.. 1!]
{ff

.1. t. .5. avejﬂwwﬁgh

.. . .r '55: 11:...» “us

I U.‘".‘IA ’1 0",’
3:11.: .3...

. . .w f.n.,..)za. H...

. 2.9.2..

:5-

. navtxtv- .L
v 13. .2":

a

.
n,:‘
‘A l

 

.n....... .5... .32
fit .31... .v.Lax«..$..\J'VL
» 2v: .\.q.a..aa. .V. ,
Dara... t r- 032“... 7.. if“.
.33. c1793...

:1

3 I...

“Pun
farf‘ .

. 0

VI

'3

"H

 

y

a .«r‘ 6......

.. ~ .v _. . . .
.112! K: .

:0; :1.

3px;} 1:151:33 .

. 1.: vH‘J‘OAl‘i 5 ~

. A in:

y 2.9.
.. I .

t":
“Jug“
'53.“
v’l' n' >

'

3."? I). . \to I:
; c . p .
r ynv i I . o.
1.1:}? hie...- 1‘“
:x.:. . .fmnﬁxvtvii
v!!! 71%. .-
v...

. a

II M

' 3:1,!)-
“5w

:1
.

‘
n:v
‘ a

, . ‘ \f’f)-
. .
. .. a... . {ﬁx ,
vhn‘oltvﬁ - . r “H. , ,
.. - . ..
. . tiﬁf- -3. X ..
. . 3 .3. its»!
_ 31...... {3.9.5-5.
. z 3.13.. .i F. i r1231?
1 ii .-,;. artiilz: .
.333... .
21...}: ,
l. : §I§.FVA

o 7.... ‘11..qu
1.5.5: .I...u(,..~.¥... 1p...;sl:aav
. . IA. L. In Ha

I:- 1.7%.: . . . x51.
:. .v 1. :5... o . . :2. , .. .1
.93.. a... ... a...» me... an... - 35:... «ivy»... we
5. . .1041... .1. _. . 5:39:37 5:...
.r i i. 1. . .. 2.72.2:71.
3:10;...“b. .i....a¢§vl '1‘
. . ab . . .
hymn.» J , .
. Pkg... , , .
. . u 1.... a
. . I v...
. . . valna¢bwnf
2 fin! . vagrﬂnnﬂxia citvbaocﬁv
v . . . if. .
is...3.5.3....32?! exist <9:
.. I ... . .fvunﬂreat .ﬁv‘. hunk... uﬁuqiﬁvﬁﬁ
.5», ,5. V . «in! .02)....an 3!... 3
a. . 1‘11? {$.11pr , Aquila.»
at} 154.. iiivis;
£339.!

.11

up... up

Ct: \ u 7 . . .

1 ... 9!.

, , , . o. 5......39.
I...» 53.3.:
:zﬂrnn . . 1.. . xylvﬂndz .
1‘. . .503”. it‘s... .5..."
.A .ziuziiiyiail. (lb

I
I
r
V
I

.p
r: ...v . . .5
:35». . 1...". ,
z 3.... . 9:39;:-
.v (7‘! a»: x r. :
00.....3439 .1 two...ova....p..f
.. . u:..n.y..onxi LI...- .
’ rI'Ii\("r£.v-l-VU&IA

'2
l

I
v
.5
‘6

’k
3%:
«if:

*i

v 5.. n ‘
.. ..~quvI.|
L74 3......

. .1... IX

,
' 3

.,.
.Vl‘

‘ '3 0

.m

ab. :1)
*w

u
,A

o

u
v .
. h. . .
I! . - 51.915292:
.. o . . . .1 v)..av...a!’!.56.)l
I X Ctr A 0‘ . . . Qvlvv 'L'II'A ‘III
2 i . , . . Xv
XII-6.. 5933313.}... . .

x 25!... .v: . . . . jut”.
. (any)! 0...!
v1.4. 1
V “VIA" Ain't;
3W4”. fxﬂl‘ : 5.1.59.1};
S v . . . . anti.) hi.»
. l'trly£'l_
, 1391.31;

¢ alavba.
vrlvyzvnthrlo
a (.1311, O
0.1 . 5.7. .3)? I 5%! .
1.. ‘ . invisuf ICC
‘ IIIWAOILri ., . it!

.1. .
.2 .52).! .5
. 2-33;... V .. . .
(1.15.: $95.13!»... .a.r.....3..})v.....$!t.l..
La..ln.v.o.¢3v..ri3£.
4.5. 5.. 5.1% .
f 3:: . 1:3,...” (FF: :2 13-13:}!
. .91.... $13.2”, 5912.536: . 33"};
$513!. 1;

.9D-I f: .
1.. .vk... ~v.‘
.mQ-Jvh. .3

a... .
k. n , . .
. 3000 ll)!
. outin- .I.l}oplb’1‘l}bla>ruvr
.5)!" (it... 1.. . .tbOO-t Iv!-

‘ ‘ ,
-34.? . .
»-\va.\b .xyA . .
.ii.‘ . . .
. . . . )D 2.1“}!
, . . . : : viilfliiu.‘
Aoavl. .‘xvlwrt I . (at): 0‘0?
. till|l.-‘ll'nl rA-iJa." nix.)
» . ¢ y.

L
. twin is». p
L. . ,1... -
rolirlnﬂh.
. r. .

‘1} cn'l
‘1' 3.13,!!!
. .- .o...$.. «a 3.5.1.) 1.7? .5063.
{s37 ‘ ‘\ .{v- ‘.tlv.: . . {33... :5. }. I
.. . 40..» . . .110: bill:
t....v.[. v. i. IL. I.
. kfllhvhltttur
‘9!»yuavruie

.v..‘:~ .Dlvr
Ir‘aal.o. . .r

:$.«..Ioahzvli.|il .r. A 4 .l. r .
215?: ..u!o}t.brrt!’!sb‘avtl . ’r’fle
. . v; I 1.9.)!

. . -01.)... II:

 

.9 ,LLL.

.....D«,.

 

THESiS

SITY LIBRARI ISE

lllllllll l H I ll

 

 

 

 

 

 

 

 

 

 

 

 

                                          

31293

This is to certify that the

thesis entitled

THE ROLE OF PROTECTIVE FACTORS IN THE RELATIONSHIP
BETWEEN RISK AND CHILD EXTERNALIZING BEHAVIOR
PROBLEMS OVER A CONTINUUM OF RISK LEVELS

presented by

Sondra R. Wilen

has been accepted towards fulfillment
of the requirements for

M.A. Psychology

 

 

degree in

QV/WPJF/«VZ

Major prgj/ essor

Dan: November 18, 1997

 

0-7 639 MSU is an Afﬁrmative Action/Equal Opportunity Institution

 

LIBRARY

University

 

Michigan State

 

 

PLACE IN RETURN BOX
to remove this checkout from your record.
TO AVOID FINES return on or before date due.

 

DATE DUE DATE DUE

DATE DUE

 

MAGIC 2

 

afﬁrms

 

.mulzm@

 

q ‘6‘ .

 

Aug 31 g 2222

 

p v

 

 

 

 

 

 

 

 

 

1/98 CJCIHC/DataDue.p&5-p.14

 

 

 

THE ROLE OF PROTECTIVE FACTORS IN THE RELATIONSHIP BETWEEN
RISK AND CHILD EXTERNALIZING BEHAVIOR PROBLEMS
OVER A CONTINUUM OF RISK LEVELS
By

Sondra R Wilen

A THESIS
Submitted to
Michigan State University

in partial fulﬁllment of the requirements
for the degree of

MASTER OF ARTS

Department of Psychology

1997

 

 

 

ABSTRACT
THE ROLE OF PROTECTIVE FACTORS IN THE RELATIONSHIP BETWEEN
RISK AND CHILD EXTERNALIZING BEHAVIOR PROBLEMS
OVER A CONTINUUM OF RISK LEVELS
By

Sondra R. Wilen

This longitudinal study examined the eﬂ‘ects of multiple risk and protective factors
over a continuum of risk levels on the development of externalizing behavior problems in a
sample of population-based preschool-aged children of alcoholics and a contrast sample of
children from non-alcoholic families. The results of this study indicate that risk and
protective factors at time I predicted child behavior problems three years later. Child
temperament moderated the eﬁ‘ects of paternal antisociality and maternal alcoholism on
child behavior problems, and maternal warmth and family cohesion mediated the effects of
maternal depression.

