K :63." ‘

.,.
~ :1 :..>.

gm

.._l a.

' ,_‘ ”L

 

* $33;
{1 t _
‘ it"

 

 

‘9 ":fl

“5 “1.; .‘-. In

‘ , - m . . wr

' 4.. v 1-v :uvthw'th'z‘l’gfl“

w. a“ ~ ’é’i’afif'~‘*"—'?“”
, ‘.

.. v A; "
.,_.
.

 

 

r"Ems

llllllllllllllllllllllllllllllllllllllllllllllllllllllll

1293 01016 2745

This is to certify that the

thesis entitled

Factors Associated with Interparent Agreement on
Child Behavior Problems Among Young Children

presented by

Frank Joseph DeMarco

has been accepted towards fulfillment
of the requirements for

M. A . Psychology

degree in

 

 

I

Major pro sor

Date ivy/37kg;

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

 

LIBRARY
Mlchlgan State
Unlverslty

 

 

 

moan RETURN aoxmmoiombmummymm
TO AVOID FINES Mum on or bdonddo duo.

DATE DUE DATE DUE DATE DUE

- -
F—l-‘W

-Eg
i—T‘T

usu loAnNflnnutlvo ActiontEqml Opportunity Imam
WA

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

FACTORS ASSOCIATED WITH INTERPARENT AGREEMENT ON CHILD
BEHAVIOR PROBLEMS AMONG YOUNG CHILDREN

By

Frank Joseph DeMarco

A THESIS

Submitted to
Michigan State University
in partial fitlfillment of the requirements
for the degree of

MASTER OF ARTS

Department of Psychology

1 994

ABSTRACT

FACTORS ASSOCIATED WITH INTERPARENT AGREEMENT ON CHILD
BEHAVIOR PROBLEMS AMONG YOUNG CHILDREN

By

Frank Joseph DeMarco

Information about parents' perceptions of child behavior problems is critical for
child psychopathological research and assessment, as parents' reports often determine
whether children are referred for services, and provide an important data base for the
assessment of children's disorders. The objective of this study is to characterize the
predictors of interparent agreement in a population-based sample of families comprising a
broad data base with considerable familial variation in psychopathology. Results suggest
that the relationship between interparent agreement and predictors such as child
disturbance level, parent psychopathology, and marital satisfaction varies depending on the
index of agreement. A negative relationship was found between level of child disturbance
and most measures of agreement, although this relationship became moderately positive
for an index correcting for greater chance agreement among parents of less disturbed
children. Moderate negative relationships were found between parent psychopathology

and agreement. The clinical significance of the findings is discussed.

To my parents, Frank and Flora

ACKNOWLEDGMENTS

I would like to thank Robert Zucker for his guidance and expertise throughout the
thesis process, and for giving me the opportunity to work on the MSU-UM Family Study.
1 would also like to thank my other committee members, Hiram Fitzgerald and Dozier
Thornton, for finding time in extremely busy schedules to serve on my committee. They
have been patient and helpful throughout this process, and I am deeply grateful.

This project would have been impossible were it not for the concerted efl‘orts of
the excellent students and staff of the MSU-UM Family Study. I am indebted to all of
them. I would especially like to thank Betty Lafl‘an and Ray Bingham for their support
and invaluable assistance with data analysis; Michael Ichiyama for developing the Marital
Satisfaction index used in this analysis; and Marcy Wallace for her help, warmth, and
humor, which I always appreciate.

Special thanks are due to Suzy Pavick, Psychology graduate secretary, for her
immeasurable help and constant moral support. I would also like to thank Roger Halley
for assistance with supplies through the Psychology business oflice.

I am grateful to Jim F uendeling and Lisa Gauthier for their unwavering friendship
and support. \Vrthout them, I would not have retained what little of my sanity remains
after three years of graduate school!

Special gratitude is due to my parents, Frank and Flora DeMarco. Without their
love and support I could not attain my goals.

Finally, I would like to thank my wife, Christine Edgar, for putting up with me for
the past five years. Her unconditional love and constant support are what keep me going.

This work was supported in part by grants to R. A. Zucker and H. E. Fitzgerald

from the National Institute on Alcohol Abuse and Alcoholism (2 R01 AA 07065).
iv

TABLE OF CONTENTS

LIST OF TABLES ...................................................................................................... vii
INTRODUCTION AND REVIEW OF THE LITERATURE ...................................... 1
INTERPARENT AGREEMENT IN THE LITERATURE ............................... 1
THE IMPORTANCE OF INTERPARENT AGREEMENT ............................. 3
FACTORS RELATED TO INTERPARENT AGREEMENT .......................... 6
EMPIRICAL INVESTIGATIONS OF THE DETERMINANTS OF
INTERPARENT AGREEMENT ..................................................................... 10
Age of the child .................................................................................... 10
Referral status and adjustment of child .................................................. 12
Domain of child behavior ...................................................................... 15
Time spent with child ........................................................................... 17
Marital discord ..................................................................................... 20
Parental psychopathology ..................................................................... 23
Overall summary of the empirical findings ............................................. 28
STATEMENT OF THE PROBLEM ................................................................ 28
HYPOTI-IESES ............................................................................................... 30
METHOD ................................................................................................................... 33
SUBJECTS ...................................................................................................... 33
PROCEDURE ................................................................................................. 34
MEASURES .................................................................................................. 34
DATA ANALYSIS .......................................................................................... 38
Dependent variable: Interparent agreement ........................................... 38
Predictor variables ................................................................................ 39

V

RESULTS ................................................................................................................... 41

DEMOGRAPHIC CHARACTERISTICS OF THE SAMPLE ......................... 41
INTERPARENT AGREEMENT ..................................................................... 44
BIVARIATE RELATIONSHIPS BETWEEN PREDICTOR
VARIABLES AND INTERPARENT AGREEMENT INDICES ..................... 47
Age of the target child .......................................................................... 47
Type of child behavior problems ........................................................... 47
Child disturbance .................................................................................. 50
Amount of time fathers spend with target child ..................................... 53
Marital satisfaction ............................................................................... 55
Parents' overall psychological fimctioning ............................................. 55
Parental depression ............................................................................... 58
Parent alcohol problems ....................................................................... 63
Parent antisocial behavior ..................................................................... 66
Summary of bivariate tests .................................................................... 66
EXPLORATORY‘FACTOR ANALYSIS ........................................................ 69
HIERARCHICAL REGRESSION ANALYSIS ............................................... 72
CLINICAL SIGNIFICANCE OF FINDINGS .................................................. 83
DISCUSSION ............................................................................................................. 92
METHODOLOGICAL CONSIDERATIONS .................................................. 92
DISCUSSION OF THE FINDINGS ................................................................ 94
SUMMARY .................................................................................................... 100
FUTURE DIRECTIONS ................................................................................. 103
APPENDIX ................................................................................................................ 105
LIST OF REFERENCES ............................................................................................ l 15

LIST OF TABLES

1. Demographic Characteristics of the Sample ............................................................. 42

2. Relationship (Pearson r's) Between Demographic Variables and Indices of
Interparent Agreement on CBCL Total Behavior Problems ..................................... 43
3. Interparent Agreement on the Description of Child Behavior Problems: Parents'
Total, Intemalizing, and Externalizing Behavior Problem Scores on the Child
Behavior Checklist .................................................................................................. 45

4. Interparent Agreement on the Description of Child Behavior Problems: Parents'
Differential Concern Index (DCI) Scores on the CBCL ........................................... 46

5. Interparent Agreement on the Description of Child Behavior Problems:
Percentage Agreement on Presence/ Absence, Effective Percentage Agreement,
and Kappa on CBCL Behavior Problems ................................................................. 48

6. Relationship Between Type of Child Behavior Problem (Intemalizing versus
Externalizing) and Interparent Agreement on the CBCL .......................................... 49

7. Relationship (Pearson r's) Between Level of Reported Child Disturbance and
Interparent Agreement on CBCL Total behavior Problems ...................................... 51

8. Relationship (Pearson r's) Between the Amount of Time Spent with Target
Child by Fathers and Interparent Agreement on CBCL Total Behavior Problems ..... 54

9. Relationship (Pearson r's) Between Parents' Marital Satisfaction and Interparent
Agreement on CBCL Total Behavior Problems ....................................................... 56

10. Relationship (Pearson r's) Between Parents' DSM-III Axis V Rating and
Interparent Agreement on CBCL Total Behavior Problems .................................... 57

11. Relationship (Pearson r’s) Between Parents' Mental Health Q-Sort Score and
Interparent Agreement on CBCL Total Behavior Problems .................................... S9

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

Relationship (Pearson r's) Between Mothers' Beck and Hamilton Depression
Scores and Interparent Agreement on CBCL Total Behavior Problems .................. 60

Relationship (Pearson r's) Between F athers' Beck and Hamilton Depression
Scores and Interparent Agreement on CBCL Total Behavior Problems .................. 62

Relationship Between Fathers' Alcoholism Type and Interparent Agreement on
CBCL Total Behavior Problems ............................................................................. 64

Relationship (Pearson r's) Between Parents' Lifetime Alcohol Problems Score
(LAPS) and Interparent Agreement on CBCL Total Behavior Problems ................. 65

Relationship (Pearson r's) Between Parents' Antisocial Behavior (ASB)
Checklist Score and Interparent Agreement on CBCL Total Behavior

Problems ............................................................................................................... 67
Exploratory Factor Analysis of Mother & Father Psychopathology Variables
(Oblimin Rotation) ................................................................................................. 70
Factor Analysis of Mother & Father Psychopathology Variables: Factor

Loadings for Two-F actor Solution (Oblimin Rotation) ........................................... 71
Correlations Between Factor Scale Scores and Outcome Variables ........................ 73

Hierarchical Regression Predicting Mothers' Differential Concern on CBCL
Total Behavior Problems ....................................................................................... 76

Hierarchical Regression Predicting Fathers' Differential Concern on CBCL
Total Behavior Problems ....................................................................................... 77

Hierarchical Regression Predicting Percentage Agreement on
Presence/Absence of CBCL Total Behavior Problems ............................................ 79

Hierarchical Regression Predicting Effective Percentage Agreement on CBCL
Total Behavior Problems ....................................................................................... 80

Hierarchical Regression Predicting Interparent Kappa on CBCL Total
Behavior Problems ................................................................................................. 82

Comparison of Top and Bottom Mother Psychopathology Quartile Groups on
the Indices of Interparent Agreement on CBCL Total Behavior Problems .............. 85

Comparison of Top and Bottom Father Psychopathology Quartile Groups on
the Indices of Interparent Agreement on CBCL Total Behavior Problems .............. 87

viii

27. Comparison of Top and Bottom Child Disturbance Quartile Groups on the
Indices of Interparent Agreement on CBCL Total Behavior Problems .................... 89

28. Comparison of Top and Bottom Marital Dissatisfaction Quartile Groups on the
Indices of Interparent Agreement on CBCL Total Behavior Problems .................... 9]

ix

CHAPTER I
INTRODUCTION AND REVIEW OF THE LITERATURE

The use of child behavior rating scales in clinical assessment and research has
burgeoned over the past fifteen years. With this increase in use has come a growing
literature assessing the validity and reliability of these instruments, with particular
emphasis on the assessment of interrater reliability. Studies have compared the ratings of
parents, teachers, observers, and children, attempting not only to assess the psychometric
properties of the instruments, but perhaps more importantly, to determine the usefulness
and validity of data obtained from different informants that interact with children in
difl‘erent situations (Achenbach, McConaughy, & Howell, 1987; Becker, 1960; Edelbrock,
Costello, Dulcan, Conover, & Kala, 1986; Loeber, Green, & Lahey, 1990; Loeber, Green,
Lahey, & Stouthamer-Loeber, 1989). Especially important is the question of interparent
agreement on child behavior measures. As Jacob, Grounds, and Haley (1982) note:
”Clarification of the relationship between parents‘ reports of their children's behavior is
essential given the pivotal role of such information in both research and clinical practice in
child psychopathology" (p. 594). The purpose of the present investigation is to explore
factors related to interparent agreement on child behavior problems in intact families with
young children, using the Child Behavior Checklist (CBCL; Achenbach & Edelbrock,
1983). The CBCL is the clinically most widely used, "most well-developed empirically
derived behavior rating scale currently available for assessing psychopathology and social
competence in children" (Barkley, 1988, p. 122).
INTERPARENT AGREEMENT IN THE LITERATURE

There have been numerous investigations of the degree to which parents agree on

the behaviors, temperament, and psychological symptoms of their children. These studies
1

2

have reported a range of correlations between parents' scores. A recent meta-analysis
(Achenbach, McConaughy, & Howell, 1987) examined 119 studies for interrater
agreement on child behavior ratings, 22 of which reported correlations between parents
scores' on these measures. Parents' correlations ranged from .03 to .75, with a mean of
.59 across the 22 studies. This mean level of interparent agreement was found to be
similar in magnitude to that of other pairs of informants sharing a common relationship to
the target child, such as pairs of teachers (mean = .64), mental health workers (mean =
.54), and observers (mean = .57), with an overall mean of .60 for similarly-situated
informants across the 119 studies. Thus parents, who have the opportunity to observe
their children's behavior over extended periods of time and across a great diversity of
settings and situations (Achenbach et al., 1991), appeared no more likely to agree on their
children's behavior than did other pairs of informants.

More recent investigations of interparent agreement have also produced mixed
findings. Achenbach, Edelbrock, and Howell (1987) used the recently developed CBCL
for Ages 2-3 (CBCL/2-3) to assess interparent agreement in a longitudinal sample of low-
birthweight and normal children and reported interparent agreement coeficients of .47
(age 2), .57 (age 3), and .58 (age 4). In a subsequent study Achenbach and his colleagues
reported a correlation between parents of .72 on Total Behavior Problems in a sample of
94 clinic-referred children aged 4-16 (Achenbach, Howell, Quay, & Conners, 1991).
Burrows and Kelley (1983) reported a relatively high agreement ratio (agreements divided
by agreements + disagreements) of .81 among parents in their non-clinic sample of
children aged 4-6. Jensen and his colleagues reported intraclass correlations between
parents' scores ranging from .41 to .67 (depending on gender and domain of behavior) for
their sample of non-referred children (Jensen, Traylor, Xenakis, & Davis, 198 8). Johnston
(1991) found correlations of .55 and .47 between parents' scores on the CBCL
internalizing and externalizing subscales, respectively, in her sample of clinic-referred
children. Thus, most studies have reported at least moderate agreement between parents

3

on child behavior measures. An exception is Reider (1991), who reported interparent
correlations of .27 for Intemalizing behaviors, .37 for Externalizing behaviors, and .26 for
Total Behavior Problems in a population-based sample of 90 families with young children
and alcoholic fathers. These coefficients are significantly lower than those reported by
Achenbach and Edelbrock (1983) in the CBCL manual (all 2'5 3 .004; Reider, 1991).

The studies discussed thus far have generally reported correlations between
parents' scale scores, which are derived by summing the item scores for the entire scale or
for a particular segment of the scale that corresponds to one or more empirically derived
factors. Some investigators, however, have explored interparent agreement at the level of
individual items or behaviors. Christensen, Margolin, and Sullaway (1992), for example,
using Cohen's (1960) kappa to assess nonchance item-level agreement between parents on
the CBCL, reported an overall kappa of only .34 in their sample of clinic-referred and non-
referred children and their parents. Similarly, Jacob et al. (1982) found that interparent
agreement on the Behavior Problem Checklist (BPC; Quay, 1977) was higher in
nonreferred families than referred families if measured correlationally, but was nearly
equivalent and very low in both types of families regarding gm behaviors (i.e., item-
level agreement). Thus, level of interparent agreement is partially a function of how one
dating ”agreement,” which is reflected in the measure of association that is used to
compute the relationship. This interesting methodological issue will be firrther explored in
the present investigation.

THE IMPORTANCE OF INTERPARENT AGREEMENT

Information about parents' perceptions of child behavior problems is critical for
child psychopathological research and assessment. Parents' perceptions often determine
whether or not children are referred for services in the first place (Achenbach &
Edelbrock, 1978; Jacob et al., 1982). Further, parents' reports provide a particularly
important data base for the assessment of children's disorders since parents have the

opportunity to observe their children's behavior over extended periods of time and across

4

diverse settings and situations (Achenbach et al., 1991). Achenbach and his colleagues, in
their 1987 meta-analysis of studies of cross-informant consistency, referred to the question
of informant agreement as "the most basic question" for empirically-based child
assessment (Achenbach, McConaughy, & Howell, 1987, p. 214). And last, for other
reasons entirely, the issue of interparent agreement raises theoretical questions pertaining
to both the accuracy of child perception by parents as well as the assessment of what the
child, parent, and family parameters are which contribute to variation in agreement.

The question of interparent agreement raises several methodological and
theoretical issues. The first of these is an issue involving measurement. If interparent
agreement on child behavior is less than perfect, to what extent does this reflect error of
measurement? Given the limitations of instruments and observers, what level of
agreement can be expected? The meta-analysis by Achenbach, McConaughy, and Howell
(1987) provides a starting point for answering this question. The mean level of similar-
informant agreement found across studies in that analysis, .60, was summarized by the
authors as "substantiaL but far from perfect" (Achenbach & McConaughy, 1987, p. 20).
Achenbach and his colleagues note, however, that this "far fi'om perfect" agreement does
not necessarily indicate unreliability in the instruments. Test-retest coeficients of behavior
rating scales are often in the 80-90 range (Achenbach & Edelbrock, 1983; Achenbach,
McConaughy, and Howell, 1987; Barkley, 1988). Different informants, however, even
those such as parents that share a similar role vis-a-vis the target child, often experience

difi‘erent samples of the child's behavior across situations (Achenbach, McConaughy, and

 

Howell, 1987). This suggests that a given informant supplies unique and valid information
that overlaps only partially with that of other informants, indicating a need to obtain data
from multiple informants that interact with the child in a variety of settings and situations
(Achenbach & McConaughy, 1987). Thus, it seems likely that the considerable variance
in parents' ratings reported in most studies is not simply due to measurement error, but

instead reflects real variance in child, rater, and situational characteristics (Christensen et

5

al., 1992). Hence, the question is not one of "maximizing" interparent agreement. Rather,
it can be asked whether it makes sense to speak of a "normative" level of interparent
agreement, such as the meta-analytic finding of .59 (Achenbach, McConaughy, & Howell,
1987), and if so, whether there are implications for child, parent, and family adjustment
when significant deviations fi'om this expected level of consistency are observed.

Related to the measurement issues raised by less than perfect interparent
agreement is the question of the interchangeability of inforrnants' reports. Despite
arguments for obtaining information on child behavior problems from a variety of sources,
it is often the case that only one parent, typically the mother, supplies the information used
by researchers and clinicians in child assessment (Phares, 1992; Phares & Compas, 1992).
Achenbach and his colleagues suggest that data from a single informant provides a
"reasonable sample" of what would be obtained from another informant playing a similar
role vis-a-vis the child (Achenbach, McConaughy, & Howell, 1987, p. 227). The veracity
of this assumption has serious implications for child assessment procedures that rely on
information obtained from only one informant. To the degree that consistency between
informants is high, the report of one informant might arguably serve as a substitute for that
of another. To the degree that consistency is low, it may reflect measurement error, or it
may indicate that informants are contributing unique information about the behavior of the
target child (Achenbach, McConaughy, & Howell, 1987). Jacob and his colleagues warn
that "the use of only one parent's data could mask important relationships that exist
between the reports of one parent (but not the other) and processes and outcomes relevant
to childhood psychopathology" (Jacob et al., 1982, p. 607). For example, suppose
depressed parents perceive their children's behavior to be more deviant than it actually is
(as some authors have suggested, e.g., Webster-Stratton, 1988). Given a dyad consisting
of one depressed and one nondepressed parent, if information fi'om the depressed parent
(but not the nondepressed parent) is used in the assessment of their child's behavior,

inaccurate diagnostic and treatment decisions could be reached, thus compromising the

6

well-being of the child and the family. Considering this, the question of interparent
agreement and the "interchangeability" of informants becomes crucial. Further, the role of
fathers in child psychopathology has been neglected in the literature relative to that of
mothers (Earls, 1976; Phares, 1992), yet fathers' behaviors and adjustment are as critical
to children's adjustment as are mothers', perhaps more so (Compas & Phares, 1992;
Compas, Phares, Banez, & Howell, 1991; Phares, 1992). Examining fathers' reports of
their children's behavior and their patterns of agreement with mothers may provide
important information for understanding fathers' effects on child adjustment.

A third major issue involves the theoretical question of whether cross-informant
consistency varies with characteristics of the informant or target child (Achenbach,
McConaughy, & Howell, 1987). Specifically, what child, parent, and family
characteristics are related to interparent agreement on child behavior, and why? The
absolute level of interparent agreement in a given population is probably less important
than the covariates of interparent agreement in that population (Achenbach & Edelbrock,
1978; Jacob et al., 1982). Establishing whether particular child, parent, and family
characteristics covary with high versus low interparent agreement is an important first step
toward determining the implications of interparent agreement for children's adjustment.
For example, Ferguson, Partyka, and Lester (1974) reported that in their sample of 112
clinic-referred and 105 non-referred children and their parents, the lowest level of
interparent agreement was found among parents of 5- to 7-year-old clinic-referred boys,
while the highest level of agreement was between parents of 5- to 7 -year-old non-referred
boys. Thus, level of interparent agreement might be a useful predictor of younger boys'
adjustment. The identification of child, parent, and family factors that are related to
interparent agreement on child behavior measures is the primary goal of the current study.
FACTORS RELATED TO INTERPARENT AGREEMENT

There are a number of factors that can be linked theoretically to interparent
agreement on child behavior problems. These factors speak primarily to the visibility of

7

behaviors, parents' opportunities for observing behaviors, and parental characteristics that
may either elicit certain types of behavior fi'om the child or bias parents' perceptions of
child behavior.

One such factor which may affect interparent agreement is the age of the child.
Infants and very young children spend more time with their mothers than with their
fathers; mothers may therefore be more apt to witness younger children's behavior
problems than would fathers, and thus report more of these problem behaviors
(Achenbach, Edelbrock, & Howell, 1987). Agreement might rise as the children get older
and fathers become more involved, but then fall again once the children enter adolescence
and spend more time away from home, thus making their behavior more difficult to judge
and cross-situationally less consistent than that of younger children (Achenbach,
McConaughy, & Howell, 1987). '

The type and severity of child behavior problems is also likely to afl'ect interparent
agreement. It has been suggested that parents are more likely to agree on the occurrence
of aggressive, acting out behaviors than they are on that of more covert, internalizing
types of problems, due to the more observable, objective, and disruptive nature of the
former compared to the latter (Earls, 1980; Christensen et al., 1992). Jensen and his
colleagues (Jensen, Traylor, Xenakis, & Davis, 1988) firrther suggest that agreement is
greater when behaviors violate social role expectations, such as when girls act
aggressively. Again, the visibility and disruptiveness of the behavior may play a key role in
determining the degree of interparent agreement.

