along with data indicating that family relationships are predictive of the clinical
course of depression (e.g., Asarnow, Goldstein, Tompson, & Guthrie, 1993), leads us to
hypothesize that these relations do have etiological significance. Nonetheless,
prospective data on both family interactions and depressive symptomatology in a clinical
sample will be necessary to address the question of directionality.

Despite these limitations, this study adds to the recent, and as yet preliminary body of
literature examining emotional regulation deficits as a component of depression, as well
as that examining the role of parenting behaviors in the development of children's
abilities to manage their own emotional arousal. The study was unusual, both in its focus
on an adolescent population and in its use of a microsocial observational methodology for
examining both parenting behaviors and adolescent emotional regulation. The study of
an adolescent sample is important because the relative neglect of this population in the
literature precludes us from understanding how adolescents develop and maintain the
skills necessary to manage negative emotional arousal--a set of skills that is related to
children's social and academic competence (e.g., Gottman & Guralnick et al., 1997).
These skills may be particularly important in adolescence, given both the greatly
expanded breadth of adolescents' social worlds and the increasing in dependence from
adult supervision and support that is one of the developmental hallmarks of this period.
Additionally, though family interactions in early childhood (the more usual focus of
emotion regulation research) contribute to the children's later ability to manage affective
arousal (see for a discussion, Cooper et al., 1998) they have fewer implications for
treatment of adolescents demonstrating emotional dysregulation. Research on more
proximal predictors is important in this regard. The use of molecular observational
approaches are also of particular benefit in that they facilitate the identification of

34
specific interactional sequences associated with dysregulated affect and are thus
beneficial in the development of skill-based family treatment interventions. The study
reported here provides an initial step in this direction. It will, of course, be necessary to
replicate both the interactional patterns and the association with depressive disorder that
emerged from this investigation. In particular, t he use of a longitudinal design will be
necessary to ascertain whether, as we expect, the observed microsocial interactional
processes are related to affective dysregulation and depressive disorder in a prospective
fashion.

In summary, the results provide preliminary support for the hypothesis that adolescent
depressive disorder is characterized, in part, by difficulties regulating negative affect. In
particular, it appears that depressed youth are less facile than are their healthy peers in
transitioning out of depressive behavioral bouts. The data also suggest that adolescents'
affective control is related to the nature of the microsocial interactional processes
occurring within their families.

ACKNOWLEDGMENTS

Data for this study were collected while Lisa Sheeber was in the Department of
Psychology at University of Cincinnati. The research was supported in part by a Summer
Research Fellowship and a University Research Council Grant awarded by the University
of Cincinnati to the first author. The authors wish to thank Hyman Hops, Anthony Alpert,
Nancy Longoria, and their team of LIFE observers for coding and managing the
observational data. We would also like to thank Elizabeth Mondulick for efforts above and
beyond in the preparation of this manuscript. Finally, we are grateful to Hyman Hops for
the numerous discussions that have influenced our thinking.

(1.) Oregon Research Institute, Eugene, Oregon.

(2.) University of Melbourne, Department of Psychology, Parkville, Victoria, Australia.

(3.) Oregon Social Learning Center, Eugene, Oregon.

(4.) Address all correspondence to Lisa Sheeber. PhD, Oregon Research Institute, 1715
Franklin Boulevard, Eugene, Oregon 97403-1983; e-mail: lsheeber@ori.org.

(5.) Readers may be concerned that the high positive correlations between reciprocity
conditionals reflect statistical artifact. Conditional z-scores adjust for the base rate of the
antecedent and consequent behaviors. High positive correlations between z-scores
representing bidirectional reciprocity (e.g., maternal reciprocity of adolescent depressive
behavior vs. adolescent reciprocity of maternal depressive behavior) occur when, across
families, the extent of maternal reciprocity of adolescent behavior is similar to that of
adolescent reciprocity to mother behavior (i.e., the conditional frequencies for mother
and adolescent reciprocity are similar). However, when the extent of reciprocity is not
similar bidirectionally, the correlations would be lower.

REFERENCES

Allison, P. D.. & Liker, J. K. (1982). Analyzing sequential categorical data on dyadic
interaction: A comment on Gottman. Psychological Bulletin, 91. 393-403.

American Psychological Association. (1994). Diagnostic and statistical manual of mental

disorders (DSM-IV) (4th ed.). Washington, DC: Author.

Asarnow, J. R., Goldstein, M. J., Tompson, M., & Guthrie, D. (1993). One-year outcomes of
depressive disorders in child psychiatric inpatients: Evaluation of the prognostic power of

35
a brief measure of expressed emotion. Journal of Child Psychology and Psychiatry and
Allied Disciplines, 34, 129-137.

Barrera, M., Jr., & Garrison-Jones, C. (1992). Family and peer social support as specific
correlates of adolescent depressive symptoms. Journal of Abnormal Child Psychology, 20,
1-16.

Beckham. E. E., & Leber, W. R. (1995). Handbook of depression (2nd ed.). New York:
Guilford.

Biglan, A., Hops, H.. Sherman, L., Friedman, L. S., Arthur, J., & Osteen, V. (1985). Problem-
solving interactions of depressed women and their husbands. Behavior Therapy 16, 431-
451,

Breen, M. P., & Weinberger, D. A. (1995). Regulation of depressive affect and

interpersonal behavior among children requiring residential or day treatment.
Development and Psychopathology, 7, 529-541.

Capaldi, D. M., Forgatch, M. S., & Crosby, L. (1994). Affective expression in family
problem-solving discussions with adolescent boys. Journal of Adolescent Research, 9, 28-
49.

Carson, J. L., & Parke, R. D, (1996). Reciprocal negative affect in parent--child interactions
and childrens' peer competency. Child Development. 67, 2217-2226.

Clark. L. A., & Watson, D. (1991). Tripartite model of anxiety and depression:
Psychometric evidence and taxonomic implications. Journal of Abnormal Psychology 100,
316-336.

Cole, P. M., & Kaslow, N. J. (1988). Interactional and cognitive strategies for affect
regulation: Developmental perspective on childhood depression. In L. B. Alloy (Ed.),
Cognitive processes in depression (pp. 310-343). New York: Guilford.

Collins, W. A. (1990). Parent--child relationships in the transition to adolescence:

Continuity and change in interaction, affect, and cognition. In R. Montemayor, G. R.
Adams. & P. Gullotta (Eds.), From childhood to adolescence: A transitional period? (pp.
85-106). Newbury Park, CA: Sage.

Cooper, M. L., Shaver, P. R., & Collins, N. L. (1998). Attachment styles, emotion
regulation, and adjustment in adolescence. Journal of Personality and Social Psychology,
74, 1380-1397.

Dawson, G., Klinger, L. G., Panagiotides, H., Hill, D., & Spieker, 5, (1992). Infants of
mothers with depressive symptoms: Electroencephalographic and behavioral findings
related to attachment status. Development and Psychopathology 4, 67-80.

Depue. R. A., & lacono, W. G. (1989). Neurobehavioral aspects of affective disorders.

Annual Review of Psychology, 40. 457-492.

Dumas. J. E., & Gibson, J. A. (1990). Behavioral correlates of maternal depressive

symptomatology in conduct-disorder children: II. Systemic effects involving fathers and
siblings. Journal of Consulting and Clinical Psychology, 58, 877-881.

Eisenberg, N., Fabes, R. A., & Murphy, B. C. (1996). Parents' reactions to children's
negative emotions: Relations to children's social competence and comforting behavior.
Child Development, 67, 2227-2247.

36
Field, T., Fox, N. A., Pickens, J., & Nawrocki, T. (1995). Relative right frontal EEG activation
in 3-6-month-old infants of "depressed" mothers. Developmental Psychology 3, 358-363,

Forgatch, M. S. (1989). Patterns and outcome in family problem solving: The disrupting
effect of negative emotion. Journal of Marriage and the Family, 51, 115-124.

Fowles, D.C. (1988). Psychophysiology and psychopathology: A motivational approach.

Psychophysiology 25, 373-391.

Fox, N. A. (1994). The development of emotion regulation. Monographs of the Society for
Research in Child Development, 59(Serial No. 240).

Garber, J., Braafladt, N., & Weiss, B. (1995). Affect regulation in depressed and
nondepressed children and young adolescents. Development and Psychopathology 7, 93-
115.

Garber, J., Braafladt, N., & Zeman, J. (1991). The regulation of sad affect: An information-
processing perspective. In J. Garber & K. A. Dodge (Eds.), The development of emotional
regulation and dysregulation (pp. 208-242). Cambridge: Cambridge University Press.

Gatchel, R. J., McKinney, M. E., & Koebernick, L. F. (1977). Learned helplessness,

depression, and physiological responding. Psychophysiology 14, 25-31.

Gilboa, E., & Gotlib, I. H. (1997). Cognitive biases and affect persistence in previously
dysphoric and never-dysphoric individuals. Cognition and Emotion, 11,517-538.

Goplerud, E., & Depue, R. A. (1985). Behavioral response to naturally occurring stress in
cyclothymia and dysthymia. Journal of Abnormal Psychology 94, 128-139.

Gotlib, I. H., & Hammen, C. L. (1992). Psychological aspects of depression: Toward a

cognitive--interpersonal integration. Chichester, UK: John Wiley and Sons.

Gottman, J. M., Guralnick, M. J., Wilson, B., Swanson, C. C., & Murray, J. D, (1997). What
should be the focus of emotion regulation in children? A nonlinear dynamic mathematical
model of children's peer interaction in groups. Development and Psychopathology 9, 421-
452.

Gottman, J. M., Katz, L. F., & Hooven, C. (1996). Parental meta-emotion philosophy and
the emotional life of families: Theoretical models and preliminary data. Journal of Family
Psychology 10, 243-268.

Gottman, J. M., Katz, L. F., & Hooven, C. (1997). Meta-emotion: How families communicate
emotionally Mahwah, NJ: Lawrence Erlbaum Associates.

Gray, J. A. (1994). Framework for a taxonomy of psychiatric disorder. In S. H. M. Van

Goozen, N.E. Van de Poll, & J. A. Sergeant (Eds.), Emotions: Essays on emotion theory.
Hillsdale, NJ: LEA.

Gross, J. J., & Munoz, R. F. (1995). Emotion regulation and mental health. Clinical
Psychology: Science and Practice, 2, 151-164.

Henriques, J. B., Glowacki, J. M., & Davidson, R. J. (1994). Reward fails to alter response
bias in depression. Journal of Abnormal Psychology 103, 460--466.

37
Hops, H., Biglan, A., Sherman, L., Arthur, J., Friedman, L., & Osteen, V. (1987). Home
observations of family interactions of depressed women. Journal of Consulting and Clinical
Psychology 55, 341-346.

Hops, H., Biglan, A., Tolman, A., Arthur, J., & Longoria, N. (1995). Living in Family
Environments (LIFE) coding system: Reference manual for coders. Eugene, OR: Oregon
Research Institute.

Hops, H., Davis, B., & Longoria, N. (1995). Methodological issues in direct observation:
Illustrations with the Living in Familiar Environments (LIFE) coding system. Journal of
Clinical Child Psychology 24, 193-203.

Hops, H., Sherman, L., & Biglan, A. (1990). Maternal depression, marital discord, and
children's behavior: A developmental perspective.

In G. R. Patterson (Ed.), Depression and aggression in family interaction (pp. 185-208).

Hillsdale, NJ: Lawrence Erlbaum.

John, L. H., Offord, D. R., Boyle, M. H., & Racine, Y A, (1995). Factors predicting use of
mental health and social services by children 6-16 years old: Findings from the Ontario
Child Health Stud y. American Journal of Orthopsychiatry; 65, 76-86.

Kaslow, N.J., Deering, C. G., & Racusin, G. R. (1994). Depressed children and their
families. Clinical Psychology Review, 14, 39-59.

Keller, M. B., Lavori, P. W., Beardslee, W. R., Wunder, I., & Ryan, N. (1991). Depression in
children and adolescents: New data on "undertreatment" and a literature review on the
efficacy of available treatments. Journal of Affective Disorders, 21, 163-171.

Lewinsohn, P. M., Clarke, G. N., Rohde, P., Hops, H., & Seeley, J. R. (1996). A course in
coping: A cognitive--behavioral approach to the treatment of adolescent depression. In E.
D. Hibbs & P. 5. Jensen (Eds.), Psychosocial treatments for child and adolescent disorders:
Empirically based strategies for clinical practice (pp. 109-135). Washington, DC: American
Psychological Association.

Lewinsohn, P. M., Gotlib, I., & Seeley, J. (1995). Adolescent psychopathology: IV.
Specificity of psychosocial risk factors for depression and substance abuse in older
adolescents. Journal of the American Academy of Child and Adolescent Psychiatry 34,
1221-1229.

Lewinsohn, P. M., Hops, H., Roberts, R. E., Seeley, J. R., & Andrews, J. A. (1993).
Adolescent psychopathology: I. Prevalence and incidence of depression and other DSM-III-
R disorders in high school students. Journal of Abnormal Psychology 102, 133-144.

Lewinsohn, P. M., Lobitz, W. C., & Wilson, S. (1973). "Sensitivity" of depressed individuals
to aversive stimuli. Journal of Abnormal Psychology 81, 259-263.

Lindahl, K. M., Clements, M., & Markman, H. (1998). The development of marriage: A 9-
year perspective. In T. N. Bradbury (Ed.), The developmental course of marital
dysfunction. New York: Cambridge University Press.

Lindahl, K. M., & Markman, H. J. (1990). Communication and negative affect regulation in
the family. In E. A. Blechman (Ed.), Emotions and the family: For better or worse (pp. 99-
115). Hillsdale, NJ: Lawrence Erlbaum Assoc.

38
McDonough, M. L., Carlson, C., & Cooper, C. R. (1994). Individuated marital relationships
and the regulation of affect in families of early adolescents. Journal of Adolescent
Research, 9, 67-87.

Montemayor, R. (1983). Parents and adolescents in conflict: All of the families some of
the time and some families most of the time. Journal of Early Adolescence, 3, 83-103.

Nolen-Hoeksema, S. (1998). Ruminative coping with depression. In J. Heckhausen, C. S.

Dweck, et al. (Eds.), Motivation and self-regulation across the life span (pp. 237-256).
New York: Cambridge University Press.

Orvaschel, H., & Puig-Antich, J. (1986). Schedule for affective disorder and schizophrenia
for school-age children. Epidemiological version: Kiddie-SADS-1 (K-SADS-E) (Technical
Report, Version 4). Pittsburgh, PA: Western Psychiatric Institute and Clinic.

Prinz, R. J., Foster, S., Kent, R. N., & O'Leary, K. D. (1979). Multivariate assessment of
conflict in distressed and nondistressed mother--adolescent dyads. Journal of Applied
Behavior Analysis, 12, 691-700.

Rohde, P., Lewinsohn, P. M., & Seeley, J. R. (1995). Are adolescents changed by an
episode of major depression? Journal of the American Academy of Child and Adolescent
Psychiatry 33, 1289-1298.

Rutter, M. (1991). Age changes in depressive disorders: Some developmental

considerations. In J. Garber & K. A. Dodge (Eds.), The development of emotional
regulation and dysregulation (pp. 273-300). Cambridge: Cambridge University Press.

Schwartz, G. E., Fair, P. L., Salt, P., Mandel, M. R., & Klerman, G. L. (1976). Facial muscle
patterning to affective imagery in depressed and nondepressed subjects. Science, 192,
489-491.

Schwartz, J. A., & Koenig, L. J. (1996). Response styles and negative affect among
adolescents. Cognitive Therapy and Research, 20, 13-36.

Sheeber, L., Hops, H., Alpert, A., Davis, B., & Andrews, J. (1997). Family support and
conflict: Prospective relations to adolescent depression. Journal of Abnormal Child
Psychology, 25, 333-344.