The relationship between risk level and child behavior problems was found to
change at a cumulative risk level of two, with the individual risk factors being predictive
of child outcome in the high risk but not the low risk group. Similarly, risk and protective
factors were found to be more negatively correlated in the high risk group than in the low
risk group, and high levels of protection were found to be unlikely within an environment
characterized by high risk. The importance of examining the relationship between risk,

protection, and child outcome over a continuum of risk is discussed.

 

 

ACKNOWLEDGMENTS

I would like to thank Dr. Hiram Fitzgerald for taking on the responsibility of
chairing my thesis committee. His knowledge of the literature and enthusiasm for the

associated project constituted a valuable source of information and inspiration. Dr.

 

Fitzgerald’s feedback as this work transformed was vital, as were his support and
guidance through its complexities. Similarly, Dr. Robert Zucker’s ambition in posing
fascinating research problems and brilliance in conceptualizing the core issues were
instrumental to the completion of this work as it stands in its present form. Dr. Zucker
has taught me much about how to conceive a problem and what tactics to employ in
creating a solution — invaluable skills I will carry with me throughout my career.

To Dr. Alytia Levendoskyl owe a tremendous debt of gratitude for her
involvement and continued support during the entire thesis process. Dr. Levendosky was
a crucial source of structure, encouragement, and feedback. She is a true mentor and has
served as a model for growth as a professional woman.

I also thank Dr. Alex VonEye and Dr. Rick DeShon for the generous giving of
their time and willingness to share their remarkable statistical knowledge. Without their
assistance, this work would simply not have been possible. Dr. Anne Bogat was willing to
oﬂ‘er words of wisdom in guidance toward completion both of this project and the

program in general. Her door has always been open to me, and for that I am grateful.

 

«mi-ammo-

 

I would also like to thank three very special friends. To Michael Semel I owe
endless thanks for his faith in my abilities, patience, and willingness to impart the wisdom
he has acquired. To Lisa Galasso, words cannot express my thanks for the countless times
she lent a kind ear and a supportive word. To Jennifer McQueeney, despite her distance in
body, her support in heart and mind was priceless.

To my husband and best ﬁiend, Brian, I extend my heartfelt thanks for his love,
patience, and support throughout both this endeavor and my life as a whole. Brian
provided endless guidance in shaping my style of writing. He has shown never-ending
faith in me, and I am grateful he has chosen to share my life.

Finally, I thank the many families who participated in this project. Their honesty
and generosity with their time has been priceless to this work. I especially thank the

children and wish them nothing but the best in their continual struggle to grow.

iv

 

TABLE OF CONTENTS

LIST OF TABLES ................................................................................ vii-viii
LIST OF FIGURES .................................................................................... ix
INTRODUCTION ...................................................................................... 1
RISK FACTORS AND CHILD OUTCOME ........................................................ 3
Parental Psychopathology ...................................................................... 3
Co-occurring Parental Diagnoses .................................................... 6

Environmental Risk Factors .................................................................... 8
Multiple Risk Factors ......................................................................... 10
PROTECTIVE FACTORS AND CHILD OUTCOME .......................................... 14
Child Characteristics .......................................................................... 17
Intelligence ............................................................................ l7

Temperament ......................................................................... 19

Family Environment ........................................................................... 21

Family Cohesion ...................................................................... 21

Home Environment .................................................................. 23

Maternal Warmth ..................................................................... 25
RELATIONSHIP BETWEEN RISK AND PROTECTIVE FACTORS ....................... 27
Orthogonality of Constructs ................................................................ 27
Implications of Sample Variability .......................................................... 29
HYPOTHESES ......................................................................................... 33
Predicted Main Effects ........................................................................ 34
Predicted Moderating Eﬂ‘ects ................................................................ 35
Predicted Mediating Effects .................................................................. 35
Predicted Eﬂ‘ects of Risk Level .............................................................. 36
METHODS ............................................................................................. 37
Subjects .......................................................................................... 37
Procedures ...................................................................................... 39
Measures ....................................................................................... 39
Measures of Risk ..................................................................... 40

Measures of Child Protection ....................................................... 44

V

 

Measures of Family Environment Protection ..................................... 45

Measure of Child Outcome ......................................................... 47
RESULTS ............................................................................................... 48
Missing Data and Outliers ................................................................... 48
Descriptive Statistics .......................................................................... 48
Multiple Regression Analyses ............................................................... 48
Main Eﬂ‘ects ........................................................................... 48
Moderating Effects ................................................................... 51
Mediating Eﬂ‘ects ..................................................................... 55
Eﬂ‘ects of Risk Level ................................................................. 57
DISCUSSION .......................................................................................... 64
Main Eﬂ‘ects of Risk Factors ................................................................. 65
Main Eﬂ‘ects of Protective Factors .......................................................... 68
Moderating Effects of Child Characteristics ............................................... 69
Mediating Eﬂ‘ects of Family Environment Characteristics ............................. 71
Eﬁ‘ects of Risk Level .......................................................................... 73
Limitations of the Current Study and Directions for Future Research ................. 79
APPENDIX A
TABLES .................................... .. ............................................................ 82
APPENDIX B
FIGURES ................................................................................................ 95
REFERENCES ....................................................................................... 102
vi

 

 

LIST OF TABLES

Table l - Descriptive Statistics for Risk Variables, Protective Variables, and Extemalizing
Behavior Problems .................................................................. 82

Table 2 - Summary of Multiple Regression Analyses Predicting Child Extemalizing Behavior
Problems (n=214): Main Eﬂ‘ects of Risk Factors and Protective Factors. . .83

Table 3 - Summary of Multiple Regression Analysis Predicting Family Cohesion (n=214):
Main Effects of Risk Factors ....................................................... 84

Table 4 - Summary of Multiple Regression Analysis Predicting Stimulating Home
Environment (n=214): Main Eﬂ‘ects of Risk Factors ............................ 85

Table 5 - Summary of Multiple Regression Analysis Predicting Maternal Warmth (n=214):
Main Effects of Risk Factors ....................................................... 86

Table 6 - Summary of Hierarchical Regression Analyses for Variables Predicting Child
Extemalizing Behavior Problems (n=214): Moderating Eﬁ‘ects of Child
Protection ............................................................................. 87

Table 7 - Summary of Hierarchical Regression Analyses for Variables Predicting Child
Extemalizing Behavior Problems (n=214): Mediating Eﬂ‘ects of F arnily
Environment Protection ............................................................. 88

Table 8 - Summary of the Relationships Between Risk and a High Level of Child
Extemalizing Behavior Problems Across Risk Level Subpopulations ........ 89

Table 9 - Summary of Multiple Regression Analyses Predicting Child
Extemalizing Behavior Problems: Main Eﬂ‘ects of Risk Factors Within Risk
Level Subgroups ..................................................................... 90

Table 10- Intercorrelations Between Individual Risk and Protective Variables Within Risk
Level Subgroups ..................................................................... 91

Table 11- Summary of the Relationships Between Individual Risk and Protective Variables
Across Risk Level Subpopulations ................................................ 92