Greater severity of child behavior problems, often equated in the literature with the
child having been clinic-referred, has been linked to lower interparent agreement (e. g.,
Ferguson et al., 1974; Christensen et al., 1992). This is perhaps due to the embeddedness
of the child's problems within a context of overall family disturbance (Margolin, 1981;
Zucker & Fitzgerald, 1991a) that includes poor interparent communication and parental
psychopathology.

8

The amount of time parents spend with their child may affect interparent
agreement on child behavior problems (Achenbach, Edelbrock, & Howell, 1987; Phares,
1992). This again seems likely to be a function of parents' opportunities to observe their
children's behavior. Even today, mothers appear to spend more time with their children
than do fathers, particularly in the younger years (Achenbach, Edelbrock, & Howell, 1987;
Phares, 1992). For this reason mothers are perhaps more likely to observe, and thus be
able to report, their children's behavior problems than are fathers (Christensen, Margolin,
& Sullaway, 1987). It seems reasonable to suggest that fathers who spend more time with
their children will have more opportunities to observe their children's behavior; this, in
turn, might be expected to pull fathers' reports of child behavior problems more in line
with mothers' reports.

Marital discord is another factor that may be associated with lower interparent
agreement on child behavior problems. This might be due simply to poor communication
between the partners regarding their child's behavior. It has also been suggested that
distressed marital partners enter into alliances with their children against the other partner
or "detour" marital conflict onto the child, thus setting the child up to act differently with
each parent and increasing the child's level of behavioral disturbance and volatility
(Christensen, Phillips, Glasgow, & Johnson, 1983; Christensen et al., 1987; Margolin,
1981). These authors suggest that, like child disturbance, marital conflict exists within a
context of overall family disturbance characterized by asymmetrical parent-child
relationships, poor interparent communication, and less cross-situational consistency in
child behavior.

Along these same lines, type and level of parental psychopathology has been linked
to interparent agreement. Perhaps most often cited in this regard is depression, especially
maternal depression. It has been suggested that depressed mothers show decreased
tolerance for normal child behavior, perceiving it as abnormal or disturbed; this in turn

results in inappropriate interventions by the parent which then elicit negative responses

9

from the child (Richters, 1992). Thus, the depressed parent starts out seeing the child's
behavior as more disturbed than does her spouse, and then enters into a cycle of negative
interactions with the child that are not mirrored by the nondepressed parent. Paternal
depression, on the other hand, has been investigated far less often, and no studies were
found that examined the relationship between paternal depression and interparent
agreement. However, there is some indication that depressed fathers rate their children
higher on behavior problems than do non-depressed fathers (Jacob & Leonard, 1986), and
there is clearly a need for studies that examine whether depression affects fathers' ratings
and interparent agreement in the same or similar ways as maternal depression.

Parental alcoholism may also affect interparent agreement on child behavior
problems. Alcoholic parents are more likely to be loosely connected to their families, to
be absent more often, to be inadequate, uninvolved parents, and to be poor role models
(Zucker & Fitzgerald, 1991a). They may show a generally impaired ability to make
judgments regarding child behavior due simply to frequent intoxication. Alcoholism is
also associated with parental depression and marital conflict (Zucker & Fitzgerald, 1991a),
which, as noted above, have been linked to lower interparent agreement on child behavior
problems.

A problem often associated with parental alcoholism is parental antisocial
personality disorder (Zucker & Fitzgerald, 199 1 a). In addition to alcoholism, parental
antisociality is likely to be associated with other problems such as marital discord and
depression, which, as already noted, are associated with lower interparent agreement on
child behavior problems. Antisocial parents are also more likely to abuse or neglect their
children, to be inadequate parents and poor role models, to be more impulsive and self-
preoccupied, and to expose their children to alcohol and other drugs both in the living
environment, and, in the case of antisocial mothers, in utero (Zucker & Gomberg, 1993).
These factors can result in a whole host of problems for the child, including earlier
exposure to a deviant peer culture due to parental neglect or lack of supervision,

10

behavioral manifestations of the physiological consequences of fetal drug or alcohol
exposure, and earlier appearance of problems in school (Zucker & Gomberg, 1993).

Thus, the parental problems associated with antisociality, such as irnpulsivity and self-
preoccupation, may reduce the antisocial parent's ability and motivation to reliably observe
and report on his or her child's behavior, in addition to increasing the range and scope of
problem behaviors that might actually be manifested by the child and thereby increase the
chances for disagreement.

EMPIRICAL INVESTIGATIONS F THE DETERMINANT F INTERPARENT

A REEMENT

This section reviews the state of our empirical knowledge regarding the child,
parent, and family characteristics discussed thus far and their relationship to interparent
agreement on child behavior problems. Although studies that specifically address the
question of interparent agreement are emphasized, other work that does not assess
agreement but yet bears on the broader issue of parents' perceptions of child behavior
problems is discussed as well, since the agreement observed in a given parental dyad is
composed of the perceptions of two individuals, and processes which affect the
perceptions of one parent but not the other can result in disagreement.

Age of the child. In their meta-analysis, Achenbach and his colleagues found that
across pairs of all types of informants agreement on child behavior was significantly
greater for children aged 6-11 years than for those aged 12-19 years (Achenbach,
McConaughy, & Howell, 1987). The authors suggest that younger children "may be
easier to judge and/or may be cross-situationally more consistent than are adolescents" (p.
227 ). These results are consistent with earlier findings for clinic-referred girls (Achenbach
& Edelbrock, 1979), but not with Achenbach's earlier results regarding clinic-referred
boys (Achenbach, 197 8), which indicated slightly lower agreement for boys 6-11 (.74)
than for boys 12-16 (.79). However, the magnitude of the difference between younger
and older boys in these earlier studies is negligible.

11

Considering even younger children, Achenbach, Edelbrock, and Howell (198 7)
reported that in their longitudinal sample interparent agreement on the CBCL/2-3 was .47
for children at age 2 (n = 61), rose to .57 at age 3 (n = 54), and was virtually unchanged
(.58) at age 4 (n = 57; sample size varied across waves due to attrition). The authors note
that the coefficients at ages 3 and 4 are similar to the mean interparent correlation of .59
found in the 1987 meta-analysis (Achenbach, McConaughy, & Howell, 1987). Achenbach
and his co-workers speculate that the lower level of agreement at age 2 reflects mothers'
greater degree of involvement with younger children's problems compared to fathers'
(Achenbach, Edelbrock, & Howell, 1987). Likewise, Field and Greenberg (1982) ‘
reported that on measures of child temperament interparent agreement was greater for
parents of toddlers and preschoolers (ages 18-60 months, 11 = 33) than for infants (ages 4-
12 months), with average correlations of .46 and .36, respectively.

Taken together, the results of these studies suggest that age bears a curvilinear
relationship to interparent agreement, with agreement being low for infants and very
young children, increasing during the grammar school years, and then decreasing again for
adolescents.

Not all investigators, however, have found a significant relationship between age
and parents' perceptions of child behavior problems. Jensen and his colleagues found no
significant correlation between age and CBCL difference scores for mothers and fathers of
80 non-clinic boys and girls aged 6-12, although there was a trend for agreement
regarding girls to rise with age (Jensen, Xenakis, Davis, & Degroot, 1988). Likewise,
although she did not directly assess interparent agreement, Johnston (1991) did find that
child's age was not significantly related to mothers' or fathers' CBCL scores considered
separately; however, this may be due to low statistical power (11 = 34). Similarly, Dawson
(1992), examining a national probability sample of 3,409 father-child and 5,892 mother-
child pairs, found that the likelihood of a child being rated by the parent as having serious

12

behavior problems was unrelated to the child's age. Again, neither Johnston's nor
Dawson's results speak directly to the question of interparent agreement.

Referral status and adjustment of child. As discussed above, Ferguson and her
colleagues found that for younger boys, degree of interparent agreement on behavior
problems discriminated between clinic-referred and non-referred children, suggesting that
agreement may be a potentially usefill predictor of child adjustment (Ferguson et al. 1974).
Similarly, Mash and Johnson (1983) found in their sample of 91 children (40 referred for
hyperactivity and 51 non-referred children) that the mothers of hyperactives rated their
children significantly higher on the Conners' Abbreviated Rating Scale (Conners, 1972)
and the Werry-Weiss-Peters Activity Scale (WWP; Rauth, Schroeder, & O'Tuama, 1974)
than did fathers, even though parents' scores on these measures were significantly
correlated. Significant difi‘erences were not reported between mothers and fathers of non-
hyperactives.

Other investigators, however, have been less able to clearly discriminate between
clinic-referred and non-referred children on the basis of interparent agreement.
Christensen and his colleagues reported that, in their sample of 81 families, mothers
reported significantly more child behavior problems than fathers across the entire sample,
but that the level of agreement did not difi‘er between parents of distressed and non-
distressed children (Christensen, Margolin, & Sullaway, 1987). Lindholrn and Touliatos '
(1981) compared parents' BPC ratings of children not in counseling (n = 1300) with those
of children in counseling (n = 57) and found no significant difi‘erences in degree of
interparent agreement between the two groups on any BPC factor except Socialized
Delinquency. On this factor the parents of children in counseling were in significantly
higher agreement than parents of non-referred children. The authors concede, however,
that a larger sample of children in counseling may have resulted in more significant results.

More recently, in a sample of children aged 3-13, Christensen, Margolin, and
Sullaway (1992) reported an interaction between the distress level of the child and the

13

index of agreement used to measure interparent agreement. Distress level of the child was
determined by both parents' average scores on the CBCL and the Becker Bipolar
Adjective Checklist (BBACL; Becker, 1960). If a child was rated in the elevated range on
either measure he or she was considered distressed. Focusing solely on the CBCL for
their results, the authors found that whether or not interparental agreement on the CBCL
predicted child distress depended on the index of agreement used. Two measures of
association, kappa and a parental discrepancies index, were calculated. Kappa was used
to assess item-by-item agreement between parents on the CBCL. The mother discrepancy
and father discrepancy indices, on the other hand, consisted of the number of
disagreements on CBCL items in which the parent in question endorsed a behavior and the
other did not, plus the number of disagreements in which the parent in question endorsed
the behavior more strongly than did the other (i.e., as a 2 rather than a 1). The authors
found that if kappa served as the measure of agreement there was no difference between
parents of distressed and non-distressed children. However, parental discrepancies did
increase with the distress level of the child. Christensen and his colleagues suggest that
the parental discrepancies index "may more accurately reflect the psychological reality of
the family than kappa" (p. 424), since due to the nature of the kappa statistic two sets of
parents evincing markedly difi‘erent patterns of discrepancy can nevertheless have identical
kappa scores.

Jacob et al. (1982) also reported an interaction between the measure of association
used to calculate interparent agreement and the referral status of the child. Their sample
consisted of 96 intact families with a male target child aged 10-18. Of these target
children, 52 were adjudicated delinquents recruited through the juvenile courts, while 6
were recruited through the schools and subsequently judged to be "disturbed" after
psychological assessment, for a total disturbed sample of 58. The "normal" sample
consisted of 38 children that were recruited for a larger research project and that did not

meet criteria for disturbance. Jacob and his co-workers found that correlational analysis

14

revealed a markedly lower level of interparent agreement among the parents of disturbed
(r = .18 on the Total BPC) versus non-disturbed (r = .61) children. I-tests, however,
revealed a statistically significant difference between the scores of parents in the mad
sample but not in the disturbed sample. When percentage of agreement was used as the
measure of association, parents in the normal sample showed better agreement than did
those in the disturbed sample (78% vs. 69%, respectively), but this figure was inflated for
parents in the normal sample due to the high rate of agreement on non-occurrences
resulting fiom the low base rate of endorsements for non-disturbed children. The
relationship was reversed when the percentage of agreement on occurrences only (i.e.,
endorsed items only) was considered; in this case parents in the normal sample agreed on
occurrences only 34% of the time compared to 53% for parents in the disturbed sample.
Finally, when the absolute difference between parents' total BPC scores was used, parents
in the disturbed sample revealed a significantly greater mean difference (9.7) than did
those in the normal sample (5.2). Hence, "the alternative reliability indices available to
represent interparent agreement yield difi‘erent patterns of results, largely as a function of
their statistical properties" (p. 606). The authors recommend the absolute difi‘erence score
as the best all-around measure of interparent agreement, in part because it can be most
easily used as a variable to investigate covariates of interparent agreement.

In summary, there exists evidence both for and against the relationship between
interparent agreement and child adjustment or referral status, although these studies are
limited by small sample sizes, particularly with respect to clinic-referred children. Further,
there is evidence that the relationship between interparent agreement and child adjustment

is partly a firnction of the index of agreement employed.

15

DMn of child behavior. Many investigations have explored whether interparent
agreement varies with the type of child behavior being rated. Most often domain of child
behavior is expressed in terms of broad-band groupings based on factor-analytic studies of
inforrnants' rating scale responses, such as the internalizing and externalizing dimensions
identified by Achenbach and his colleagues (e.g., Achenbach, 1978; Achenbach &
Edelbrock, 1979). This dichotomy, reflecting "a distinction between fearful, inhibited,
overcontrolled behavior [i.e., internalizing], and aggressive, antisocial, undercontrolled
behavior [i.e., externalizing] " (Achenbach & Edelbrock, 1983), has been identified in
numerous multivariate studies (see Achenbach & Edelbrock, 1978, for a review) and has
been variously termed personality problems versus conduct problems, inhibition versus ,
aggression, and overcontrolled versus undercontrolled (Achenbach & Edelbrock, 1983).

It has generally been found that informants agree more on the presence of
externalizing than internalizing behaviors (Achenbach, McConaughy, & Howell, 1987);
this is typically attributed to the putatively more observable, objective, and disruptive
nature of externalizing problems as opposed to the more covert, subjective, and
unobtrusive internalizing problems (Earls, 1980; Christensen et al., 1992). Christensen
and his colleagues (Christensen et al., 1992) had judges rate each CBCL behavior problem
item on its objectivity, observeability, molarity, social undesirability, and disturbance, and
found that CBCL items that load on the externalizing scale consistently across age x sex
groups were rated as significantly more objective, observable, and socially undesirable
than items that load on the internalizing scale consistently across age x sex groups.
Further, for items that load on the externalizing and internalizing scales for at least one age
x sex group, interparent agreement (kappa) was significantly higher for externalizing
items.

For 108 families of US Army personnel, Jensen and his co-workers found that, for
boys, the intraclass correlation (ICC) between parents' CBCL internalizing scores was .49,

compared to .53 for externalizing scores. The difl‘erence between scales was much more

16

pronounced for girls, however, with ICC's of .41 versus .67 on the internalizing and
externalizing scales, respectively. The authors speculate that this contrasting result for
boys versus girls suggests that interparent agreement is increased when reporting overt,
observable behaviors that violate social role expectations, as when girls exhibit
externalizing problems such as aggression (Jensen, Traylor, Xenakis & Davis, 1988).

Other research has produced similar findings. In their sample of 1,300 non-clinic
children and their parents, Lindholm and Touliatos (1981) reported a correlation between
parents of .72 on the BPC conduct problems factor compared to .65 for personality
problems. Garrison and Earls (1985) reported interparent correlations of .56 for CBCL
externalizing problems and .40 for internalizing problems in their sample of 62 non-clinic
6-7—year-olds. Mash and Johnson (1983) found correlations of .74 for externalizing
scores and .67 for internalizing scores between parents of 91 hyperactive and normal
children (interparent agreement was not reported for the two groups separately). More
recently, Achenbach et al. (1991) reported interparent externalizing and internalizing
correlations of .81 and .62, respectively, among the parents of 94 clinic children aged 4-
16. Thus, much empirical work shows a general trend for undercontrolled, externalizing
behaviors to be agreed upon more often by parents than are overcontrolled, internalizing
behaviors.

Other work, however, questions the strength of this trend. In their meta-analysis,
Achenbach, McConaughy, and Howell (1987) were unable to show a statistically
significant difference between parents' reported agreement on internalizing and
externalizing problems, although such a difference did exist for other pairs of informants.

Other studies, such as Becker (1960), Kazdin, Esveldt-Dawson, Unis, and
Rancurello (1983), and Schaughency and Lahey (1985) have likewise failed to show a
clear pattern of greater interparent agreement on undercontrolled behaviors. Indeed,
Johnston (1991) found just the opposite pattern for parents of 34 clinic-referred children
aged 4-13; the interparent correlation for CBCL internalizing behaviors was .55,

17

compared to .47 for externalizing behaviors. These investigations, however, are limited by
small sample sizes and low statistical power; the largest sample size in this group is 71 in
Becker (1960).

Finally, Jacob et al. (1982) showed that, as was the case regarding referral status,
the pattern of agreement across domains of behavior depends upon the index of agreement
employed. When assessed correlationally, interparent agreement in their total sample of
96 children showed a pattern similar to most other studies, with a correlation of .73
between parents' BPC conduct problems scores compared to .43 for personality problems.
A similar pattern was observed when either absolute difference scores or percentage of
agreement on occurrences were used to measure agreement, but the opposite was found
when t-tests were employed to assess differences between parents' scores. In the latter
case, a significant difference was found for parents' scores on conduct problems, but not
personality problems.

In summary, most studies suggest that interparent agreement is greater for ratings
of undercontrolled, externalizing behavior problems than for overcontrolled, internalizing
problems. Some investigations, most notably Achenbach and colleagues' meta-analysis
(Achenbach, McConaughy, & Howell, 1987), have failed to show such an efiect for
parent's agreement, although they did show a difference for other pairs of informants.

Time mt with child. Studies that report discrepancies between mothers' and
fathers' ratings of child behavior often attribute these disagreements to difi‘erences in the
amount and/or quality of time spent with children by mothers and fathers, but few have
actually investigated this question empirically. It is assumed that parental discrepancies,
with mothers typically reporting more problem behaviors than fathers, are in part the result
of mothers' greater time commitment to the children and involvement with child-rearing
tasks (Achenbach, Edelbrock, & Howell, 1987; Christensen et al., 1987; Christensen et al.,
1992; Schaughency & Lahey, 1985). While this may or may not be a reasonable
assumption, the relationship between the amount and quality of time parents spend with

18

children and interparent agreement on child behavior problems requires further empirical
investigation. While it does appear that mothers are more involved with their children
than fathers, especially in the earlier years, fathers' time commitment is substantial as well,
and fathers contribute in important ways to their children's development (Earls, 1976;
Phares, 1992; Phares & Compas, 1992). Further, Phares (1992) suggests that the amount
of time spent by fathers is perhaps not as important as the quality of time spent, and that
any investigation of differences between mothers and fathers must take into account the
relative amount of actual caretaking responsibilities of each parent, not just the amount of
time spent.

A few studies have attempted to measure time spent with child and relate it to
interparent agreement. In his sample of 100 three-year-olds, Earls (1980) found that
fathers tended to underreport problem behaviors relative to mothers. However,
comparing the reports of working mothers to those of fathers he found that working
mothers' reports were more similar to fathers' reports than to the reports of non-working
mothers. Assuming that fathers and working mothers spend a similar (i.e., lower) amount
of time with their children, Earls concluded that the time factor may help explain the
overall discrepancy between parents' reports. Jensen and his colleagues also used mothers'
employment as an indicator of time spent with child, along with fathers' recent absences
(e.g., on business trips), but found neither of these measures to be significantly related to
parental discrepancies in CBCL ratings of their children (Jensen, Xenakis, Davis, &
Degroot, 1988).

Burrows and Kelley (1983) employed a more direct indicator of time spent with
child, using a situational assessment questionnaire to measure daily child/parent
interactions and relating it to parents' ratings of their children's behavior on the Children's
Behavior Checklist (CBC; Ferguson et al., 1974). The authors found relatively good
interparent agreement on the CBC (average ratio of agreements to total responses = .81),
but found that those dyads demonstrating the highest levels of agreement did not

l9

necessarily spend the largest amount of time in similar situations with their child as
measured by the situational assessment questionnaire. Burrows and Kelley suggest that
these parents have developed "a shared familiarity with their own child's behavior across a
variety of situations" (p. 47) without actually experiencing the same situations with the
child. Hence, some other factor such as the quality of interparent communication may be
important for explaining interparent agreement on child behavior ratings (Kazdin et al.,
1983; Schaughency & Lahey, 1985). Burrows and Kelley did not directly measure this
"shared familiarity" factor, however; further, their results should be interpreted cautiously
due to a very small sample size (17).

Fitzgerald, Zucker, and Maguin (in press) examined interparent agreement in 107
high-risk families with young sons in which parents were divided into low, medium, and
high-contact groups based on a revised version of the Oregon Social Learning Center
(OSLC) Parent Daily Report (PDK Chamberlain-Waechter, 1980). The PDR is designed
to assess low-frequency behaviors (e.g., stealing, fighting) by asking parents for yes or no
responses to a series of items pertaining to the past 24 hours. In addition, parents are
asked about how much time they (and their spouse) spent with the target child during the
past 24 hours. Parents are contacted by phone daily for six days (three calls to mother and
three to father), including one weekend day. The authors examined interparent agreement
regarding aggressive and acting-out behavior problems as measured by three different
child behavior rating scales. They found that high-contact parents demonstrated
significantly higher agreement (1 = .64) than did medium- (r = .40) or low-contact (r =
.41) parents on the Child Behavior Rating Scale (CBRS) aggression score. Similarly, they
found better agreement among high-contact parents (r = .54) than among low-contact
parents (r = .35) on the Conners Parent Questionnaire conduct problems score. However,
the authors reported no significant differences in level of agreement among parents in the
three contact groups on the CBCL aggression score. The results are best explained by
variation in fathers' time spent with the target child. The authors found that, overall,

20

mothers rated their sons as more aggressive on the CBRS than did fathers, and that
mothers' mean rating of their sons' aggressiveness did not vary significantly among contact
groups. However, fathers perceptions of their sons' aggressive behavior did vary by
contact group: fathers who spent more time with their sons perceived them to be more
aggressive. Fathers spending the most time with their children (i.e., the high-contact
group) perceived them to be as aggressive as did mothers.

In summary, some studies that have investigated the relationship between
interparent agreement and the amount and quality of time parents spend with their child
failed to measure the time variable directly. Two studies that did directly measure the
amount of time spent with the child report mixed results, but overall suggest that the
greater the amount of time fathers spend with their children, the greater is their
concordance with mothers' reports of child behavior problems. More work is needed to
clarify this relationship. Such work should measure the time factor as directly as possible
rather than relying on crude indicators such as mothers' employment status, and should
take into account the quality of interparental communication.

Marit_al discord. It is often assumed that child psychopathology exists within the
context of a disturbed family system characterized by marital problems and other family
stressors (Margolin, 1981). It is important to empirically investigate the relationship
between marital discord and child behavior problems, particularly when assessment of the
latter is based largely upon parents' reports. There are a number of studies that have
examined the effect of marital discord on interparent agreement, with varying results.