Sheeber, L., Hops, H., Andrews, J. A., Alpert, A., & Davis, B. (1998). Interactional
processes in families with depressed and nondepressed adolescents: Reinforcement of
depressive behavior. Behaviour Research and Therapy, 36, 417-427.

Sheeber, L., & Sorensen, E. (1998), Family relationships of depressed adolescents: A

multimethod assessment. Journal of Clinical Child Psychology, 27, 268-277.

Thompson, R. A. (1990). Emotion and self-regulation. In R. A. Thompson (Ed.).

Socioemotional development: Nebraska symposium on motivation (pp. 367-467). Lincoln,
NE: University of Nebraska Press.

Tomarken, A. J., & Keener, A. D. (1998). Frontal brain asymmetry and depression: A self-
regulatory perspective. Cognition and Emotion, 12, 387-420.

COPYRIGHT 2000 Plenum Publishing Corporation

COPYRIGHT 2001 Gale Group

39
Familial Predictors of Treatment Outcome in Childhood Anxiety Disorders.
(Statistical Data Included)

Author/s: A. Melissa Crawford

Issue: Oct, 2001

ABSTRACT

Objective: To determine whether family factors are predictive of outcome in children with
anxiety disorders who are receiving cognitive-behavioral treatment. Method: Participants
were 61 children aged 8 to 12 years (mean = 10.0, SD = 1.4) with Axis I anxiety disorders
who had been referred to a large Toronto children's hospital. Parents and children
completed measures assessing family functioning, parenting stress, parental frustration,
and parental psychopathology before and after treatment. Outcome measures included
clinician-rated functioning (Children's Global Assessment Scale) and self-and parent-rated
anxiety (Revised Children's Manifest Anxiety Scale). Results: Child ratings of family
dysfunction and frustration predicted clinician-rated improvement (total [R.sup.2] = 0.28,
less than .001). Mother and father reports of family dysfunction, and maternal parenting
stress, predicted mother-rated child improvement (total [R.sup.2] - 0.18, less than .01).
Father-rated somatization and child reports of family dysfunction and frustration
predicted child-rated improvement (total [R.sup.2] -0 25 less than 001). Several family
factors improved with treatment. Conclusion: Family dysfunction appears to be related to
less favorable treatment outcome in children with anxiety disorders. J. Am. Acad. Child
Adolesc. Psychiatry, 2001, 40(10):1 182-1189. Key Words: childhood anxiety disorders,
family functioning, predictors of treatment outcome.

Anxiety disorders are the most prevalent psychological problem of childhood (Bernstein
and Borchardr, 1991) and have been linked to academic difficulties, low selfesteem, peer
relationship problems, and depression (Strauss et al., 1987). Cognitive-behavioral
treatments are effective for many anxious children, but some respond less favorably than
others (Kendall, 1994; Mendlowitz et al., 1999). Factors related to treatment nonresponse
have not been previously examined. Identifying and addressing such factors could
improve treatment outcome (March and Curry, 1998).

This study will examine the relationship between family and parent characteristics,
previously linked to childhood anxiety; and outcome in a cognitive-behavioral treatment
program. Changes in these characteristics during the course of treatment will also be
examined. Multiple informants (clinician, mother, child, father) will be used to rate child
anxiety before and after treatment, inasmuch as many studies show low correspondence
among raters (Manassis et al., 1997). It is recognized that other factors (e.g., child
characteristics, fit between child and treatment, etc.) may also influence treatment
outcome, but examination of these factors is beyond the scope of this study.

40
FAMILY FACTORS PREVIOUSLY LINKED TO ANXIETY DISORDERS

Family Functioning

Problematic family relationships are more prevalent in anxious than nonanxious children
and have been related to increased impairment in childhood anxiety disorders (Manassis
and Hood, 1998; Messer and Beidel, 1994). Although poor family functioning is a correlate
of childhood anxiety, the causal relationship between the two is nor yet known. Families
of anxious children are more involved, more controlling, more rejecting, and less intimate
than those of comparison children without anxiety (Dadds et al., 1996; Rapee, 1997;
Siqueland et al., 1996; Stark et al., 1990).

Research also suggests that parents reinforce their child's insecure, avoidant, and
anxious behavior, which may play a central role in the development and maintenance of
anxiety disorders (Dadds et al., 1996). It is still unknown whether parents promote
anxious behavior in their children, or whether anxious children elicit more protective
responses from their parents, or both. Siqueland et al. (1996) observed that parents of
children with anxiety disorders encouraged less autonomy and were less accepting than
comparison parents. Similarly, Dadds et al. (1996) found that parents of anxious children
tended to promote cautious and avoidant child behavior by modeling avoidance,
vocalizing doubt in their child's abilities, and providing acceptance and comfort when
their child displayed avoidant behavior. These findings suggest that parent- child
relationships and family functioning may affect treatment outcome.

Parental Psychopathology

Children of parents with anxiety disorders show elevated levels of anxiety (Bernstein and
Borchardt, 1991; Last et al., 1991), suggesting a familial transmission of anxiety. It is
unknown whether this transmission is due to heredity, environment, or both (Bernstein
and Borchardt, 1991).

Last et al. (1991) found that first-degree relatives (especially males) of children with
anxiety disorders were significantly more likely to have an anxiety disorder than children
with ADHD or no psychopathology; Other evidence for a genetic contribution to anxiety
disorders comes from twin studies. Thapar and McGuffin (1995) found, in a twin sample of
children and adolescents, that anxious symptoms had a heritability estimate of 59%.
Similarly, Warren et al. (1999) found that genetic variables accounted for about one third
of the variance of self-reported physiological and social anxiety scores in 7-yearold twins.
Thus it appears that anxious symptoms may be genetically influenced (Warren et al.,
1999).

Messer and Beidel (1994) examined both parental psychopathology and family
environment in children with anxiety disorders. They found that fathers of anxious
children showed more obsessive-compulsive, depressive, and global pathological
symptoms than the fathers of children without anxiety. Anxious children also tended to
have more controlling parents who promoted less independence than comparisons, and
family characteristics moderately predicted the presence of anxiety disorders over and
above the child's characteristics alone (Messer and Beidel, 1994). They hypothesized that
family environment indirectly affects the acquisition of anxiety because parental
psychopathology promotes more conflict and less cohesion in the home, which in turn
contributes to the maintenance and/or enhancement of child anxiety (Messer and Beidel,
1994). Parental psychopathology and family environment will both be examined in this
study.

41
Parenting Stress

Child behavior, life events, and parental personality can make parenting stressful (Abidin,
1995). Parenting stress has been found to adversely affect the quality of caregiving,
parent-child interactions, and child behavior. It is unclear how parenting stress influences
family interactions and whether it is a result or cause of child behavior problems. Adverse
effects on caregiving quality include the risk of insecure attachment (Manassis et al.,
1994), maternal unresponsiveness, and the use of negative parenting styles (less
affection, supervision, and autonomy) (Onatsu-Arvilommi et al., 1998; Ritchie and Holden,
1998).

Parenting stress is also related to poorer child adjustment at school (Onatsu-Arvilommi et

al., 1998) and more parent-rated child psychopathology (Abidin et al., 1992). In a
longitudinal study examining infants at ages 1 and 4 years, Abidin et al. (1992) found that
child gender, life stress (Life Stress Events scale of the Parenting Stress Index [PSI]), child
characteristics (PSI Child Domain), and maternal characteristics (Depression and
Competence subscales of the PSI) accounted for 39% of the variance in maternal-rated
child adjustment 4 years later. Parenting stress was not significantly related to anxiety in
this study.

The major domains of parenting stress identified by Abidin (1995) (Total Stress, Parent
Domain, Child Domain, and Life Stress) will be included as predictors in this study.

FAMILY FACTORS RELATED TO OUTCOME

There is some support for our hypothesis that pretreatment family functioning can affect
treatment outcome. Outcome studies with anxious adults show that individuals are
significantly more likely to relapse if they have only moderately good family functioning
(as opposed to very good family functioning) (Scheibe and Albus, 1997), poor social
interactions (Scheibe and Albus, 1997), or tension and friction in their marital relationship
(Durham et al., 1997; Lelliott et al., 1987). Similar studies have nor been done in children.
March and Curry (1998) have identified the need for treatment studies in pediatric
anxiety disorders that include variables other than the disorder itself (e.g., socioeconomic
status, parental psychopathology; intelligence) in the prediction of outcome.

STUDY OBJECTIVES

The main objective of this study is to examine whether family dysfunction, parental
frustration (a further indication of troubled parent-child relationships), parental
psychopathology and parenting stress predict the treatment outcome of children with
anxiety disorders. Second, we will examine whether family functioning changes over the
course of treatment. It is expected that children with poorer pretreatment family
relationships will improve less with treatment than children with good family
relationships. Furthermore, we hypothesize that family factors will improve with
treatment, inasmuch as parents are involved in the child's treatment and having a less
anxious child may reduce family stress.

METHOD

Participants

Participants were 61 children (34 males, 27 females) aged 8 to 12 years (mean = 10.0,
SD = 1.4) and their parents. Of the children participating in this study, primary diagnoses
included generalized anxiety disorder (65.6%), separation anxiety disorder (21.1%),
simple phobia (5.6%), social phobia (2.2%), panic disorder (1.1%), and other (such as
trichorillomania and selective mutism; 4.4%). The majority of the sample was white

42
(85%); the rest were of African-American or Asian descent (15%), approximating the local
census population. Eighty-five percent of the parents had some postsecondary education.
Subsets of the total sample having no missing data on the outcome measures were used
for each analysis. Fathers completed fewer questionnaires than children and mothers
because mothers generally brought their children to the assessments.

The families were referred for treatment to the Anxiety Disorders Clinic of a large
children's hospital by physicians and mental health professionals. All children met the
criteria for at least one DSM-IV anxiety disorder, and this disorder accounted for the main
clinical problem presented. Children having psychotic disorders, a medical condition
which would interfere with treatment, or who were not proficient in the English language
were excluded from participation. Children with IQs less than 80 or who had learning
problems that would interfere with their understanding and participation in the treatment
(based on school information and clinician judgment) were also excluded. The six children
in the sample receiving psychoactive medication were included but kept on a constant
dosage throughout treatment to minimize the effects of medication on outcome.

Procedure

Children and parents were seen separately for a psychiatric assessment. Psychiatrists
administered to participants (children and parents) a semistructured diagnostic interview
with questions from the Diagnostic Inventory for Children and Adolescents-Revised-Parent
Version (Reich and Welner, 1988), which had been revised to meet the DSM-IV criteria.
The treatment process and research study were explained to children and parents. If
interest was expressed in the research, parental consent and child assent were attained.
Children and parents then completed questionnaires assessing a variety of family
functioning factors and the child's anxiety symptoms. The order of the presentation of the
questionnaires was counterbalanced. After the initial assessment, children were randomly
assigned to either individual or group treatment conditions. Children in both conditions
received a cognitive-behavioral treatment program and parents received a parent-
training program. Treatment attendance was very good, with an average attendance of
10.5 of 12 sessions for children and parents. After treatment, children and parents
completed the same questionnaires as those given before treatment. All predictor
variables in the regression analyses are based on the initial assessment measures.

No significant differences in treatment outcome were found between children who

completed individual or group therapy; this finding implies that these groups of children
are comparable and do not require separate analyses. Similarly, correlations between age
and treatment outcome were not significant, so all age groups were included in the
analysis.

Treatment Manuals

The Coping Bear Workbook (Scapillato and Mendlowitz, unpublished, 1993) is an

adaptation for group therapy of the Coping Cat Workbook developed by Kendall (1990).
This treatment program consists of 12 sessions teaching children how to identify their
physical reactions to anxiety, how to relax, how to change maladaptive selftalk, and how
to reinforce their adaptive coping responses.

Group therapy for parents is modeled after the book Keys to Parenting Your Anxious Child
(Manassis, 1996). It consists of strategies to help parents understand and deal with their
child's anxiety and teaches them how to help their child cope with anxiety-provoking
situations. The content of this program roughly parallels the content of the child's
treatment. Both child and parent treatment include homework assignments, group
exercises, and group problem-solving to reinforce the strategies in the manuals. (Manuals
can be obtained by writing to the first author.)

43
Outcome Measures

Revised Children's Manifest Anxiety Scale. The Revised Children's Manifest Anxiety Scale
(RCMAS) (Reynolds and Richmond, 1985) measures the nature and severity of a child's
anxiety disorder. Children (aged 6-19) and parents are asked to rate 37 items about the
child's anxious symptoms as either true or false. The RCMAS yields a total anxiety score,
but also has scales measuring physiological symptoms, worry/ oversensitivity, social
concerns/concentration, and social desirability. Internal consistency reliabilities (that
examine the consistency of test scores to justify good test measurement) of the present
sample were 0.83, 0.76, and 0.84 for child, mother, and father reports of child anxiety,
respectively.

Children's Global Assessment Scale. The Children's Global Assessment Scale (CGAS)
(Shaffer et al., 1983) gives clinician ratings of children's (aged 4-16) adaptive functioning
during the previous month. It is rated on a 100-point scale, with 1 being most impaired
and 100 least impaired. Descriptors are given for each 10-point interval. Clinicians select
the interval that best describes the child's current functioning, then assign an exact
rating within that interval. To obtain an unbiased rating, three clinicians not involved in
the study estimated the children's global functioning before and after treatment using all
clinical data from the initial (for pretreatment CGAS ratings) and posttreatment (for
posttreatment CGAS ratings) assessments. They were blind to the pretreatment versus
posttreatment status.

Measures of Family Functioning Completed by Child and Parents

Brief Family Assessment Measure-III. The Brief Family Assessment Measure-III (FAM)
(Skinner et al., 1995) is a 14-item child- and parentreported measure of overall family
functioning. Participants are asked to answer questions about how family members relate
to each other on a 4-point scale ranging from "strongly disagree" to "strongly agree."
Items are summed to yield a total score, which is converted into a T score. Internal
consistencies for the current sample were 0.69 for child report, 0.84 for mother report,
and 0.80 for father report. This study will use both child- and parent-reported family
functioning as separate predictors in the analysis.

Parental Frustration Questionnaire. The Parental Frustration Questionnaire (PFQ) (Bradley,

unpublished, 1996) is an eight-item childand parent-rated measure assessing how
frustrated mothers and fathers are with their child's behavior. It examines separately how
often ("more than once a day," "more than once a week," "more than once a month,"
"less than once a month," and "never") mothers and fathers get angry, irritable, or critical
with their child. The child-raced version examines both maternal and paternal frustration.
Child responses are summed to yield a Total Maternal Frustration score and Total
Paternal Frustration score (with low scores indicating low frustration). Responses on the
parent-rated version are summed to yield a Total Frustration score. Our sample had
internal consistencies of 0.79, 0.85. and 0.85 for child, mother, and father ratings,
respectively.

Measures of Family Functioning Completed by Parents Only

Brief Symptom Inventory. The Brief Symptom Inventory (BSI) (Derogatis, 1992) is a 53-
item self-reported list of psychological symptoms in adults, rated on a 5-point scale
ranging from "not at all" to "extremely." Individuals rate how distressing each symptom is
for them. Responses are summed to yield nine subseales including Anxiety, Depression,
Hostility, Obsessive-Compulsive, Somatization, Interpersonal Sensitivity, Phobic Anxiety,
Paranoid Ideation, and Psychoticism. The BSI yields three global indices of distress. Only
the Global Severity Index and the nine subscales will be used in this study. Internal
consistency reliabilities of 0.71 to 0.85 have been found for the nine dimensions of the
BSI (Derogatis and Lazarus, 1994). The Global Severity Index showed internal

44
consistencies for the current sample of 0.97 for mother-, and 0.96 for father-reported
distress.