 

 

Table 12- Summary of Multiple Regression Analyses Predicting Child Reactivity: Main
Eﬁ‘ects of Risk Factors Within Risk Level Subgroups ........................ 93

Table 13- Summary of Multiple Regression Analyses Predicting Family Cohesion: Main
Eﬂ‘ects of Risk Factors Within Risk Level Subgroups ........................ 94

 

 

LIST OF FIGURES

Figure l - Model of Hypothesized Relationships Between Risk, Protection, and Child
Outcome .............................................................................. 95

Figure 2 - Hypothesized Relationship Between Risk and Extemalizing Behavior Problems
Over the Continuum of Risk Levels ............................................... 96

Figure 3 - Hypothesized Inverse Relationship Between Risk and Protection
Over the Continuum of Risk Levels ............................................... 97

Figure 4- Moderating Effect of Child Reactivity on the Relationship Between
Maternal Alcoholism and Child Extemalizing Behavior Problems ............ 98

Figure 5 - Moderating Eﬂ‘ect of Child Attention on the Relationship Between
Paternal Antisociality and Child Extemalizing Behavior Problems ............ 99

Figure 6- Moderating Eﬂ‘ect of Child Approach-Mthdrawal/Adaptability on the Relationship
Between Maternal Alcoholism and Child Extemalizing Behavior

Problems ........ . ..................................................................... 100
Figure 7- Estimation of Breakpoint in Slope of the Regression Equation
Predicting Extemalizing Behavior Problems .................................... 101
ix

 

 

 

 

 

 

INTRODUCTION

Research has consistently indicated that children of alcoholics are at a higher risk
than the general population for the development of a variety of behavior problems in
childhood (e. g. Fitzgerald et al., 1993; Zucker & Fitzgerald, 1991; Jansen, Fitzgerald,
Ham, & Zucker, 1995; Hill & Muka, 1996; West & Prinz 1987; Werner, 1986).
However, studies also have consistently found that a substantial number of children of
alcoholics never develop problems despite evidence of having been placed at an increased
risk. As a result, many researchers have attempted to determine what diﬂ‘erentiates those
children of alcoholics who experience dimculties from those who are resilient (e.g.,
Fitzgerald et al., 1993, Werner, 1986).

Although this area has been studied extensively, an understanding of what protects
some children of alcoholics from poor outcOmes has not yet been fully achieved; this
inadequacy of understanding has impeded preventive eﬂ‘orts. Much of the previous
research has utilized retrospective or cross-sectional designs with samples which do not
include children at the highest end of the continuum of risk and therefore cannot
adequately address the role of protective child characteristics and the family environment
as moderators or mediators of the development of problems in children of alcoholics
(Rogosch, Chassin, & Sher, 1990; Sher & Trull, 1996). While functionally more attractive

to researchers, it does not appear that these methods can successfully explain the complex

 

2

relationships between multiple risk factors and protective intervening variables which
result in diﬂ‘erential outcomes in children of alcoholics.

The purpose of this study is to examine the relationship between multiple risk and
protective factors and the role of these variables over time in the development of
externalizing behavior problems in children. The current investigation hypothesizes that
child characteristics, such as intelligence and temperament, as well as aspects of the family
environment, such as family cohesion and maternal warmth, will mitigate the adverse
impact of parental psychopathology and environmental risk factors on'child development.
This investigation ﬁrrther hypothesizes that there exists a diﬂ‘erential relationship between
risk and protection among risk status subpopulations, with these constructs becoming
decreasingly independent as risk increases. Additionally, it is anticipated that variability in
the proportion of outcome variance accounted for over a range of risk statuses will be
found, with the predictability of child externalizing behavior problems increasing at the
extremes on the continuum of risk (i.e., Bernstein & Hans, 1994). Therefore, this study
will investigate the eﬁ‘ects of multiple risk and protective factors on children’s
development of externalizing behavior problems over time in order to begin to explain and

predict the variations in outcome which are observed in children of alcoholics.

 

 

RISK FACTORS AND CHILD OUTCOME

Considerable research has addressed the relationship between risk factors in a
child’s environment and adverse child outcomes. Research has found risk factors such as
parental psychopathology and marital conﬂict consistently to be associated with the
development of both externalizing andinternalizing child behavior problems, although
these relationships are neither unavoidable nor correspond exclusively to speciﬁc risk
factors and speciﬁc child outcomes (Hill & Muka, 1996; Warner, Muﬁ'son, & Weissman,
1995; Katz & Gottman, 1993; Downey & Walker, 1992; Werner, 1986; Steinhausen,
Gobel, & Nestler, 1984). While research has found a variety of factors to be associated
with adverse child outcomes, this investigation will focus on those risk factors which have
been associated with externalizing behavior problems in children.
PM Psychopathology

Research has consistently found that children whose parents suffer from a mental
illness are at an increased risk for the development of externalizing behavior problems,
such as oppositional disorder, conduct disorder, and attention-deﬁcit hyperactivity
disorder (Jansen et al., 1995; Stanger, Achenbach, & McConaughy, 1993; Hammen,
Burge, & Stansbury, 1990; Walker, Downey, & Bergman, 1989). Whereas studies have
investigated the impact of a variety of parental psychopathologies on child development,
some areas have been more widely researched than others. Parental alcoholism is one risk

factor which has been studied extensively (Frick et al., 1992; Jansen et al., 1995; Rubio-

 

4
Stipec, Bird, Canino, Bravo, & Alegria, 1991; Steinhausen, Gobel, & Nestler, 1984;

Werner, 1986; West & Prinz, 1987; Zucker & Fitzgerald, 1991).

There is a consensus in the literature that children of alcoholics not only are at a
higher risk than the general population for the development of behavior problems but also
have a 4 to 5 times greater chance for the development of alcoholism in their lifetime than
the children of nonalcoholics (Donovan, 1986; Johnson, Sher, & Rolf, 1991). In a review
of the literature investigating the relationship between paternal characteristics and child
and adolescent psychopathology, Phares and Compas (1992) report that studies have
found a number of childhood problems to be associated with paternal alcoholism,
including hyperactivity, conduct disorder, substance abuse, and aggressive behavior. For
example, Rubio-Stipec et al. (1991) examined the relationship between parental
alcoholism and child outcomes in a community sample of parents and found that the
children of alcoholics were signiﬁcantly more likely to exhibit behavior problems than the

° ren of non-substance abusers. In addition, Jansen et al. (1995) found that those
preschool children who exhibited clinically signiﬁcant levels of behavior problems were
more likely to have parents who had a history of alcohol problems and antisocial behavior
than those children in the nonclinical range.

Children as young as 3 years of age exhibit the adverse eﬁ‘ects of parental

alcoholism (Zucker & Fitzgerald, 1991). ucker and Fitzgerald (1991) report that, with

-— —.d'".

’m-ﬁ“ M—un—d/

the exception of deprESSiom preschool children between the ages of 3 and 5 who have an

 

alcoholic parent were statistically more impaired in every area of symptomatology
measured by the Child Behavior Checklist (CBCL) than a matched control group of

children. In particular, three times as many high-risk children were rated in the clinical

 

5
range for the CBCL “overall problem behavior” category and the global measures of

externalizing symptoms and aggressive behavior (Zucker & Fitzgerald, 1991). The
researchers argue that the results of this study, which utilized a sample of preschool
children, parallel the results of studies which have found similar relationships between
parental alcoholism and behavior problems in older children (e.g., Werner, 1986; Jacob &
Leonard, 1986).