Some investigations have found little or no relationship between marital
adjustment, parents' reports of child behavior problems, and interparent agreement. In
their sample of 28 clinic-referred children and their parents, Thompson and Bemal (1982)
found no relationship between marital adjustment as measured by the Locke-Wallace
Marital Adjustment Test (MAT; Locke & Wallace, 1959, cited in Thompson & Bernal,

1982) and parents' scores on the Becker Attitude Inventory (Becker, 1960). Likewise,

21

Schaughency and Lahey (1985) failed to demonstrate a significant association between
parents' MAT scores and child behavior ratings on the Conners Parent Rating Scale (PRS;
Conners, 1973, cited in Schaughency & Lahey, 1985). Parents' MAT scores did correlate
between -.30 and -.39 with several PRS factors (9 < .05), but the authors did not consider
these coeflicients to be statistically significant when controlling the experiment-wise error
rate at the .05 level. Their small sample size (n = 41) may have contributed to their failure
to find significant results. Finally, Johnston (1991) failed to find a significant relationship
between parents' scores on a short form of the MAT and their ratings of child behavior on
the CBCL. Note that these three studies did not assess interparent agreement per se, but
instead related the marital adjustment measures to mothers' and fathers' behavior ratings
separately. Further, all three studies suffered from low statistical power, with sample sizes
of 28, 41, and 34, respectively.

Jensen and his colleagues did assess interparent agreement, but failed to directly
measure marital discord, instead relating agreement to a more global measure of child and
family stressors, the Coddington Life Events Record (LER; Coddington, 1972, cited in
Jensen, Xenakis, Davis, & Degroot, 1988), to mother-father CBCL difi'erence scores for
108 non-referred children. The authors found that LER scores were not significantly
related to mother-father CBCL difference scores (Jensen, Xenakis, Davis, & Degroot,
1988). Christensen et al. (1987) likewise found no significant difference in the size of
CBCL difi‘erence scores between parents in distressed versus non-distressed families (as
defined by cutoff scores on child and marital distress measures; n = 81). In an earlier
study, however, Christensen and his colleagues did find marital adjustment to be
significantly negatively related to parents' scores on the BBACL; unfortunately,
interparent agreement was not reported in this investigation (Christensen et al., 1983).

More recent work by Christensen et al. (1992) suggests that the relationship
between interparent agreement on child behavior problems and marital/family distress is
partially a function of the index of agreement. Using cutoff scores on the CBCL, BBACL,

22

and two measures of marital adjustment, the Dyadic Adjustment Scale (DAS; Spanier,
1976) and Areas of Change Questionnaire (ACQ; Weiss & Birchler, 1975), Christensen
and his co-workers defined 46 non-distressed families, 41 "single-system" distressed
families (i.e., above the clinical cutofl‘ on at least one of the marital or; child distress
measures) and 50 "dual-system" distressed families (i.e., above the cutoff on at least one
marital at one child measure). The authors found that parental discrepancies (i.e., the
number of disagreements on occurrence and frequency of occurrence of behavior
problems) increased with the distress level of the family, but kappa did not. Christensen
and his colleagues note that the discrepancies index may more accurately reflect the
dynamics of the marital dyad than kappa, since due to the nature of the kappa statistic,
dyads evincing markedly different patterns of discrepancy can nevertheless show identical
kappa scores. O

Other researchers, too, have found a relationship between marital adjustment and
parents' reports of child behavior problems. Dadds and Powell (1991) recemly reported
that mothers' scores on the DAS predicted their ratings of aggression in clinic and non-
clinic children of both sexes and their ratings of anxiety in clinic and non-clinic boys, but
their analysis did not include fathers or interparent agreement. Earls (1980) did assess
interparent agreement and found that agreement was positively related to mothers' rating
of marital satisfaction and negatively related to fathers' reports of recent stressful life
events. Likewise, in their sample of 18 inpatient children and their parents, Rosenberg and
Joshi (1986) reported significant correlations between a clinician's rating of marital discord
and parental differences on the CBCL externalizing and total behavior problem scores, and
between the clinician's rating of parents' communication difficulties and their differences
on the CBCL internalizing and total scores.

In summary, there are a number of studies that have failed to find a significant
relationship between marital and family discord and parents' reports of child behavior
problems, but many of these investigations have failed to assess interparent agreement per

23

se and have been plagued by small sample sizes. There is evidence that the index of
agreement used can afi‘ect the magnitude of the relationship reported between interparent
agreement and marital maladjustment. Margolin (1981) has said that empirical work
based on major theoretical frameworks such as family systems theory, social learning
theory, and role strain theory has offered only limited support for explanations of the ways
in which marital discord relates to child behavior problems. She calls for longitudinal
studies that can assess the long-term reciprocal effects of marital and child problems
across stages of child development. Further work in this area is necessary to clarify these
relationships.

Parental psychopathology. In addition to marital discord, parents' psychological
problems have been associated with child psychopathology and family disturbance (Phares
& Compas, 1992). There have been a number of studies investigating the relationship
between parents' psychiatric symptoms and their reports of child behavior problems. The
majority of these have reported results for mothers and fathers separately, however, not
measuring directly the impact of parental psychopathology on interparent agreement.
Depressign. Parental (particularly maternal) depression has often been implicated as
afl‘ecting parents' perceptions of child behavior problems. F riedlander, Weiss, and Traylor
(1986) found in their sample of 33 clinic and 21 non-clinic children that mothers' scores on
the Beck Depression Inventory (BDI; Beck, Ward, Mendelson, Mock, & Erbaugh, 1961)
were significantly correlated with their broad- and narrow-band CBCL scores except for
Somatic Complaints. This suggests that mothers' ratings of their children's behavior are
significantly related to mothers' depressive symptomatology, but the authors do not
compare mothers' perceptions to those of fathers. Johnston (1991) measured depression
in both mothers and fathers of 34 clinic-referred children aged 4-13. She reported that
mothers' BDI scores were significantly related to CBCL internalizing scores, but not to
externalizing scores, in contrast to the results of F riedlander and colleagues. F urtherrnore,
the magnitude of the correlation coeficients reported by Johnston was about halfthat

24

reported by Friedlander et al. (.30 vs. .60 for internalizing, and .27 vs. .56 for externalizing
behaviors, respectively). There was not a significant relationship between BDI and CBCL
scores for fathers in Johnston's sample. Likewise, Webster-Stratton (1988) found
significant correlations between BDI and CBCL scores for mothers but not for fathers in
her sample of 85 clinic-referred children aged 3-8. Johnston's and Webster-Stratton's
results suggest that depression affects mothers' reporting of child behavior problems
differently than it does fathers'; unfortunately, neither study directly assessed the effect of
depression on interparent agreement.

In a rare study assessing paternal (but not maternal) depression, Jacob and
Leonard (1986) found that, for children aged 10-18, parental dyads with depressed fathers
reported significantly more internalizing and total behavior problems for boys and more
internalizing, externalizing, and total behavior problems for girls than did control dyads.
This finding contrasts with the results of Johnston (1991) and Webster-Stratton (1988) in
its finding of a significant effect of depression on fathers' reports of child behavior
problems. However, it appears that the fathers in Jacob and Leonard's depressed sample,
all of whom had to meet the Research Diagnostic Criteria (RDC; Spitzer et al., 1975) for
depressive disorder, were more depressed than those in Johnston's and Webster-Stratton's
samples; fathers in both of the latter samples showed mean BDI scores of around 5, which
is not significantly elevated.

Other work has failed to show a significant relationship between depression and
parents’ reports of child behavior problems. Christensen et al. (1983) reported non-
significant correlations between parents' BBACL ratings and their scores on the lung
Self-Rating Depression Scale (Zung, 1973, cited in Christensen et al., 1983). Likewise,
Schaughency and Lahey (1985) found no significant relationship between parents' BDI
scores and their ratings of their child on the Conners PRS once the correlations were

adjusted to control the experiment-wise error rate at the .05 level. Again, small sample

25

sizes (n = 36 in Christensen et al. and 41 in Schaughency & Lahey) and sub-clinical levels
of depression may account for these non-siglificant results.

Only one study directly assessed the relationship between depression and
interparent agreement on behavior problems. Webster-Stratton and Hammond (1988)
found that depressed mothers (defined as BDI score 2 10) perceived their children as
having more externalizing, internalizing, and depressive behavior problems on the CBCL
than did their (non-depressed) spouses; the same discrepancy was not observed for non-
depressed mothers and their husbands. Unfortunately, no studies were found that directly
examined the efl‘ects of paternal depression on interparent agreement.

Recently, Richters (1992) critically reviewed the literature that claims that
depressed mothers have a "distorted" view of their children's behavior and give inflated
reports of their offspring's behavior problems. He argues (persuasively) that none of the
research making the case for distortion is able to show that depressed mothers are more
likely than non-depressed mothers to disagree with some "objective" criterion rating of
their child's behavior, or even that such a criterion rating exists and is a more accurate
measure of children's behavior than mothers' ratings. However, the question of interparent
agreement on child behavior ratings makes no assumptions about the veracity of one
informant's report over the other's. Rather, the concern is with whether or not the level of
agreement between parents is associated with important child, parent, or family
characteristics and would therefore. serve as an index or predictor of child adjustment.

The question of the veracity of inforrnants' information is more usefirlly framed in terms of
the usefirlness and validity of data obtained fi'om difi‘erent informants that interact with
children in different situations, as opposed to determining who is "right" or "wrong"
(Achenbach, McConaughy, & Howell, 1987). As Reid and his co-workers put it, "a
report's veracity misses the point: Ifan involved adult reports that there is a problem with
a child, whether this represents perceived or actual dificulties, there is in fact a problem”
(Reid, Kavanagh, & Baldwin, 1987, p. 458). Hence, it remains an important and largely

26

unanswered question whether and how interparent agreement is affected by parental
depression.

Alcohol problems. Alcohol abuse/dependence has been investigated in relation to parents'
reports of child behavior problems. Jacob and Leonard (1986) averaged mothers' and
fathers' CBCL scale scores to yield composite parental scores and compared groups with
alcoholic, depressed, and control fathers. 296 children were rated (100 with alcoholic
fathers, 91 with depressed fathers, and 105 controls). Results showed that parents in the
alcoholic and depressed groups rated boys higher on the CBCL internalizing and total
behavior problems scales than did controls, but that the alcoholic and depressed groups
did not difi'er significantly from each other. For girls, on the other hand, significant
differences were found on the externalizing, internalizing, and total problem scales, with
alcoholics rating their daughters as more disturbed than controls, and depressives rating
their girls as more disturbed than alcoholics. More recently, Dawson (1992), examining
National Center for Health Statistics data on 3,409 father-chfld and 5,892 mother-child
pairs (for children aged 5-17), found that the presence of an alcoholic father (but no
alcoholic mother) increased by two-thirds the odds of the target child being rated in the
top 10% of a behavior problem distribution based on parents' reports; the odds of this
occurring were more than doubled if the mother (but not the father) was an alcoholic.
Further, in the mother-child sample, if an adult heavy drinker other than the mother was
present in the household, the odds of the child being rated as disturbed were increased
nearly two and one-half times. Thus, there is evidence that parental alcoholism is
associated with more disturbed ratings on child behavior measures, but it remains to be
directly demonstrated whether alcoholism is associated with reduced interparent
agreement on these measures.

W Parental alcoholism has been associated with parental antisociality
and deviance (Zucker & Fitzgerald, 1991a). Unfortunately, studies relating parental
antisocial behavior and criminality to parents' perceptions of child behavior problems are

27

conspicuously absent from the literature, despite evidence that these risk factors carry
significant implications for child adjustment (Loeber & Dishion, 1983; Phares & Compas,
1992; Zucker & Fitzgerald, 1991a). One study attempted to measure this relationship. In
their sample of 36 families with children aged 4-12, Christensen et al. ( 1983) found that
parents' scores on the social non-conformity subscale of the Psychological Screening
Inventory (PSI; Lanyon, 1970, cited in Christensen et al., 1983) were not related to their
ratings of their child on the BBACL.

A more thorough investigation of the effect of parental antisociality on perceptions
of child behavior problems is needed. Such an investigation should include a reliable and
valid measure of parents' antisocial behavior problems and sufiicient statistical power.
Overfl psychologipg firnctioning. Two studies assessed interparent agreement on
behavior problems as it relates to parents' general psychological adjustment. Earls (1980)
found that mothers' psychological maladjustment as assessed via clinical interview was
positively related to mother-father discrepancies in reported child behavior problems,
although the same was not true for fathers' maladjustment. More recently, Jensen and his
colleagues (Jensen, Traylor, Xenakis & Davis, 1988) administered a multiphasic symptom
inventory, the Hopkins Symptom Checklist (HSCL; Derogatis et al., 1974) to parents and
related these scores to their difi‘erence scores on the CBCL internalizing, externalizing,
and total behavior problem scales, with interesting results. They found that, for boys,
mothers' total HSCL symptoms were significantly correlated with mother-father CBCL
differences on the internalizing and total behavior problem scales. Mothers' symptoms
were not related to mother-father CBCL differences for girls, however. On the other
hand, fathers' HSCL symptoms were not significantly correlated with CBCL differences
for boys' behavior ratings, but were quite significantly related to mother-father difi‘erences
regarding girls' behavior problems. Hence, fathers' symptoms were associated with
interparent disagreement regarding daughters, while mothers' symptoms were associated
with disagreement regarding sons.

28

Overall summary of the errrnirical findings. The literature presents mixed results
regarding the relationship between interparent agreement on child behavior problems and
the various child, parent, and family characteristics that have been discussed. Generally,
the disagreement among studies tends to be one of " some efi‘ect" versus "no effect," rather
than for studies to report contradictory effects. That is, some consistent findings do
emerge from the literature, and those studies that fail to concur with these general findings
more often report no significant findings than they do findings that contradict the general
trends. Further, the studies that fail to report significant findings almost invariably suffer
from low statistical power, with sample sizes of less than 100 and in most cases less than
50. The only exception is the finding in Achenbach, McConaughy, and Howell's (1987)
meta-analysis of no significant difi‘erence between parents' reports on externalizing versus
internalizing behavior problems. However, their analysis did show the predicted efi‘ect for
other pairs of informants, and many other studies do report higher agreement between
parents on externalizing problems. Thus, the weight of the evidence suggests that
agreement tends to be higher for externalizing behaviors across informants.

This review of the literature suggests a need for studies with larger samples that
directly measure interparent agreement, and that validly measure a range of clinical
phenomena such as parental depression, alcoholism, and antisocial behavior. The present
investigation hopes to contribute to this need.

STATEMENT OF THE PROBLEM

As reviewed above, most studies that have examined interparent agreement on
child behavior measures have reported at least moderate correlations between parents'
scores. The 1987 meta-analysis of Achenbach, McConaughy, and Howell, which
aggregates the results of 22 studies and hence provides some of the more reliable figures
on interrater reliability available, found a mean interparent agreement coeficient of .59
across the 22 studies. In the manual for the CBCL, the instrument that is the focus of the

present study, Achenbach and Edelbrock (1983) reported interparent correlations of .77

29

on the Intemalizing scale, .74 on the Externalizing scale, and .75 on the Total Behavior
Problems scale for 33 clinic-referred boys aged 4-5.

However, as discussed earlier, the results of an analysis by Reider (1991) of data
from the MSU Longitudinal Study (Zucker et al., 1984) contrast sharply with those found
by Achenbach and his colleagues. Recall that for 90 alcoholic fathers and their spouses,
the interparent correlations were .27 for Intemalizing behaviors, .37 for Externalizing
behaviors, and .26 for Total Behavior Problems. These coefiicients are significantly lower
than those reported by Achenbach and Edelbrock (1983) in the CBCL manual (all p's s
.004; Reider, 1991). This considerable variation in reported interparent agreement is of
primary concern in the present investigation.

Hence, because of the importance to child assessment of interparent agreement on
child behavior problems, it is crucial to understand the conditions under which agreement
varies. The objective of the present study is to better characterize the predictors of
interparent agreement in a population-based sample of families in which the breadth of the
data base, as well as the familial variation in psychopathology, is very large. On both
counts, the opportunity to track sources of agreement (and disagreement) is great, and the
representative nature of the sample (albeit one with substantial variation in
psychopathology) assures that the findings will be more generalizable than would be the
case with a clinic-derived study. Another need is for research which articulates the
difl‘erences between mother versus father influences as they pertain to both veracity and
systematic distortions in parents' perception of their child. Only when a data base assesses
both of these sources of variation can conclusions be drawn; the present study hopes to
address this clear research need.

Jacob and his colleagues noted over ten years ago that "the study of parent
agreement, its correlates and consequences, has barely gone beyond the preliminary stages
of investigation" (Jacob et al., 1982, p. 606). A review of the literature clearly suggests

that the need for firrther exploration of this question remains great.

3O

HYPOTHESES
Based upon a review of the relevant literature, the following hypotheses will be
tested in this investigation.

Child chgacteristics:

l) Irpegparent agreement on child behpvior problems will be garter for older children M
for younger. The children in our sample range from 2-7 years old; in this age range
agreement is predicted to rise with age, given the findings in the literature (Achenbach,
McConaughy, & Howell, 1987; Achenbach, Edelbrock, & Howell, 1987). This is
perhaps due to the child's growmg repertoire of observable behaviors and the
increasing involvement of fathers as children get older.

2) Inte_rparent mment will be ggeater for externalizing behavior problems thln for
internalizing. This has been widely reported in the literature, and may be due to the
putatively more observable, objective, and disruptive nature of externalizing problems
as opposed to the more covert, subjective, and unobtrusive internalizing problems
(Christensen et al., 1992).

3) Greater severity of child beh_avior problems will be associated with lower interparent

 

agzeement. Studies have shown that interparent agreement on child behavior problems
is lower for parents of clinic-referred children than for those of non-referred children
(Ferguson et al., 1974; Christensen et al., 1992), suggesting that greater severity of
reported problems is associated with poorer agreement. Because the children in the
present sample are not clinic-referred, severity of behavior problems will be measured
as a continuous variable using the sum of both parents' CBCL Total Behavior

Problems scores.

Parenting chmcteristics:

4) Wter the more time fathers spend with the child. The
relative amount of actual child-rearing responsibilities of each parent would perhaps be

 

a more powerfirl predictor (Phares, 1992), but is not measured in our data set.

31

However, it seems reasonable to suggest that, ceteris paribus, as the amount of time
fathers spend with their children increases, so will their opportunities to observe their
children's behavior, thus increasing interparent agreement on child behavior problems.

5) Marital maladjustment will be aasociated with lower interparent agreement on child
behavior problems. There is empirical as well as theoretical support for this hypothesis
(Christensen, 1992). A poor marital relationship may be associated with poorer
interparent communication, fewer shared beliefs regarding child-rearing and the
definition of problem behavior, and asymmetrical parent-child relationships
(Christensen et al., 1987; Margolin, 1981).

Parental individufl differences/psychopathology:

6) A ggeater level of overall parental pgchopathology will be associated with lower
intarparent went. Work by Jensen and his colleagues (Jensen, Traylor, Xenakis

 

& Davis, 1988) and Earls (1980) suggests that parents' general psychological
adjustment is positively related to agreement on child behavior measures.

7) A gpeater level pf parental dapression will be associated with lower inte_rparent
aggeement. Whether due to "distortion" on the part of the depressed parent or to
increased child volatility in response to parental depression, this relationship has been
widely reported in the literature for mothers (e.g., Webster-Stratton & Hammond,
1988) and will be investigated as it pertains to both parents in the present study in an
attempt to begin to redress the lack of information about the effects of fathers'
depression on interparent agreement.

8a) Parents in the overt alcoholic gr_'oup will shpw lowar inte_rpgent aggeement than those
inth v r nrol ou s d ar tsinthe vert holic will how
lower ampem thaa those in the control group; and, i

b) skater severity pf alcohol problems will be associated with lower intarparent
meta. Alcoholic parents are more likely to be loosely connected to their families,

 

 

 

to be absent more often, to be inadequate, uninvolved parents, and to be poor role

9)

32

models (Zucker & Fitzgerald, 1991a). They may show a generally impaired ability to
make judgments regarding child behavior due simply to frequent intoxication.
Alcoholism is also associated with parental depression and marital conflict (Zucker &
Fitzgerald, 1991a), which, as noted above, have been linked to lower interparent
agreement on child behavior problems.

Parental antisociality will be associated with lower interparental agreement. The need
for an empirical test of this question remains great (Phares & Compas, 1992); it is
posited that the parental problems associated with antisociality, such as irnpulsivity and
self-preoccupation (Zucker & Gomberg, 1993), may reduce the antisocial parent's
ability and motivation to reliably observe and report on his or her child's behavior, in
addition to increasing the range and scope of problem behaviors that might actually be
manifested by the child and thereby increase the chances for disagreement.

CHAPTER 11
METHOD
SUBJECTS

Subjects are 298 families participating in wave one of the Michigan State
University Longitudinal Study (Zucker & Fitzgerald, 1991b), a longitudinal investigation
of the factors affecting child development and health in the children of alcoholic and non-
alcoholic parents. At wave one the sample consists of 298 boys aged approximately 2-7
and both of their biological parents. One exception is a family that includes a stepmother
rather than the child's biological mother; however, in this case the stepmother has cared
for the child since he was a few months old and the child has had little contact with the
biological mother.

The sample is divided into three groups of families: a court-recruited "overt"
alcoholic group (n=166), a community-comparison "covert" alcoholic group (n=42), and a
community-comparison non-alcoholic (control) group (n=90). The overt alcoholic
subjects are men recruited from the district courts in four mid-Michigan counties. All
eligible subjects a) were convicted of drunk driving and had a blood alcohol concentration
of at least .15 (.12 if a repeat ofi‘ender) at the time of arrest; b) had biological sons
between 3.0 and 6.0 years of age at the time of initial contact (wave one); c) were living in
intact families with the target child and his biological mother at the time of initial contact;
and d) were of non-Hispanic Caucasian heritage. The men were asked by their probation
oficers for permission to release their names and phone numbers to the MSU F arnily
Study, and were assured that the study had no connection to the courts and that all
information obtained would remain confidential. Seventy-nine percent of those asked

agreed to be contacted by the study and 91% of those contacted agreed to participate.
33

34

Subsequent assessments were conducted to confirm that the fathers met the Diagnostic
and Statistical Manual of Mental Disorders, Third Edition--Revised (DSM-IH-R;
American Psychiatric Association, 1987) criteria for alcohol dependence or abuse.

The community-comparison groups were recruited using door-to-door survey
methods and were matched to the core alcoholic group for race, intactness of family, and
presence of a male child aged 3-6. The parents in 90 of these families (the community
controls) did not meet diagnostic criteria for either alcohol or other drug abuse or
dependence. F orty-two of these families (the covert alcoholics) were found to have a
father who met a probable or definite alcohol-related diagnosis, although none of these
fathers had been convicted of drunk driving or drug-related ofi‘enses.