Parenting Stress Index. The PSI (Abidin, 1995) is a 120-item questionnaire measuring
parent, child, and environmental attributes that contribute to parenting stress and
dysfunctional parenting (Abidin, 1995). It includes four subseales: Child Domain (assesses
aspects of children that may make them difficult to parent), Parent Domain (assesses the
parent's functioning, how competent the parent feels, and aspects of life that make
parenting difficult and create dysfunctional child-parent interactions), Total Stress score,
and Stressful Life Events scale (Abidin, 1997). The Child Domain and Parent Domain are
rated on a 5-point scale ranging from "strongly agree" to "strongly disagree," whereas the
Stressful Life Events scale is a forced-choice (yes/no) format. Previously reported internal
consistencies were 0.90 for the Child Domain, 0.93 for the Parent Domain, and 0.95 for
the Total Stress scale (Abidin, 1997).

Statistical Analyses

Pearson product-moment correlations and intercorrelations were determined for all

predictor measures (FAM, BSI, PSI, and PFOJ and posttreatment outcome measures
(CGAS, child-reported anxiety [RCMAS], mother-rated child anxiety [RCMAS], and father-
rated child anxiety [RCMAS]). Only those subscales of each measure that significantly
correlated with the outcome measures were included in the subsequent analyses.

Multiple linear regressions were performed to predict posttreatment child anxiety, using
the subseales significantly correlated with them. This method allows an estimate of the
relative contribution of individual factors to the variance in children's treatment outcome.

To isolate the contribution of each predictor variable, a three-step enter/remove

procedure was used. First, the entire predictor set was entered into a regression equation
for the outcome measure. Second, to control for the pretreatment anxiety rating which
would significantly contribute to posttreatment anxiety ratings, the pretreatment reports
of anxiety were removed from the equation. The resulting equation gives an estimate of
the proportion of the variance accounted for by the predictor variables (without the
influence of pretreatment ratings). Third, the contribution of each predictor variable to
this equation (adjusted partial [R.sup.2]) was calculated by removing the variance due to
all predictors but the predictor of interest. This procedure avoids obscuring the
contribution of certain predictors. Thus if two predictors are both highly correlated with
the outcome measure and with each other, the three-step enter/remove procedure will
include both and determine the unique contribution of each.

Finally, paired t tests were conducted to examine whether family functioning changed
with treatment. The Bonferroni correction for multiple t tests was applied to evaluate the
significance of the data.

RESULTS

PREDICTORS OF TREATMENT OUTCOME

Correlational Analyses

Clinician-Rated Children's Global Assessment Scale. Child reports of family dysfunction

and child-rated parental frustration were significantly associated with clinicianrated global
functioning. Partial correlations are negative because lower CGAS scores indicate more
impairment. Table 1 shows the significant partial correlations between predictor variables
and CGAS.

45
Mother Reports of Child Anxiety. Mother and father reports of family dysfunction and
mother-rated total parenting stress were significantly related to posttreatment mother
reports of child anxiety. All variables were also significantly intercorrelared. Significant
partial correlations between predictors and mother-rated RCMAS are shown in Table 1.

Child Reports of Anxiety. Child reports of family dysfunction, father-rated somatization

symptoms, and childreported maternal frustration were significantly related to
posttreatment self-reported anxiety. Child-rated family dysfunction was significantly
associated with child reports of maternal frustration. Table 1 shows the significant partial
correlations between predictor variables and childreported anxiety.

Father Reports of Child Anxiety. None of the predictor variables was significantly related
to father-rated child anxiety.

Regression Analyses

Clinician-Rated Children's Global Assessment Scale. The top portion of Table 2 presents
the results of the regression analyses of the predictors of treatment outcome as
measured by the clinician's rating of posttreatment child functioning and shows the
contribution of the individual predictor variables. The set of predictor variables, including
the pretreatment CGAS, entered together accounted for 63% of the total variance in
posttreatment CGAS scores. When the pretreatment CGAS scores were removed from the
regression, the predictor variables accounted for 28% (p [less than] .001) of the total
variance in posttreatment CGAS scores. When the contribution of each predictor variable
(adjusted partial [R.sup.2]) was calculated, child reports of family dysfunction made the
greatest contribution, followed by child-rated parental frustration. Because of an
intercorrelation between child-rated family dysfunction and child-reported parental
frustration, the sum of the adjusted partial [R.sup.2] values is greater than the [R.sup.2]
for the model. Thus these predictors make unique but overlapping contributions to the
model.

Mother Reports of Child Anxiety. The predictor variables, along with the pretreatment
mother-rated RCMAS scores, accounted for 20% of the variance in posttreatment mother
reports of child anxiety. When the contribution of pretreatment mother-rated RCMAS
scores was removed from the equation, the predictors accounted for 18% (p [less
than] .01) of the variance in posttreatment mother-rated RCMAS scores. When the
contribution of each predictor (adjusted partial [R.sup.2]) was calculated, father-rated
family dysfunction made the most significant contribution, followed by mother-rated total
parenting stress and mother reports of family dysfunction. Because of significant
intercorrelations among all predictor variables, the sum of the adjusted partial [R.sup.2]
values is greater than the [R.sup.2] for the model. Thus these predictors make unique but
overlapping contributions to the model. Table 2 (middie portion) shows the results of the
regression of the predictors of outcome as measured by mother-rated child anxi ety.

Child Reports of Anxiety. Table 2 (bottom) shows the results of the regression analyses of
the predictors of treatment outcome, with child ratings of posttreatment anxiety as the
outcome measure. The predictors, along with the pretreatment child-rated RCMAS scores,
accounted for 27% of the variance in posttreatment child-reported anxiery When the
contribution of the pretreatment RCMAS scores was removed from the equation, the set
of predictor variables accounted for 25% (p [less than] .001) of the variance in
posttreatment child-rated anxiety. When the contribution of each predictor (adjusted
partial [R.sup.2]) was calculated, father somatization made the greatest contribution,
followed by child-rated maternal frustration and child ratings of family dysfunction. Due
to the intercorrelation between child FAM scores and child-reported maternal frustration,
the sum of the adjusted partial [R.sub.2] values is greater than the [R.sub.2] for the
model. Thus these predictors make unique but overlapping contributions to the model.

CHANGES IN FAMILY FUNCTIONING WITH TREATMENT

46
Paired t tests for all family functioning variables (i.e., FAM, BSI, PSI, PFQ) were used to
examine whether family functioning changed during the course of treatment. Pre- and
posttreatment mother, child, and father ratings of these factors were used in the analysis
(sample sizes for father ratings vary considerably as fathers were less consistent in their
attendance). Table 3 shows the means and standard deviations for all pre- and
posttreatment scores. When the Bonferroni correction for multiple t tests was applied (p
[less than] .0036), only mother and father reports of frustration and mother-rated
psychopathology remained significant. Child-reported family dysfunction and mother
reports of having a difficult child (PSI Child Domain) also significantly improved with
treatment but were not significant after the Bonferroni correction was applied.

DISCUSSION

This study found that a variety of family factors predicted child treatment outcome.
Furthermore, parental frustration and maternal psychopathology improved over the
course of the treatment program. Predictors of our-come are discussed separately for
each reporter (clinician, mother, child, and father), as correspondence between child
anxiety ratings for different reporters is only fair for internalizing disorders (Manassis et
al., 1997).

Child reports of family dysfunction and parental frustration were significant predictors of
clinician ratings of less improvement in child functioning. These findings suggest that
children who perceive more problems in their families may be less likely to improve with
treatment. Thus, as hypothesized by many investigators of anxiety (Manassis and
Bradley, 1994; Messer and Beidel, 1994; Rapee, 1997), it is possible that family
dysfunction maintains a child's anxiety, which in turn hinders a child's response to
intervention. It is conceivable, however, that family dysfunction and treatment response
both relate to another factor that was not examined in this study.

The frustration-related finding suggests that perceived frustration may exacerbate a

child's anxiety and limit the success of treatment or that children with these perceptions
of their parents (accurate or not) are less responsive to treatment. Anger directed toward
the child, which may be linked to conflict and control, may make the child more fearful
and uncertain of his/her surroundings. Research examining inconsistent caregiving and
control suggests that these factors contribute to a child's anxiety (Chorpita et al., 1998;
Rapee, 1997; Siqueland et al., 1996). In addition, children may perceive that they are the
cause of their parents' frustration and negative family interactions, which in turn may
exacerbate their anxiety.

Mother and father reports of family dysfunction and mother-rated total parenting stress
were significant predictors of maternal reports of improvement in child anxiety. Mothers
who experienced high parenting stress and perceived their family as dysfunctional
continued to see their child as anxious after treatment. High levels of maternal parenting
stress may be related to more dysfunctional parent-child interactions (as hypothesized by
Abidin, 1995), which in turn may interfere with a child's treatment outcome. Alternatively,
mothers who reported family dysfunction and little improvement in the child's anxiety
with treatment may have been more pessimistic reporters.

Child reports of poor family functioning and maternal frustration and father reports of
somatization were predictive of lack of improvement in child-rated anxiety. Father-rated
somatic complaints were most predictive of their child's perceived treatment outcome.
Possible explanations for this finding include child modeling of father maladaptive
behavior (Messer and Beidel, 1994) or encouragement of avoidant behavior by
somatizing fathers (Dadds et al., 1996). Alternatively, anxiety in a family member may
simply reflect higher constitutional vulnerability to anxiety. Messer and Beidel (1994)
hypothesized that paternal psychopathology may also negatively impact family
functioning, which hinders a child's improvement.

47
The results of our research show consistency with outcome studies in adults. Scheibe and
Albus (1997) found that adults with panic disorder who had relapsed reported lower-
functioning households and lower life satisfaction than patients who had not relapsed. In
addition, both Durham et al. (1997) and Lelliott et al. (1987) found that poorer marital
adjustment was predictive of poorer outcome in anxious adults. While marital
relationships do not correspond exactly to family functioning, these adult outcome
studies show that factors outside of an individual's personal characteristics relate to
outcome.

In summary, in all regression analyses, pretreatment family dysfunction predicted poorer

treatment outcome of children with anxiety disorders. For clinician and child outcome
ratings, perceived parental frustration with the child also predicted reduced child
improvement. Finally, paternal somatization was predictive of child-rated outcome. These
results lend credence to the hypothesis that family dysfunction and parental
psychopathology are involved in the development and maintenance of childhood anxiety
(Messer and Beidel, 1994; Rapee, 1997) and suggest that family functioning should be a
target of intervention in anxious children.

Child reports of family dysfunction, mother and father ratings of frustration with their
child, mother psychopathology, and mother reports of having a difficult child (PSI Child
Domain) all significantly improved with treatment. There are two possible explanations
for these findings. First, given that parents underwent a parenting program teaching
them how to understand and help manage child anxiety, it is likely that parents became
more tolerant and less frustrated by their child's behavior. This, in turn, may help to
create more positive parent-child interactions, consequently decreasing family
dysfunction. A second explanation is that the child's anxiety before treatment was
contributing to family dysfunction. Thus, while the treatment was reducing the child's
symptoms, it also created more harmonious parent-child relationships. It is difficult to
ascertain whether a child's treatment outcome is due to treatment-related anxiety
reduction, to the gains in family functioning, or to both.

Although the parenting component of treatment did not target parental psychopathology,
mothers' psychological distress decreased. It is possible that by understanding their
child's symptoms, the mothers gained a better understanding of how to manage their
own distress. Alternatively, it is possible that the interaction between the reduction of
child anxiety and the effects of the parenting component created less psychopathology in
the mother. The direction of these effects is not known.

In conclusion, this investigation demonstrates that aspects of family functioning are

related to anxious children's treatment response. Furthermore, a treatment that does not
specifically address family functioning but includes substantial parent involvement
appears to contribute to more harmonious family interactions.

Limitations

As this was an exploratory study, and little research to date has examined the familial
predictors of treatment outcome in anxious children, these results must be replicated
with a larger sample size. This study, like many others (Manassis et al., 1997; Stark et al.,
1990), found low correspondence among raters of child anxiety. This low agreement
makes it difficult to integrate the findings. Future research should examine the
interpretation of such discrepant findings or a means of combining reports from multiple
raters. The direction of various familial influences (i.e., does poor family functioning cause
or result from child anxiety?) also requires further elucidation. The absence of a control
group prevents us from establishing whether family functioning would have changed
without treatment. Mendlowitz et al. (1999) found no change in anxiety with an 8-week
wait-list control period, but did not examine changes in family functioning. Finally,
samples from more culturally and educationally diverse populatio ns would greatly

48
improve the generalizability of these findings, as the majority of the parents were white
(85%) and had postsecondary education (85%).

Clinical Implications

This study shows that family factors and father-rated somatization can be predictive of a
child's treatment outcome. If other studies find a link between family dysfunction and
poorer treatment outcome, it may be possible to identify, prior to treatment, children who
are at increased risk for treatment failure (Durham et al., 1997). These children may
benefit from more intensive intervention, more parental involvement, treatment of
parental psychopathology or longer-term support (Durham et al., 1997). These results
also suggest that in the treatment of childhood anxiety disorders, therapists must
address issues within the family that may be contributing to and maintaining child
anxiety. It is likely that addressing these family factors will, in turn, increase the
effectiveness of the treatment. The results of this study also support the inclusion of a
parenting component in the treatment of childhood anxiety disorders, as family
functioning improved from pre- to posttreatment.

Ms. Crawford is a doctoral student, Department of Human Development and Applied

Psychology, Ontario Institute for Studies in Education, Toronto. Dr. Manassis is a staff
psychiatrist, Department of psychiatry, Hospital for Sick Children, Toronto, and Associate
Professor of Psychiatry, Department of Psychiatry, University of Toronto.

The authors gratefully acknowledge support from the Ontario Mental Health Foundation.
Special thanks to Lisa Fiksenbaum, David Avery, and Bess Crawford for their help on this
project.