While an association between paternal alcoholism and child behavior problems has
been found repeatedly, the impact of maternal alcoholism has not been as widely
investigated. Although theliterature has not thoroughly addressed the relationship
between maternal alcoholism and child outcomes, many researchers have suggested that
because of the critical role that mothers play in child development, maternal
psychopathologies such as alcoholism and depression may signiﬁcantly impact children
(Hill & Muka, 1996; Fitzgerald et al., 1993; Werner, 1986). Fitzgerald et al. (1993) found
that while father’s depression, lifetime alcohol problems, and antisocial behavior did not
predict child outcomes, the presence of psychopathology in the mother predicted
children’s total behavior problems and externalizing behavior problems. In a longitudinal
study of the children of alcoholics, Werner (1986) investigated the characteristics of the
child and the caregiving environment which differentiated between those children who
manifested psychopathology and those who did not. This study found that while equal
numbers of children of alcoholic fathers exhibited resiliency or problems, the majority of
the children of alcoholic mothers exhibited problem behaviors by the age of 18 (Werner,

1986).

6
In order to address the paucity of maternal alcoholism studies, Hill and Muka

(1996) investigated the relationship between maternal alcoholism and children’s

psychiatric diagnoses. The researchers utilized a sample of alcoholic women who were

receiving treatment for substance abuse and a control sample who were screened for a

family history of alcoholism. The results of this study revealed signiﬁcantly higher rates of

psychiatric diagnoses in the children of maternal alcoholics than in the children of controls.
[2 . P l D .

Research has found that the adverse eﬁ‘ects of parental psychopathology on
children are compounded by the presence of multiple parental psychopathologies (Hill &
Muka, 1996; Merikangas, Prusoﬁ‘, & Weissman, 1988; Warner et al., 1995). Frick et al.
(1991) found higher rates of both substance abuse and antisocial personality disorder in
the parents of children who met criteria for conduct disorder or oppositional deﬁant
disorder than in the parents of control children. Studies also have revealed increased rates
of psychopathology in the spouses of individuals suﬁ‘ering ﬁ'om a psychological disorder
such as alcoholism (Fitzgerald et al., 1993; Downey & Coyne, 1990; Merikangas et al.,
1988; Jacob & Leonard, 1986). For example, Fitzgerald et a1. (1993) found an elevated
level of maternal psychopathology, involving alcohol consumption and depression, in the
wives of alcoholics, and Merikangas et al. (1988) found signiﬁcantly higher rates of
psychopathology in the spouses of depressed parents compared to controls.

Warner et al. (1995) argue that previous research has revealed the importance of
examining combinations of risk factors in the prediction of psychiatric disorders. These
researchers conducted a study in which they attempted to identify possible mediating and

independent variables which were predictive of psychiatric diagnoses in the offspring of

 

7
parents suffering from a psychiatric disorder (Warner et al., 1995). The results of this

study revealed that the risk of Major Depressive Disorder (MDD) in oﬁ‘spring was
signiﬁcantly increased by the presence of recurrent MDD in the parent (W amer et al.,
1995). Additionally, the researchers found that alcohol abuse in coparents signiﬁcantly
increased the risk for disorder in oﬁ‘spring (Warner et al., 1995). Warner et al. (1995)
suggest that a possible explanation for this ﬁnding is that risk for psychiatric disorder in
offspring is signiﬁcantly increased as a result of the additive eﬁ‘ects of multiple parental
diagnoses.

Studies not only have revealed that the spouses of alcoholics have signiﬁcantly
higher rates of psychopathology but also have highlighted the important role which the
compounded eﬂ’ect of simultaneous maternal and paternal psychopathology may play in
child outcome. Therefore, studies whiCh control for the presence of psychopathology in
one parent may not accurately represent the risk mechanisms which impact child
development in the general population (F rick et al., 1992; Walker, Downey, & Bergman,
1989). For example, in a study excluding families in which maternal psychopathology was
present, Johnson et al. (1989) did not ﬁnd any signiﬁcant diﬁ‘erences between the
drinking habits or attitudes of the adolescent children of paternal alcoholics, depressives,
or controls. The researchers suggest that one possible explanation for the lack of
signiﬁcantly diﬁ‘erent drinking habits may be that not all children of alcoholics are at risk
for the development of alcoholism due to the moderating eﬁ‘ects of other child and parent
variables (Johnson et al., 1989). However, prior research which has found a signiﬁcant
increase in psychopathology in the Spouses of alcoholics suggests that by eliminating those

families in which psychopathology was present in the mother, this sample of children of

 

 

8

alcoholics was not representative. Because it may have been biased toward a more
functional family environment than is typically found in the families of alcoholics, the
sample utilized in this study may have precluded the detection of the possible moderating
variables which Johnson et al. (1989) hypothesize play a crucial role in the risk for
alcoholism. -

Eny_I_r;Q° mm Risk Factors

Adverse environmental factors, such as marital conﬂict and low socioeconomic
status, are associated with parental psychopathology as well as with poor child outcomes,
such as aggression and conduct problems (Bolger, Patterson, Thompson, & Kupersmidt,
1995; Bond & McMahon, 1984; Corrao, Busellu, Valenti, & Lepore, 1993; Frick et al.,
1992; Hill & Muka, 1996; Jacob & Leonard, 1986; Jansen et al., 1995; Chassin, Rogosch,
& Barrera, 1992; Rubio-Stipec et al., 1991; Jouriles, Pﬁmrer, & O’Leary, 1988;
Merikangas et al., 1988). Jacob and Leonard (1986) found not only that the fathers and
mothers of the most severely impaired children of alcoholics reported more psychological
problems, but also that the presence of alcohol-related problems, such as occupational and
marital diﬁculties, were inﬂuential in the development of severe child impairment.

In a review of the literature concerning the children of depressed parents, Downey
and Coyne (1990) report that individuals suffering ﬁom depression tend to experience
interpersonal problems and to marry people who either themselves also suﬂ‘er ﬁ'om a form
of psychopathology or have a family history of psychiatric illness. These researchers also
found that the presence of psychiatric illnesses in both spouses is associated with more
severe symptomatology, marital disturbance, and family disturbance (Downey & Coyne,

1990). Additionally, Downey and Coyne emphasize that the child problems associated

 

9

with parental depression are not speciﬁc to parental depression and, in fact, the children of
depressed parents are similar to other children who experience signiﬁcant parental illness
and the accompanying environmental disturbances. Similarly, Rubio-Stipec et al. (1991)
found that children of parents with a psychiatric diagnosis were exposed to a greater
number of adverse family environmental factors, such as marital discord and family
dysﬁrnction, than children whose parents did not meet criteria for a disorder. Studies not
only have found that environmental variables, such as marital discord, are more common
in the families of parents suﬁ‘ering from a psychopathology, but also have revealed that
both environmental factors and parental psychopathology are associated with poor child
outcomes (Downey & Coyne, 1990; Rubio-Stipec et al., 1991). Thus, it often remains
unclear whether environmental factors aﬂ‘ect child adjustment more directly than simply
the presence or absence of a speciﬁc parental psychopathology.

Chassin, Rogosch, and Barrera (1992) state that while most research has revealed
that children of alcoholics exhibit higher rates of externalizing behavior problems, studies
have not adequately addressed whether this has resulted ﬁ'om the presence of parental
alcoholism or from other factors, such as co-occurring parental psychopathology or
disruptions in the family environment. These researchers addressed this issue in a large-
scale community and court recruited sample of children of alcoholics and matched
community controls and found that while there were unique eﬁ‘ects of environmental
stress, parental antisociality, and parental depression, parental alcoholism did not exert
unique eﬁ‘ects on child externalizing problems.