PR EDURE

Child, parent, and family data are collected by a trained team of professional stafl‘,
graduate students in clinical and developmental psychology and social work, and
undergraduate students. The full assessment protocol involves nine data collection
sessions with the family in their home, at the University, or by phone, and one data
collection session with teachers. Assessment staff are blind to the family's group
membership (i.e., overt alcoholic, covert alcoholic, or control).

MEASOES

For the purposes of this investigation, the following measures from the assessment
protocol will be used.

Ohild behavior problem; Child behavior problems are measured using the Child
Behavior Checklist (CBCL; Achenbach & Edelbrock, 1983), a 118-item checklist that
yields standardized scores on eight narrow-band and two broad-band factors. For this
investigation the focus will be on the more reliable broad-band Intemalizing and
Externalizing factors, and on the Total Behavior Problem score. The CBCL has been
normed on children 4-16 years old, but there is evidence that the scale can be validly used
with three-year-olds as well (Fitzgerald et al., 1993). The items on the scale are scored

35

from 0 - 2, with 0 indicating that the item is not true of the target child, 1 indicating that it
is somewhat or sometimes true, and 2 indicating that it is very true or often true. The
CBCL is filled out independently by each parent. Achenbach and Edelbrock (1983) report
a median one-week test-retest reliability of .89 for CBCL scale scores. Type or domain of
child behavior problems will be defined according to the broad-band Intemalizing and
Externalizing factors. Severity of child disturbance will be measured by summing mothers'
and fathers' CBCL Total Behavior Problems scores.

Amount of time spent with the target child. The amount of time fathers spend with
the target child is assessed using a revised version of the Oregon Social Learning Center
(OSLC) Parent Daily Report (PDR; Chamberlain-Waechter, 1980). The PDR is designed
to assess low-fiequency behaviors (e.g., stealing, fighting) by asking parents for "yes or
no" responses to a series of items pertaining to the past 24 hours. In addition, parents are
asked about how much time they and their spouse spent with the target child during the
past 24 hours. Parents are contacted by phone daily for six days (three calls to mother and
three to father), with one of the calls to each parent involving a weekend day. The tirne-
spent index will be averaged over the three days that the father reports. Since there is no
significant disagreement among mothers and fathers in this sample regarding how much
time fathers spend with the target child either in terms of total time (t=.87, us.) or time
spent in actual activities (EU, as), fathers' self-reports were used in the analyses.

Marital ralationship. The spouses' interpersonal relationship is assessed using
Benjamin's (1983) Intrex Questionnaire based upon her Structural Analysis of Social
Behavior (SASB) model of interpersonal relationships. The respondent rates his behavior
toward himself and his spouse, as well as his spouse's behavior toward him, on a 0-100
point scale for each item. A computer program translates questionnaire responses into a
three-level description of self- and other-perception and interpersonal interaction for each
spouse. Prior work using the Intrex system has produced interjudge agreement ranging
fi'om kappa = .66 (Humphrey, Apple, & Kirschenbaum, 1986) to kappa = .91 (Henry,

 

36

Schacht, & Strupp, 1986). Ichiyama (1994) has recently developed an index of marital
satisfaction based on the SASB clusters that incorporates both the self and other foci of
the Intrex questionnaire. The appropriate face-valid clusters are combined into indices if
marital discomfort and marital comfort for each respondent, and an index of overall
marital satisfaction is derived by computing the difference between each respondent's
marital comfort and discomfort scores. This measure of marital satisfaction will be
employed in the present investigation.

Parents' overall psychological functioning. Two measures of parents' overall
psychological functioning will be examined. The first is each parent's score on Axis V of
the DSM-III, involving social competence. Specifically, this axis measures the parent's
highest level of adaptive functioning during the past year. Information regarding the
subject's social, occupational, and leisure time functioning is obtained during a clinical
interview employing the Diagnostic Interview Schedule (DIS; Robins, Helzer, Ratcliff, &
Seyfried, 1982). An interrater reliability of .85 for Axis V has been established in the
present study (Zucker & Fitzgerald, 1991b).

The second measure of general firnctioning is the Composite Psychological Health
Q-Sort (Livson & Peskin, 1967; 1981). This measure utilizes the California Q-Sort
(Block, 1978), a set of 100 statements about the subject's personality and social behavior.
The interviewer who administers the DIS sorts the items into a fixed, normally distributed
set, such that they range from those statements that are most descriptive of that person to
those that are least descriptive. A measure of psychological health is derived by
correlating the actual Q-sort for each subject with a composite, expert-derived Q-sort that
depicts an ideal or "psychologically healthy" personality. Prior experience with this
measure in our research group has produced interobserver reliabilities of at least .78

(Zucker & Fitzgerald, 1991b).

37

Depression. Two measures of depression are employed in the study. The
Hamilton Rating Scale for Depression (HRSD; Hamilton, 1960), like the DSM-III Axis V
rating, is coded from responses to the DIS by the clinician who conducts the interview.
This inventory permits judgments to be made both about the respondent's worst-ever
depressive symptomatology and about his current level of depression. Interrater
reliabilities of .78 for current and .80 for worst-ever depression have been obtained in our
research group (Reider, 1991).

The short form of the Beck Depression Inventory (BDI; Beck et al., 1961) is a 13-
item self-report instrument that emphasizes the psychological rather than the physiological
aspects of depression and is more sensitive to mild or moderate levels of depression than
to severe. Beck and his colleagues report that the short and long form of the BDI
correlate between .89 and .97 (Beck, Steer, & Garbin, 1988).

Severity of alcohol problems. This is measured by the Lifetime Alcohol Problems
Score (LAPS; Zucker, 1988). The LAPS is a composite score derived from a) the
squared inverse of the age at which the respondent reports first having drank enough to
get drunk; b) the number of areas of life in which drinking is reported to be a problem; and
c) the interval between the respondent's earliest and most recent alcohol problems,
corrected for current age.

Antisocial behavior. Parents' antisociality is measured using the Antisocial
Behavior Checklist (ASB; Zucker & Noll, 1980a), a 46-item questionnaire that assesses
nine content areas of antisocial behavior ranging from truancy and other forms of child
delinquency to theft, sexual deviance, and other types of adult antisocial behavior. The
ASB rates each behavior on a four-point scale ranging from "never" to "often." A series
of reliability and validity studies has produced a four-week test-retest reliability of .91 and
has shown that the instrument difi‘erentiates among groups with difi‘erent levels of known

deviance such as prisoners, minor offenders, and college students; further, the ASB

38

difi‘erentiates between alcoholic and nonalcoholic men in the present study (Zucker &
Fitzgerald, 1991b).
DATA ANALYSIS

The design of this analysis is cross-sectional, limited to data collected at wave one
on boys living with both biological parents except as otherwise noted (see page 32,
above).

Igapeadant va_ri_able: Intemarenpageement. Following Jacob et al. (1982) and
Christensen et al. (1992), several difi‘erent measures of interparent agreement will be
calculated. These are the following:

a) Percentage agreement on presence/fiance of child behavior problems, which
consists of the percentage of total observations on which the parents agree that a behavior
problem exists or does not exist (i.e., m rate the behavior as a either 1 or 2, or bo_th rate

the behavior as 0).

 

 

b) Efl‘ective percentage agreement (Jacob et al., 1982), which excludes agreement
on nonoccurrence of behaviors (i.e., agreement on 0's) fiom the calculation. This corrects
for the inflation of the percentage agreement index that occurs in less severe cases where a
high proportion of agreement on nonoccurrence would be expected simply by chance due
to the high base rate of "0" endorsements. Agreement as measured by this index is
essentially agreement on le_v_e_l of behavior problem (i.e., my parents rate the behavior as
a 1, or goth parents rate the behavior as a 2).

c) Cohen's (1960) kappa, used to assess the proportion of nonchance agreement
between parents (Christensen, et al., 1992).

d) Mother and Father Differential Concern Index (DCI); the equivalent of the
parent discrepancy index developed by Christensen, et al. (1992), this measure consists of
the number of disagreements on CBCL items in which a given parent endorses a behavior
and the other does not, pha the number of disagreements in which the parent in question
endorses the behavior more grow than does the other (i.e., as a 2 rather than a l).

39

Unlike kappa, the DCI can indicate the direction as well as the degree of disagreement,

 

and hence may more accurately reflect the "psychological reality" of the family
(Christensen, et al., 1992). The phrase "differential concern" implies a difference in the
direction of a greater level of endorsement relative to the other parent, and is therefore
preferable to "discrepancy" since the latter does not specify the direction of the difference.
Further, the term "discrepancy" implies an objective judgment about the rater's accuracy in
rating the child's "true" behavior, and it must be stressed that no such judgment is involved
in this index. The purpose of the DCI is to elucidate "the specific nature of the
disagreement between parents. ..[, i.e.,] whether disagreement between parents is random
(neither parent is more likely to endorse problem behavior than the other) or systematic
(one parent is more likely to endorse problem behavior than the other is)" (Christensen et
al., 1992, pg. 421). '

Each of these measures of agreement will be calculated for parents' Intemalizing,
Externalizing, and Total Behavior Problem scale scores. Note that due to differing
numbers of items in the CBCL Intemalizing (n=46) and Externalizing (n=40) subscales
and the fact that these differences are not reflected in the calculation of the DCI, each
respondent's Externalizing DCI score is weighted by a factor of 1.15 (i.e., 46/40) to
correct for this difference.

Predictor variablas. Data analysis will begin by examining the bivariate
relationship between each of the predictor variables (child's age, domain of child behavior,
severity of child disturbance, amount of time fathers spent with child, marital satisfaction,
parents' overall psychological adjustment, parents' depression, presence and severity of
parental alcohol problems, and parental antisocial behavior) and each of the indices of
interparent agreement.

Next, following Moses (1992), an exploratory factor analysis of the predictor
variables will be undertaken to determine the extent to which the predictors comprise

coherent factors.

40

Finally, factor-derived scale scores will be computed and used as variables in a
hierarchical regression analysis to predict interparent agreement. A model will be

developed and tested for each index of interparent agreement (the dependent variable).

CHAPTER III
RESULTS
DEMOGRAPHIC CHARACTERISTICS OF THE SAMPLE

Table 1 reports the demographic characteristics of the sample. Complete
demographic data were available for 260 of the 298 families. In this sample, mothers are
on average younger than fathers by about two years (t=-8.32, p<.001). There is no
significant difi‘erence between mothers and fathers in years of education, but mothers do
demonstrate a significantly higher estimated Full Scale IQ than do fathers (1:369,
p<.001). The average age of target children is 4.4 years.

Table 2 reports the relationship between the demographic variables and the indices
of interparent agreement on total behavior problems that serve as dependent variables in
this investigation. (For this and subsequent analyses, nearly all contrasts were virtually
identical when total behavior problems were broken down into internalizing and
externalizing subscales. On these grounds, unless otherwise noted only data pertaining to
total behavior problems are presented.) Parents' age and education are not significantly
related to interparent agreement. Higher 1Q among fathers is associated with fewer
mother discrepancies (r-=-. 15, p<.05), as is higher family income (_r=-.20, p<.01).
Likewise, both family income and family SES are positively associated with percentage
agreement on the presence/absence of child behavior problems (r's=.14 and .13,
respectively, both p's<.05). Age of the target child, contrary to what was hypothesized, is
negatively related to percentage agreement on the presence/absence of behavior problems
(r=—. 15, p<.05), but is not significantly related to any of the other indices of agreement.
Finally, family income is associated with higher overall interparent agreement as measured

by kappa (5.14, p<.05).
41

Table 1

42

Demographic Characteristics of the Sample (N=260 families)

 

 

 

 

Mothers Fathers t3 p
Parents' Age
Mean 30.5 32.4 -8.32 .000
(SD) (4.4) (5.2)
Parents' Education
(Years)
Mean 13.2 13.3 -1.05 .295
(SD) (1.9) (2.3)
Parents' IQb
Mean 99.4 95.8 3.69 .000
(SD) (15.4) (15.7)
Family SES (Duncan)
Mean 329
(SD) (150)
Family Income
Mean $31,714
(SD) ($16,822)
Number of Children in
the Home
Mean 2.3
(SD) (0.9)
Age of Target Child
Mean 4.4
(SD) (1.1)

 

a I-values for difl‘erences between mothers' and fathers' mean age, edueation, and IQ.
b Estimated IQ fi'om WAIS-R information and Digit Symbol Subtests.

43
Table 2

Relationship (Pearson r's) Between Demogmm Variables and Indices of Integgent Agreement on
CBCL Total Behavior Problems (N =260 families)

 

 

Mother Father Percentage Effective
Difi‘erential Differential Agreement Percentage Kappae
Concern lndexa Concern Indexb (Pres/Abs)c Agreementd

 

Mothers' Age -.09 -.01 .07 -.09 -.02
Fathers' Age -. 12 .03 .07 -.02 .06
Mothers' Ed. -.05 -.02 .04 .04 .04
Fathers' Ed. -. 10 -.03 .08 -.02 .00
Mothers' 1Qf -.01 -.01 -.04 .06 .05
Fathers' IO -.15* .00 .10 .03 .07
Family SES -.09 -.08 .13“ .02 .05
(Duncan)

Family Income -.20** .00 .14* .09 .14*

No. of Children

 

.00 -.02 .02 -.04 -.02
Age of Target .11 .04 -.15* -.07 -.09
Child
3 Number of items which mother endorses and father does not, plus number of items she endorses more
strongly than does father.
9 Number of items which father endorses and mother does not, plus number of items he endorses more
strongly than does mother.

c Percentage agreement on the presence/absence of child behavior problems.
9 Effective percentage agreement on child behavior problems.

e Cohen's Kappa.

f Estimated IQ from WAIS-R Information and Digit Symbol Subtests.

*p < .05 ”p < .01

44

INTERPARENT AGREEMENT

Table 3 provides descriptive data on parents' CBCL scores. Means and standard
deviations for mothers' and fathers' CBCL Behavior Problems scores for internalizing,
externalizing, and total behavior problems are presented. Although mothers do, on the
average, report more internalizing, externalizing, and total behavior problems than fathers,
the differences are slight and not significant. However, as noted above (pg. 41) and
reported in Reider (1991), the Pearson correlations between parents' scores, while
significant at the .001 level, are of much lower order than those reported for the CBCL
standardization sample (Achenbach & Edelbrock, 1983).

Table 4 reports the means and standard deviations for mothers and fathers on the
Differential Concern Index (DCI). It can be seen that, within all three categories of
behavior problems (internalizing, externalizing, and total), mothers evince greater
differential concern over child behavior problems than do fathers. These differences are
small, however, approaching significance only for total behavior problems (_t=1.66, p<.10).
Parents' DCI scores are negatively correlated for all three behavior problem scales (all p's
< .001), indicating that high difl‘erential concern in one parent is related to lower
difl‘erential concern in the other parent. This result may at first seem predictable, but it is
not mathematically necessary that mothers' and fathers' DCI be inversely related. For
instance, a mother and father can both demonstrate high difi‘erential concern (i.e., rating
behaviors as more problematic than the other parent does), but for different child
behaviors; thus, both parents would show high DCI scores, and their scores would be

 

pasifiyaly correlated. The fact that the scores are inversely related indicates that, in many
families, high difi‘erential concern is found in only one or the other parent; this finding in

itself is a marker of disagreement.

45
Table 3

Integra_re_nt Ament on the Deacriptioa of Chil_d Behavior Problems: Parents' Total, Intemalizing, and
Externalizing Behavior Problem Scojras on the Chilg Behavior Caecklist (CBCL) m=298 families)

 

 

CBCL Behavior Problems Scores (sums):

Mothers Fathers ta [b
Total Behavior
Problems
Mean 31.2 29.3 1.59 27*"
(SD) (17.7) (17.0)
Intemalizing Behavior
Problems
Mean 10.6 10.1 1.12 23*"
(SD) (7.2) (6.9)
Externalizing Behavior
Problems
Mean 14.5 13.9 1.13 38*"
(SD) (8.6) (8.4)

 

a I-values for difference in means between parents' Behavior Problem Scores.
b Pearson {s for correlation between parents' Behavior Problem Scores.

seep < .001

46
Table 4

lnterwent Agraament on the Descn'ptipn of Child Behavior Problems: Parents' Differenfll Concern
Mex (DCI)a Scorea on th_e CBCL (N=298 farmilies)

 

Differential Concern Index

Mothers Fathers 1" g9
Total Behavior
Problems
Mean 14.5 12.8 1.66 -.45***
(SD) (10.5) (10.8)
Intemalizmg Behavior
Problems
Mean 5.7 5.2 1.13 -.34"‘**
(SD) (4.5) (4.8)
Externalizing Behavior
Problems
Mean 6.0 5.4 1.26 -.53"'**
(SD) (4.8) (4.7)

 

a Number of items which parent A endorses and parent B does not, plus number of items which parent A
endorses more strongly than parent B.

b I-values for difference in means between parents' Differential Concern Index Scores.

C Pearson ['5 for correlation between parents' Differential Concern Index Scores.

**.2 < .001

47

Table 5 provides descriptive data on the three remaining indices of agreement.
First, percentage agreement on presence/absence of child behavior problems averages
nearly 80% for total behavior problems, 78% for internalizing problems, and 76% for
externalizing problems. However, efl'ective percentage agreement (which eliminates
agreement on zeroes) is much lower, averaging 29% (total), 24% (internalizing), and
33% (externalizing). Thus, parents are relatively likely to agree on whether or not their
child demonstrates a particular problem at a gross level, but correspondence decreases
when agreement on non-occurrences is eliminated and parents are compared only on their
agreement about relative level of problems, i.e., whether the problem is "somewhat true"
or "very true" of their child. Finally, kappa, which assesses parents' nonchance agreement
on both presence and level of behavior problems, averages .37 (total), .29 (internalizing),
and .36 (externalizing), indicating a low-to-moderate level of agreement between parents
in the sample.
BIVARIATE RELATIONSHIPS BETWEEN PREDICTOR VARIABLES AND
INTERPARENT AOREEMENT INDICES

Ag; of tha target child. As noted above in Table 2, age of the target child,
contrary to what was hypothesized, is negatively related to percentage agreement on the
presence/absence of behavior problems (_r=-.15, p<.05), but the relationship is of very low
order. Age is not significantly related to any of the other indices of agreement.

Type of child behavior problems. It was hypothesized that agreement would be
greater for externalizing than for internalizing behavior problems. As Table 6 indicates,

support for this hypothesis varies depending upon which index of agreement is considered.

48
Table 5

Integflent Agleement on the Descrintlo_r_r of Child Behavior Problems: Percentage Agreement on
Presence/Absence, Effective Percentage Agreement, and Kappa on CBCL Behavior Problems CN=298

families)

 

 

 

Percentage Effective
Agreement Percentage
(Pres/Abs)a Agreementb Kappac
Total Behavior
Problems
Mean 79.8 28.9 .37
(SD) (7.9) (10.8) (.12)
Intemalizing Behavior
Problems
Mean 78.3 24.0 .29
(SD) (10.6) (13.7) (. 18)
Externalizing Behavior
Problems
Mean 75.8 32.8 .36
(SD) (9.4) (15.3) (.17)

 

a Percentage agreement on the presence/absence of child behavior problems.
9 Efl‘ective percentage agreement on child behavior problems.

C Cohen's Kappa.

49
Table 6

Relationship Betweea Time of Child Behavior Problem (Intemalizing versus Externalizing) and
Integglent Agleement on the CBCL (N=298 families)

 

 

Intemalizing Externalizing
Index of Behavior Behavior
Interparent Agreement Problems Problems :3
Mother
Differential Concern indexb
Mean 5.7 6.0 -1.51
(SD) (4.5) (4.8)
Father
Differential Concern Indexc
Mean 5.2 5.4 -1.17
(SD) (4.8) (4.7)
Percentage Agreement
(Pres/Abs)d
Mean 78.3 75.8 4.27"”""l
(SD) (10.6) (9.4)
Efi'ective Percentage
Agreemente
Mean 24.0 32.8 -8.38***
(SD) (13.7) (15.3)
Kappal
Mean .29 .36 -5.l7"”"*
(SD) (.18) (.17)

 

a I—values for difierence in means between parents' Intemalizing and Externalizing Behavior Problem
Scores.

b Number of items which mother endorses and father does not, plus number of items she endorses more
strongly than does father.

c Number of items which father endorses and mother does not, plus number of items he endorses more
strongly than does mother.

‘1 Percentage of agreement on the presence/absence of child behavior problems.

e Effective percentage agreement on child behavior problems.

f Cohen's Kappa.

"*2 < .001

50

The table shows that neither mothers nor fathers evinced a greater level of differential
concern regarding externalizing behaviors than they did regarding internalizing behaviors.
However, contrary to hypothesis, a higher percentage of agreement on presence/absence
of behavior problems was found for internalizing problems than for externalizing problems
(_t_=4.27, p<.001), although the difference is of very low order: This difi‘erence in the
direction of better agreement on internalizing problems may be due to the fact that there is
a lower base rate for endorsements of internalizing items by both mothers and fathers (see
Table 3), which increases the chances of agreement on nonoccurrence and therefore
inflates the index. On the other hand, the hypothesis that agreement would be better on
externalizing problems was supported when effective percentage agreement was
considered, with agreement being greater for externalizing problems than for internalizing
(t=-8.38, p<.001). Likewise, when kappa was used as the measure of agreement, there
was better correspondence between parents on externalizing problems than on
internalizing (1=-5.17, p<.001). In summary, these results indicate that while parents are
slightly more likely to agree on the presence/absence of internalizing behavior problems,
they are significantly more likely to agree on the l_eye_l or se_ve_rig of externalizing
problems, and also evince greater overall correspondence on externalizing behaviors as
measured by kappa.

Child distugbance. It was hypothesized that agreement would be lower between
parents of reportedly more disturbed children (as defined by the sum of both parents'
CBCL Total Behavior Problems scores) than between parents reporting fewer problems.
As was the case with type of behavior problem, the relationship between severity of
reported child disturbance and interparent agreement (Table 7) varies depending upon the

index of agreement.

5 1
Table 7

Relationship (Pearson r's) Betweerievel of Reported Child Disturbance and lntegrar_ent Aggeement on
CBCL Total behavior Proble_ms (N= 298 families)

 

 

Index of

Interparent Agreement Level of Reported Child Disturbancea
Mother

Differential .44***

Concern Indexb

Father
Differential 44*“
Concern Indexc

Percentage
Agreement -.72***

(Pres/Abs)d

Effective
Percentage 23*”
Agreement"

Kappaf .02

 

3 Sum of mother's and father's CBCL Total Behavior Problem Scores.

9 Number of items which mother endorses and father does not, plus number of items she endorses more
strongly than does father.

C Number of items which father endorses and mother does not, plus number of items he endorses more
Strongly than does mother.