REFERENCES

Abidin RR (1995), Parenting Stress Index: Professional Manual, 3rd ed. Odessa, FL:
Psychological Assessment Resources

Abidin RR (1997), Parenting Stress Index: a measure of the parent-child system. In:
Evaluating Stress: A Book of Resources, Zalaquette PC, Wood RJ, eds. Lanham, MD:
Scarecrow Press, pp 277-291

Abidin RR, Jenkins CL, McGaughey MC (1992), The relationship of early family variables to
children's subsequent behavioral adjustment. J Clin Child Psychol 21:60-69

Bernstein GA, Borchardt CM (1991), Anxiety disorders of childhood and adolescence: a

critical review. J Am Acad Child Adolesc Psychiatry 30:519-532

Chorpita BF, Brown TA, Barlow DH (1998), Perceived control as a mediator of family
environment in etiological models of childhood anxiety. Behav Ther 29:457-476

Dadds MR, Barrett PM, Rapee RM, Ryan S (1996), Family process and child anxiety and
aggression: an observational analysis. J Abnorm Child Psychol 24:715-734

Derogatis LR (1992), BSI: Administration, Scoring and Procedures Manual-II. Baltimore:

Clinical Psychometric Research

Derogatis LR, Lazarus L (1994), SCL-90-R, Brief Symptom Inventory, and matching clinical
rating scales. In: The Use of Psychological Testing for Treatment Planning and Outcome
Assessment, Maruish ME, ed. Hillsdale, NJ: Erlbaum, pp 217-248

Durham RC, Allan T, Hackett CA (1997), On predicting improvement and relapse in

generalized anxiety disorder following psychotherapy. Br J Clin Psychol 36(pt 1):101-119

49
Kendall PC (1990), The Coping Cat Workbook. Merion Station, PA: Temple University

Kendall PC (1994), Treating anxiety disorders in children: results of a randomized clinical

trial. J Consult Clin Psychol 62:100-110

Last CG, Hersen M, Kazdin A, Orvaschel H, Perrin S (1991), Anxiety disorders in children
and their families. Arch Gen Psychiatry 48:928-934

Lelliott PT, Marks IM, Monteiro WO, Tsakiris F, Noshirvani H (1987), Agoraphobies 5 years
after imipramine and exposure: outcome and predictors. J Nerv Ment Dis 175:599-605

Manassis K (1996), Keys to Parenting Your Anxious Child. New York: Barrons Educational
Series

Manassis K, Bradley SJ (1994), The development of childhood anxiety disorders: toward

an integrated model. J Appl Dev Psychol 15:345-366

Manassis K, BradleyS, Goldberg S, Hoodj, Swinson RP (1994), Attachment in mothers with

anxiety disorders and their children. J Am Acad Child Adolesc Psychiatry 33:1106-1113

Manassis K, Hood J (1998), Individual and familial predictors of impairment in childhood

anxiety disorders. J Am Acad Child Adolesc Psychiatry 37:428-434

Manassis K, Tannock R, Mendlowitz S, Laslo D, Masellis M (1997), Distinguishing anxiety

disorders psychometrically. J Am Acad Child Adolesc Psychiatry 36:1645-1646

March JS, Curry JF (1998), Predicting the outcome of treatment. J Abnorm Child Psychol
26:39-511

Mendlowitz SL, Manassis K, Bradley S, Scapillaro D, Miezitis S, Shaw BF (1999), Cognitive-

behavioral group treatments in childhood anxiety disorders: the role of parental
involvement. J Am Acad Child Adolesc Psychiatry 38:1223-1229

Messer SC, Beidel DC (1994), Psychosocial correlates of childhood anxiety disorders. J Am

Acad Child Adolesc Psychiatry 33:975-983

Onatsu-Arvilommi T, Nurmi J, Aunola K (1998), Mothers' and fathers' wellbeing, parenting

styles, and their children's cognitive and behavioral strategies at primary school. Eur J
Psychol Educ 13:543-556

Rapee RM (1997), Potential role of childrearing practices in the development of anxiety

and depression. Clin Psychol Rev 17:47-67

Reich W, Welner Z (1988), Diagnostic Inventory for Children and Adolescents-Revised-

Parent Version. Toronto: Multi-Health Systems

Reynolds CR, Richmond BO (1985), Revised Children's Manifest Anxiety Scale ManuaL Los
Angeles: Western Psychological Services

Ritchie K.L, Holden GW (1998), Parenting stress in low income battered and community
women: effects on parenting behavior. Early Educ Dev 9:97-112

Scheibe G, Albus M (1997), Predictors and outcome in panic disorder: a 2-year

prospective follow-up study. Psychopathology 30:177-184

50
Shaffer D, Gould MS, Brasic J et al. (1983), A children's global assessment scale (CGAS).
Arch Gen Psychiatry 40:1228-1231

Siqueland L, Kendall PC, Steinberg L (1996), Anxiety in children: perceived family

environments and observed family interaction. J Clin Child Psychol 25:225-237

Skinner H, Steinhauer P, Santa-Barbara J (1995), Manual for Family Assessment Measure-

III Toronto: Multi-Health Systems

Stark KD, Humphrey LL, Crook K, Lewis K (1990), Perceived family environments of
depressed and anxious children: child's and maternal figure's perspectives. J Abnorm
Child Psychol 18:527-547

Strauss CC, Frame CL, Forehand R (1987), Psychosocial impairment associated with
anxiety in children. J Clin Child Psychol 16:235-239

Thapar A, McGuffin P (1995). Are anxiety symptoms in childhood heritable? J Child Psychol
Psychiatry 36:439-447

Warren SL, Schmitz S, Emde RN (1999), Behavioral genetic analyses of self-reported

anxiety at 7 years of age. J Am Acad Child Adolesc Psychiatry 38:1403-1408

TABLE 1

Partial Correlations Between Outcome

Measures and Predictor Variables

Outcome Variable

Measure CGAS 1

1. Child-rated FAM -0.59 [**]

2. Child-rated Frustration -0.33 [*] 0.23

Mother
RCMAS 1

1. Mother-rated FAM 0.34 [*]

2. Father-rated FAM 0.40 [**] 0.51 [**]
3. Mother-rated PSI Total Stress 0.32 [*] 0.37 [**]

Child
RCMAS 1

1. Child-rated FAM 0.30 [*]

2. Child-rated Maternal Frustration 0.32 [*] 0.31 [*]
3. Father-rated BSI Somatization 0.35 [*] -0.01

Measure

1. Child-rated FAM
2. Child-rated Frustration

51
 2

1. Mother-rated FAM
2. Father-rated FAM
3. Mother-rated PSI Total Stress 0.36 [**]

1. Child-rated FAM
2. Child-rated Maternal Frustration
2. Father-rated BSI Somatization -0.05

Note: CGAS = Children's Global Assessment

Scale; FAM = Brief Family Assessment Measure-
III; RCMAS = Revised Children's Manifest
Anxiety Scale; BSI = Brief Sympton Inventory;
PSI = Parenting Stress Index.

(*)p [less than] .05;

(**)p [less than] .01.

TABLE 2

Results of Regression on Each Outcome Variable

Partial [R.sup.2] t

Clinician-rated CGAS [a]

Child-rated FAM 0.23 15.21 [**]

Child-rated Frustration 0.12 24.72 [**]

Mother-rated child anxiety (RCMAS) [b]

Mother-rated FAM 0.09 2.33 [*]

Father-rated FAM 0.15 2.92 [**]
Mother-rated FAM 0.12 2.65 [**]

Child-rated anxiety (RCMAS) [c]

Child-rated FAM 0.08 2.25 [*]

Child-rated Maternal Frustration 0.09 2.40 [*]
Father-rated BSI somatization 0.12 2.67 [**]

Note: CGAS = Children's Global Assessment

Scale; FAM = Brief Family Assessment Measure-
III, RCMAS = Revised Children's Manifest
Anxiety Scale; BSI = Brief Symptom Inventory;
PSI = Parenting Stress Index.

(a)Total Sample (n = 44); df = 2,41;

[R.sup.2] for model = 0.28.

52
(b)Total Sample (n = 44); df = 3,40;
[R.sup.2] for model = 0.18.

(c)Total sample (n = 48); df = 3,44;

[R.sup.2] for model = 0.25.

(*)p [less than] .05;

(**)p [less than] .01.

TABLE 3

Means and Standard Deviations for Pre-

and Posttreatment Family Functioning
Variables

Posttreatment
Measure n Mean SD

Child-rated FAM 61 42.6 10.9

Mother-rated FAM 55 52.5 9.3
Father-rated FAM 47 51.5 8.4
Child-rated Frustration 58 11.2 6.1
Mother-rated Frustration 55 17.2 5.2
Father-rated Frustration 46 16.9 6.0
Mother BSI Global Severity Index 54 29.6 27.3
Father BSI Global Severity Index 45 19.1 20.5
Mother-rated Total Parenting Stress 52 66.6 27.7
Father-rated Total Parenting Stress 39 58.8 28.3
Mother-rated PSI Child Domain 52 78.3 24.2
Father-rated PSI Child Domain 42 77.8 24.6
Mother-rated PSI Parent Domain 53 50.5 28.8
Father-rated PSI Parent Domain 40 38.5 24.3

Posttreatment
Measure Mean SD

Child-rated FAM 38.8 [*] 13.5

Mother-rated FAM 51.4 10.1
Father-rated FAM 50.1 9.9
Child-rated Frustration 11.0 6.2
Mother-rated Frustration 13.8 [***] 5.6
Father-rated Frustration 13.6 [***] 6.1
Mother BSI Global Severity Index 22.8 [***] 21.0
Father BSI Global Severity Index 20.6 22.8
Mother-rated Total Parenting Stress 62.8 30.0
Father-rated Total Parenting Stress 61.4 26.6
Mother-rated PSI Child Domain 71.5 [**] 28.2
Father-rated PSI Child Domain 76.0 24.2
Mother-rated PSI Parent Domain 52.7 28.8
Father-rated PSI Parent Domain 44.3 25.9

Note: FAM = Brief Family Assessment

Measure-III; BSI = Brief Symptom

53
Inventory; PSI = Parenting Stress Index.

(*)p [less than] .05;

(**)P [less than] .01;

(***)p [less than] .0036 (Bonferroni

Continued from page 13

correction).

COPYRIGHT 2001 Lippincott/Williams & Wilkins

COPYRIGHT 2001 Gale Group

Conceptions of Relationships in Children with Depressive and Aggressive

Symptoms: Social-Cognitive Distortion or Reality?(Statistical Data Included)

Author/s: Karen D. Rudolph

Issue: Feb, 2001

Karen D. Rudolph [1,2]

Alyssa G. Clark [1]

This research tested skill-deficit and cognitive-distortion models of depression and

aggression in 615 fifth- and sixth-grade children. Children completed a measure of their
generalized conceptions of relationships in the peer domain and their level of depressive
symptoms. Teachers completed measures of social competence, social status, and
aggression. As anticipated, children with higher levels of depressive symptoms, either
alone or in combination with aggression, demonstrated more negative conceptions of
both self and peers than did nonsymptomatic children. Conceptions of relationships did
not differentiate between aggressive and nonsymptomatic children. Children with
depressive symptoms and children with aggressive symptoms displayed unique profiles
of social competence deficits and problematic status in the peer group. Analysis of the
accuracy of children's conceptions of relationships revealed support for both skill-deficit
and cognitive-distortion models. Consistent with a skill-deficit model, children with
depressive and depressive-aggressive symptoms were sensitive to actual differences in
their social status. In contrast, aggressive children showed an insensitivity to social cues.
Consistent with a cognitive-distortion model, children with depressive and depressive-
aggressive symptoms had more negative conceptions than would be expected given their
social status, whereas aggressive-unpopular children demonstrated a self-enhancement
bias. These findings indicate the importance of integrated cognitive-interpersonal models
of depression and aggression that incorporate multiple pathways among social-cognitive,
interpersonal, and emotional functioning.

KEY WORDS: Depression; aggression: cognitions; interpersonal competence; children.

Cognition-Distortion and Skill-Deficit Models of Depression

Both cognitive and interpersonal models have been proposed to elucidate the processes
underlying the development of multiple forms of psychopathology. Interestingly, these
two perspectives may yield contradictory predictions regarding the etiology and
perpetuation of disorder. Cognitive models focus on maladaptive or biased thought
processes, whereas interpersonal models focus on social difficulties and stressful
interpersonal environments as precipitants of psychopathology. Accordingly, cognitive
models often presume that negative views of the self and the world represent

54
inaccuracies or distortions in the appraisal of personal competencies and the social
environment, whereas interpersonal models often presume that negative views of the self
and the world represent accurate reflections of skill deficits and aversive social
circumstances. Despite recent growing attention to integrative cognitive-interpersonal
approaches to developmental psychopathology (Dodge, 1993; Gotlib & Hammen, 1992;
Hammen & Rudolph, 1 996; Rudolph, Hammen, & Burge, 1997; Shirk, Boergers, Eason, &
Van Horn, 1998), this apparent paradox has not been adequately addressed. The goal of
the present research was to reconcile cognitive-distortion and skill-deficit models of
depression and aggression in childhood.

In light of this interest in the intersection between children's cognitive and interpersonal
worlds, a particular focus was placed in this study on children's conceptions of their
interpersonal relationships. Conceptualized under the guise of a variety of constructs,
such as "internal working models" (Bowlby, 1973), "interpersonal or relational schemas"
(Baldwin, 1992; Safran, 1990; Shirk, Van Horn, & Leber, 1997), and "cognitive
representations of relationships" (Rudolph, Hammen, & Burge, 1995), these internalized
constructions of relationships are viewed as cognitive templates that contain generalized
expectations and assumptions about relationships and that guide the processing of
incoming social information. Negative conceptions of self and significant others within an
interpersonal context have been implicated as a vulnerability factor for depression (e.g.,
Cummings & Cicchetti, 1990; Kaslow, Rehm, & Siegel, 1984; Rudolph et al., 1997; Shirk et
al., 1997, 1998). For example, children who believe that th ey are unworthy of positive
social attention and who view their peers as untrustworthy or hostile may be susceptible
to low self-worth, anhedonia, hopelessness, sad affect, and other symptoms of
depression. Moreover, children who enter novel interpersonal situations with pessimistic
expectations may selectively attend to and recall negative aspects of these encounters
and may show negatively biased interpretations of interpersonal transactions. This focus
on negative information may then stimulate depressive symptoms. In support of this
proposal, poor social self-concept, negative views of others, and biased interpersonal
information processing have been linked to concurrent (Armsden & Greenberg, 1987;
Kaslow et al., 1984; Quiggle, Garber, Panak, & Dodge, 1992; Rudolph et al., 1997; Shirk et
al., 1997) as well as future (Shirk et al., 1998) depressive symptoms.

Typically, conceptions of relationships have been viewed as developmental sequelae

stemming from early socialization experiences (e.g., Bowlby, 1973; Main, Kaplan, &
Cassidy, 1985; Rudolph et al., 1995). Consequently, the negative perceptions
characteristic of depressed children often are assumed to represent biases or distortions
in thinking that are generalized from early adverse interpersonal encounters. However,
interpersonal theories of depression (e.g., Barnett & Gotlib, 1988; Coyne, 1976;
Lewinsohn, 1974) highlight the critical role of ongoing interpersonal difficulties in the
onset and maintenance of disorder. Indeed, a wealth of research demonstrates the
presence of significant social impairment in depressed youngsters (see Gotlib & Hammen,
1992; Hammen & Rudolph, 1996; Weisz, Rudolph, Granger, & Sweeney, 1992, for
reviews). Depression has been found to be associated with maladaptive interpersonal
problem-solving styles, including hostility and withdrawal (e.g., Kennedy, Spence, &
Hensley, 1989; Qui ggle et al., 1992; Rudolph, Hammen, & Burge, 1994), lower rates of
prosocial activity and higher rates of aversive behavior in the peer group (Altmann &
Gotlib, 1988; Rudolph et al., 1994), and poorer quality friendships (Goodyer, Wright, &
Altham, 1990). Moreover, several studies have revealed that depressed children elicit
negative reactions from peers during dyadic interactions (Baker, Milich, & Manolis, 1996;
Connolly, Geller, Marton, & Kutcher, 1992; Rudolph et al., 1994) and are less accepted by
peers than are nondepressed children (e.g., Cole, 1990; Patterson & Stoolmiller, 1991).
Thus, considerable evidence indicates that depressed children demonstrate social
competence deficits and encounter negative interpersonal environments in their
everyday lives.

Taken together, these two patterns of findings--the presence of negative conceptions of

self and others within interpersonal relationships and the presence of multiple social
difficulties--present an intriguing puzzle for researchers interested in integrative

55
cognitive-interpersonal models of depression: Do the negative conceptions of
relationships displayed by depressed children represent biased evaluations of themselves
and the world around them or veridical reports of personal incompetencies and
interpersonal realities? Given the large body of research demonstrating social difficulties
in depressed youngsters, a reasonable conclusion may be that these negative
conceptions are quite accurate. In fact, Cole's (1991; Cole, Martin, & Powers, 1997; Cole
& Turner, 1993) competence-based theory of depression posits that the self-deprecating
beliefs of depressed children reflect the internalization of negative feedback from the
environment that stems from competence deficits. In support of this model, self-percei
ved competencies (in multiple domains, including peer relationships) have been found to
account for the association between competence appraisals by others and depression
(Cole & Turner, 1993; Cole et al., 1997). Alternatively, however, the appraisals of
depressed children may represent exaggerated accounts of true negative
circumstances--that is, depressed children may distort social information above and
beyond their actual interpersonal difficulties or may focus more on negative interpersonal
feedback than do nondepressed children (see Weisz et al., 1992).