Research has found marital discord to be associated with the parent-child

relationship and poor child outcomes (Black & Pedro-Carroll, 1993; Bond & McMahon,

 

10
1984; Dadds & Powell, 1991; Katz & Gottman, 1993; Jouriles, Pﬁmrer, & O’Leary,

1988). Erel and Burman (1995) conducted a meta-analytic review of the literature and
found that the quality of the marital relationship is positively related to the quality of the
parent-child relationship. Jouriles, Pﬁﬂ‘ner, and O’Leary (1988) found both general
marital discord and overt marital conﬂict to be related to maternal reports of toddler
conduct problems and waiting-room obserVations of child deviance. Katz and Gottman
(1993) found that speciﬁc patterns of marital conﬂict predicted speciﬁc child behavior
problems 3 years later, with hostile conﬂict resolution predicting externalizing behavior
problems and angry, emotionally distant conﬂict resolution predicting anxiety and social
withdrawal in children (Katz & Gottman, 1993).

Similarly, studies also have found low socioeconomic status to be associated with
child behavior problems (Bolger et al., 1995; Jansen et al., 1995; Lempers, Clark-
Lempers, & Sirnons, 1989). Jansen et al. (1995) found that preschool boys who scored in
the clinical range on the “Total Behavior Problems” scale of the CBCL were more likely
to have an alcoholic parent and to be of lower socioeconomic status than children who
scored in the normal range. Bolger et al. (1995) found that family economic hardship was
associated with poor child outcomes, including conduct problems and low self-esteem. In
fact, Sameroﬁ‘ and Seifer (1983) have argued that socioeconomic status may have at least
as signiﬁcant an inﬂuence on children’s development as other factors that are frequently
used to determine risk-status in children, such as parental psychopathology.

Multiple Risk Factors
Research has found that numerous risk factors are similarly inﬂuential on child

outcomes and often coexist. The exact nature of the relationship between speciﬁc risk

 

1 1
factors and child outcomes therefore appears to be unclear. Frick et al. (1992) state that

the eﬂ‘ects of family risk factors on child development often are confounded, as these
variables typically are not independent of each other. Similarly, Walker et al. (1989) argue
that the inability to ﬁnd speciﬁc child outcomes associated with speciﬁc risk factors has
resulted ﬁ'om the fact that risk factors often co-occur and have additive eﬁ'ects.

As a result, many researchers have emphasized the importance of investigating the
impact of multiple risk factors. on child development (Blackson & Tarter, 1994; Frick et
al., 1992; Hill & Muka, 1996; Walker et al., 1989; Sameroﬁ' et al., 1993; Zucker, 1989).
Frick et al. (1992) point out that investigating family variables in isolation can not account
for the additive or interactional inﬂuence of multiple risk factors operating together to
inﬂuence child outcomes. Blackson and Tarter (1994) also stress the need to consider a
“multifactorial model of liability” which“. accounts for individual, family, and peer variables
(p. 819).

In order to address the apparent paucity of studies examining the impact of
multiple, simultaneous risk factors on child development, Walker et al. (1989) investigated
both the direct and interactive effects of several risk factors, such as parental
psychopathology and child maltreatment, on childhood behavior problems. The
researchers report that the ﬁndings from their study highlight the need to investigate
simultaneously the impact of multiple risk factors, since both main and interactive eﬂ‘ects
of parental psychopathology and child maltreatment were found. This study revealed a
signiﬁcant interaction between parental psychopathology and child maltreatment, with
children exposed to both maltreatment and parental psychopathology exhibiting increases

in externalizing behavior problems over time. The researchers concluded that their

12
investigation of both the direct and interactive eﬂ‘ects of multiple risk factors allowed them

to identify more accurately the relationships between risk factors and child behavior
problems than the investigation of only main eﬁ‘ects would have allowed (Walker et al.,
1989).

Researchers have also addressed the impact of multiple risk factors by
investigating the relationship between child outcome and the number of environmental risk
factors to which a child is exposed. Zucker (1994) argues that investigating the impact of
a single risk factor «does not make theoretical sense from a developmental perspective, as it
is not the presence of a single risk factor but rather the accumulation of risk over time
which is necessary for the development of a disorder. He points out that studies have
found more than a 50 percent increase in the probability for a poor outcome when multiple
risk factors are present over time (Zucker, 1989).

Sameroﬂ‘ et al. (1993) addressed the impact of multiple risk factors in a
longitudinal study in which they measured the IQ scores and various social and family risk
factors for children at age 4 and then again at age 13 years. By summing the number of
high-risk factors each child experienced, such as low socioeconorrric status and maternal
psychopathology, the researchers calculated a multiple environmental risk score for each
child (Sameroff et al., 1993). Upon examining the relationship between the multiple
environmental risk score and IQ at age 4, the researchers found a linear relationship, with
those children with no risk factors displaying an average IQ score over 2 standard
deviations above those children with the greatest number of risk factors (Sameroff et al.,
1993). Furthermore, this study revealed that when the children had reached the age of 13

years, the multiple risk score explained 37% of the variance in child IQ and a multiple

 

l3
regression approach utilizing all 10 risk variables accounted for 50% of the variance

(Sameroﬁ‘ et al., 1993). Additionally, Sameroﬁ‘ et al. (1993) determined that the impact of
multiple risk factors was independent of both social status and the possible genetic
inﬂuence of mother’s IQ, as a major eﬁ’ect was still found for the multiple risk score after
both SES/race and mother’s IQ had been covaried.

Overall, the ﬁndings of Sameroff et al. (1993) revealed that it was not the type of
risk factor a child encountered that determined outcome, but rather it was the number of
risk factors which was inﬂuential (Sameroﬁ’ et al., 1993). Sarneroff et al. (1993) state that
the results of their study ﬁt a “multicausal model of development,” with similar child
outcomes being inﬂuenced by various patterns of risk factors rather than a single variable
(Sameroﬁ‘ et al., 1993, p. 81). Furthermore, other researchers (e.g., Jessor et al., 1995;
Williams, Anderson, McGee, & Silva, 1990) have also found the number of risk factors to
be more predictive of childhood psychopathology than the presence of any particular risk

factor.

Protective Factors and Child Outcome

Although this is an area which historically has been neglected in the literature, the
importance of investigating factors that protect at-risk individuals has been emphasized
recently by many researchers (Sroufe & Rutter, 1984; Rutter, 1987; Garmezy, 1983;
Jessor et al., 1995; Grizenko & Fisher, 1992; Rak & Patterson, 1996; Werner, 1986;
Garmezy et al., 1984; Wyrnan et al., 1992; Seifer, Sameroﬂ‘, Baldwin, & Baldwin, 1992).
Garmezy (1983) states that protective factors are those characteristics that diminish the
likelihood of poor outcomes and encourage positive outcomes in individuals who are at
risk. Garmezy et al. (1984) argue that the eﬂ‘ects of protective factors, such as IQ, can be
conceptualized as “coping” mechanisms that emerge in the presence of stress and reduce
the negative impact of the risk factor. When individuals avoid negative outcomes despite
having been exposed to stress or risk factors, they are considered to be resilient (Rutter,
1987).

Jessor et al. (1995) state that protective factors can have both a direct effect on
behavior and moderate the relationship between risk factors and behavior. In this way, a
protective factor can decrease the likelihood of a poor outcome independent of the
presence of risk factors, as well as interact with risk factors to attenuate the relationship
between risk and negative outcomes (Jessor et al. 1995). For example, high child
intelligence not only may protect a child ﬁom negative outcomes but also may decrease

the negative impact of the risk factor of paternal alcoholism, thus altering the relationship

14

15

between risk and outcome.