‘1 Percentage of agreement on the presence/absence of child behavior problems.

e Effective percentage agreement on child behavior problems.

f Cohen's Kappa.

seep < .001

 

52

As predicted, both mother DCI and father DCI were positively related to greater reported
child behavior problems (both ['s=.44, p<.001). Thus, the greater the overall reported
child disturbance, the more likely were parents to evince differential concern over behavior
problems relative to their spouses. Also as predicted, percentage agreement on
presence/absence showed a strong negative relationship with level of reported behavior
problems (yr-.72, p<.001). Thus, the greater the level of overall child behavior problems
reported, the less likely were parents to agree on the presence/absence of specific
behaviors. This finding, however, may be artifactual; a higher base rate of agreement on
nonoccurrences (zeroes) would be expected in cases of lower child disturbance, since
there would be fewer non-zero endorsements reported by the parents. Thus, for statistical
reasons alone, lower disturbance could be expected to be associated with higher
percentage agreement. In fact, once agreement on nonoccurrences was controlled for by
effective percentage agreement, it was found that greater reported child disturbance was
actually associated with somewhat graatg agreement ([=.23, p<.001), which is contrary to
the hypothesis. Thus, while the parents of reportedly more disturbed children appeared
less likely to agree on the prasence/absence of specific behaviors (a finding that may be
artifactual), they were more likely to agree on the se_ve_rm of those behaviors that at least
one parent endorsed as being present in their child once agreement on nonoccurrence was
controlled for. Finally, there was no significant relationship between kappa and level of
reported child disturbance.

53

Amount of time fathers spend with ta_rget child. It was hypothesized that the more
time fathers spend with the target child, the greater would be the interparent agreement on
child behavior problems. As shown in Table 8, this hypothesis was not supported. For
both total time spent and time spent in activities, no significant relationship was found with
any of the indices of interparent agreement. Similar findings were reported by Fitzgerald,
Zucker, and Maguin (in press), who found no significant differences in interparent
correlations on CBCL Aggression as a function of whether fathers spend low, medium, or

high amounts of time with the target child.

54
Table 8

Relationship (Pearson r'slBetween the Amoun_t of Time Spant with Target Child by Fathers and
Intemnt Aggment on CBCL Total Behavior Problems (N=213 families)

 

 

 

Total Time Activity Time
Index of Father Spent with Childa Father Spent in with Childb
Interparent Agreement
Mother
Differential -.02 -.09
Concern Indexc
Father
Differential .10 .06
Concern Indexd
Percentage
Agreement -.07 .05
(Pres/Abs)°
Effective
Percentage . -.03 .07
Agreement1
Kappag -.03 .09

 

a Fathers' report of the ml amount of time spent with the target child during a 24 hour period (averaged
over three days), from the Parent Daily Report.

b Fathers' report of the amount of time spent doiag somethipg with the target child during a 24 hour
period (averaged over three days), from the Parent Daily Report.

c Number of items which mother endorses and father does not, plus number of items she endorses more
strongly than does father.

d Number of items which father endorses and mother does not, plus number of items be endorses more
strongly than does mother.

e Percentage of agreement on the presence/absence of child behavior problems.

f Efl‘ective percentage agreement on child behavior problems.

8 Cohen's Kappa.

55

Marital satisfaction. It was hypothesized that greater marital satisfaction would be
related to greater interparent agreement on child behavior problems. Table 9 indicates
that, for mothers, greater marital satisfaction meant less differential concern over behavior
problems relative to fathers (5.26, p<.001), a higher percentage agreement on the
presence/absence of behavior problems (_r=.22, p<.001), and higher overall agreement as
measured by kappa (_r=.14, p<.05). Mothers' marital satisfaction was unrelated to father
DCI or to effective percentage agreement.

For fathers, marital satisfaction was significantly related only to percentage of
agreement on presence/absence of behavior problems (l_'=.19, p<.01) and kappa (r=. 16,
p<.05). Fathers' marital satisfaction did not predict fathers' (or mothers') DCI, nor was it
related to effective percentage agreement.

Parents' overall psychological fungioning. It was predicted that parents' overall
psychological functioning would be positively related to interparent agreement on child
behavior problems. Table 10 presents the relationship between the first measure of overall
functioning, the DSM-III Axis V Rating, and interparent agreement. For mothers, a
higher Axis V rating (indicating better overall adaptive functioning) was associated with
less difl'erential concern over child behavior problems relative to fathers (1:24, p<.001).
F athers' higher adaptive functioning likewise was related to lower mother DCI. However,
neither parents' Axis V rating was significantly related to father DCI. F athers' (but not
mothers') Axis V rating was positively related to percentage of agreement on
presence/absence of child behavior problems (_r=. 13, p<.05). Neither parents' Axis V

rating was related to effective percentage agreement or to kappa.

56
Table 9

Relationship (Pearson r's) Between Parents' Marital Satisfaction and lnterpgent Amment on CBCL
Total Behayipr Problems (N=240 families)

 

 

 

Index of Mothers' Fathers'
Interparent Agreement Marital Satisfactiona Marital Satisfaction
Mother

Differential -.26*** -.1 1
Concern Indexb

Father

Differential .00 -. 10
Concern IndexC

Percentage

Agreement .22" .19“
(Pres/Abs)d

Effective

Percentage .09 .08
AgreementC

Kappaf .14* .16*

 

3 Marital Satisfaction Index derived from Benjamin Intrex Questionnaire.

9 Number of items which mother endorses and father does not, plus number of items she endorses more
strongly than does father.

C Number of items which father endorses and mother does not, plus number of items he endorses more
strongly than does mother.

C Percentage of agreement on the presence/absence of child behavior problems.

C Effective percentage agreement on child behavior problems.

f Cohen's Kappa.

*p < .05 “p < .01 “*p < .001

5 7
Table 10

Relationship (Pearson r's) Between Parents' DSM-III Axis V Rating and Intement Agreement on CBCL
Total Behavior Problems (5:298

 

 

 

 

families)

Index of Mothers' Fathers'
Interparent Agreement Axis V Ratinga Axis V Rating
Mother

Differential -.24*** -.16**
Concern Indexb

Father

Diflemnfial .10 -.03

Concern IndexC

Percentage
Agreement .06 .13*

(Pres/Abs)d

Effective
Percentage -.03 .04
AgreementC

Kappaf .01 . 10

 

a Global assessment of functioning taken from DSM-III Axis V Rating.

C Number of items which mother endorses and father does not, plus number of items she endorses more
strongly than does father.

C Number of items which father endorses and mother does not, plus number of items be endorses more
strongly than does mother.

‘1 Percentage of agreement on the presence/absence of child behavior problems.

C Effective percentage agreement on child behavior problems.

f Cohen's Kappa.

‘p< .05 **p< .01 "*p< .001

58

Table 11 presents the relationship between interparent agreement and parents'
Composite Psychological Health Q—Sort scores, the second measure of overall
psychological functioning. Since the Psychological Health Q-Sort score is a correlation
between the subject's sort and a consensually defined "ideal" or "mentally healthy" sort,
higher scores are indicative of greater overall psychological functioning. As shown in the
table, both mothers' and fathers' Q-Sort scores are related to lower mother DCI (g's=-.15
and -.13, respectively, both p's<.05). However, parents' Q-Sort scores were not related to
any other indices of interparent agreement. Thus, the results shown in Tables 10 and 11
indicate that, in general, parents' overall mental health is related to lower mother DCI, but
not to the other indices of interparent agreement.

Parental depression. Parental depression was hypothesized to be negatively related
to interparent agreement on child behavior problems. Table 12 summarizes the results for
mothers. Greater self-reported maternal depression as measured by the Beck Depression
Inventory (BDI) was associated with greater mother differential concern over behavior
problems relative to fathers (5.27, p<.001) and a lower percentage agreement on
presence/absence of child behavior problems (_r=-.20, p<.01), but was unrelated to the
other agreement indices. Clinician-rated depression, as measured by mothers' current
Hamilton depression score, also was associated with greater mother DCI (5.21, p<.001).
Interestingly, mothers' current Hamilton score was negatively related to fathers' DCI,
albeit weakly (r-=—. 12, p<.05). Thus, while mothers' current depression as measured by the
Hamilton was related to mothers' greater concern over child behavior problems relative to
fathers, it was associated with a m tendency for fathers to report greater concern

relative to mothers.

59
Table 11

Relationship (Pearson r's) Between Parents' Mental Health O-Sort Score and Integmnt Amment on
CBCL Total Behavior Problems (5:298 families)

 

 

Index of Mothers' Fathers'
Interparent Agreement Mental Health Q-Sort a Mental Health Q-Sort
Mother

Differential -. 15* -. 13*

Concern Indexb

Father
Differential .03 -.04
Concern IndexC

Percentage
Agreement .06 .12

(Pres/Abs)d

Efi‘ective
Percentage -.01 .03
Agreemth

Kappaf .04 .08

 

a Composite Psychological Health Q—Sort.

b Number of items which mother endorses and father does not, plus number of items she endorses more
strongly than does father.

C Number of items which father endorses and mother does not, plus number of items be endorses more
strongly than does mother.

9 Percentage of agreement on the presence/absence of child behavior problems.

C Effective percentage agreement on child behavior problems.

f Cohen's Kappa.

‘p< .os

60
Table 12

Relationship (Pearson r's) Between Mothers' Beck and Hamilton Depression Scores__ and Integgent
Ageament an CBCL Total Behavior Problems (N=298 families)

 

 

 

Mothers' Mothers'
index of Mothers' HRSDC Score HRSD Score
Interparent Agreement BDIa Score (Current) (Worst Ever)
Mother
Difierenfial .27*** .21*** .17”
Concern IndexC
Father
Differential -.02 -. 12* -.08
Concern IndexC
Percentage
Agreement -.20** -.08 -.05
(Pres/Abs)C
Effective
Percentage . .03 -.02 .01
Agreement1

We -.07 -.06 -.01

 

a Beck Depression Inventory.

b Hamilton Rating Scale for Depression.

C Number of items which mother endorses and father does not, plus number of items she endorses more
strongly than does father.

C Number of items which father endorses and mother does not, plus number of items he endorses more
strongly than does mother.

C Percentage of agreement on the presence/absence of child behavior problems.

f Efi‘ective percentage agreement on child behavior problems.

8 Cohen's Kappa

‘p < .05 “p < .01 “*p < .001

61

Finally, mothers' worst ever Hamilton score was positively related to mother DCI (F. 17,
p<.01). None of the measures of mothers' depression were related to effective percentage
of agreement or to kappa.

Table 13 presents the results for fathers' depression. As hypothesized, fathers' BDI
score was associated positively with both father (gr-.20, p<.001) and mother DCI (_r=.14,
p<.05), and negatively with percentage of agreement on presence/absence of behavior
problems (_r=-.27, p<.001) and kappa (_r=-.12, p<.05), but was unrelated to effective
percentage agreement. In contrast, fathers' current Hamilton score was not related to any
of the agreement indices. F athers' worst ever Hamilton score was modestly associated
with greater father DCI (5.14, p<.05) and lower percentage agreement on
presence/absence of behavior problems (_r=-. 14, p<.05). Overall, the results summarized
in Tables 12 and 13 show that, as hypothesized, parental depression is related to
interparent agreement as measured by mother and father DCI and percentage of
agreement on presence/absence of behavior problems, but in general (with the exception

of fathers' BDI score) is not associated with efi‘ective percentage of agreement or kappa.

62
Table 13

Relationship (Pearsoar’s) Between Fathers' Beck and Hamilton Depresaian Scores and Intel’palent
Angment on CBCL Total Behavior Prob_le_ms (N=298 families)

 

 

Fathers' Fathers'
Index of Fathers' I-IRSDb Score HRSD Score
Interparent Agreement BDIa Score (Current) (Worst Ever)
Mother
Differential . 14* .04 .05
Concern IndexC
Father
Differential .20*** .06 .14*
Concern IndexC
Percentage
Agreement 4 -.27*** -.07 -.14*
(Pres/Abs)C
Effective
Percentage _ -.03 .00 .00
Agreementt
Kappag -. 12* -.03 -.02

 

3 Beck Depression Inventory.

b Hamilton Rating Scale for Depression.

C Number of items which mother endorses and father does not, plus number of items she endorses more
strongly than does father.

C Number of items which father endorses and mother does not, plus number of items he endorses more
strongly than does mother.

C Percentage of agreement on the presence/absence of child behavior problems.

f Effective percentage agreement on child behavior problems.

g Cohen's Kappa.

*p < .05 "*p < .001

63

Parent alpohol problems. Parent alcohol problems are measured in two ways in
this investigation. First, fathers' alcoholism type reflects the basis for the family's inclusion
in the study and is measured as a discrete variable with three categories that can be
conceptualized as lying along a continuum of pathology: community controls (no
evidence of father alcohol abuse/dependence), "covert" alcoholics (fathers met criteria for
alcohol abuse/dependence but did not come to the attention of the study through the
courts), and "overt" alcoholics (court-recruited; alcohol abuse/dependence has resulted in
the significant social/legal problem of DUI conviction). It was hypothesized that group
membership farther along this alcoholism continuum would be associated with lower
interparent agreement on child behavior problems. Table 14 presents ANOVA results for
each of the indices of interparent agreement. Significant difi‘erences in the hypothesized
direction were found between the control and overt alcoholic groups on percentage
agreement on presence/absence of behavior problems (P=5.55, p<.01) and kappa (P=4.45,
p=.01). No group differences were found on mother or father DCI or on effective
percentage agreement. There were no difi‘erences on any of the agreement indices
between the covert alcoholic group and the control or overt alcoholic groups.

The second measure of parent alcohol problems is the Lifetime Alcohol Problems
score (LAPS), a continuous measure of the severity of mothers' and fathers' lifetime
alcohol problems. It was hypothesized that greater severity of alcohol problems would be
associated with lower interparent agreement. Table 15 indicates that, consistent with the
hypothesis, mothers' LAPS is associated with greater mother DCI (gr-.18, p<.01) and a
lower percentage agreement on presence/absence of child behavior problems (_r=-.16,
p<.01). Likewise, fathers' LAPS is associated with greater father DCI (ls-.18, p<.01) and
a lower percentage agreement on presence/absence of child behavior problems (_r=-.15,
p<.01). Neither mothers' nor fathers' LAPS was related to effective percentage agreement

or to kappa.

Table 14

Relationship Between Fathers' A1coholis_m Type and Integralent Agreement on CBCL Total Behavior
Problems

 

 

 

Overt Covert
Alcoholics Alcoholics Controls
(N=166) (N=42) (N=90) E 2
Mother Differential
Concern IndexC
Mean 15.6 12.5 13.4 2.29 .10
(SD) (11.7) (9.5) (7.9)
Father Differential
Concern Indexb
Mean 13.8 13.7 10.5 2.89 .06
(SD) (11.9) (9.9) (8.3)
Percentage Agreement
(Pres/Abs)C
Mean 78.51 80.3 81.91 5.55 .00
(SD) (9.1) (6.7) (5.6)
Efi‘ective Percentage
AgreementC
Mean 27.6 30.2 31.6 2.79 .06
(SD) (11.4) (9.1) (10.3)
Karmac
Mean .351 .39 .391 4.45 .01
(SD) (.13) (.11) (.11)
a Number of items which mother endorses and father does not, plus number of items she endorses more
strongly than does father.
b Number of items which father endorses and mother does not, plus number of items he endorses more
strongly than does mother.

C Percentage of agreement on the presence/absence of child behavior problems.
C Effective percentage agreement on child behavior problems.
C Cohen's Kappa.

T Subgroups sharing the same superscript symbol difler significantly fiom each other at p < .05.

65
Table 15

Relationship (Pearson r's) Between Parents' Lifetime Alcohol Problems Score (LAPS) and lntewrcnt
Agreement on CBCL Total Behavior Problems m=298 families)

 

 

Index of Mothers' Fathers'
Interparent Agreement LAPSa LAPS
Mother

Difl‘erential . 18" .02

Concern Indexb

Father
Differential .03 . 18* *
Concern Indexc

Percentage
Agreement -. 16“ -.15**

(Pres/Abs)d

Efi‘ective
Percentage .05 .05
Agreement6

Kappaf .00 -.04

 

a Lifetime Alcohol Problems Score.

b Number of items which mother endorses and father does not, plus number of items she endorses more
strongly than does father.

c Number of items which father endorses and mother does not, plus number of items he endorses more
strongly than does mother.

d Percentage of agreement on the presence/absence of child behavior problems.

e Effective percentage agreement on child behavior problems.

f Cohen's Kappa.

“p < .01

66

fluent antisocial behavior. It was hypothesized that parental antisociality would be
linked to lower interparent agreement on child behavior problems. As Table 16 shows,
mothers' antisociality is associated with higher mother DCI (_r_=.27, p<.001) and lower
percentage agreement on presence/absence of behavior problems (r_'=-. l6, p<.01).
Similarly, fathers' antisociality is linked to higher father DCI (1:21, p<.001) and lower
percentage agreement on presence/absence of behavior problems (,r_=-.22, p<.001).
Neither parents' ASB scores were related to effective percentage agreement or to kappa.

Summm of bivariate tests. The results of the bivariate analyses were at least
partially consistent with all of the research hypotheses except that concerning the amount
of time fathers spend with the target child. In each case, however, the significance and
direction of the relationship depended on the measure of interparent agreement that was
employed, and in some cases a result emerged that was contrary to what was
hypothesized. In general, the magnitude of significant efi‘ects was low to moderate, with
the exception of the child disturbance variable which showed a moderate to strong
relationship with the indices of interparent agreement.
Mother Difi‘erenjfil Concern. The bivariate analyses indicate that, as hypothesized,
mothers who report greater concern over problem behaviors relative to their spouses tend
to be more dissatisfied with their marriages, to be more depressed, to have greater lifetime
alcohol problems, to evince more antisociality, and to be less psychologically well-adjusted
overall than mothers with lower DCI scores. Higher mother DCI is also associated with
greater depression and poorer overall psychological adjustment among fathers, and with
greater overall reported child disturbance. Contrary to hypothesis, mothers report more
difl‘erential concern over externalizing behavior problems than over internalizing problems.
Child's age, the amount of time father spends with the child, and father's alcoholism type

were not significantly related to mother DCI.

67
Table 16

Relationship (Pearson r's) Between Parents' Antisocial Behavior (ASB) Checklist Score and Interparent
Agreement on CBCL Total Behavior Problems (N=298 faLmilies)

 

 

 

Index of Mothers' F athers'
Interparent Agreement ASBa ASB
Mother

Difl‘erential .27*** .07

Concern Indexb

Father
Differential -.02 .21 ***
Concern IndexC

Percentage
Agreement -. 16" -.22***
(Pres/Abs)d

Effective
Percentage .01 -.02
Agreemente

Kappa" -.04 -.06

 

a Antisocial Behavior Checklist Score.

b Number of items which mother endorses and father does not, plus number of items she endorses more
strongly than does father.

C Number of items which father endorses and mother does not, plus number of items be endorses more
strongly than does mother.

‘1 Percentage of agreement on the presence/absence of child behavior problems.

e Effective percentage agreement on child behavior problems.

f Cohen's Kappa.

”p < .01 "*9 < .001

68

Father Differential Concern. As hypothesized, fathers who report greater concern over
problem behaviors relative to their spouses tend to be more depressed, to have
experienced greater lifetime alcohol problems, and to evince more antisociality than
fathers with lower DCI scores. Higher father DCI is also associated with greater overall
reported child disturbance. Contrary to hypothesis, but similarly to mothers, fathers report
greater difl‘erential concern over externalizing behavior problems than over internalizing

problems. Also unexpected was the finding that a father DCI is lower if the mother is

 

 

depressed. Child's age, the amount of time father spends with the child, marital
satisfaction, parents' overall psychological functioning, and father's alcoholism type were
not significantly related to father DCI.

Pmentage of agreement on presence/absence of behavior problems. As hypothesized,

 

parents in this sample are more likely to agree on whether or not their child displays a
particular behavior problem if the child is less disturbed, if parents are satisfied with their
marriage, if fathers are psychologically well-adjusted in general, if parents are not
depressed, if the family is in the control group as opposed to the overt alcoholic group, if
parents have had fewer alcohol problems, and if parents evince less antisociality. Contrary
to hypothesis, parents in this sample are also more likely to agree on whether or not their
child displays a particular behavior problem if the problem is of an internalizing nature, and
if the child is younger. Amount of time father spends with the target child was not
significantly related to agreement on the presence/absence of behavior problems.

Efl‘mive percengge of agreement. As hypothesized, parents in this sample are more
likely to evince a higher efi‘ective percentage of agreement (i.e., whether the problem is
present and whether it is "somewhat true" or "very true" of their child) when the problem
is of an externalizing nature. Contrary to hypothesis, greater child disturbance was also
associated with higher effective percentage of agreement. None of the other predictors

was significantly related to efl‘ective percentage of agreement.

69

Kappa. As hypothesized, the interparent kappa score was higher for externalizing
behavior problems, and was higher if parents are satisfied with their marriage, if fathers are
not depressed, and if the family is in the control group as opposed to the overt alcoholic
group. None of the other predictors was significantly related to kappa.

EXPLORATORY FACTOR ANALYSIS .

Moses (1992), analyzing a subsample (n=l72) of the present data set, found that
parent alcohol problems, antisociality, and depression clustered together into a coherent
risk factor. Following Moses, an exploratory factor analysis was conducted to determine
whether the various measures of parent psychopathology employed in this analysis would
comprise coherent factors. Standard unweighted least squares extraction was used to find
the factor solution which would best fit the data, and oblique rotation was used because it
does not impose the condition of orthoganality upon the factor solution. A two-factor
solution yielded the most clearly interpretable and coherent pattern matrix. The first
factor ("mother psychopathology") is comprised of mothers' depression, antisociality,
alcohol problems, and (inverted) overall psychological functioning measures. The second
factor ("father psychopathology") is comprised of fathers' depression, antisociality, alcohol
problems, and (inverted) overall psychological firnctioning measures. Table 17 reports the
eigenvalues and percentage of variance accounted for by the two factors; together the
factors account for 45% of the total variance within the factor space. Table 18
summarizes the pattern matrix, including each variable's loading on both factors, after

oblique rotation.

70
Table 17

Egploratog Factor Analysis of Mother & Father Psvchopathology Variables (Oblimin Rotation) (N=245
families)

 

 

 

% Variance Cumulative %
Eigenvalue Acct. For Variance
Factor 1 (Mother Psychopathology) 5.11 - 36.5 36.5

Factor 2 (Father Psychopathology) 1.19 8.5 45.0

 

7 1
Table 18

Factor Analysis of Mother & Father Psychopathology Variables: Factor Loadings for Two-Fmr Solution
(Oblimin RotatioanN=245 families)

 

 

FACTOR 1 FACTOR 2
Variable Mother Psychopathology Father Psychopathology
M-HAMILTON CURRENTa .35 -. 14
M-HAMILTON WORSTal .79 -.06
M-DSM-III AXIS vb .77 .10
M-MENTAL HLTI-l Q.SORIc .75 -.01
M-ASB CHECKLISTd .40 .20
M-BECK DEPRESSIONe .38 .09
M-LAPSf . 27 .17
F-LAPSf -. 14 .so
F-ASB CHECKLIST“ -. 11 .70
F-DSM-III AXIS vb .27 .55
F-HAMILTON WORSTa .08 .63
F-HAMILTON CURRENTa .20 .51
F-MENTAL HLTH Q-SORTC .31 .49
F-BECK DEPRESSIONC .08 .43

 

Note: Item loadings used to define the factors are shown in bold type.

a Mothers' Hamilton Rating Scale for Depression (Current & Worst Ever).
b Mothers' and Fathers' DSM-III Axis v (Global Functioning).

c Mothers' and Fathers' Mental Health Q-Sort.

d Mothers' and Fathers' Antisocial Behavior Checklist.

e Mothers' and Fathers' Beck Depression Inventory.

f Mothers' and Fathers' Lifetime Alcohol problems.