Few investigators have sought to discriminate these alternatives. Moreover, three studies
that did directly examine the accuracy of social appraisals in depressed children yielded
contradictory findings. Proffitt and Weisz (1992) found that depression was associated
with accurate, albeit negative, self-perceptions of social competence, whereas Kendall,
Stark, and Adam (1990) found that depression was associated with the underestimation
of competence, including popularity. In the only longitudinal study to address this issue
(Cole, Martin, Peeke, Seroczynski, & Hoffman, 1998), depression was linked to the
underestimation of social competence, although this cognitive tendency was found to be
a sequelae rather than a predictor of depression. The present study included an in-depth
examination of conceptions of relationships and teacher-rated interpersonal competence,
and used a novel approach to assess the accuracy of depressed children's conceptions,
as described later.

Cognition-Distortion and Skill-Deficit Models of Aggression

Cognitive models of aggression (Crick & Dodge, 1994; Dodge, 1986, 1993) also implicate
deficits and distortions in the processing of interpersonal information and dysfunctional
interpersonal schemas as precursors of disorder. According to these models, aggression
is associated with maladaptive patterns of encoding, interpretation, and retrieval of
interpersonally relevant information. A wealth of research has examined information-
processing biases and social perceptions in aggressive children (see Crick & Dodge, 1994;
Dodge, 1993; Garber, Quiggle, Panak, & Dodge, 1991, for reviews). A number of studies
have documented no differences between aggressive and nonaggressive children in
general social self-concept (e.g., Hymel, Bowker, & Woody, 1993; Hymel, Rubin, Rowden,
& LeMare, 1990; Hughes, Cavell, & Grossman, 1997; Patterson, Kupersmidt, & Griesler,
1990), general appraisals of peers (Hughes et al., 1997; Rabiner, Keane, & MacKinnon-
Lewis, 1993), or appraisals of self and peers within specific interpersonal situations
(Lochman & Dodge, 1998).

Although studies have not revealed differences in the absolute level of social self-concept
in aggressive and nonaggressive children, aggression has been linked to inaccuracies in
self-perception. In contrast to the deprecating self-conceptions held by depressed
children, aggressive children, particularly those who experience peer rejection, have been
shown to possess inflated self-conceptions relative to the appraisals of others. For
instance, unpopular or rejected aggressive children overestimate their social competence
and acceptance by peers (e.g., Boivin, Poulin, & Vitaro, 1994; Hughes et al., 1997; Hymel
et al., 1993; Patterson et al., 1990) and underestimate their social rejection (Zakriski &
Coie, 1996) and peer-directed aggression (Lochman, 1987; Lochman & Dodge, 1998).
Moreover, aggressive children consistently have been found to display a tendency toward
negatively biased processing of information about their peers, particularly in the context
of ambiguous, self-relevant social information (e.g., Dodge & Frame, 1982; Quiggle et al.,
1992; see Crick & Dodge, 1994, Dodge, 1993, for reviews).

56
In stark contrast to these optimistic self-conceptions of social relatedness, research has
revealed significant social difficulties in aggressive children. Aggression is associated with
maladaptive interpersonal problem-solving styles characterized by more coercive and
aggressive responses and fewer assertive and prosocial responses (e.g., Dodge, Pettit,
McClaskey, & Brown, 1986; Quiggle et al., 1992; see Crick & Dodge, 1994; Garber et al.,
1991, for reviews). Not surprisingly, aggressive children display higher levels of aversive,
impulsive, and uncooperative behavior and lower levels of prosocial behavior in the peer
group (e.g., Bierman, Smoot, & Aumiller, 1993; Hymel et al., 1993) and experience
decreased peer acceptance and increased peer rejection (e.g., Boivin & Hymel, 1997;
Little & Garber, 1995; see Asher & Coie, 1990; Dodge & Richard, 1985, for reviews).

Once again, integrating cognitive and interpersonal aspects of aggression requires a

consideration of the origins of interpersonal information-processing styles and
conceptions of self and others. On the one hand, the social-cognitive orientation of
aggressive children may have its roots in early socialization experiences. According to
this perspective, expectations of hostility from others that emerge from negative early
interpersonal transactions are inappropriately generalized to novel social encounters,
leading to misattributions about the intentions of others (Dodge, 1993; Lochman &
Dodge, 1998; Rogosch, Cicchetti, & Aber, 1995). Moreover, self-enhancement biases and
denial of social inadequacy may be viewed as a by-product of self-protective mechanisms
developed in response to early experiences with interpersonal threat (Hughes et al.,
1997; Zakriski & Coie, 1996). This self-protective bias may then lead to an underdetection
or underutilization of social cues in the formation of self-perceptions. For i nstance,
aggressive children have been found to be insensitive to self-directed rejection feedback
(Zakriski & Coie, 1996) and to base self-perceptions of aggression on their prior
expectations rather than on their actual behavior (Lochman & Dodge, 1998). This lack of
integration of situational cues into their self-conceptions may account for the
maintenance of optimistic self-views in the face of disconfirmatory feedback, such as
rejection by peers.

On the other hand, social perceptions may emerge from the internalization of ongoing
interpersonal experiences (Boivin & Hymel, 1997; Crick & Ladd, 1993; Zakriski & Coie,
1996). In this case, the hostile attributions displayed by aggressive children may merely
reflect their aversive peer-relationship history and current negative feedback from peers.
For instance, aggressive children are more frequent recipients of aggression and
victimization by peers (Boivin & Hymel, 1997; Hughes et al., 1997). Interestingly, it also
has been suggested that the positive self-conceptions of aggressive-rejected children
may in part result from the inhibition of feedback by peers to this particular subgroup due
to fear of retaliation. Thus, peer attitudes of disliking may not always be overtly conveyed
to aggressive children, allowing for the perpetuation of a positive self-image (Zakriski &
Coie, 1996).

Unraveling these complex linkages between cognitive and interpersonal aspects of

aggression therefore requires the simultaneous consideration of children's actual
interpersonal experiences and their perceptions of these experiences. How can research
reveal both comparable levels of self-perceptions in aggressive and nonaggressive
children, as well as self-enhancement biases in aggressive children? In fact, the observed
lack of discrimination in self-perceptions, in the context of the problematic social
circumstances of aggressive children, actually indicates a perceptual bias. That is, a
failure to acknowledge very real interpersonal difficulties, either because of a lack of
sensitivity or a reinterpretation of social cues, may account for the inflated self-
perceptions of aggressive-rejected children. To examine whether aggressive children
were sensitive to social cues (i.e., whether their perceptions reflected their social status),
the present study compared the conceptions of aggressive-accepted and aggres sive-
unpopular children. To examine whether aggressive children overestimated their social
competence in the context of low peer acceptance (i.e., whether they engaged in
cognitive distortion), the present study compared the conceptions of aggressive-
unpopular and nonsymptomatic-unpopular children. This joint examination of sensitivity

57
to skill deficits and engagement in cognitive distortion united these two lines of research
to provide a unique perspective on aggression.

Overview of the Present Research

The goal of this study was to examine the validity of skill-deficit and cognitive-distortion
models of depression and aggression. To determine the accuracy of children's
conceptions of relationships, we evaluated children's generalized perceptions of self and
peers in comparison to teacher reports of the social experiences encountered in
children's everyday lives, as reflected in their status in the peer group. In contrast to prior
research, which often has used sociometric categories derived from peer nomination
procedures as a proxy for children's interpersonal environments (e.g., Hymel et al., 1993;
Rabiner et al., 1993), we used teacher reports of peer social status. Because peer-based
sociometric measures reflect only the attitudes of the peer group, it is possible that
children categorized into unpopular groups are not subject to the overt expression of
negative feedback (Boiven & Hymel, 1997). Given our interest in children's use of social
cues in the formation of conceptions of relationships, it was essential to employ an index
that more likely captured behavioral manifestations of unpopularity. Children identified
by teachers as unpopular were likely to be overtly exposed to aversive social
environments, either because they clearly had few friends and playmates or because
peers were actively mean or hostile toward them. This approach also allowed us to refute
the possibility that positive self-perceptions in aggressive children were due to an
absence of negative feedback from peers (see Zakriski & Coie, 1996).

The first step was to examine symptom group differences in children's conceptions of
relationships, social behavior, and social status. We expected that children with
depressive symptoms would demonstrate more negative conceptions of self and peers
than would nonsymptomatic children. In light of evidence suggesting that aggressive and
nonaggressive children do not differ in their general social self-concept or general
appraisals of peers, we expected that conceptions of relationships in aggressive children
would be consistent with those in nonsymptomatic children. We anticipated that teachers
would report significant social impairment in children with depressive and aggressive
symptoms. Both groups were expected to demonstrate lower levels of prosocial behavior
and to be less popular than were nonsymptomatic children, according to teachers.
Teacher reports also were expected to reveal that children with depressive symptoms
demonstrate higher levels of withdrawn behavior and are neglected by peers, whereas ag
gressive children demonstrate higher levels of aggressive and disruptive behavior and
are rejected by peers.

The second step was to examine the accuracy of children's conceptions of relationships in
light of their interpersonal experiences, as assessed by teachers. Consideration of
children's social environments afforded the opportunity to detect whether the
conceptions of depressed and aggressive children reflected a realistic or distorted
account of their interpersonal circumstances. Consistent with a skill-deficit model, if
conceptions of self and peers are an accurate depiction of children's personal
competencies and interpersonal experiences, we would expect that conceptions would
vary across teacher-reported social status categories, with unpopular children showing
more negative conceptions than accepted children. Consistent with a cognitive-distortion
model, if conceptions of self and peers are a biased depiction of reality, we would expect
that conceptions would diverge in children receiving similar social feedback, as reflected
in their teacher-reported social status, with depressed and aggressive children showing
different conceptions from their nonsymptomatic counterparts.

Based on theory and research suggesting that depressed children internalize negative
feedback from the environment (e.g., Cole et al., 1997), we predicted that these children
would be sensitive to social cues provided by peers and, therefore, their conceptions of
relationships would be somewhat consistent with their everyday interpersonal
experiences, as reflected in their status in the peer group. However, because depressed
children have been found to display interpersonal information-processing biases even

58
when presented with similar social information (e.g., Rudolph et al., 1997; Shirk et al.,
1997), we also expected that their conceptions would be more negative than was
warranted by their social status.

Based on theory and research suggesting that aggressive children underutilize feedback
from the environment (Dodge & Newman, 1981; Zakriski & Coie, 1996), we expected that
these children would be insensitive to social cues and, therefore, their conceptions of
relationships would not be consistent with their everyday interpersonal experiences, as
reflected in their status in the peer group. Moreover, we expected that aggressive-
unpopular children would show a self-enhancement bias, reflected in more positive
conceptions of self than their nonsymptomatic counterparts. However, in light of research
demonstrating that the biases of aggressive children are specific to situations or
feedback relevant to the self (e.g., Lochman, 1987; Zakriski & Coie, 1996), we did not
anticipate a bias in the peer conceptions of aggressive-unpopular children.

The final issue addressed in this study concerned the role of co-occurring symptoms in
skill-deficit and cognitive-distortion models of depression and aggression. Children with
co-occurring depression and aggression may manifest several different patterns of
cognitive and interpersonal characteristics. For instance, these children may demonstrate
features of one or both symptom groups. Alternatively, this group may demonstrate
features that are quantitatively or qualitatively different from children with either
depression or aggression alone (see Garber et al., 1991). Minimal research is available
regarding the social-cognitive styles of children with both depression and aggression. In
one study (Quiggle et al., 1992), this group was found to demonstrate the cognitive
characteristics of both groups. In the current study, however, making predictions about
the conceptions of relationships in children with co-occurring symptoms was more
complex. For example, we expected that children with depressive symptoms an d
children with aggressive symptoms would show opposite distortions in self-conceptions,
namely self-deprecation in depressed children and self-enhancement in aggressive
children. Because the conceptions of aggressive children were hypothesized to result
from a cognitive deficit (i.e., lack of sensitivity to social cues) whereas the conceptions of
depressed children were hypothesized to result from a cognitive excess (i.e.,
oversensitivity to negative social cues), we predicted that the conceptions of children
with both depression and aggression would be more likely to mirror those of depressed
children. In line with prior research on the interpersonal competence of children with co-
occurring depression and externalizing symptoms, such as aggression and conduct
problems (e.g., Asarnow, 1988; Cole & Carpentieri, 1990; Rudolph et al., 1994), we
predicted that this group would demonstrate the highest level of teacher-rated social
impairment, particularly hostile and aversive behavior and peer rejection.

METHOD

Participants

The participants included 615 fifth and sixth graders (313 girls, 302 boys; Mage = 11.5
years, SD = .68) recruited from elementary and secondary schools in several school
districts in the midwest. This sample represented 93% of fifth and sixth graders in the
targeted schools. The ethnic composition was 67.0% Caucasian, 28.5% African American,
2.0% Asian American, 1.3% Latino/a, .2% Native American, and 1.0% other. Based on the
available data, 50.3% of the children were receiving free or reduced cost lunch, indicating
heterogeneity in the socioeconomic status of the sample.

Procedures

Trained research assistants administered questionnaires to children during two or three

classroom administration sessions. Questions were read aloud, while children provided
written responses. Teachers completed questionnaires on 99% of the participating
children, yielding the present sample.

59
Measures

Conceptions of Relationships

Conceptions of relationships were assessed with the Perceptions of Peers and Self
Questionnaire (POPS; Rudolph et al., 1995). The peer subscale measures children's
generalized perceptions of peers and friendships, along dimensions such as
dependability, supportiveness, and empathy (e.g., "Other kids cannot be trusted."
"Friends usually stick up for you when you're in trouble."). The self subscale measures
children's generalized perceptions of social self-worth (e.g., "It's a waste of other kids'
time to be friends with me.") and social self-competence (e.g., "I am good at making
other kids laugh."). The original version of this measure included 12 peer items and 15
self items; this revised version consisted of 15 items on each subscale. Children rated on
a 4-point scale (1 = Not at All to 4 = Very Much) the degree to which each statement was
descriptive of their peers or themselves. Scores were calculated as the mean of the 15
peer items ([alpha] = .74) and the 15 self items ([alpha] = .78). Higher scores re flect
more negative conceptions of relationships.

The POPS has been used in prior research examining children's conceptions of
relationships (Rudolph et al., 1995, 1997). Both the self and peer subscales showed
significant stability over a 6-month interval in the present sample (rs = .47 - .62, ps [less
than] .001). Significant correlations have been found between the POPS and other
measures of conceptions of relationships, including interpersonal expectancies and
appraisals of peer social support (Rudolph et al., 1995).

Social Behavior

Children's social behavior in the peer group was assessed with the Teacher Assessment
of Social Behavior (TASB; Cassidy & Asher, 1992). Factor analyses of the TASB (Cassidy &
Asher, 1992) have yielded four factors: Prosocial (e.g., "This child is friendly and nice to
other children."), Withdrawn (e.g., "This child is shy/withdrawn."), Aggressive (e.g., "This
child is mean to other children."), and Disruptive (e.g., "This child disrupts other children's
activities."). For each item, teachers rated children on a 5-point scale (1 = Very
Uncharacteristic to 5 = Very Characteristic). Scores were calculated as the mean of the
three items on the Prosocial ([alpha] = .89), Withdrawn ([alpha] = .61), Aggressive
([alpha] = .86), and Disruptive ([alpha] = .88) subscales. Because the Aggressive and
Disruptive subscales were highly correlated, r(612) = .78, p [less than] .001, and we
hypothesized that similar results would emerge for the two subscales, they were
averaged into a single score. Higher scores indicate higher levels of each type of social
behavior.