While protective factors are hypothesized to aﬂ‘ect outcomes directly as well as to
inﬂuence the relationship between risk factors and outcomes, Rogosch et al. (1990) argue
that the distinction of the mediating or moderating role of protective factors often remains
unclear. These researchers assert that mediator variables account for the “how” or “why”
of the relationship between a risk factor and an outcome variable, whereas moderator
variables address the strength and direction of that relationship (Rogosch et al., 1990).
Rogosch et al. (1990) argue that many of the studies which have investigated the role of
protective factors in child development have not been designed adequately or have not
utilized appropriate statistical analyses, leaving the distinction of the mediating or
moderating role of protective factors unclear. In an attempt to correct the limitations of
prior research, these researchers investigated the role of protective child personality
variables in children of alcoholics and found that child characteristics, such as positive
temperament, should be considered moderators rather than mediators of the relationship
between risk and outcome (Rogosch et al., 1990).

Downey and Walker (1992) argue that interest in those characteristics which may
moderate risk often emerged from the observation that despite elevated risk for a poor
outcome given a risk factor, not all at-risk children inevitably develop a disorder. For
example, studies have found positive outcomes in children who are exposed to risk factors
such as parental psychopathology, marital discord, and low socioeconomic status (Werner,
1986; Seifer et al., 1992; Wyman et al., 1992; Warner et al., 1995; Katz & Gottman,
1993). Werner conducted a longitudinal study which followed over 600 children from

birth to the age of 32. Approximately one-third of the original sample of children were at

16
high-risk for poor outcomes due to poverty, perinatal stress, and a variety of family risk

factors, such as parental psychopathology and marital discord (Werner, 1992). However,
Werner (1992) reports that despite having encountered four or more risk factors, one-
third of the children classiﬁed as high-risk were resilient and developed into healthy,
competent adults. Similarly, Wyrnan et al. (1992) found that a signiﬁcant number of high-
risk urban children exhibited resilient outcomes despite having been exposed to a minimum
of four and an average of nine stressﬁrl life events and circumstances, such as parental
alcoholism, family discord, and poverty.

Researchers have attempted to identify those factors that serve a protective
function in the development of at-risk children who are resilient and do not experience
poor outcomes, such as behavior problems (Garmezy, 1983). Garmezy (1983) argued
that the protective factors identiﬁed by Various studies can be organized into three general
categories of factors: positive child characteristics, positive family environment, and
external support systems. Subsequently, this has been supported by numerous researchers
(Werner, 1986; Werner, 1990; Grizenko & Fisher, 1992; Garmezy, et al., 1984; Rak &
Patterson, 1996; Wyrnan et al., 1992; Chase-Landsdale, Wakschlag, & Brooks-Gum,
1995; Smith & Prior, 1995). The following discussion of protective factors will be
organized into two sections, child characteristics and family environment, based on the
categories set forth by Garmezy (1983). External support systems, although important to
the development of children, will not be discussed in this review, as these protective
factors typically are not available to children until they have reached school-age and have

access to sources of support independent of their immediate family.

 

17

Research consistently has identiﬁed several personal attributes as characteristic of
resilient children (Garmezy, 1983; Garmezy et al., 1984; Werner, 1986; Rogosch, Chassin,
& Sher, 1990; Rutter, 1987;1(yrios & Prior, 1991; Herrenkohl, Herrenkohl, & Egolf,
1994; Fergusson & Lynskey, 1996;.Masten et al., 1988; Smith & Prior, 1995; Wyman et
al., 1991; Tschann, et al., 1996; Seifer et al., 1992; Varni, Rubenfeld, Talbot, &
Setoguchi, 1990; Corbo-Richert, 1994). Two such attributes to be discussed in this
review due to their relevance to the development of externalizing behavior problems in
children are high intellectual ability and positive temperament characteristics. Studies have
found these factors to play a protective role in the development of children exposed to a
variety of risk factors, ranging from parental alcoholism (Werner, 1986; Rogosh et al.,
1990) to physical disorders (Lavigrre & Faier-Routman, 1993; Vami et al., 1990; Corbo-
Richert, 1994).

Intelligmg.

Research has found high intellectual ability (high IQ) to be associated with better
outcomes as well as to be protective in the development of children exposed to risk
(Garmezy et al., 1984; Masten et al., 1988; Herrenkohl et al., 1994; Fergusson & Lynskey,
1996). In a study using a community sample of children, Masten et al. (1988) investigated
the moderating effect of hypothesized protective factors on the relationship between stress
and children’s competence in school. The researchers utilized several competency
dimensions, including classroom disruptiveness, engagement in school activities, and
academic achievement, and found that the relationship between stress and children’s

competence was dependent upon individual characteristics as well as the competence

18
criterion. The results of this study revealed that high IQ protected against the adverse

impact of stress on the competency variable of classroom disruptiveness whereas low IQ
protected against the adverse impact of stress when the competency criterion of classroom
engagement was utilized (Masten et al., 1988). The researchers hypothesize that while a
high IQ protects children against the adverse effects of stress on classroom disruptiveness,
the effects of stress are apparent in the lives of high IQ children when the realm of
classroom engagement is considered, with the opposite outcome being found for low IQ
children (Masten et al., 1988).

Other researchers have investigated the protective function of high intellectual
ability in at-risk children’s development of emotional problems. Fergusson and Lynskey
(1996) conducted a sixteen-year longitudinal study of a birth cohort of 940 children in an
attempt to identify those factors which were characteristic of children who had been
exposed to high levels of family adversity, such as marital conﬂict and low SES, and did
not experience any externalizing behavior problems, such as substance use, conduct
disorder, and juvenile delinquency. The researchers found that resiliency in children was
associated with a higher IQ. Furthermore, similar to other studies addressing the impact
of multiple risk factors (i.e. Sameroﬂ‘ et al., 1993), Fergusson and Lynskey (1996)
addressed the impact of multiple protective factors by investigating the relationship
between child outcome and the combination of protective factors present in a child’s
development. The researchers report that while individual protective factors were only
modestly associated with resilient outcomes, a combined resiliency factor, consisting of
measures of family adversity, IQ at age eight, novelty seeking, and aﬂiliations with

delinquent peers, was strongly predictive of resilience. This ﬁnding is consistent with

 

19

those of Jessor et al. (1995), who found a summative index of protective factors to
moderate the relationship between risk and poor outcomes.

Although numerous studies have found high intellectual ability to serve a
protective ﬁrnction for children exposed to risk, Downey and Walker (1992) report that
the protective function of child characteristics needs to be considered in conjunction with
other variables. In a study which investigated the role of both child and family
characteristics in at-risk children’s development of aggression and depression, these
researchers found that IQ was not associated with outcomes when family-level inﬂuences,
such as parental psychopathology and child maltreatment, were controlled.

Temperament.

Studies have also found positive temperament to be associated with better
outcomes in children (Fergusson & Lynskey, 1996; Werner, 1986; Varni et al., 1990;
Corbo-Richert, 1994; Smith & Prior, 1995; Wyman et al., 1991; Tschann et al., 1996;
Kyrios & Prior, 1991). Characteristics of positive temperament include ﬂexibility, social
responsiveness, autonomy, and positive aﬁ‘ect. Wyman et al. (1991) investigated those
factors which were protective in the lives of high-risk urban children who had been
exposed to a minimum of four and an average of nine stressﬁrl life events and
circumstances, such as parental alcoholism, family discord, and poverty. The researchers
classiﬁed children as stress-resilient at ages 10-12 based on high scores on both parent and
teacher ratings of adjustment in domains such as classroom behavior and peer
relationships. The results of this study revealed that easy temperament ratings in infancy
diﬂ‘erentiated between stress-resistant and stress-aﬁected children (Wyman et al., 1991).

Similarly, Werner (1986) found those children of alcoholics who did not develop a coping

 

 

20
problem such as delinquency to be characterized by easy temperament and average to

above average intelligence.