72

Scales corresponding to the two parent psychopathology factors were constructed
by converting parents' scores on the separate depression, antisociality, alcohol problems,
and (inverted) overall psychological functioning measures to standard scores and summing
them. Reliability analysis yielded scale coeflicient alphas of .75 for mother
psychopathology and .72 for father psychopathology. The correlation between the scales
is .53 (p<.001). Table 19 presents correlations between the factor scale scores and the
indices of interparent agreement. Mother psychopathology shows a moderate positive
relationship to mother discrepancies (l_'=.32, p<.001) and a low-order but significant
negative correlation with percentage agreement on presence/absence of child behavior
problems (r=-.15, p<.05). Similarly, father psychopathology shows a low-order but
significant positive relationship to father discrepancies (r=. 16, p<.05) and a low-order but
significant negative correlation with percentage agreement on presence/absence of child
behavior problems (_r=-.20, p<.01).

HIERARCHICAL REGRESSION ANALYSIS

A hierarchical regression analysis was performed separately for each of the indices
of interparent agreement. The models contained the composite parent psychopathology
variables derived from the above factor analysis, a composite marital satisfaction variable
composed of the mean of both parents' (within-family) marital satisfaction scores, and the
child variables (age and disturbance level). Fathers' time spent with the target child was
not included in the models because it was shown to be unrelated to the outcome variables.
The analyses presented are based on interparent agreement on total behavior problems
only, as separate analyses conducted on the internalizing and externalizing subscales
yielded virtually identical results to the analyses based on total score.

Table 19

 

73

Correlations Between Faggor Scale Scoreaand Olpcome Va_ria_bles CN=Z45 families)

 

 

 

Mother Father Percent Effective
Difl‘erential Differential Agreement Percent Kappa
Concern Index Concern Index (Pres/Abs) Agreement
Mother
Psychopathology 32"" -.09 -.15* .03 -.04
Father
Psychopathology . 12 . 16* -.20** -.02 -.08
*p < .05 "p < .01 “"‘p < .001

74

Mother Difi‘ertLtial Concern. A four-step regression was performed to test the
efi‘ects of the five predictor variables (mother psychopathology, father psychopathology,
marital satisfaction, child's age, and child's disturbance level) on mother DCI. Mother
psychopathology was entered first, on the assumption that mothers' characteristics are
most proximal to mothers' ratings. Marital satisfaction was entered second since it is a
dyadic measure involving both mothers and fathers and is therefore less proximal to
mothers' ratings than mothers' characteristics, but more proximal than fathers'
characteristics. Father psychopathology was entered third, on the assumption that fathers'
characteristics are more distal to mothers' ratings than mothers' or dyadic characteristics.
Finally, the two child variables were entered as a block in the last step in order to gauge
what the effects of child characteristics would be after parent characteristics were already
accounted for in the model. The results of this analysis are presented in Table 20. As
shown, mother psychopathology was a significant predictor of mother DCI, accounting
for over 10% of the variance (F/change=24.76, p<.001). Once the efl‘ect of mother
psychopathology was controlled for in the model only the child variables produced a
significant improvement in R2 (F/change=23.9, p<.001); most of this effect can be
attributed to the child disturbance variable (beta=.43). Neither marital satisfaction nor
father psychopathology contributed significantly to the model. Adjusted R2 for the entire
model was .28 (F[5,198]=16.74, p<.001).

Father Differential Concern. A four-step regression was performed to test the

 

effects of the five predictor variables on father DCI. Following the same logic as the
analysis for mother DCI, father psychopathology was entered first, marital satisfaction
second, and mother psychopathology third, followed by the two child variables entered as
a block. The results of this analysis are presented in Table 21. Father psychopathology
was a weak but significant predictor of father DCI, accounting for 2% of the variance
(F/change=5.79, p<.05). Even with the effect of father psychopathology controlled for in
the model, mother psychopathology was a stronger predictor of father DCI than was

75

father psychopathology, but, contrary to expectations, the relationship was negative
(beta= -.25); greater overall maternal psychopathology was thus associated with less
father differential concern (F/change=8.25, p<.01). As before, the child variables
produced the most significant improvement in R2 (F/change=22.12, p<.001). Marital
satisfaction again did not contribute significantly to the model. Adjusted R2 for the entire
model was .22 (F [5,198]=12.40, p<.001).

Table 20

76

Hierarchical Regression Predictim. M_9thers' Differgrlaal Concern on CBCL Total Behavior Problems

LN=204 families)

 

Dependent Variable: Mother Differential Concern Index

 

Block -
No. Predictor Beta-1n RZ/Ch R2 Adj. R2 F/Ch
1. Mother
Psychopathology .33 . l l .1 1 .1 l 2476*"
2. Marital
Satisfaction -.12 .01 . 12 .11 2.70
3. Father
Psychopathology -.10 .01 .13 .11 1.45
4. Child
Age . 12
Child
Disturbance .43 .17 .30 .28 2390*“

 

”*p < .001

Table 21

Hierarchical Regreasion Predicting Fathers' Differengll Concern on CBCL Total Behavior Problems

(N=204 families)

 

Dependent Variable: Father Difl'erential Concern Index

 

Block
No. Predictor Beta-In R2/Ch R2 Adj. R2 F/Ch
1. Father

Psychopathology .17 .03 .03 .02 579*
2. Marital

Satisfaction .05 .00 .03 .02 .40
3. Mother

Psychopathology -.25 .04 .07 .05 8.25"
4. Child

Age .05

Child

Disturbance .45 .17 .24 .22 2212*"

 

‘p < .05 "p < .01

*ttp < 001

78

Percentage Agreement (Presence/Absence). A three-step regression was
performed to test the effects of the five predictor variables on percentage agreement on
presence/absence of child behavior problems. Since this index is based on the ratings of
both parents, mother and father psychopathology were entered as a block on the first step,
followed by marital satisfaction entered singly, and then the two child variables entered as
a block. (The same logic is followed on the models for effective percentage agreement
and kappa, below.) The results of this analysis are presented in Table 22. Parent
psychopathology is inversely related to percentage agreement on presence/absence of child
behavior problems (F/change=6.50, p<.01), but is a fairly weak predictor, accounting for
only 5% of the variance of the outcome variable. Marital satisfaction was again
nonsignificant. By far the largest contribution to the model was made by the child
variables, which together accounted for about 45% of the variance in percentage
agreement on presence/absence of behavior problems (F/change=93.79, p<.0001).
Adjusted R2 for the entire model was .51 (F[5,198]=43.76, p<.0001).

Effective Percemge Agreement. A three-step regression was performed to test
the effects of the five predictor variables on efi‘ective percentage agreement. Mother and
father psychopathology were entered as a block on the first step, followed by marital
satisfaction entered singly, and then the two child variables entered as a block. The results
of this analysis are presented in Table 23. Only the child variables, particularly child's
disturbance level, contributed significantly to the model (F/change=5.93, p<.01). As was
the case for the bivariate test, it was found that the greater the overall reported child
disturbance, the more likely were parents to agree on the level or severity of specific child
behavior problems as measured by efi‘ective percentage agreement. Though significant,
however, child disturbance accounts for only about 6% of the variance in the outcome

measure. Adjusted R2 for the entire model was .05 (F[5,198]=3.09, p=.01).

79
Table 22

Hierarchical Regzession Predicting Percen_tage Agreemera on Presence/Absence of CBCL Total Behavior
Problems (N=204 families)

 

Dependent Variable: Percentage Agreement (Presence/Absence)

 

Block
No. Predictor Beta-In R2/Ch R2 Adj. R2 F/Ch
1. Mother

Psychopathology -.03

Father

Psychopathology -.25 .06 .06 .05 6.50"
2. Marital

Satisfaction .14 .01 .07 .06 3.05
3. Child

Age -. 16

Child's

Disturbance -.71 .45 .52 .51 9379*"

 

"p < .01 ***p < .0001

80
Table 23

Hierarchical Regression Predicting Effective Percentage Agreemen_t on CBCL Total Behavior Problems
(N=204 families)

 

Dependent Variable: Efl‘ective Percentage Agreement

 

Block
No. Predictor Beta-In R2/Ch R2 Adj. R2 F/Ch
1. Mother

Psychopathology .03

Father

Psychopathology -.03 .00 .00 -.Ol . 12
2. Marital

Satisfaction . 14 .02 .02 .00 3.16
3. Child

Age -. 10

Child

Disturbance .23 .06 .07 .05 5.93"

 

”p < .01

81

Kappa A three-step regression was performed to test the effects of the five
predictor variables on overall interparent agreement as measured by kappa. Again, mother
and father psychopathology were entered as a block on the first step, followed by marital
satisfaction entered singly, and then the two child variables entered as a block. The results
of this analysis are presented in Table 24. In a trend unlike that found in the other
regression models, marital satisfaction made a statistically significant (though weak)
contribution, accounting for 2% of the variance in kappa (F/change=3.91, p<.05). No
other block of variables contributed to a significant improvement in R2. Adjusted R2 for
the entire model was .02 (F [5, 198]=1.92, n.s.).

82
Table 24

Hierarchical Regzession Predlating Interparent Kappa on CBCL Total Behavior Problems (N=204
families)

 

Dependent Variable: Kappa

 

Block
No. Predictor Beta-In RZ/Ch R2 Adj. R2 F/Ch
1 . Mother

Psychopathology .01

Father

Psychopathology -. 12 .01 .01 .00 1.38
2. Marital

Satisfaction .16 .02 .03 .02 3.91 "
3. Child

Age -. 12

Child

Disturbance .03 .Ol .05 .02 1.41

 

‘p < .05

83

CLINICAL SIGNIFICANCE OF FINDINGS

Jacobsen, Follette, and Revenstorf (1984) emphasize the distinction between
statistical significance and practical clinical significance in reporting psychological findings.
They suggest that researchers emphasize the magnitude of the effect size of a given
treatment or observed relationship. This is particularly germane to the present study, as a
number of the findings are low-level yet statistically significant. To more clearly elucidate
the clinical implications of these results, subsamples were defined that correspond to the
top and bottom quartiles of the sample with respect to mother psychopathology, father
psychopathology, marital dissatisfaction, and level of reported child disturbance. It is
assumed that parents falling in the upper 25% of this sample on the psychopathology
measures are likely to be experiencing clinically significant levels of disturbance, while
those in the lowest quartile are assumed to approximate normality. Similar assumptions
are made about marital functioning and level of child disturbance. (Note that marital
functioning is expressed in terms of high and low marital dfisatisfaction for these analyses,
so that "high" consistently refers to a greater degree of distress or dysfunction, while
”low" refers to a lesser degree of distress or dysfunction.) AN OVA's for each index of
interparent agreement were performed, comparing the two groups defined as high and low
on each predictor variable (mother psychopathology, father psychopathology, marital
dissatisfaction, and level of reported child disturbance). In addition, following Jacobson et
al. (1984), a measure of effect size was calculated for each difference in means, to better
illustrate the magnitude of the difference. The efl'ect size is simply a standard score,
calculated as the difference between the two group means divided by the pooled standard
deviation of the lowest-quartile group. Dividing by the standard deviation of the lowest

group allows the efi‘ect size to represent the magnitude of the effect relative 19 the lowest-

 

quartile group. Put difi‘erently, it illustrates the size of the effect of being a member of the
highest-quartile group rather than of the lowest-quartile group, and is similar to Jacobson's
et a1. (1984) reliable change (RC) index for psychotherapy outcome. The RC index

84

expresses the magnitude of change on an outcome variable observed in the treatment
group, relative to the distribution of scores in a normal control group. Thus, expressing
effect size in this way allows the mean interparent agreement score of the highest-quartile
group to be compared to the distribution of scores in the lowest-quartile group, expressed
in percentiles. For example, an effect size of 1.00 means that memberShip in the high
group is associated, on average, with being about one standard deviation unit higher on
the dependent variable than the average person in the low group; thus, someone at the
mean in the high group is comparable to someone at the 84th percentile in the low group.
The results of these analyses are presented in Tables 25 through 28.

For groups defined by high and low mother psychopathology, presented in Table
25, the only statistically significant group difference is found on mothers' differential
concern index: mothers in the high mother psychopathology group had an average DCI
score of 18.9, compared with only 9.8 for mothers in the low psychopathology group
(F[1,132]=23.2, p<.0000). The effect size was 1.21, which means that an average mother
(i.e., 50th percentile) in the high psychopathology group will evince a degree of
difi‘erential concern over child behavior problems (relative to her spouse) that falls at
about the 88th percentile of the low psychopathology group distribution of difl‘erential
concern. This mean difference of nearly forty percentile points indicates that this finding
may have practical significance in addition to being statistically reliable. Mothers in the
high group evince nearly twice as much differential concern as mothers in the low group,
and it is also worth noting that the maximum mother DCI score observed in the high
group was 93, while in the low group the highest observed score was only 31. It appears
therefore that mothers sufi‘ering high levels of distress are much more likely than their
nondistressed peers to experience greater concern over their child's behavior problems

than their husbands as reflected by a greater number of CBCL endorsements.

85
Table 25

Compa_n'son of Top and Bottom Mother Pg'chomthology we 1e Groups on the Indices of Integflent
Agrfiment on CBCL Totaflehavior Proble_ms (N=134 families)

 

 

 

High Low
Mother Mother Effect
Psychopathology Psychopathology E p Sizet
(N=67) (N=67)
Mother Differential
Concern Indexa
Mean 18.9 9.8 23.20 .000 1.21
(SD) (13.6) (7.5)
Father Differential
Concern Indexb
Mean 11.5 14.9 2.42 .12 -0.21
(SD) (8.1) (16.5)
Percentage Agreement
(Pres/Abs)°
Mean 78.4 81.1 2.53 .11 -0.24
(SD) (8.8) (1 1.2)
Eflective Percentage
Agreementd
Mean 29.7 28.2 .63 .43 0.13
(SD) (10.9) (11.7)
Kappae
Mean .37 .37 .05 .83 .00
(SD) (.13) (.13)

 

a Number of items which mother endorses and father does not, plus number of items she endorses more
strongly than does father.

b Number of items which father endorses and mother does not, plus number of items he endorses more
strongly than does mother.

c Percentage of agreement on the presence/absence of child behavior problems.

d Efi'ective percentage agreement on child behavior problems.

e Cohen's Kappa

f Efi‘ect size = (mean of high group - mean of low group) / standard deviation of low group.

86

Table 26 shows the results for the groups defined by the top and bottom quartile of
father psychopathology. Three statistically significant effects were found. First, mothers
in the high father psychopathology group have higher DCI scores than mothers in the low
group (F[1,126]=4.66, p<.03), although the magnitude of the effect (0.44, 67th percentile
relative to low group) is not nearly as high as that found when the groups were defined by
mother psychopathology. Second, fathers in the high father psychopathology group have
higher DCI scores than their low-group counterparts (F[l,126]=6.53, p<.01). The effect
size for fathers was the same as for mothers (0.44, 67th percentile). Is a difference of
seventeen percentile points between the two group distributions clinically significant?
Obviously, a precise answer to this question would be somewhat arbitrary, but based on
these findings it appears reasonable to argue that high father psychopathology makes a
clear and reliable difference in the amount of differential concern over child behavior
problems shown by both mothers and fathers. Finally, parents in the high father
psychopathology group evinced a lower percentage agreement on presence/absence of
child behavior problems than did their peers in the low group (E[1,126]=14.99, p<.000).
This was the largest effect size observed in this comparison (-O.65), indicating that parents
in the high paternal psychopathology group fall on average at the 24th percentile of their
low-group peers on the percentage agreement measure. Although this is a respectable
difference in a statistical sense, one must bear in mind that parents in both groups agree
over 77 percent of the time on the presence/absence of behavior problems, with the
difference between the means of the two groups amounting to only 4.3 percentage points.

Thus, the clinical significance of this difi‘erence remains questionable.

87

 

 

 

 

 

 

Table 26
Comgrisom ngd Bott_om Father Psychopathology Quartile Groaps on the Indices of Irl_ter1)arent
Aggeement on CBCL Toti Behavior Problem_st==128 families)
High Low
Father Father Effect
Psychopathology Psychopathology E p Sizet
(N=64) (N=64)
Mother Differential
Concern Indexa
Mean 16.4 12.7 4.66 .03 0.44
(SD) (10.7) (8.5)
Father Differential
Concern Indexb
Mean 14.9 10.7 6.53 .01 0.44
(SD) (9.2) (9.5)
Percentage Agreement
(Pres/Abs)c
Mean 77.6 81.9 14.99 .000 -0.65
(SD) (5.9) (6.6)
Effective Percentage
Agreementd
Mean 29.5 29.2 .02 .88 0.03
(SD) (11.3) (10.8)
KaPPa°
Mean .36 .37 .46 .50 -0.08
(SD) (.12) (.12)
a Number of items which mother endorses and father does not, plus number of items she endorses more
strongly than does father.
Number of items which father endorses and mother does not, plus number of items he endorses more
strongly than does mother.

c Percentage of agreement on the presence/absence of child behavior problems.
‘1 Effective percentage agreement on child behavior problems.

e Cohen's Kappa.
f Effect size = (mean ofhigh group - mean oflow group) / standard deviation oflow group.

88

As was the case in the earlier analyses, level of reported child disturbance was
significantly associated with interparent agreement, and these effects were further
pronounced when groups were defined based on the highest and lowest quartile of
reported disturbance. As Table 27 shows, group difi‘erences were highly statistically
significant for all measures of interparent agreement except kappa. For the significant
findings, effect sizes ranged from 0.68 (75th percentile) for effective percentage agreement
(indicating better agreement on more disturbed children once agreement on nonoccurrence
is controlled) to -3.33 (<.01 percentile) for percentage agreement on presence/absence
(not surprising, since this index includes agreement on nonoccurrence which is strongly
inversely related to reported child disturbance). Thus, the results for effective percentage
agreement indicate that parents reporting higher overall levels of child disturbance are
better able to agree on the fine distinctions among levels of child disturbance than are
parents whose children are less disturbed. Moving to the DCI, parents in the high child
disturbance group were much more likely to express differential concern about their child's
problems than were parents in the low group; the effect size for mothers was 1.59 (95th
percentile) and for fathers was 2.17 (99th percentile). Thus, greater overall disturbance is
linked with significantly more differential reporting of problems relative to the other

parent.

89
Table 27

Comparison of Top and Bottom Child Disturbance Quartile Groups on the Indices of Integflent
Agzeement on CBCL Total Behavior Problems (N=154 families)

 

 

 

High Low
Child Child Effect
Disturbance Disturbance E p Size1
(N=76) (N=78)
Mother Differential
Concern Indexa
Mean 19.5 9.5 31.53 .000 1.59
(SD) (14.5) (6.3)
Father Differential
Concern Indexb
Mean 19.1 7.8 41.45 .000 2.17
(SD) ( 14.7) (5.2)
Percentage Agreement
(Pres/Abs)°
Mean 73.3 86.3 112.13 .000 -3.33
(SD) (10.0) (3.9)
Effective Percentage
Agreementd
Mean 30.9 23.4 19.75 .000 0.68
(SD) (9.8) (11.1)
Kappa!e
Mean .35 .33 .79 .37 0.14
(SD) (. 10) (. 14)
a Number of items which mother endorses and father does not, plus number of items she endorses more
strongly than does father.
b Number of items which father endorses and mother does not, plus number of items he endorses more
strongly than does mother.

c Percentage of agreement on the presence/absence of child behavior problems.
‘1 Effective percentage agreement on child behavior problems.

e Cohen's Kappa.
f Effect size = (mean ofhigh group - mean oflow group) / standard deviation oflow group.

90

Finally, groups based on the highest and lowest quartiles on marital dissatisfaction
were compared on interparent agreement. Table 28 presents the results of this analysis.
The strongest effect of group membership was found for mother DCI (£[1,118]=9.97,
p<.002; efi‘ect size=0.96, 83rd percentile). Thus, mothers in less satisfactory marriages
evince greater differential concern over child behavior problems relative to their husbands,
with the average mother in the maritally dissatisfied group falling above the 80th percentile
of the DCI distribution of the nondistressed group. This difference of over thirty
percentile points argues for clinical, as well as statistical, significance. A smaller
significant effect was found for percentage agreement on presence/ absence of behavior
problems (efi‘ect size=-O.51). Most notable, however, is that groups difi‘erences in marital
satisfaction were associated with differences in kappa (F[1,118]=7.73, p<.01); parents in
the higher marital dissatisfaction group had lower interparent kappa scores than parents
that were less dissatisfied with their marriage (effect size=-O.55, 30th percentile).
However, one must wonder at the practical significance of the .06 mean difference in
kappa between the two groups. As the regression analysis in Table 24 showed, while
marital satisfaction was related to kappa, it was able to explain only about 2% of the

variance in kappa. Hence, the clinical significance of this difference must be questioned.

9 1
Table 28

Comparison of Topand Bottam Marital Dissatisfaction Quartile Groups on the Indices of Inteflrent
Amment on CBCL Totflchayior Problems (N=120 families)

 

 

 

High Low
Marital Marital Effect
Dissatisfaction Dissatisfaction E p Sizet
(N=60) (N=60)
Mother Differential
Concern Indexa
Mean 17.9 11.5 9.97 .002 0.96
(SD) , (14.1) (6.6)
Father Diflerenfial
Concern Indexb
Mean 13.6 11.7 .78 .38 0.13
(SD) (9.1) (14.3)
Percentage Agreement
(Pres/Abs)c
Mean 77.2 82.4 9.28 .003 -0.51
(SD) (8.3) (10.3)
Effective Percentage
Agreementd
Mean 29.3 32.0 1.98 .16 -.025
(SD) (10.2) (10.7)
KaPPa"
Mean .36 .42 7.73 .006 -.55
(SD) (.13) (.11)
a Number of items which mother endorses and father does not, plus number of items she endorses more
strongly than does father.
9 Number of items which father endorses and mother does not, plus number of items he endorses more
strongly than does mother.

° Percentage of agreement on the presence/absence of child behavior problems.

d Effective percentage agreement on child behavior problems.

e Cohen's Kappa.

f Effect size = (mean of high group - mean of low group) / standard deviation of low group.