Scores on the TASB subscales were relatively stable across a 6-month interval in the
present sample (rs = .32 - .55, ps [less than] .001). In prior research, strong correlations
have been found between teacher reports on the TASB and peer reports of the same
social behaviors (rs = .40-.75, ps [less than] .001) Cassidy & Asher, 1992). Moreover,
sociometric groups determined from peer nomination procedures have been found to
differ in the expected ways on prosocial, withdrawn, aggressive, and disruptive behavior
as assessed with the TASB (Cassidy & Asher, 1992).

Social Status

Teachers completed two measures of children's social status in the peer group. First,
teachers were asked to endorse one of five mutually exclusive social status categories for
each child: (a) Social Star (n = 141), (b) Average (n = 299), (c) Rejected or Disliked (n =
40), (d) Neglected or Ignored (n = 75), and (e) Controversial (n = 60) [3]. For each
category, descriptions were provided that mapped onto traditional definitions of these
categories as reflected in peer nomination procedures. For example, a social star was
described as a child "who is liked by most playmates and who has almost no real

60
enemies." These groups were collapsed in line with the hypotheses for the present study.
Because our central goal was to compare accepted and unpopular children, we combined
the social stars and average groups into a single accepted category, and we combined
the neglected and rejected groups into a single unpopular category. Second, teachers
rated each child's level of popularity, neglect, and rejection on a 7-poi nt scale (1 = Not at
All to 7 = Extremely).

In support of our group formation, social stars versus average children and neglected
versus rejected children did not differ in their conceptions of self, ts [less than] 1.72, ns,
or peers, ts [less than] 1.37, ns. Furthermore, the accepted group was more popular, less
neglected, and less rejected than the unpopular group, ts [greater than] 13.07, ps [less
than] .001, based on the social status ratings. The two social status categories also
differed in the expected ways on social behavior (i.e., the accepted group showed more
prosocial behavior and less withdrawn and aggressive/disruptive behavior than did the
unpopular group, ts [greater than] 6.07, ps [less than] .001). In the present sample, social
status ratings completed by different teachers were significantly correlated across a 6-
month interval spanning a transition to a new grade (rs = .26 -.45, ps [less than] .001).
Moreoever, teacher ratings of popularity, neglect, and rejection were significantly
associated in the expected direction with children 's perceptions of the degree of stress
experienced within peer relationships (rs = .22 - .26, ps [less than] .001). These
categorical and continuous teacher-report measures of social status also have been
validated in past research (e.g., Cole, 1990; Rudolph et al., 1994, 1997). For example,
teacher and peer ratings of popularity have been found to be highly correlated, r(107)
= .64, p [less than] .001 (Jacobsen, Lahey, & Strauss, 1983).

Depressive Symptoms

Depressive symptoms were assessed with the Children's Depression Inventory (CDI;
Kovacs, 1980/81). Each of the 27 items presents three alternatives representing varying
levels of symptom severity. Children indicated which level best described their
experiences in the past two weeks. The CDI has well-established reliability and validity
(Kovacs, 1980/81; Smucker, Craighead, W. E., Craighead, L. E., & Green, 1986). High
internal consistency was found in this sample ([alpha] = .90). Scores ranged from 0 to 50
(M = 10.24, SD = 8.53), indicating a wide range of depressive symptoms

Aggressive Symptoms

Aggressive symptoms were assessed with the Aggression subscale of the Teacher Report
Form of the Achenbach Child Behavior Checklist (TRF; Achenbach, 1991). Teachers rated
on a 3-point scale (0 = Not True to 2 = Very True or Often True) the severity and/or
frequency of conduct problems displayed by children (e.g., "destroys property belonging
to others," "disobedient at school," "bragging/boasting"). Given the range of problems
included in this subscale, the overlap between aggressive symptoms and aggressive
behavior, as assessed by the TASB, was minimal. This measure yields T scores with a
mean of 50 and a standard deviation of 10. Scores ranged from 50 to 100 (M = 55.25, SD
= 8.46), indicating a wide range of aggressive symptoms.

RESULTS

Symptom Group Formation

Cut-off scores were selected on the CDI and TRF for the purpose of forming symptom
groups. Following recommendations by the authors (Kovacs, 1983) and previous research
(e.g., Altmann & Gotlib, 1988; Quiggle et al., 1992), a score of 13 and above on the CDI
was used as a cut-off for moderate depressive symptoms. A score of 60 and above (i.e., 1
SD above the mean) on the TRF was used as a cut-off for moderate aggressive symptoms
(see Kendall & Fischler, 1984). Four symptom groups were created based on these cut-off
scores: (a) Nonsymptomatic: children scoring below the cutoffs on both depression and

61
aggression (n 347; Mean CDI = 5.30, SD 3.42; Mean TRF = 51.60, SD = 2.58), (b)
Depressed: children scoring at or above the cut-off on depression and below the cut-off
on aggression (n = 140; Mean CDI = 19.91, SD =7.07; Mean TRF = 52.12, SD = 2.95), (c)
Aggressive: children scoring at or above the cut-off on aggression and below the cut-off
on depression (n = 79; Mean CDI = 7.18, SD = 3.21; Mean TRF = 67.78, SD = 8 .23), and
(d) Depressed-Aggressive: children scoring at or above the cut-offs on both depression
and aggression (n = 49; Mean CDI = 22.56, SD = 8.13; Mean TRF = 69.90, SD = 11.41).
Notably, the mean depression scores in the depressed and depressed-aggressive groups
and the mean aggression scores in the aggressive and depressed-aggressive groups
indicated levels of symptoms that typically are viewed as severe. [4]

Preliminary Analyses on Demographic Variables

We conducted a series of chi-square analyses to examine the association between

demographic variables (i.e., sex, ethnicity, and school lunch status) and group placement.
Significant associations were found between ethnicity and symptom group, [[chi].sup.2]
(l5) = 66.91, p [less than] .001. Specifically, African Americans were overrepresented,
relative to their representation in the total sample, in the aggressive (58%) and
depressed-aggressive (51%) groups. Significant associations also were found between
school lunch status and both symptom group, [[chi].sup.2](3) = 12.70, p [less than] .01,
and social status category, [[chi].sup.2](1) = 11.04, p [less than] .005. Specifically,
children receiving a free or reduced cost lunch were overrepresented, relative to their
representation in the total sample, in the aggressive (63%) and depressed-aggressive
(74%) groups, and were slightly underrepresented in the nonsymptomatic group (44%).
These children also were overrepresented in the unpopular group (64%) and were
underrepres ented in the accepted group (44%). No significant associations were found
for sex. In light of these group differences, we examined the impact of ethnicity and
school lunch status in later analyses.

Symptom Group Differences in Conceptions of Relationships, Social Behavior, and Social

Status

The first set of analyses examined symptom group differences in children's conceptions
of relationships, social behavior, and social status. Three multivariate analyses of
variance (MANOVAs) were conducted with symptom group (Nonsymptomatic, Depressed,
Aggressive, Depressed-Aggressive) as a between-subjects factor and conceptions of self
and peers, dimensions of social behavior, and continuous ratings of social status as
dependent variables. Initially, ethnicity and lunch status were included as between-
subjects factors. Neither ethnicity nor lunch status significantly interacted with symptom
group in the prediction of conceptions of relationships, social behavior, or social status.
Moreover, MANOVAs that controlled for these variables yielded the same results. Thus, all
analyses collapsed across demographic variables. Significant multivariate effects using
the Wilks' lambda criterion were followed with post hoc comparisons (Tukey's HSD) of the
four symptom groups. Because directional hypotheses were made, o ne-tailed
significance levels are reported.

Conceptions of Relationships

A significant multivariate effect was found for conceptions of relationships, F(6, 1212) =
21.04, p [less than].001 (see Table I for means, standard deviations, pairwise
comparisons, and effect sizes). As predicted, comparisons revealed that children in the
depressed and depressed-aggressive groups demonstrated more negative conceptions of
both self and peers than did nonsymptomatic and aggressive children. Group differences
were moderate to large, as reflected in the effect sizes (.54 - 1.02). As anticipated,
conceptions of self and peers did not differentiate between the nonsymptomatic and
aggressive groups or between the depressed and depressed-aggressive groups, as
reflected in the small effect sizes (.00 - .15).

62
Social Behavior

A significant multivariate effect was found for social behavior, F(9, 1482) = 53.39, p [less
than].001 (see Table II for means, standard deviations, pairwise comparisons, and effect
sizes). Consistent with predictions, teachers reported that nonsymptomatic children
displayed the highest level of prosocial behavior and children in the depressed-aggressive
group displayed the lowest level of prosocial behavior, although the depressed-
aggressive group did not significantly differ from the aggressive group. Children with
depressive symptoms displayed significantly more prosocial behavior than did aggressive
children. Furthermore, as expected, children with depressive symptoms displayed more
withdrawn behavior than did nonsymptomatic children, and marginally more withdrawn
behavior than did aggressive children. The depressed and depressed-aggressive groups
did not differ significantly in their level of withdrawn behavior. Finally, consistent with
predictions, children in the depressed-aggressive group displayed t he highest level of
aggressive/disruptive behavior, although they did not significantly differ from the
aggressive group, and nonsymptomatic children displayed the lowest level of
aggressive/disruptive behavior. Aggressive children displayed significantly more
aggressive/disruptive behavior than did children with depressive symptoms. Once again,
effect sizes for significant differences were moderate to large (.25 - 2.50), whereas effect
sizes for nonsignificant differences were small (.03 - .17).

63
Social Status

A significant multivariate effect was found for social status ratings, F(9, 1480) = 8.66, p
[less than].001 (see Table III for means, standard deviations, pairwise comparisons, and
effect sizes). As predicted, nonsymptomatic children were rated as the most popular and
children in the depressed-aggressive group were rated as the least popular, although the
depressed-aggressive group did not significantly differ from the aggressive group.
Depressed and aggressive children did not significantly differ in their level of popularity.
Furthermore, nonsymptomatic children were rated as the least neglected but, contrary to
expectations, children in the depressed-aggressive group were rated as the most
neglected, although they did not significantly differ from the aggressive group. Depressed
and aggressive children did not significantly differ in their level of neglect. Finally,
consistent with predictions, children in the depressed-aggressive group were rated as the
most rejected, although they did not significantly differ from the aggressive group, and
nonsymptomatic children were rated as the least rejected. Aggressive children were rated
as significantly more rejected than were children with depressive symptoms. Effect sizes
for significant differences were moderate to large (.30-.99), whereas effect sizes for
nonsignificant differences were small (.17-.25).

Accuracy of Conceptions of Relationships

The second set of analyses examined the accuracy of children's conceptions of

relationships. As noted earlier, social status categories were employed as a proxy for
children's daily social environments. Because we were interested in examining children's
sensitivity to social cues, and the cues received by the controversial group were
ambiguous (that is, they likely received both positive and negative cues), we focused in
the accuracy analyses on the accepted and unpopular groups. Dropping the controversial
group also restricted the analyses to those that were most important to testing our
hypotheses, thereby reducing the number of comparisons.

We conducted a 4 x 2 MANOVA with symptom group (Nonsymptomatic, Depressed,

Aggressive, Depressed-Aggressive) and social status category (Accepted, Unpopular) as
independent variables, and conceptions of self and peers as dependent variables. This
analysis yielded a significant multivariate interaction, F(6, 1084) = 2.07, p [less than] .05,
suggesting that conceptions of relationships in the two social status categories differed
across symptom groups. To test skill-deficit and cognitive-distortion models, a series of
planned comparisons was conducted across and within social status categories (see Table
IV for means, standard deviations, planned comparisons, and effect sizes).

Skill-Deficit Model

To test a skill-deficit model, we compared the negativity of depressed and aggressive

children's conceptions of relationships across social status categories. The presence of
differences in conceptions of relationships that mapped onto social status categories
would suggest that children are sensitive to social cues provided by their peer
environment. In this case, negative conceptions of self and peers may be, at least in part,
a realistic appraisal of competence problems and an aversive social environment. As
predicted, comparisons revealed that depressed-unpopular children endorsed
significantly more negative conceptions of self and peers than did depressed-accepted
children, and depressed-aggressive-unpopular children endorsed significantly more
negative conceptions of self than did depressed-aggressive-accepted children (effect
sizes = .48-.69). Conceptions of relationships did not differentiate significantly between
aggressive children across the two social status categories (effect sizes = .05-.26).

Because our hypothesis concerning the conceptions of aggressive children reflected the
null hypothesis (i.e., the conceptions of aggressive-accepted versus aggressive-unpopular
children would not differ), we conducted some additional planned interactions to examine
whether the discrepancy between social status categories was significantly greater in the

64
depressed and depressed-aggressive groups than in the aggressive group. As expected,
Symptom Group (Depressed, Aggressive) x Social Status Category (Accepted, Unpopular)
interactions were found for conceptions of self, F(1, 182) = 7.19, p [less than] .01, and
peers, F(1, 182) = 2.36, p = .06. To demonstrate the magnitude of these differences, the
discrepancy between the mean score in the accepted category and the mean score in the
unpopular category was calculated within each symptom group. The discrepancies for self
and peers in the depressed group (.31 and .21, respectively) were two to three times as
big as the discrepancies for self and peers in the aggre ssive group, which were in the
opposite direction and close to zero (-.11 and -.02, respectively). Similarly, a Symptom
Group (Depressed-Aggressive, Aggressive) x Social Status Category (Accepted,
Unpopular) interaction was found for conceptions of self, F(1, 82) = 5.20, p [less
than] .05. Specifically, the discrepancy for self in the depressed-aggressive group (.36)
was more than three times as big as the discrepancy for self in the aggressive group
(-.11).

Cognitive-Distortion Model

To test a cognitive-distortion model, we compared the negativity of children's

conceptions of relationships within social status categories. The presence of differences
between conceptions of relationships in nonsymptomatic and symptomatic children
within the same social status category would suggest that some children may be more or
less negative about themselves and their peers than is warranted by their social
environment. As expected, comparisons revealed that depressed children endorsed
significantly more negative conceptions of self and peers than did nonsymptomatic
children within the same social status category (ts [greater than] 3.56, ps [less
than] .001; effect sizes = .60-1.29). That is, the conceptions of self and peers in the
depressed-accepted and depressed-unpopular groups, respectively, were more negative
than those in the nonsymptomatic accepted and nonsymptomatic-unpopular groups.
Similarly, children in the depressed-aggressive group endorsed more negative
conceptions of self and peers than did nonsymptomatic children within the same social
status category (ts [greater than] 2.43, ps [less than] .01; effect sizes = .65 to 1.20).
Conceptions of self and peers in the aggressive children did not significantly differ from
those of the nonsymptomatic children within the same social status category (ts [less
than] 1.01, ns; effect sizes = .l0-.33), with one exception. As predicted, aggressive-
unpopular children endorsed significantly more positive conceptions of self than did
nonsymptomatic-unpopular children (t = 1.98, p [less than] .05; effect size = .67); the
level of their self-conceptions was, in fact, very similar to that of the nonsymptomatic-
accepted group.

Once again, because our hypotheses concerning the conceptions of self and peers in
aggressive-accepted children and the conceptions of peers in aggressive-unpopular
children reflected the null hypothesis, we examined the discrepancy scores to provide
information about group differences. Specifically, the discrepancy between the mean
score in each symptomatic group and the mean score in the nonsymptomatic group was
calculated within each social status category. In accepted children, the discrepancies
between the depressed and depressed-aggressive groups versus the nonsymptomatic
group (.24 to .37) were four to nine times as big as the discrepancies between the
aggressive and nonsymptomatic groups, which were close to zero (.04 and .06). For
conceptions of peers in unpopular children, the discrepancies between the depressed and
depressed-aggressive groups versus the nonsymptomatic group (.26 and .32) were two to
three times as big as the discrepancy between the aggressive and nonsymptomatic
groups (-.11).