Smith and Prior (1995) applied an intra-family‘ design to determine those
characteristics of school-age children which protected them from negative outcomes
despite high levels of familial psychosocial stress, such as ﬁnancial problems and drug
addiction. ‘The researchers classiﬁed children as resilient if all scores were in the non-
clinical range on both the CBCL and the Teacher’s Report Form in the areas of social
competence, home and school behavioral adjustment, and adaptive functioning (Smith &
Prior, 1995). Smith and Prior (1995) found that teachers tended to rate children more
positively than mothers. These researchers report that teacher ratings of a temperament
factor, consisting of characteristics such as easiness and likability, discriminated more
accurately between those children who were classiﬁed as resilient and those who
experienced behavior problems than any other variable investigated in this study, including
maternal warmth and the number of negative life events experienced by the child.

While many of the studies investigating the role of protective factors in children’s
outcomes have historically utilized school-age children, researchers have begun to examine
the protective impact of child characteristics on outcomes in preschool children (Kyrios &
Prior, 1991; Tschann et al., 1996). Tschann et al. (1996) found that pre-school children
with diﬁicult temperaments experienced more externalizing and internalizing behavior
problems overall than children with easy temperaments. They also report that the eﬂ‘ects
of family conﬂict on preschoolers behavior problems were moderated by an easy
temperament (Tschann et al., 1996). The researchers hypothesize that an easy

temperament may protect children in high-conﬂict environments by decreasing the

21
likelihood that they will be the target of negative responses from family members.

Similarly, Kyrios and Prior (1991) found that positive temperament characteristics
decreased the adverse effects of family risk factors, such as marital problems and parental
psychopathology, on preschool children’s behavioral adjustment, such as aggressiveness
and overactivity. A potential problem with this study is the degree of overlap which exists
between the measures of temperament and child behavioral adjustment, each of which
assess many of the same characteristics, such as manageability and activity level.
MW

Considerable research has also found characteristics of the family environment to
be associated with positive outcomes in children who are exposed to risk factors (Smith &
Prior, 1995; Seifer et al., 1992; Pettit & Bates, 1989; Grossman et al., 1992; Rubenstein,
Heeren, Housman, Rubin, & Stechler, 1989; Weist, Freedman, Paskewitz, Proescher, &
Flaherty, 1995; Bradley et al., 1995; Bradley, Caldwell, & Rock, 1988; Bradley &
Caldwell, 1981). Characteristics of the family environment which will be discussed in this
review are family cohesion, maternal warmth, and a stimulating physical and psychological
home environment, including factors such as cleanliness and availability of books. Similar
to the results found regarding protective child characteristics, studies have found family
environment factors to play a protective role in the development of children exposed to a
variety of risk factors, ranging from parental alcoholism (Zucker & Gomberg, 1986) to
pediatric medical problems (Phipps & Mulhem, 1995; Varni et al., 1990).

Fm]! Cohesion,

Research has found family cohesion to be associated with better outcomes and to

be protective in the development of children exposed to risk (Phipps & Mulhern, 1995;

 

22
Seifer et al., 1992; Pettit & Bates, 1989; Grossman et al., 1992; Rubenstein, Heeren,

Housman, Rubin, & Stechler, 1989; Weist, Freedman, Paskewitz, Proescher, & Flaherty,
1995). Family cohesion is a concept that addresses a family’s emotional closeness, mutual
supportiveness, and involvement in activities together. While a family environment
characterized by cohesion may be more likely to be characterized by the absence of family
risk factors such as family conﬂict, it is theoretically possible for a family to be cohesive
and to engage in conﬂictual interactions. In fact, Moos and Moos (1986) report that while
family cohesion and family conﬂict subscales of a family environment scale are modestly
inversely correlated (-.46), these subscales measure distinct, although related, aspects of
the family environment. Phipps and Mulhern (1995) investigated the relationship between
aspects of the family environment and resilience to the stress of bone marrow transplant in
children. These researchers report that-pediatric bone marrow transplant provides an
excellent opportunity for the study of resiliency in children because it is comprised of
several stressful components, including the isolation from ﬁiends and family which often
accompanies the procedure, the life-threatening nature of the procedure, and the duration
and intensity of the procedure (Phipps & Mulhern, 1995). Phipps and Mulhern (1995)
measured child adjustment in the areas of behavior problems, social competence, and self-
esteem both prior to hospital admission and 6-12 months after the procedure was
completed. The results of this study revealed that while family cohesion was not related to
measures of adjustment prior to hospitalization, it served as a protective factor against the
stress of bone marrow transplant and was predictive of children’s adjustment at follow-up.
Research has also found that the role of family cohesion as a protective factor may

vary by the gender of the child (Weist et al., 1995; Grossman et al., 1992). Weist et al.

23
( 1995) attempted to identify factors that played a protective role in the behavioral

adjustment and school performance of at-risk urban children. The results of this study
revealed that family cohesion was the only variable that protected boys exposed to
stressful life events against adverse consequences, such as externalizing behavior
problems. Family cohesion was not found to play a protective role in the lives of girls;
instead, it was associated with diﬁculties at school. The researchers hypothesized that
family cohesion did not appear to play a protective role in the lives of girls because girls
may be more likely to focus on social relationships during times of stress, and, therefore,
may encounter diﬁculties at school as their attention is diverted away from schoolwork
and towards relationships with family members and peers. Because the results of this
study were based on a cross-sectional sample, it is important to keep in mind that
inferences can not be made regarding the causal relationships among variables.

Grossman et al. (1992) also investigated the relationship between family variables
and positive outcomes in at-risk children. These researchers report that identiﬁed
protective factors, such as family cohesion and good communication with parents, did not
interact with and mitigate the negative eﬂ‘ects of risk factors, but rather, had a direct
impact on outcome independent of risk (Grossman et al., 1992). These researchers found
that family cohesion protected girls against delinquency, poor grades, distressed mood,
and low self-esteem. However, the results of this study revealed that family cohesion was
only protective against delinquency and low self-esteem for boys.

W

Studies have also found that a stimulating physical and psychological home

environment, including characteristics such as cleanliness, adequate living space, and the

24

provision of toys and reading materials, is associated with better outcomes in children
(Bradley et al., 1995; Bradley, Caldwell, & Rock, 1988; Bradley, Caldwell, Rock, &
Harris, 1986). Although oﬁen inversely related to SES, the quality of the home
environment has been found to be inﬂuential in children’s development independent of
SES (Bradley and Caldwell, 1984). Bradley et al. (1994) found that the home
environments of resilient children living in poverty who did not exhibit behavior problems
or developmental delays were more stimulating, safer, and less crowded than those of
comparably at—risk children who were not functioning adequately in these areas.

Bradley et al. (1988) conducted a study in which they assessed the home
environments of children when they were infants and then reassessed these characteristics
of the home when the children were 2 years old and when they were between 10 and 11
years of age. These researchers then investigated the relationship between aspects of the
home environment over time and children’s classroom behavior in areas such as hostility
and distractibility at the age of 10 or 11. They found that having a variety of
developmentally stimulating materials in the home environment at all three points in time
was signiﬁcantly related to children’s classroom behavior at the age of 10. Furthermore,
the researchers found that the organization of the physical environment at age 2 was
related to children’s classroom behavior at age 10.

Bradley et al. (1995) also investigated the relationship between aspects of the
home environment of low birth weight children and adaptive social behavior, such as
disruptiveness and compliance, over time. These researchers found that the quality and
the quantity of stimulation available to the child in the home environment, including a

variety of developmentally stimulating experiences and learning materials, was signiﬁcantly

 

25
predictive of adaptive social behavior in this at-risk group of children (Bradley et al.,

1995). Furthermore, Bradley et al. (1995) also reported that aspects of the home
environment appeared to be predictive of adaptive behavior independent of SES or family
structure.

Mategrﬂ Warmth.