CHAPTER IV
DISCUSSION
lvfliTHODOLOGICAL CONSIDERATIONS

This study is unique in its opportunity to track sources of interparent agreement
and disagreement in a population-based sample of families comprising a broad data base
with considerable familial variation in psychopathology. Because the sample is not clinic-
derived and the developmental age of the families is young, the alcoholic parents in the
sample are typically not yet in treatment for alcohol or other drug problems (Zucker &
Fitzgerald, 1991b), and thus it is likely that the findings are more generalizable than would
be the case with a clinic-based sample (Christensen et al., 1992). Additionally, due to the
large variation in familial psychopathology, the effects of specific disorders or problems-
be they drug or alcohol problems, depression, marital problems, and so on--can be studied
within the context of their natural comorbidity with other problems, further increasing the
potential utility of the study (Zucker & Fitzgerald, 1991b). Having said this, however, it
should be emphasized that the findings of this study cannot be assumed to generalize
beyond young, non-Hispanic Caucasian families with boys primarily between 3 and 6 years
old.

The cross-sectional nature of this analysis precludes drawing conclusions about
causality or directionality in the observed relationships. It is likely that parental
psychopathology of any kind represents sequelae of processes that began well before the
establishment of the parents' dyadic relationship, which in turn precedes the birth of
children and subsequent child behavior problems. Indeed, such assumptions about
primacy informed the development of the hierarchical regression models tested in this

analysis. However, longitudinal analyses are needed to determine more concretely which
92

93

factors afi‘ect interparent agreement in a more proximate sense and which are more distal,
while at the same time establishing the changing ways in which parent, child, and family
factors affect agreement over the course of the child's development. Fortunately, such
longitudinal analyses will be possible as subsequent waves of data collection are
completed.

This analysis relies primarily on parent self-reports for the assessment of parent
psychopathology, marital satisfaction, and child behavior problems. Future investigations
should supplement parents' reports with more observational and clinician-derived data in
order to develop a richer and more "Objective" operationalization of these variables. For
example, the marital satisfaction index derived from the SASB system (Benjamin, 1983;
Ichiyama, 1994) could be supplemented by data fi'om a videotaped marital problem
solving task. Of course, observer ratings are not flee from bias and distortion, and not all
of the phenomena of interest lend themselves to direct observation or clinical rating
(Loeber et al., 1989), but triangulation of methods would nevertheless provide a richer
picture of the ways in which parent, child, and family variables are related to interparent
agreement on child behavior problems.

The most important area in which to supplement parents' reports with observers'
ratings is that of child behavior problems. Indeed, the most robust predictor of interparent
agreement in this analysis was the overall level of behavior problems reported by parents
on the CBCL. It would be desirable to triangulate parents' reports with those of trained
observers in order to better characterize the relationship between child disturbance and
interparent agreement. If it is indeed the case that the most disturbed children are also
those whose parents agree less on child behavior problems, as found in this and other
studies (e. g., Ferguson et al., 1974), then a special challenge is presented to service

providers who must rely heavily on parent reports in the assessment of child pathology.

94

DISCUSSION OF THE FINDINGS

The ability to predict interparent agreement varied depening on the index of
aggeement. Different results were obtained in these analyses depending on which index of
agreement served as the dependent variable. In the hierarchical regression analyses,
mother and father psychopathology, marital satisfaction, and child's age and reported
disturbance level were able to explain nearly a third of the variance in mothers' differential
concern (Table 20), and a fourth of the variance in fathers' differential concern (Table 21).
These predictors accounted for over half of the variance in parents' percentage agreement
on presence/absence of child behavior problems (Table 22), yet explained only around 5%
of the variance in effective percentage of agreement (Table 23). Finally, the model
accounted for only about 2% of the variance in kappa (Table 24). These results indicate
that it is inappropriate to consider interparent agreement to be a unitary construct about
which one formulates general hypotheses. Rather, it must be recognized that the various
indices used to compute agreement provide distinct qualitative information about
agreement, as well as differing in their statistical properties, and will therefore relate
difi‘erently to a given set of predictors. As a result of these considerations, each index has
its own unique strengths and weaknesses. The simple (uncorrected) percentage agreement
score is easy to calculate and straightforward to interpret, but does not correct for
agreement due to chance. Further, since the amount of chance agreement will vary across
difi‘erent outcome measures and coding schemes, percentage agreement scores are ofien
not comparable across studies (Bakeman & Gottman, 1987). Efi‘ective percentage
agreement corrects for some of this chance inflationnthe portion attributable to agreement
on nonoccurrences in cases of less severe child behavior problems. It does so, however, at
the expense of throwing away a good deal of information, as it treats agreement on
nonoccurrence as unimportant. Further, since the index is based on a smaller number of
endorsements, it is less stable than its uncorrected counterpart (Jacob, et al., 1982). The

parent difi‘erential concern indices share some of the problems of inflation due to chance

95

that their cousin the percentage agreement score does, but provide important qualitative
information about the direction and extent of interparent agreement and disagreement that
is not captured in those measures that combine parents' information into a single index
(Christensen et al., 1992). Given this qualitative advantage, the parent DCI indices are
probably more usefirl measures of interparent agreement than percentage agreement.

The use of kappa in these analyses warrants special discussion. Kappa is a
rigorous measure of reliability because it takes into account all of the raters' agreement at
an item-by-item level, rr_ri_n_pa the portion of observed agreement that_is estimated to be due
to chance (Cohen, 1960). What remains amounts to a lower limit of actual agreement
between the raters, and is therefore a very conservative estimate of interrater reliability.
However, by removing the portion of observed agreement due to chance, the computation
of kappa truncates the variance of the outcome variable. One efl‘ect of this truncation is to
lower the power of subsequent statistical tests, since statistical power is lost whenever
there is a decrease in variability in a measure (Cohen, 1988). As a result, correlations
between kappa and other measures will be truncated. It may be for this reason that, with
very few exceptions, kappa was unrelated to the various predictor variables such as parent
psychopathology and child disturbance. The only significant effect detected ill the
regression analyses was a positive relationship with marital satisfaction, but even this
factor could explain only about 2% of the variance in kappa. These negative findings for
kappa mirror the results obtained by Christensen et al. (1992). Such a truncation effect
may explain the low and often nonsigrrificant correlations observed for effective
percentage agreement as well, since it too is a stringent measure of agreement that loses
considerable variability once agreement on nonoccurrences is removed. These caveats
aside, however, and following Christensen and his colleagues (1992), it must be concluded
that kappa is the most "statistically correct" (pg. 424) measure of interrater agreement,

since it is the best measurement of "actual" (non-chance) agreement. Investigators should

96

continue to attempt to identify factors that are associated with kappa in order to

rigorously specify the correlates of interparent agreement.

Severig of reported child disturbance was the strongest overall predictor of
integparent agreement. First, greater reported child disturbance was associated with

 

greater difl‘erential concern on the part of both mothers and fathers (both fs=.44, p<.001).
It may be that greater child disturbance is associated with disturbance across a number of
family subsystems (e.g., child, parent, and marital subsystems). In the present study, for
example, level of reported child disturbance was positively correlated with both mother
psychopathology (r=.29, p<.001) and father psychopathology (_r=.35, p<.001), and was
negatively related to parents' marital satisfaction (5.24, p<.001). Christensen, et al.
(1992) found that greater distress in any one family subsystem was related to greater
parent difi‘erential concern, and that distress in more than one subsystem was associated
with still greater differential concern. Thus, child disturbance is may often be intertwined
with greater parental psychopathology and marital distress, all of which may contribute to
poorer agreement via such mechanisms as parents' perceptual biases or poor interparent
communication. In addition, the finding that greater reported child disturbance is
associated with greater differential concern, considered alongside the fact that mothers'
and fathers' DCI scores are negatively correlated (_r=-.45, p<.001), indicates considerable
situational specificity in child behavior problems. That is, the child exhibits a high overall

level of problem behavior but exhibits difi‘erent behaviors around mother than he does

 

around father, thus causing each parent to have difl'erential concern about specific
behaviors relative to the other parent. ~

Second, greater reported child disturbance was strongly related to lower
percentage agreement on the presence/absence of behavior problems (1:272, p<.0001).
This is not surprising, since, as discussed earlier, this index is inflated for less severe cases

due to the high number of agreements on nonoccurrence.

97

Third, while greater child disturbance was related (as predicted) to greater parent
difl‘erential concern and lower percentage agreement on presence/absence, it was
positively associated with efl‘ective percentage agreement (which controls for the higher

probability of agreement on non-occurrence among parents of less-disturbed children).

Hence, once chance aggeement on non-occurrence is controlled, parents of more disturbed

children are actually better able to agree on the severig of their child's problems than are
parents of less disturbed children. This may be due to a greater amount of discussion

about child behavior problems required of parents of more disturbed children, or perhaps
professional intervention that has increased the parents' awareness of finer gradations
among levels of disturbance for those problems that are mutually recognized. Finally,
level of child disturbance was unrelated to kappa, both in the bivariate test and in the
hierarchical regression analysis. As noted above, the relationship between child
disturbance and interparent agreement found in these analyses indicates the need for
further investigation employing non-parent-report measures of child disturbance for
purposes of replicating and extending these findings.

InteM’ ° g versus Externalizing Behavior Problems. It was predicted that
agreement would be better for externalizing problems as they are more visible and
disruptive, and hence more observable, than internalizing problems. Again, the findings
were mixed and depend upon the particular method for measuring agreement (see Table
6). Contrary to expectations, no difference between parents' DCI scores was found for
externalizing versus internalizing behavior problems. Also contrary to prediction,
percentage of agreement on presence/absence of behavior problems was lower for
externalizing behaviors. On the other hand, efl‘ective percentage of agreement and kappa
indicated better agreement on externalizing problems, as predicted. These results are
similar to the results for severity of child behavior problems discussed above. Recall that
greater child disturbance was related to greater parent DCI scores and lower percentage

agreement on presence/absence, but was related to higher efi‘ective percentage agreement;

98

these results nrirror the findings for externalizing behavior problems, indicating that
perceived severity of disturbance may be closely linked with externalizing, as opposed to
internalizing, problems. It may be precisely the disruptive nature of these problems that is
related to disagreement between parents, since acting out by the child may be associated
with a generally chaotic fanrily environment characterized by poor interparent
communication, marital problems and other family stressors (Margolin, 1981; Christensen
et al., 1992). On the other hand, the differential agreement on internalizing versus
externalizing problems may also reflect situational and developmental specificity in child
behavior, with the child acting in different ways and displaying different sorts of behavior
problems with mother than he does with father. Finally, higher efl‘ective percentage
agreement on externalizing behaviors indicates that, once severity of child behavior
problems is corrected for, parents are indeed better able to agree on more overt, acting
out behaviors than they are on internalizing problems.

&e of child was fourg to be largely unrelated to interparent agrgeement. Age was

unrelated to all indices of interparent agreement except for a low-level negative correlation

 

(_r=-.15, p<.05) with percentage agreement on presence/absence of behavior problems.
Although the literature suggests that interparent agreement improves as children get older
(until adolescence), the results of the studies reviewed for this investigation may not be
firlly generalizable to the present sample. For example, none of the studies cover the same
age range of children in our sample (2.2 - 7.1, with 90% between 3.0 and 6.0). Also, ours
is the only population-based sample of children with parents assessed for clinical levels of
psychopathology. Thus, the lack of significant findings regarding the relationship between
age and interparent agreement should not be viewed as discrepant with the literature, but
as a unique finding in its own right. Statistically, it may be that the failure to find
significant relationships between age and the other indices of interparent agreement is due

to the restricted range of child ages in this sample. This highlights the importance of doing

99

longitudinal research that can elucidate the changes in interparent agreement across a
broad range of child ages.

Amount of time spent with child by fathers was unrelated to agreement. It was
hypothesized that if fathers spend more time with their child and are thereby afforded an
increased opportunity to observe their child's behavior, their reports should fall more into
line with mothers' reports of child behavior problems. This hypothesis was not supported
by the data. This may be because the measure of time used in this analysis is an unreliable
indicator of the actual amount of time fathers spend with the child; the three-day sampling
of time that is used to compute the measure is a small sample, and may not be
representative of how much time fathers spend with the child in general. Another
possibility, however, is that the child behaves quite differently with father than he does
with mother-once again, the question of situational specificity. Achenbach and his
colleagues (Achenbach, McConaughy, & Howell, 1987) suggest that children's behavior is
likely to be situation-specific, and that we cannot neglect "the possibility that different
informants validly contribute different information" (pg.213). Thus, even if fathers do
spend more time with their children, they may observe a systematically difl‘erent sample of
child behavior than mothers do. Observational analyses of the child's behavior with father
compared to mother would be useful in order to test this possibility.

Marita Satisfaction. It is notable that parents' marital satisfaction was one of the
few predictors significantly related to kappa, both in the bivariate tests for each parent
separately, and in the hierarchical regression analysis for the parental dyad: as predicted,
higher marital satisfaction was associated with a higher interparent kappa score (all
p's<.05). In all instances the relationship was of low order, but these results nevertheless
point to the potential impact of the quality of the marital relationship upon interparent
agreement. More direct self-report measures of marital satisfaction such as the Dyadic
Adjustment Scale (Spanier, 1976), combined with observational ratings of marital

interaction, may prove useful in firture investigations of interparent agreement. It may be

100

that a strong marital relationship characterized by better interparent communication is the
key to parents developing, as Burrows and Kelley (1983) put it, "a shared familiarity with
their own child's behavior across a variety of situations" (pg. 47) without actually having
to Observe the same samples of the child's behavior (Kazdin et al., 1983; Schaughency &
Lahey, 1985).

Parental Psychopathology. Parents' general psychological well-being, as measured
by the Psychological Health Q-Sort and DSM-III Axis V rating, was related to lower
mother differential concern; likewise, specific problems such as depression, alcohol
problems, and antisociality were predictive of both mother and father difl‘erential concern,
as well as percentage agreement on presence/ absence. These effects were low to
moderate in strength and in the predicted direction, with greater levels of psychopathology
associated with lower interparent agreement. Only fathers' self-reported depression and
alcoholism type (control vs. overt alcoholic) were related to kappa. When combined into
factor-derived scale scores and entered into a hierarchical regression, the results for
parental psychopathology were much the same as in the bivariate analyses. Thus, the
forms of parental psychopathology examined in this study (a cluster which tends to be
found in alcoholic families) were predictive of parents' tendency to systematically report
higher levels of behavior problems relative to their spouse, and parents' failure to agree on
the presence/absence of specific behaviors, but were in general not significantly related to
kappa.

SUMMARY

The results of these analyses indicate that, once agreement due to chance factors is
controlled for, there exists considerable disagreement between parents regarding child
behavior problerrrs. Some of this disagreement is due to random measurement error, but
some of it is systematic disagreement arising from a variety of sources. Two such sources
have been suggested by this study: situational specificity of child behavior, and parents'
perceptual biases. As discussed throughout this study, children's behavior may be

101

situation-specific, thus making it possible that different informants experience (_li_f_fe_ran_t
sammes of the child's behavior across situations (Achenbach, McConaughy, and Howell,
1987). This suggests that a given informant may supply unique and valid information that
overlaps only partially with that of another informant. This results in disagreement when
informants' endorsements on behavior rating scales are compared, even though both
informants may be contributing valid information about the child's behavior problems
(Achenbach, McConaughy, & Howell, 1987). Second, parents' perceptual biases
associated with parent psychopathology may also lead to disagreement, such as when a
depressed parent demonstrates decreased tolerance for normal child behavior and
perceives this behavior as problematic, while the other (nondepressed) parent does not. In
addition to situational specificity and parent bias, a third source of disagreement may stem
simply fiom differences in parents' abilities to observe, or in their observational habits
(Thornton, personal communication). For example, an attention deficit, an impulsive style
of responding to environmental stimuli, or a tendency to ignore or "screen out" the child's
behavior on the part of one parent may result the acquisition of different information about
the child's behavior compared to the other parent. In summary, it seems likely that the
considerable variance in parents' ratings is not due simply to measurement error, but
instead reflects real variance in child, rater, and situational characteristics (Christensen et
al., 1992).

Clinical sigpljj' cance of findings. The analysis summarized in Table 25 indicates
that mothers experiencing clinical levels of psychopathology are much more likely than
their nondistressed peers to express greater difl‘erential concern over child behavior
problems relative to their spouses. Not only is the relationship statistically significant, but
the effect size is such that the average mother (50th percentile) in the distressed group
would fall near the 90th percentile on differential concern in the nondistressed group,
indicating a strong link between mother psychopathology and difi‘erential concern over
child behavior problems.

102

Similarly, fathers experiencing clinical levels of psychopathology (Table 26) are
more likely than their nondistressed peers to express greater concern over child behavior
problems relative to their spouses; further, the spouses of distressed fathers are also more
likely to have high DCI scores, indicating that both partners are expressing high levels of
concern, but over difierent child behavior problems. The size of these effects was smaller
than for groups defined by level of maternal pathology (e.s.=.44, 67th percentile).

A clinically significant level of child disturbance (Table 27) was found to be
strongly related to higher parent DCI scores and lower percentage agreement on
presence/absence of behavior problems, and to higher effective percentage agreement, as
was the case in earlier analyses. This finding underscores the strong link between child
disturbance and parent differential concern, while again indicating that once agreement on
nonoccurrence is controlled for by effective percentage agreement, parents of more
disturbed children are better able to distinguish among levels of child behavior problems.

Finally, mothers in the lowest quartile on marital satisfaction expressed more
difl‘erential concern over child behavior problems than did mothers that are satisfied with
their marriage, and parents in the maritally distressed group showed poorer agreement as
measured by percentage agreement and kappa.

The findings of this investigation are similar to those of Christensen et al. (1992)
and Jacob, et al. (1982); parental differential concern and percentage of agreement on
presence/absence of behavior problems are significantly related to the disturbance level of
both child and parents, while kappa is generally not affected by child and parent
pathology. Christensen and his colleagues (1992) argue that, while kappa may be the
most statistically rigorous measure of interrater reliability, the mother and father DCI "may
more accurately reflect the psychological reality of the family than kappa" (pg. 424). They
point out that two families can evince almost identical interparent kappa scores while
difl‘ering markedly on the number and pattern of differential concerns (called

"discrepancies" by the authors).

103

In one nondistressed family, the mother endorsed 16 and the father endorsed 11
problem behaviors, but 11 of the mother's endorsements and 6 of the father's were
not agreed upon by the other parent. The kappa was .30. In one dual-system
distressed [i.e., marital and child distress] family, the mother endorsed 75 and the
father endorsed 57 problem behaviors, but 41 of the mother's endorsements and 13
of the father's were not agreed upon by the other parent. The kappa was .31. We
would argue that despite almost identical kappa values, these families have a
different experience of parental agreement about child behavior problems.
Because parents in the second case have more than three times as many cases in
which they disagree about problem behavior in their child, they are more likely to
be aware of their different views of their child and are more likely to openly
disagree about him or her (pg. 424).

A similar but even more striking example can be found in the present sample. Two
families with extremely low kappa scores of -.01 and -.05 were compared. If this were all
one knew about the families, one might assume that both sets of parents evinced equally
poor interparent agreement. But when the DCI scores were compared, a striking
difl‘erence emerged. For the first family, the mother DCI was 1 and the father DCI was 4,
indicating that one of the mother’s endorsements and four of the father's were not agreed
upon by the other parent. For the second family, the mother DCI was 93 and the father
DCI was 3, indicating that 93 of the mother's endorsements, but only 3 of the father's
endorsements, were not agreed upon by the other parent. Like Christensen and
colleagues' example quoted above, this finding indicates that parents in these two families
have a very different experience of interparent agreement on their child's behavior
problems. Thus, the results of the present study in many ways replicate previous findings
while extending them to a more representative, population-based sample of families. It
appears increasingly clear that child, family, and parental disturbance are related to
systematic differences between parents in the reporting of child behavior problems.
FUTURE DIRECTIONS

The future directions suggested by this study primarily involve addressing some of
the methodological concerns discussed earlier. Longitudinal analyses are needed to

determine more concretely the relative primacy of parent, child, and family factors that

104

affect interparent agreement, and to document the changing ways in which these factors
may affect agreement over the course of the child's development. Future investigations
should also supplement parents' reports with more observational and clinician-derived data
in order to develop a richer and more "Objective" operationalization of the predictor
variables. Along these lines, it would be particularly important to triangulate parents'
reports of child behavior problems with those of trained observers in order to better

characterize the relationship between child disturbance and interparent agreement.

APPENDIX

APPENDIX

EXPANDED TABLES SHOWING RELATIONSHIPS
BETWEEN PREDICTORS AND INTERPARENT AGREEMENT
FOR TOTAL, INTERNALIZIN G, AND EXTERNALIZING BEHAVIOR PROBLEMS

l 05
Table A1

Relationship (Pearson r's) Between Demographic Variables and Indices of Intewrent Aggeement on
CBCL Total, Intemalizing, and Externalizing Behavior Problems (N=260 families)

 

 

Mother DCI Father DCI % Agree. (P/A) Eff. % Agyee. Kappa
Total Total Total Total Total

Intemalizing Intemalizing Intemalizing Intemalizing Intemalizing
Externalizing Externalizing Externalizing Externalizing Externalizing

 

-.09 -.01 .07 -.09 -.02
Mothers' Age -.09 . 00 .05 -. 15* -.13*
-.07 -.05 .08 -.07 -.02
-. 12 .03 .07 -.02 .06
Fathers' Age -. 10 .03 .05 -.08 -.05
-. 10 .00 .09 -.03 .03
-.05 -.02 .04 .04 .04
Mothers' Ed. -.07 -.01 .04 .00 .03
-.04 -.O3 .04 .05 .09
-. 10 -.03 .08 -.02 .00
Fathers' Ed. -. 12* .02 .07 -.07 -.03
-.10 -.06 .08 -.04 .03
-.01 -.01 -.O4 .06 .05
Mothers' IQ -.Ol .01 -.03 .07 .05
-.03 -.04 .00 .03 .07
-. 15* .00 .10 _ .03 .07
Fathers' 1Q -. 12 .02 .06 -.04 .02
-.16* -.01 .ll -.01 .06
-.09 -.08 . 13* .02 .05
Family SES -.09 -.07 .12* -.06 -.01
(Duncan) -.09 -.09 .1 1 .01 .08
-.20** .00 .14* .09 .14*
Family Income -.20** -.02 .17** .06 . 10
-.18** .01 .10 .04 .11
.00 -.02 ' .02 -.04 -.02
No. of Children .06 .00 -.04 -.1 1 -.11
.00 -.05 .02 .01 -.01
.1 l .04 -. 15* -.07 -.09
Age of Target .19** .08 -.24*** -.11 -.19**
Child .09 -.03 -.05 -.02 -.03

 

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

Table A2

Relationship (Pearson r's) Between Levle Reported Chil_d Disturbance and Integpagent Agreement on
CBCL Total, Intemalizing, and Externalizing Behavior Problems (N= 298 families)

106

 

 

 

Index of

Interparent Agreement Level of Reported Child Disturbance
Mother DCI

Total . 44***
Intemalizing .43***
Externalizing .3 1 ***
Father DCI

Total .44***
Intemalizing .46***
Externalizing .34***
% Aime. (PIA)

Total -.72***
Intemalizing -.70***
Externalizing -.46***
Eff. % Ag1_'ee.