65
Supplemental Analyses

In a supplemental set of analyses, we used an alternative approach to examine the role

played by children's social environment, as reflected in their social status, in accounting
for the association between conceptions of relationships and depressive and aggressive
symptoms. To this end, we conducted partial correlations to determine the effect of
controlling for social status ratings (i.e., popularity, neglect, rejection) on the relation
between self and peer conceptions and symptoms.

Depressive Symptoms

As expected, negative conceptions of self, r(607) = .49, p [less than] .001, and peers,
r(610) = .33, p [less than] .001, were found to be significantly associated with higher
levels of depressive symptoms, adjusted for aggressive symptoms. After controlling for
the social status ratings, the associations between negative conceptions of both self,
r(604) = .47, p [less than] .001, and peers, r(607) = .31, p [less than] .001, and
depressive symptoms remained significant and almost unchanged, suggesting that
children's social status did not explain the relation between negative conceptions of
relationships and depressive symptoms. Thus, this finding is consistent with the previous
pattern of results indicating that children with depressive symptoms show negative
conceptions above and beyond those that would be expected given their social status.

Aggressive Symptoms

Negative conceptions of self, r(607) = - .09, p [less than] .05, but not of peers, r(610)
= .05, ns, were significantly associated with aggressive symptoms. Notably, more
negative conceptions of self were associated with lower levels of aggressive symptoms,
and this association increased slightly after controlling for the social status ratings, r(604)
= -.13, p [less than] .005. Consistent with the accuracy results, this suppressor effect
suggests that once children's actual social status is considered, aggressive symptoms are
associated even more strongly with positive self-conceptions, although the effect is
modest.

DISCUSSION

Skill-deficit models and cognitive-distortion models of psychopathology have, for the most
part, emerged in the context of two independent lines of theory and research.
Consequently, reconciling the discrepant predictions and empirical findings yielded by
these models has been difficult to achieve. The present study sought to unite these two
conceptual approaches, with the goal of developing more integrative cognitive-
interpersonal models of depression and aggression in childhood. Support was gained for
both types of models.

Consistent with past research, children with depressive symptoms displayed negative
conceptions of self and peers within relationships. Specifically, children with depressive
symptoms, both with and without co-occurring aggression, were more likely to view
themselves as incompetent and unworthy in peer relationships and to believe that peers
and friends are untrustworthy and hostile. As anticipated, aggressive children did not
differ from nonsymptomatic children in their overall level of self and peer conceptions.
These patterns of negative conceptions of relationships in children with depressive
symptoms and nondiscriminating conceptions of relationships in aggressive children add
to a growing body of research on the interpersonal perceptions associated with
depression (e.g., Rudolph et al., 1997; Shirk et al., 1997) and aggression (e.g., Hymel et
al., 1993; Patterson et al., 1990; Rabiner et al., 1993). As expected, both children with
depressive symptoms and aggressive children demonstrated social impairme nt, but the
specific nature of the interpersonal profiles differed across groups. Based on teacher
report of competence, children with depressive symptoms showed moderate levels of

66
prosocial and aggressive/disruptive behavior and were most withdrawn. They also
experienced moderate levels of peer popularity, neglect, and rejection. Aggressive
children and children with co-occurring depressive and aggressive symptoms experienced
more severe social difficulties, including low rates of prosocial behavior and high rates of
aggressive/disruptive behavior and alienation from the peer group.

Although informative, these two sets of findings leave key questions unanswered. Do the
negative conceptions of depressed children merely reflect their problematic social
worlds? Do the neutral conceptions of aggressive children, in concert with their obvious
social disturbance, reflect a lack of awareness or lack of acknowledgment of real
interpersonal difficulties? A crucial aspect of the current study was the direct examination
of the accuracy of children's conceptions of relationships.

Consistent with a skill-deficit model, the conceptions of children with depressive

symptoms mapped onto teacher perspectives of their social environment: Depressed-
unpopular children endorsed more negative conceptions of self and peers than did
depressed-accepted children. These results support the proposal that depressed children
are sensitive to social cues and incorporate feedback from the environment into their
social perceptions. Specifically, depressed-unpopular children are more likely to be
receiving feedback from their peers that they are not well-liked, causing them to feel that
they are socially incompetent and unworthy of positive peer attention and that other
children are unsupportive and cannot be counted on. It is important to note that the
conceptions of relationships assessed here were not parallel to the teacher ratings of
social status. That is, children's ratings of multiple dimensions of self and peer
perceptions were not directly comparable to teacher ratings of children's acceptance or
unpopularity in the peer group. Thus, children's conceptions do not provide a pure
measure of cue sensitivity.

Moreover, teachers may not have full access to children's daily experiences in the peer
group. Future research would therefore benefit from the use of additional methods of
assessing children's environments, such as observations of specific peer interactions
reflecting acceptance versus rejection, that would provide a more immediate link
between perceptions and experiences. Despite these qualifications, the findings from the
present study are consistent with theories of "depressive realism" (Alloy & Abramson,
1988), which suggest that the negative views of depressed individuals are in fact
accurate appraisals of aversive circumstances.

However, this social realism does not preclude the possibility that depressed children
possess negative beliefs above and beyond their interpersonal experiences. Indeed,
comparisons of the conceptions of depressed and nonsymptomatic children within the
same social status categories revealed that the conceptions of children with depressive
symptoms were more negative than was warranted by their social status. Thus, the
conceptions of children with depressive symptoms are not entirely based on reality, as
they tend to be more negative than the conceptions of nondepressed children even when
the two groups experience a similar social context. The negativity of their views may
therefore be somewhat exaggerated. This pattern was confirmed by supplemental
analyses indicating that the association between negative conceptions of relationships
and depression was not accounted for by teacher ratings of children's social status. These
results are consistent with research indicating that depressed children show biased pa
tterns of interpersonal information processing even when presented with similar
information (Rudolph et al., 1997; Shirk et al., 1997). Of course, the negativity of
depressed children's views was judged relative to that of nonsymptomatic children. It is
possible that the observed differences reflect an overestimation of competence by
nonsymptomatic children, but studies of cognitive distortion typically have not revealed
overestimation of social competence in normative groups (e.g., Hughes et al., 1997;
Hymel et al., 1993).

A distinctly different social-cognitive profile emerged for aggressive children. In contrast

to the depressed group, the conceptions of aggressive children did not map onto teacher

67
perspectives of their social status and, in fact, the discrepancies between the two social
status categories were close to zero. This lack of discrimination is consistent with the
hypothesis that aggressive children are insensitive to social cues in their environment.
Indeed, aggressive-unpopular children endorsed inflated self-perceptions relative to their
nonsymptomatic counterparts. In line with prior research (e.g., Hughes et al., 1997;
Zakriski & Coie, 1996), therefore, the self-conceptions of aggressive-unpopular children
may be particularly impenetrable to rejection feedback, perhaps due to a self-protective
bias that prevents accurate processing of incoming social information. These findings
belie the feasibility of the alternative explanation that aggressive-rejected children
maintain positive self-perceptions because pee rs fail to provide behavioral feedback
about their negative attitudes. In the present study, it was likely that teachers based their
social status categorization on overt manifestations of neglect or rejection. Consequently,
the most parsimonious explanation for this self-enhancement bias is that aggressive-
unpopular children fail to use feedback from others in the formation of judgments about
their personal competence. Consistent with prior research (Lochman, 1987; Zakriski &
Coie, 1996), this bias appeared to be activated only in the formation of self-appraisals but
not appraisals of peers.

Depressed and aggressive children clearly demonstrated distinct patterns of social-cue

detection deficits or distortions. An interesting question, therefore, concerned the social-
cognitive orientation of children with symptoms of both depression and aggression.
Overall, the pattern of conceptions in the depressed-aggressive group most closely
paralleled that of the depressed group, including a tendency to view the self and peers
more negatively than their nonsymptomatic counterparts. Moreover, unpopular children
with co-occurring symptoms showed more negative conceptions of self than their
accepted counterparts, reflecting some sensitivity to social cues. A tentative conclusion
would therefore be that the oversensitivity to negative cues characteristic of depressed
children may counteract the lack of sensitivity to negative cues characteristic of
aggressive children, but additional research with larger samples and alternative
measurement approaches is needed to resolve this issue.

Moreover, because aggressive symptoms were assessed with teacher report whereas
depressive symptoms were assessed with self report, it is unclear to what extent the
source of information affected the observed findings. Because past research has
confirmed the validity of teacher reports of aggression (e.g., Hughes et al., 1997;
Lochman & Dodge, 1998) and self reports of depression (e.g., Kendall, Cantwell, &
Kazdin, 1989; Reynolds, 1994), we chose to rely on these specific sources. To provide a
more comprehensive test of the skill-deficit and cognitive-distortion models, future
research will need to incorporate multiple informant reports of both symptoms and social
competence.

Although these data were cross-sectional and, therefore, cannot provide definitive
information about the cause-and-effect links among conceptions of relationships, social
impairment, and psychopathology, they do provide a basis for conceptualizing possible
developmental pathways to depression and aggression. In terms of depression,
preexisting negative conceptions of self and peers may cause children to act in ways that
create stress or rejection in close relationships, which in turn increases risk for
depression. In this case, these negative conceptions would initially be viewed as
distortions, shaped perhaps by early experiences, but then erroneously generalized to
later relationships. Unfortunately, interpersonal rejection and depression may lead to
even more dysfunctional conceptions of relationships and social incompetence. At this
point, children's conceptions may become, in part, based on reality. Alternatively,
negative conceptions may stem directly from ongoing aversive interpersonal
experiences. In this case, conceptions would be viewed as an accurate acknowledgment
of interpersonal difficulties. However, depressive symptoms resulting from this
pessimistic, albeit realistic, view of the world and from negative interpersonal feedback
may then precipitate biased appraisals of future social encounters and ensuing
competence deficits. In either case, a self-perpetuating cycle of cognitive distortion,
interpersonal impairment, and depression may emerge over time.

68
The development and progression of aggression may follow a similar type of reciprocal-
influence pathway. For example, insensitivity to social cues may produce maladaptive
interpersonal styles characterized by low levels of prosocial behavior and high levels of
aggressive behavior, which elicit rejection from peers. Despite these negative
interpersonal circumstances, a self-protective bias may allow aggressive-rejected children
to maintain overly positive self conceptions. This idealized self-image would then hinder
behavioral change and provoke even more negative reactions from the peer group (see
Bukowski, Sippola, Verlan, & Newcomb, 1993; Hughes et al., 1997; Zakriski & Coie, 1996),
leading to an escalating cycle of aggressive behavior and peer rejection.

Although future research will need to use longitudinal designs to disentangle these
multiple pathways, findings from the present study have implications for the efficacy of
intervention efforts with depressed and aggressive children. Most importantly, this study
revealed that depressed and aggressive children exhibit both cognitive distortions and
skill deficits that potentially may create a cycle of dysfunction. Consequently, intervention
efforts will need to be directed toward both altering maladaptive thought processes as
well as enhancing children's social skills. Finally, in the event that the temporal primacy
or salience of cognitive distortions versus skill deficits varies across children, intervention
approaches will need to be tailored to the needs of particular subgroups.

ACKNOWLEDGMENTS

We are grateful to the children, teachers, and principals who contributed their time and
efforts to this study. This research was supported by a University of Illinois Research
Board Beckman Award, William T. Grant Foundation Faculty Scholars Award, and National
Institutes of Mental Health Grant MH 56327-01 awarded to Karen D. Rudolph.

(1.) Department of Psychology, University of Illinois, Urbana-Champaign, Illinois.

(2.) Address all correspondence to Karen D. Rudolph, Department of Psychology,

University of Illinois, 603 E. Daniel St., Urbana-Champaign, Illinois 61820: e-mail:
krudolph@uiuc.edu.

(3.) As discussed later, the controversial group was not included in the analyses that
examined the accuracy of conceptions of relationships across and within social status
categories. Thus, this group was excluded from other analyses that involved social status
categories, but was included in all analyses that did not involve social status categories.

(4.) It is important to note that the symptom groups are defined according to elevations
on symptom levels, and are not necessarily comparable to clinical groups defined
according to diagnostic criteria.

REFERENCES

Achenbach, T. M. (1991). Manual for the Teacher's Report Form and 1991 profile.
Burlington, VT: University of Vermont, Department of Psychiatry.

Alloy, L. B., & Abramson, L. Y. (1988). Depressive realism: Four theoretical perspectives.
In L. B. Alloy (Ed.), Cognitive processes in depression (pp. 223-265). New York: Guilford.

Altmann, E. O., & Gotlib, I. H. (1988). The social behavior of depressed children: An
observational study. Journal of Abnormal Child Psychology, 16, 29-44.

Armsden, G. C., & Greenberg, M. T. (1987). The Inventory of Parent and Peer Attachment:
Individual differences and their relationship to psychological well-being in adolescence.
Journal of Youth and Adolescence, 16, 427-454.

69
Asarnow, J. R. (1988). Peer status and social competence in child psychiatric inpatients: A
comparison of children with depressive, externalizing, and concurrent depressive and
externalizing disorders. Journal of Abnormal Child Psychology, 16, 151-162.

Asher, S. R., & Coie, J. D. (1990). Peer rejection in childhood. New York: Cambridge
University Press.

Baker, M., Milich, R., & Manolis, M. B. (1996). Peer interactions of dysphoric adolescents.
Journal of Abnormal Child Psychology, 24, 241-255.

Baldwin, M. W. (1992). Relational schemas and the processing of social information.

Psychological Bulletin, 112, 461-484.

Barnett, P. A., & Gotlib, I. H. (1988). Psychosocial functioning and depression:

Distinguishing among antecedents, concomitants, and consequences. Psychological
Bulletin, 104, 97-126.

Bierman, K. L., Smoot, D. L., & Aumiller, K. (1993). Characteristics of aggressive-rejected,

aggressive (nonrejected), and rejected (nonaggressive) boys. Child Development, 64,
139-151.

Boivin, M., & Hymel, S. (1997). Peer experiences and social self-perceptions: A sequential
model. Developmental Psychology, 33, 135-145.

Boivin, M., Poulin, F., & Vitaro, F. (1994). Depressed mood and peer rejection in childhood.
Development and Psychopathology, 6, 483-498.

Bowlby, J. (1973). Attachment and loss: Vol. 2. Separation. New York: Basic.

Bukowski, W. M., Sippola, L. K., Verlan, P., & Newcomb, A. F. (1993). The association
between aggression, popularity, and narcissism and children 's feelings of social well-
being. Paper presented at the biennial meeting of the Society for Research in Child
Development, New Orleans, LA.

Cassidy, J., & Asher, S. R. (1992). Loneliness and peer relations in young children. Child
Development, 63, 350-365.

Cole, D. A. (1990). Relation of social and academic competence to depressive symptoms

in childhood. Journal of Abnormal Psychology, 99, 422-429.

Cole, D. A. (1991). Preliminary support for a competency-based model of depression in

children. Journal of Abnormal Psychology, 100, 181-190.

Cole, D. A., & Carpentieri, S. (1990). Social status and the comorbidity of child depression
and conduct disorder. Journal of Consulting and Clinical Psychology, 58, 748-757.

Cole, D. A., Martin, J. M., Peeke, L. G., Seroczynski. A. D., & Hoffman, K. (1998). Are
cognitive errors of underestimation predictive or reflective of depressive symptoms in
children: A longitudinal study. Journal of Abnormal Psychology, 107, 48196.

Cole, D. A., Martin, J. M., & Powers, B. (1997). A competency-based model of child
depression: A longitudinal study of peer, parent, teacher, and self-evaluations. Journal of
Child Psychology and Psychiatry, 38, 505-514.