Total .23***
Intemalizing . 14*
Externalizing .28***
Kapg

Total .02
Intemalizing -.07
Externalizing .05

 

¢s¢2<001

‘p<.05 "p<.01

Table A3

107

Relationship (Pearson r's) Between the Amouat of Time Spera with Target Chil_d by Fathers app
Integpgent Aggeement on CBCL Total, Intemalizing, and Externalizing Behavior Problems (N=213

 

 

 

 

 

families)

Index of Total Time Activity Time
Interparent Agreement Father Spent with Child Father Spent in with Child
Mother DCI

Total -.02 -.09
lnterrralizing -.02 -.1 1
Externalizing -.04 -. 10
Father DCl

Total .10 .06
Intemalizing .09 .09
Externalizing .11 .05
% Agree. (P/A)

Total -.07 .05
Intemalizing -.04 .03
Externalizing -.06 .08
Elf. % Agree.

Total -.03 .07
Intemalizing .02 .09
Externalizing -.01 .08
I_<_arzp_a

Total -.03 .09
Intemalizing .01 .08
Externalizing -.03 .11
*p<.05 **p<.01 ***p<.001

108

Table A4

Relationship (Pearson r's) Between Parents' Marital Satisfaction and Inteflgent Agreemerp on CBCL
Total, Intemalizing, Ed Externalizing Behavior Problems (N=240 families)

 

 

 

 

 

Index of Mothers' Fathers'
Interparent Agreement Marital Satisfactiona Marital Satisfaction
Mother DCI

Total -.26*** -. ll
Intemalizing -.22** -. 10
Externalizing -.23*** -.09
Father DCI

Total .00 -. 10
Intemalizing -.04 -. 15*
Externalizing .02 -.09
% AgreeJP/A)

Total .22" .19**
Intemalizing .20** .21“
Externalizing .20** . 14*
Eff. % Agree.

Total .09 .08
Intemalizing .08 .07
Externalizing .01 .02
Kappa

Total .14* .16*
Intemalizing .08 .l l
Externalizing .06 .09

 

‘p<.05 "p<.01 ***p<.001

109

Table A5

Relationship (Pearson r's) Between Parents' DSM-III Axis V Rating and Integparent Aggeement on CBCL
Total, Intemalizing, and Externalizing Behavior Problems (N=298 families)

 

 

 

Index of Mothers' Fathers'
Interparent Agreement Axis V Rating Axis V Rating
Mother DCI

Total -.24*** -.16**
Intemalizing -.24*** -.17**
Externalizing -.21*** -.14*
Father DCI

Total .10 -.03
Intemalizing .09 -.03
Externalizing .09 -.04
% Aggee. (P/A) _

Total , .06 .13*
Intemalizing .09 . 15**
Externalizing .03 .07
Efl‘. % Agrg.

Total -.03 .04
Intemalizing .03 .08
Externalizing -.08 -.07
Kappa

Total .01 .10
Intemalizing .03 .1 l
Externalizing -.04 .01

 

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

Table A6

110

Relationship (Pearson r's) Between Parents' Mental Hglth Q-Sort Score and Intemnt Agreement on
CBCL Total, Intemalizing, and Externalizing Behavior Problems m=298 families)

 

 

 

Index of Mothers' Fathers'
Interparent Agreement Mental Health Q-Sort a Mental Health Q-Sort
M_other_D_C_I

Total -.15* -. 13*
Intemalizing -.21** -.13 *
Externalizing -. 10 -. 12
Father DCI

Total .03 -.04
Intemalizing .04 -.04
Externalizing .01 -.04
% Ame/A)

Total .06 .12
Intemalizing . 10 . 14*
Externalizing .05 .10
Eff. % Aggee.

Total -.01 .03
Intemalizing .05 .1 1
Externalizing -.05 -.07
Q1292

Total .04 .08
Intemalizing .05 .12
Externalizing .00 -.01

 

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

Table A7

Relationship (Pearson r's) Between M_others' Beck and Hamilton Depresaion Scorgand Integpagnt
Agreement on CBCL Total, Intemalizing, and Externalizing Behavior Problems (N=298 families)

 

 

 

Mothers' Mothers'
Index of Mothers' Hamilton Score Hamilton Score
Interparent Agreement Beck Score (Current) (Worst Ever)
MLherLCl
Total .27*** .21*** .17**
Intemalizing .23*** .23*** .16"
Externalizing .24** .15** .15*
Father DCI
Total -.02 -. 12* -.08
Intemalizing .00 -.13* -.05
Externalizing -.02 -.1 1 -.09
% Agree. (P/A)
Total -.20** -.08 -.05
Intemalizing -.18** -.07 -.07
Externalizing -.18** -.05 -.01
Efl‘. % Agree.
Total .03 -.02 .01
Intemalizing .04 -.06 -.02
Externalizing . 05 .05 .06
Karim
Total -.07 -.06 -.01
Intemalizing -.01 -.03 -.04
Externalizing -.02 .00 .01

 

*p<.05 "p<.01

***p<.001

112

Table A8

Relationship (Pearson r's) Between Fathers' Beck and Hamilton Depres_sion Scores Ed Intemgent
Agreemeat on CBCL Total, Intemalizing, and Externalizing Behavior Problems (N=298 families)

 

 

 

Fathers' F athers'
Index of Fathers' Hamilton Score Hamilton Score
Interparent Agreement Beck Score (Current) (Worst Ever)
Mother DCI
Total .14* .04 .05
Intemalizing . 14* .03 .04
Externalizing .08 .02 .02
Father DCI
Total .20*** .06 .14*
Intemalizing .18** .05 .12*
Externalizing .21*** .08 .14*
% Agree. (P/A)
Total -.27*** -.07 -.14*
Intemalizing -.25*** -.05 -.11*
Externalizing -. 19** -. 10 -.1 1*
BE. % Aggee.
Total -.03 .00 .00
Intemalizing -.01 .01 -.01
Externalizing .02 .05 .04
Kappa -. 12* -.03 -.02
Total -.09 .00 -.05
Intemalizing -.07 -.03 -.02
Externalizing

 

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

113

Table A9

Relationship (Pearson r '5) Between Pa_________rents' Lifetime Alcohol Problems Score (LAPS) and Intemrent
meement on CBCL Total, Intemalizing, and Externalizing Behavior Problems (N= 298 families)

 

 

 

Index of Mothers' F athers'
Interparent Agreement LAPS LAPS
Mother DCI

Total .18** .02
lnternalizing . 16** .04
Externalizing . 19** -.01
Father DCI

Total .03 .18**
Intemalizing .07 .16"
Externalizing .00 .2 1 ** *
% AgrmJ/A)

Total -. 16** -.15**
Intemalizing l9** -. 17**
Externalizing 10 -.12*
Eff. % Aggee.

Total .05 .05
Intemalizing -.05 .01
Externalizing .12 .06
Kapg

Total .00 -.04
lnternalizing -.08 -.03
Externalizing .06 -.O4

 

‘p<.05 Mp<.01 m*p<.001

114

Table A10

Relationship (Pearson r's) Between Parents' AntiSOCfl Behavior (ASB) Checklist Score and Interparent
Aggeement on CBCL Total, Intemalizing, and Externalizing Btfilaw'or Problems (N=298 families)

 

 

Index of Mothers' Fathers'
Interparent Agreement ASB ASB
Mother DCI

Total .27*** .07
Intemalizing .21*** .06
Externalizing .30*** .08
Father DCI

Total -.02 .21***
Intemalizing .05 .20***
Externalizing -.08 .20***

% Agree. (P/A)
Total
Intemalizing
Externalizing

Eff. % Aggee.
Total

Intemalizing
Externalizing

1541293

Total
Intemalizing
Externalizing

-.16**

‘21***

-.10

.01
-.07
.10

-.04
-.13*
.02

‘22***

-21***

m22*..

-.02
-.04
-.00

-.06
-.06
-.09

 

*p<.05 "p<.01

tit2<1)01

LIST OF REFERENCES

LIST OF REFERENCES

Achenbach, T. M. (1978). The child behavior profile: 1. Boys aged 6-11. Journal of
Consultirgand Clinical Psychology, 59, 478—488.

Achenbach, T. M., & Edelbrock, C. S. (1978). The classification of child
psychopathology: A review and analysis of empirical efforts. Psychological Bulletin,
85, 1275-1301.

Achenbach, T. M., & Edelbrock, C. S. (1979). The child behavior profile: 11. Boys aged
12-16 and girls aged 6-11 and 12-16. Journal of Consulting and Clinical Psychology,
41, 223-233.

Achenbach, T. M., & Edelbrock, C. S. (1983). Manual for the Child Behavior Checklist
aad Revised Child Behavior Profile. Burlington: University of Vermont, Department
of Psychiatry.

Achenbach, T. M., Edelbrock, C ., & Howell, C. T. (1987). Empirically based assessment
of the behavioral/emotional problems of 2- and 3-year-old children. Journal of
Abnormal Child Psychology, .12, 629-650.

Achenbach, T. M., Howell, C. T., Quay, H. C., & Conners, C. K. (1991). National survey
of problems and competencies among four- to sixteen-year-olds: Parents' reports for
normative and clinical samples. Monogzaphs of the Society for Research in Child
Development. _5_6 (3, Serial NO. 225).

Achenbach, T. M., McConaughy, S. H., & Howell, C. T. (1987). Child/adolescent
behavioral and emotional problems: Implications of cross-informant correlations for
situational specificity. ngchological Bulletin. 10L 213-232.

American Psychiatric Association (1987). Diagpostic and statistical manual of mental
disorders. third edition. revised. Washington, D. C .: Author.

Bakeman, R., & Gottman, J. M. (1987). Applying observational methods: A systematic

view. In J. D. Osofsky (Ed), _Handbookarf 111ng Development (2nd ed.) (pp. 818-
854). New York: Wiley and Sons.

115

116

Barkley, R. A. (1988). Child behavior rating scales and checklists. In M. Rutter, A. H.
Tuma, & 1. S. Lann (Eds), Assessment and Diagaosis in Child Psychopathology (pp.
113-155). New York: Guilford.

Beck, A. T., Steer, R. A., & Garbin, M. G. (1988). Psychometric properties of the Beck
Depression Inventory: Twenty-five years of evaluation. Clinical Psychology Review,
§, 77-100.

Beck, A. T., Ward, C. H., Mendelson, N., Mock, J ., & Erbaugh, J. (1961). An inventory
for measuring depression. Archives of General Psychiatgg, 5, 53-63.

Becker, W. C. (1960). The relationship of factors in parental ratings of self and each other
to the behavior of kindergarten children as rated by mothers, fathers, and teachers.
Journal of Counseling Psychology, E, 507-527.

Benjamin, L. S. (1983). Intrex user's manual. Madison, WI: Intrex Interpersonal
Institute.

Block, J. (1978). The Q-sort method in personalig) assessment and psychiatric research.
Palo Alto: Consulting Psychologists Press.

 

Burrows, K. R., & Kelley, C. K. (1983). Parental interrater reliability as a function of
situational specificity and familiarity of target child. Journal of Abnormal Child

Pachology, 1_1, 41-48.

Chamberlain-Waechter, P. (1980) Standardization of aparent report maasure.
Unpublished doctoral dissertation, University of Oregon, Eugene, OR.

 

Christensen, A., Margolin, G., & Sullaway, M. (1987, April). Parental agreement on the
child behavior checkliat. Paper presented at the annual convention of the
Southwestern Psychological Association, New Orleans, LA.

Christensen, A., Margolin, G., & Sullaway, M. (1992). Interparent agreement on child
behavior problems. Psychological Assessment, A, 419-425.

Christensen, A, Phillips, S., Glasgow, R E., & Johnson, S. M. (1983). Parental
characteristics and interactional dysfunction in families with child behavior problems:
A preliminary investigation. Journal of Abnormal Child Psyphplpgy, 1_l_, 153-166.

Coddington, R. D. (1972). The significance of life events as etiologic factors in the
diseases of children: A study of a normal population. Journal of Psychosomatic
Research, 16, 205-213.

Cohen, J. (1960). A coefficient of agreement for nominal scales. Educational and
chhological Measurement, _2_Q, 37-46.

117

Cohen, J. (1988). SLatisticaapower analysis for the behavioral sciences (2nd ed.).
Hillsdale, NJ: Erlbaum.

Compas, B. E., Phares, V., Banez, G. E., & Howell, D. C. (1991). Correlates of
internalizing and externalizing behavior problems: Perceived competence, causal
attributions, and parental symptoms. Journal of Abnormal Child Psychology, 1_9, 197-
218.

Conners, C. K. (1972). Pharrnacotherapy of psychopathology in children. In H. C. Quay
and J. S. Werry (Eds), Psychopathological disorders of childhood. New York: Wiley.

Dadds, M. R, & Powell, M. B. (1991). The relationship of interparental conflict and
global marital adjustment to aggression, anxiety, and immaturity in aggressive and
nonclinic children. Joumal Of Abnorml Child Psychology, 12, 553-567.

Dawson, D. A. (1992). The effect of parental alcohol dependence on perceived children's
behavior. Joulal of Substame Abuse. :1, 329-340.

Derogatis, L. R. (1983). SCL-90-R administratiop, scoring, and procedures manual.
Towson, MD: Clinical Psychometric Research.

 

Derogatis, L. R, Lipman, R. S., Richels, K., Uhlenhuth, E. H., & Covi, L. (1974). The
Hopkins Symptom Checklist (HSCL): A self-report symptom inventory. Behavioral
m 19 1-15.

9 —’

Earls, F. (1976). The fathers (not the mothers): Their importance and influence with
infants and young children. Psychiatgy, 32, 209-226.

Earls, F. (1980). The prevalence of behavior problems in 3-year-old children. Journal of
the American Academy of Child Psychiatgy, L9, 439-452.

Edelbrock, C., Costello, A. J., Dulcan, M. K., Conover, N. C., & Kala, R. (1986). Parent-
child agreement on child psychiatric symptoms assessed via structured interview.
Journal of Child Psychiatgy and Psychology, _21, 181-190.

Ferguson, L. R, Partyka, L. B., & Lester, B. M. (1974). Patterns of parent perception
differentiating clinic from nonclinic children. Journal of Abnormal Child Pachology,
;, 169-181.

Fitzgerald, H. E., Sullivan, L. A., Ham, H. P., Zucker, R. A., Bruckel, S., Schneider, A.
M., & Noll, R B. (1993). Predictors of behavior problems in three-year-old sons of
alcoholics: Early evidence for the onset of risk. Child Development, fl, 110-123.

118
Fitzgerald, H. E, Zucker, R. A, & Maguin, E. (in press). Time spent with child and

parental agreement on ratings of child behavior among preschoolers. Perceptual and
Motor Skills, in press.

Friedlander, S., Weiss, D. S., & Traylor, J. (1986). Assessing the influence of maternal
depression on the validity of the Child Behavior Checklist. Jouraal of Abnoraral Child

Psychology, 1_‘1, 123-133.

Hamilton, M. (1960). A rating scale for depression. Journal of Neurology. Neurosurgegy,
and Psychiatgg, 23, 56-62.

Henry, W. P., Schacht, T. E., & Strupp, H. H. (1986). Structural analysis of social
behavior: Application to a study of interpersonal process in differential
psychotherapeutic outcome. Journal of Consulting and Clinical Psychology, 25, 27-
31.

Humphrey, L. L., Apple, R. F., & Kirschenbaum, D. S. (1986). Differentiating bulirnic-
anorexic fi'om normal families using interpersonal and behavioral observation systems.
Journal of Consulting and Clinical Psychology, 54 190-195.

_’

Ichiyama, M. (1994). Development of a marital satisfaction index. Unpublished
instrument, Michigan State University, East Lansing, MI.

Jacob, T., Grounds, L., & Haley, R (1982). Correspondence between parents' reports on
the Behavior Problem Checklist. Journal of Abnormal Child Psychology, _1_(_)_, 593-
608.

Jacob, T., & Leonard, K. (1986). Psychosocial functioning in the children of alcoholic
fathers, depressed fathers, and control fathers. Journal of Studiea on Alcohol. fl,
373-3 80.

Jacobson, N. S., Follette, W. C., & Revenstorf, D. (1984). Psychotherapy outcome
research: Methods for reporting variability and evaluating clinical significance.
Behavior Therapy, 15 336-352.

———’

Jensen, P. S., Traylor, J ., Xenakis, S. N., & Davis, H. (1988). Child psychopathology
rating scales and interrater agreement 1. Parents' gender and psychiatric symptoms.

Jourral of th_e American Academy of Child Ed Adolegent chhiatry, fl, 442-450.

Jensen, P. S., Xenakis, S. N., Davis, H., & Degroot, J. (1988). Child psychopathology
rating scales and interrater agreement 11. Child and family characteristics. J puma] of

tha American Acadepy of Child and Adolescent Paychiatry, _2_7_, 451-461.

Johnston, C. (1991). Predicting mothers' and fathers' perceptions of child behavior
problems. Canadian Joumal of Behaviouraa Science. 23, 349-357.

 

 

 

119

Kazdin, A. E., Esveldt-Dawson, K., Unis, A. S., & Rancurello, M. D. (1983). Child and
parent evaluations of depression and aggression in psychiatric inpatient children.
Journal of Abnormal Child Psychology, 11 401-413.

—,

 

Lindholm, B. W., & Touliatos, J. (1981). Parents' and teachers' perceptions of children in
counseling and not in counseling. The Journal of Psycholggy, _108, 283-289.

Loeber, R., & Dishion, T. J. (1983). Early predictors of male delinquency: A review.
Psychological Bulletin, 24, 68-99.

Loeber, R., Green, S. M., & Lahey, B. B. (1990). Mental health professionals' perception
of the utility of children, mothers, and teachers as informants on childhood
psychopathology. Journal of Clinical Child Paychology, 1_9_, 136-143.

 

Loeber, R, Green, S. M., Lahey, B. B., & Stouthamer-Loeber, M. (1989). Optimal
Inforrnants On Childhood Disruptive Behaviors. Development and Psychopathology,
1, 317-337.

Margolin, G. (1981). The reciprocal relationship between marital and child problems. In J.
P. Vincent (Ed). Advances inFJnin Irltgwergion. Assessment, & Theog; A
Research Annual (Vol. 2). Greenwich, CT: Jai Press.

Mash, E. J ., & Johnston, C. (1983). Parental perceptions of child behavior problems,
parenting self-esteem, and mothers' reported stress in younger and older hyperactive
and normal children. Journal of Con_sulting and Clinical Psychology, 5;, 86-99.

Moses, H. D. (1992). Contextual moderators of the relationship between father alcohol
problems and child behavior problems in a high-risk population. Unpublished master's
thesis, Michigan State University, East Lansing, MI.

Phares, V. (1992). Where's Poppa? The relative lack of attention to the role of fathers in
child and adolescent psychopathology. American Psychologiat, fl, 656-664.

Phares, V., & Compas, B. E. (1992). The role of fathers in child and adolescent
psychopathology: Make room for Daddy. Pachological Bulletin, l_l_1, 387-412.

Reid, J. B., Kavanagh, K., & Baldwin, D. V. (1987). Abusive parents' perceptions of child
problem behaviors: An emple of parental bias. Journal of Abnormal Child
Psycholpgy, Q, 457-466.

Reider, E. E. (1991). Relationships between pargntal pgychgpathplogy, family conflict,
and child behavior problefl in young alcoholic families. Unpublished doctoral

dissertation, Michigan State University, East Lansing, MI.

120

Richters, J. E. (1992). Depressed mothers as informants about their children: A critical
review of the evidence for distortion. Psychological Bulletin. 112. 485-499.

Robins, L. N., Helzer, J. E., Ratclifl‘, K. S., & Seyfried, W. (1982). Validity of the
Diagnostic Interview Schedule, Version 11: DSM-III diagnoses. Psychological
Medicine _1_2, 855-870.

 

Rosenberg, L. A., & Joshi, P. (1986). Effect of marital discord on parental reports on the
Child Behavior Checklist. Psychological Reports 59 1255-1259.

’—,

Routh, D. K., Schroeder, C. S., & O'Tuama, L. C. (1974). Development of activity level
in children. Developmengal Psychology, 19, 163-168.

Schaughency, E. A., & Lahey, B. B. (1985). Mothers' and Fathers' Perceptions of Child
Deviance: Roles of Child Behavior, Parental Depression, and Marital Satisfaction.
Journal of Consulting and Clinical Psycholggy, _5_3_, 718-723.

Spanier, G. B. (1976). Measuring dyadic adjustment: New scales for assessing the quality
of marriage and similar dyads. J ourral of Marriage and the Family, 3, 15-28.

Spitzer, R. L., Endicott, J ., & Robins, E. (1975). Clinical criteria for psychiatric diagnosis
and DSM-III. American Journal of Psychiatg, 132, 1187-1192.

 

Thompson, R J ., & Bernal, M. E. (1982). Factors associated with parent labeling of
children referred for conduct problems. Journal of Abnormal Child Psychology, m,
191 -202.

Webster-Stratton, C., & Hammond, M. (1988). Maternal depression and its relationship to
life stress, perceptions of child behavior problems, parenting behaviors, and child

conduct problems. Journal of Abnormal Child Paychology, 16, 299-315.

Weiss, R L., & Birchler, G. R. (1975). Areas of change. Unpublished manuscript,
University of Oregon, Eugene, OR.

Zucker, R. A. (1988). The Lifetime Alcohol Problems Score (LAPS). Unpublished
paper, Department of Psychology, Michigan State University, East Lansing, MI.

Zucker, R A, & Fitzgerald, H. E. (1991a). Early developmental factors and risk for
alcohol problems. Alcohol Health and Research World, 15, 18-24.

Zucker, R. A., & Fitzgerald, H. E. (1991b). Risk and coping in children of alcoholics:
Years 6 to 10 of the Michigan State University Longitudinal Study. Grant proposal
submitted to the National Institute on Alcohol Abuse and Alcoholism.

121

Zucker, R. A., & Gomberg, E. S. L. (under review). Alcoholism and antisocial
comorbidity in women: A note on a hot spot and some hypotheses.

Zucker, R. A., & Noll, R. B. (1980a). The antisocial behavior checklist. Unpublished
instrument, Michigan State University, East Lansing, MI.

Zucker, R. A., Noll, R., Draznin, T., Baxter, J ., Wei], C., Theado, D., Greenberg, G.,
Charlot, C., & Reider, E. (1984). The ecology of alcoholic families: Conceptual
framework for the Michigan State University Longitudinal Study. Paper presented at
the NCA National Alcoholism Forum, Detroit, MI.