Cole, D. A., & Turner, J. E., Jr. (1993). Models of cognitive mediation and moderation in
child depression. Journal of Abnormal Psychology, 102, 271-281.

70
Connolly, J., Geller, S., Marton, P., & Kutcher, S. (1992). Peer responses to social
interaction with depressed adolescents, Journal of Clinical Child Psychology, 21, 365-370.

Coyne, J. C. (1976). Depression and the response of others. Journal of Abnormal

Psychology, 85, 186-193.

Crick, N. R., & Dodge, K. A. (1994). A review and reformulation of social information-
processing mechanisms in children's social adjustment. Psychological Bulletin, 115, 74-
101.

Crick, N. R., & Ladd, G. W. (1993). Children's perceptions of their peer experiences:
Attributions, loneliness, social anxiety, and social avoidance. Developmental Psychology,
29, 244-254.

Cummings, E. M., & Cicchetti, D. (1990). Toward a transactional model of relations

between attachment and depression. In M. Greenberg, D. Cicchetti, & E. M. Cummings
(Eds.), Attachment in the preschool years: Theory, research, and intervention (pp. 339-
372). Chicago: University of Chicago Press.

Dodge, K. A. (1986). A social information processing model of social competence in

children. In M. Perlmutter (Ed.), The Minnesota Symposium on Child Psychology (Vol. 18,
pp. 77-125). Hillsdale, NJ: Erlbaum.

Dodge, K. A. (1993). Social-cognitive mechanisms in the development of conduct disorder

and depression. Annual Review of Psychology, 44, 559-584.

Dodge, K. A., & Frame, C. L. (1982). Social cognitive biases and deficits in aggressive
boys. Child Development, 53, 620-635.

Dodge, K. A., & Newman, J. P. (1981). Biased decision making processes in aggressive
boys. Journal of Abnormal Psychology, 90, 375-379.

Dodge, K. A., Pettit, G. S., McClaskey, C. L., & Brown, M. M. (1986). Social competence in
children. Monographs of the Society for Research in Child Development, 51, 1-85.

Dodge, K. A., & Richard, B. A. (1985). Peer perceptions, aggression, and the development
of peer relations, In J. D. Day & J. B. Pryor (Eds.), The development of social cognition (pp.
35-58). New York: Springer.

Garber, J., Quiggle. N. L., Panak, W., Dodge, K. A. (1991). Aggression and depression in
children: Comorbidity, specificity, and cognitive processing. In D. Cicchetti & S. Toth
(Eds.), Internalizing and externalizing expressions of dysfunction (pp. 225-264). Hillsdale,
NJ: Erlbaum.

Goodyer, I., Wright, C., & Altham, P. (1990). The friendships and recent life events of
anxious and depressed school-age children. British Journal of Psychiatry, 156, 689-698.

Gotlib, I. H., & Hammen, C. L. (1992). Psychological aspects of depression: Toward a

cognitive-interpersonal integration. London: Wiley.

Hammen, C., & Rudolph, K. D. (1996). Childhood depression. In E. J. Mash & R. A. Barkley
(Eds.), Child psychopathology (pp. 153-195). New York: Guilford Press.

Hughes, J. N., Cavell, T. A., & Grossman, P. B. (1997). A positive view of self: Risk or
protection for aggressive children? Development and Psychopathology, 9, 75-94.

71
Hymel, S., Bowker, A., & Woody, E. (1993). Aggressive versus withdrawn unpopular
children: Variations in peer and self-perceptions in multiple domains. Child Development,
64, 879-896.

Hymel, S., Rubin, K. H., Rowden, L., & LeMare, L. (1990). Children's peer relationships:
Longitudinal prediction of internalizing and externalizing problems from middle to late
childhood. Child Development, 61, 2004-2021.

Jacobsen, R. H., Lahey, B. B., & Strauss, C. C. (1983). Correlates of depressed mood in
normal children. Journal of Abnormal Child Psychology, 11, 29-40.

Kaslow, N. J., Rehm, L. P., & Siegel, A. W. (1984). Social-cognitive and cognitive correlates
of depression in children. Journal of Abnormal Child Psychology, 12, 605-620.

Kendall, P. C., Cantwell, D. P., & Kazdin, A. E. (1989). Depression in children and
adolescents: Assessment issues and recommendations. Cognitive Therapy and Research,
13, 109-146.

Kendall, P.C., & Fischler, G. L. (1984). Behavioral and adjustment correlates of problem
solving: Validational analyses of interpersonal cognitive problem-solving measures. Child
Development, 55, 879-892.

Kendall, P. C., Stark, K. D., & Adam, T. (1990). Cognitive deficit or cognitive distortion in
childhood depression. Journal of Abnormal Child Psychology, 18, 255-270.

Kennedy, E., Spence, S. H., & Hensley, R. (1989). An examination of the relationship
between childhood depression and social competence amongst primary school children.
Journal of Child Psychology and Psychiatry, 30, 561-573.

Kovacs, M. (1980/81). Rating scales to assess depression in school-aged children. Acta

Paedopsychiatry, 46, 305-315.

Kovacs, M. (1983). The Children Depression Inventory: A self-rated depression scale for
school-aged youngsters. Unpublished manuscript.

Lewinsohn, P. M. (1974). A behavioral approach to depression. In R. Friedman & M. Katz

(Eds.), The psychology of depression: Contemporary theory and research (pp. 157-185).
Washington, DC. Winston-Wiley.

Little, S. A., & Garber, J. (1995). Aggression, depression, and stressful life events
predicting peer rejection in children, Development and Psychopathology, 7, 845-856.

Lochman, J. E. (1987). Self and peer perceptions and attributional biases of aggressive
and nonaggressive boys in dyadic interactions. Journal of Consulting and Clinical
Psychology, 55, 404-410.

Lochman, J. E., & Dodge, K. A. (1998). Distorted perceptions in dyadic interactions of

aggressive and nonaggressive boys: Effects of prior expectations, context, and boys' age.
Development and Psychopathology, 10, 495-512.

Main, M., Kaplan, N., & Cassidy, J. (1985). Security in infancy, childhood, and adulthood: A
move to the level of representation. In I. Bretherton & E. Waters (Eds.), Growing points in
attachment theory and research. Monographs of the Society for Research in Child
Development, 50(1-2, Serial No. 209), 66-104.

72
Patterson, C. J., Kupersmidt, J. B., & Griesler, P. C. (1990). Children's perceptions of self
and of relationships with others as a function of sociometric status. Child Development,
61, 1335-1349.

Patterson, G. R., & Stoolmiller, M. (1991). Replications of a dual failure model for boys'
depressed mood. Journal of Consulting and Clinical Psychology, 59, 491-498.

Proffitt, V. D., & Weisz, J. R., (1992). Perceived incompetence and depression in childhood:
Cognitive distortion or accurate appraisal. Unpublished manuscript.

Quiggle, N. L.. Garber, J., Panak, W. F., & Dodge, K. A. (1992). Social information
processing in aggressive and depressed children. Child Development, 63. 1305-1320.

Rabiner, D. L.. Keane, S. P., & MacKinnon-Lewis, C. (1993). Children's beliefs about
familiar and unfamiliar peers in relation to their sociometric status, Developmental
Psychology, 29, 236-243.

Reynolds, W. M. (1994). Assessment of depression in children and adolescents by self-

report questionnaires. In W. M. Reynolds & H. F. Johnston (Eds.), Handbook of depression
in children and adolescents (pp. 209-234). New York: Plenum.

Rogosch, F., Cicchetti, D., & Aber, J. L. (1995). The role of child maltreatment in early
deviations in cognitive and affective processing abilities and later peer relationship
problems. Development and Psychopathology, 7, 591-609.

Rudolph, K. D., Hammen, C., & Burge, D. (1994). Interpersonal functioning and depressive
symptoms in childhood: Addressing the issues of specificity and comorbidity. Journal of
Abnormal Child Psychology, 22, 355-371.

Rudolph, K. D., Hammen, C., & Burge, D. (1995). Cognitive representations of self, family,
and peers in school-age children: Links with social competence and sociometric status.
Child Development, 66, 1385-1402.

Rudolph, K. D., Hammen, C., & Burge, D. (1997). A cognitive-interpersonal approach to

depressive symptoms in preadolescent children. Journal of Abnormal Child Psychology,
25, 33-45.

Safran, J. D. (1990). Towards a refinement of cognitive therapy in light of interpersonal

theory I. Theory. Clinical Psychology Review, 10, 87-105.

Shirk, S. R., Boergers, J., Eason, A., & Van Horn, M. (1998). Dysphoric interpersonal
schemata and preadolescents' sensitization to negative events. Journal of Clinical Child
Psychology, 27, 54-68.

Shirk, S. R., Van Horn, M., & Leber, D. (1997). Dysphoria and children's processing of
supportive interactions. Journal of Abnormal Child Psychology, 25, 239-249.

Smucker, M. R., Craighead, W. E., Craighead, L. W., & Green, B. J. (1986). Normative and
reliability data for the Children's Depression Inventory. Journal of Abnormal Child
Psychology, 14, 25-39.

Weisz, J. R., Rudolph, K. D., Granger, D. A., & Sweeney, L. (1992). Cognition, competence,
and coping in child and adolescent depression: Research findings, developmental
concerns, therapeutic implications. Development and Psychopathology, 4, 627-653.

73
Zakriski, A. L., & Coie, J. D. (1996). A comparison of aggressive-rejected and
nonaggressive-rejected children's interpretations of self-directed and other-directed
rejection. Child Development, 67, 1048-1070.

Conceptions of Relationship by
Symptom Groups
Nonsymptomatic Depressed Aggressive
group [1] group [2] group [3]
Conceptions of self 1.78 2.19 1.79
(.38) (.49) (.40)
(n = 345) (n = 140) (n = 78)
Conceptions of peers 2.09 2.37 2.13
(.39) (.46) (.41)
(n = 347) (n = 140) (n = 78)
Depressed-aggressive Pairwise
group [4] comparisons p-Value Effect size
Conceptions of self 2.19 1 vs. 2 .000 0.99
(.53) 1 vs. 3 ns 0.03
(n = 48) 1 vs. 4 .000 1.02
2 vs. 3 .000 0.87
2 vs. 4 ns 0.00
3 vs. 4 .000 0.88
Conceptions of peers 2.44 1 vs. 2 .000 0.68
(.43) 1 vs. 3 ns 0.10
(n = 49) 1 vs. 4 .000 0.89
2 vs. 3 .000 0.54
2 vs. 4 ns 0.15
3 vs. 4 .000 0.74
Note. Higher values indicate more
negative conceptions of relationships.
Standard deviations are noted in
parentheses. Tukey's HSD procedure
was used for pairwise comparisons.
Social Behavior by Symptom Groups
Nonsymptomatic Depressed Aggressive
group [1] group [2] group [3]
(n = 347) (n = 139) (n = 79)
Prosocial behavior 4.26 3.95 2.87
(.74) (.74) (.89)
Withdrawn behavior 2.00 2.23 1.97
(.90) (.98) (.76)
Aggressive/disruptive 1.64 1.84 3.48
behavior (.78) (.78) (.83)
Depressed-aggressive
group [4] Pairwise
(n = 49) comparisons p-Value Effect size
Prosocial behavior 2.84 1 vs. 2 .000 0.42
(.82) 1 vs. 3 .000 1.81
1 vs. 4 .000 1.89
2 vs. 3 .000 1.35
2 vs. 4 .000 1.46
3 vs. 4 ns 0.03
Withdrawn behavior 2.07 1 vs. 2 .031 0.25
(.84) 1 vs. 3 ns 0.03
1 vs. 4 ns 0.08
2 vs. 3 .081 0.29
2 vs. 4 ns 0.17
3 vs. 4 ns 0.13
Aggressive/disruptive 3.57 1 vs. 2 .029 0.26

74
behavior (.72) 1 vs. 3 .000 2.33
1 vs. 4 .000 2.50
2 vs. 3 .000 2.05
2 vs. 4 .000 2.26
3 vs. 4 ns 0.11
Note. Higher values indicate higher levels of prosocial, withdrawn,
and aggressive/disruptive behavior. Standard
deviations are noted in parentheses.
Tukey's HSD procedure was used for pairwise comparisons.
Social Status Ratings by Symptom Groups
Nonsymptomatic Depressed Aggressive Depressed-aggressive
group [1] group [2] group [3] group [4]
Popularity 4.68 4.14 3.92 3.55
(1.30) (1.29) (1.40) (1.61)
(n = 347) (n = 140) (n = 79) (n = 49)
Neglect 2.35 2.81 3.10 3.43
(1.44) (1.59) (1.55) (1.54)
(n = 347) (n = 140) (n = 78) (n = 49)
Rejection 2.27 2.70 3.28 3.67
(1.38) (1.50) (1.53) (1.64)
(n = 347) (n = 140) (n = 78) (n = 49)
Pairwise
comparisons p-Value Effect size
Popularity 1 vs. 2 .000 0.42
1 vs. 3 .000 0.58
1 vs. 4 .000 0.84
2 vs. 3 ns 0.17
2 vs. 4 .021 0.43
3 vs. 4 ns 0.25
Neglect 1 vs. 2 .006 0.31
1 vs. 3 .000 0.51
1 vs. 4 .000 0.74
2 vs. 3 ns 0.18
2 vs. 4 .030 0.39
3 vs. 4 ns 0.21
Rejection 1 vs. 2 .009 0.30
1 vs. 3 .000 0.72
1 vs. 4 .000 0.99
2 vs. 3 .012 0.38
2 vs. 4 .000 0.63
3 vs. 4 ns 0.25
Note. Higher values indicate higher levels of peer popularity,
neglect, and rejection. Standard deviations
are noted in parentheses. Tukey's HSD
procedure was used for pairwise comparisons.
Conceptions of Relationships by Social Statusand Symptom Groups
Accepted Unpopular Pairwise comparisons t
Conceptions of self
Nonsymptomatic [1.76.sup.ab] [1.93.sup.cde] 1 vs. 2 2.76
(.39) (.32)
(n = 289) (n = 43)
Depressed [2.12.sup.a] [2.43.sup.c] 1 vs. 2 3.31
(.46) (.46)
(n = 100) (n = 33)
Aggressive 1.82 [1.71.sup.d] 1 vs. 2 .98
(.45) (.35)
(n = 35) (n = 18)
Depressed-aggressive [2.04.sup.b] [2.40.sup.e] 1 vs. 2 1.97
(.51) (.53)
(n = 15) (n = 18)

75
Conceptions of peers
Nonsymptomatic [2.08.sup.ab] [2.21.sup.ce] 1 vs. 2 2.08
(.39) (.35)
(n = 290) (n = 44)
Depressed [2.32.sup.a] [2.53.sup.c] 1 vs. 2 2.38
(.44) (.27)
(n = 100) (n = 33)
Aggressive 2.12 2.10 1 vs. 2 .19
(.44) (.27)
(n = 35) (n = 18)
Depressed-aggressive [2.45.sup.b] [2.47.sup.e] 1 vs. 2 .16
(.46) (.50)
(n = 15) (n = 19)
p-Value Effect size
Conceptions of self
Nonsymptomatic .003 .45
Depressed .001 .67
Aggressive ns .26
Depressed-aggressive .029 .69
Conceptions of peers
Nonsymptomatic 0.19 .34
Depressed .010 .48
Aggressive ns .05
Depressed-aggressive ns .04

Note. Means with the same superscript within each column differ at p [less than] .05 (ts
[greater than] 1.98). Comparisons were made only between the nonsymptomic group
versus each of the symptom groups. Higher values indicate more negative conceptions of
relationships, Standard deviations are noted in parentheses.

COPYRIGHT 2001 Plenum Publishing Corporation

COPYRIGHT 2001 Gale Group

76