Researchers have found another aspect of the family environment, maternal
warmth, to be associated with better outcomes and to be protective in the development of
children exposed to risk (Smith & Prior, 1995; Bradley et al., 1991; Pettit & Bates, 1989;
Bradley & Caldwell, 1981). Maternal warmth is a concept that addresses mother’s
positive aﬂ’ect directed at the child, such as pride, aﬁ‘ection, and positive physical contact.

In a study which utilized naturalistic observation methods to assess family
interaction patterns, Pettit and Bates (1989) investigated the relationship between aspects
of the family relationship and child behavior problems over time. These researchers
derived a composite variable, mother positivity, ﬁ'om measures of maternal affectionate
contact, aﬂ‘ectionate teaching, and verbal stimulation. The results of this study revealed
that mother positivity assessed over a child’s ﬁrst two years of life was protective against
behavior problems at the age of four (Pettit & Bates, 1989). Furthermore, Pettit and
Bates (1989) report that mother positivity early in life is predictive of future behavioral
outcomes independent of age for measures of maternal positivity. However, the results of
this study should be interpreted with caution, as an assessment of child behavior problems
independent of potentially confounding maternal variables was not obtained.

A study which had been discussed in this review in relation to the protective

function of positive temperament also found maternal warmth to be associated with

26
positive child outcomes (Smith & Prior, 1995). Smith and Prior (1995) utilized an intra-

farnily design to detemrine those characteristics of school-age children exposed to
psychosocial stressors which protected them from negative outcomes. These researchers
not only discovered that a positive temperament served a protective role in the lives of the
resilient at-risk children, but they also found that mother-child warmth, which assessed
mother’s verbal and emotional responsivity to the child, was associated with child
competence both at home and at school (Smith & Prior, 1995). The researchers argue
that the importance of a warm mother-child relationship is emphasized by the cross-

situational protection it provided in the lives of the resilient children in this study.

Relationship Between Risk and Protective Factors

Orthogonality of Constructs

Many researchers argue that protective factors, although often negatively
correlated with risk factors, are not merely at the opposite end of a continuum of risk from
risk factors, but are separate concepts which cart, and often do, coexist (Jessor et al.,
1995; Rutter, 1987). Jessor et al. (1995) state that while empirical studies often ﬁnd high
risk environments lacking in protective factors, it is possible for high protection to occur
with high risk. To illustrate this point, the researchers suggest that it is logically possible
for an adolescent to have delinquent friendships while maintaining high academic
involvement and achievement (Jessor et al., 1995). Jessor et al. (1995) argue that risk and
protective factors should therefore be conceptualized as orthogonal variables, despite the
fact that they are often experienced as inverse concepts. Tabachnick and Fidel] (1996)
deﬁne orthogonality as a “perfect nonassociation between variables,” with knowledge of
the value of one factor providing no information regarding the value of another factor (p.
9).

In order to provide empirical support for the conceptual argument that risk and
protection should be treated as discrete categories, Jessor et al. (1995) investigated the
relationship between risk and protective factors for adolescent problem behavior in a

sample recruited from the 7th, 8th, and 9th grades of a large metropolitan school district.

27

28
The researchers report that the initial participation rate for their study was not optimal, as

only 67% of the middle school students and 49% of the high school students participated
(Jessor et al., 1995). Jessor et al. (1995) utilized several statistical methods to explore the
relationship between risk and protective variables and to provide support for their
hypothesis that these factors should be treated as orthogonal variables rather than as
inverse correlates. First, the researchers investigated the proportion of variance shared by
a summative index of risk and a summative index of protection and found that these
indexes shared only a modest proportion (18%) of common variance. The researchers
then investigated the degree to which individual risk and protective factors which seemed
to be on opposite ends of a continuum, such as exposure to ﬁiends who model problem
behavior and exposure to ﬁiends who model conventional behavior, were correlated and
found only a modest relationship (r = -.20). Additionally, Jessor et al. (1995) found that
these potentially nonorthogonal individual risk and protective factors retained signiﬁcant
beta weights at the ﬁnal step of hierarchical multiple regressions for outcome variables
such as delinquent behavior and problem drinking. Furthermore, Jessor et al. (1995)
found that the degree to which risk and protective variables correlated with other
measures, including both independent and outcome variables, varied. The researchers
conclude that high risk does not necessarily imply low protection and, therefore, a
distinction between protection and risk remains a valuable objective.

However, Jessor et al. (1995) also report that the generalizability of their ﬁndings
may be limited by diﬁiculties with their sample, such as a low initial participation rate and
a high attrition rate. The researchers state that while conceptually a distinction between

risk and protection appears meaningful and empirically their study provides preliminary

 

29
support for this diﬂ‘erentiation, ﬁrrther research utilizing samples which include extreme

scores on both risk and protective factors would provide additional information regarding
the relationship between these constructs (Jessor et al., 1995).
Implications of Sm]; Vm’abilig

While Jessor et al.'s (1995) study provides support for the coexistence of risk and
protection in some populations, the issues raised by the researchers regarding problems
with their sample and consequently the generalizability of their results should not be
dismissed. Sher and Trull (1996) emphasize that although many researchers often ﬁnd it
necessary to utilize convenience samples, the population from which a sample is chosen
has many important implications for the research ﬁndings. These researchers caution that
nonprobability sampling not only is often subject to selection bias but also jeopardizes the
external validity of research ﬁndings (site: & Trull, 1996). It seems that adolescents
engaging in problem behaviors such as delinquency and substance use, the outcome
variables utilized by Jessor et al. (1995), may not be accurately represented in a volunteer
sample of students.

Additionally, Kazdin and Kagan (1994) argue that risk factors often do not exist in
isolation but rather are interrelated or appear in “packages” (p. 39). For example, these
researchers state that the presence of a risk factor for child psychopathology, such as low
SES, increases the likelihood of the presence of other child risk factors, such as poor
parenting, as these factors are frequently positively related. As those adolescents
characterized by extremely high risk scores may not have volunteered or obtained parental

consent to participate in Jessor et al.'s (1995) study, the generalizability of these ﬁndings

to high-risk samples is potentially limited.

30
Kazdin and Kagan (1994) state that developmental psychopathology research

typically investigates main effects and linear relationships between variables such as risk
factors and outcomes. However, the researchers argue that these assumptions are often
violated, as many nonlinear relationships and interaction effects have been identiﬁed in
studies of development and psychopathology. They therefore suggest that the impact of
multiple variables on outcome be conceptualized nonlinearly in order to more accurately
capture the relationship among variables. However, McClelland and Judd (1993) state
that there often exist signiﬁcant statistical dimculties in detecting interactions and
moderator eﬁ‘ects in ﬁeld studies which often do not exist in experiments. The researchers
argue that often the distributions of independent variables in ﬁeld studies are more
restricted, resulting in lower residual variance of the interaction term. In this way, it can
be seen that restrictions in the extreme'ranges of variables such as risk factors in ﬁeld
studies can produce results which do not accurately reﬂect relationships between
variables, such as moderator effects.

While the possibility of diﬂ‘erential relationships between risk and protection over a
range of risk severity could not be determined in the sample utilized in the Jessor et al.
(1995) study, a study conducted by Bernstein and Hans (1994) illustrates the potential
implications of a restricted range of risk on a study’s ﬁndings. Bernstein and Hans (1994)
found that for a high-risk group, determined by prenatal exposure to methadone, a
cumulative risk variable predicted four developmental child outcomes, such as child
attention and cooperativeness, although this variable did not predict child outcome in the
low-risk comparison group. This study revealed consistently higher correlations between

the high-risk group’s cumulative risk index and child outcome measures than for the low-

 

3 1
risk group or the entire sample as a whole (Bernstein & Hans, 1994). In discussing this

ﬁnding, the researchers argue that the higher correlations for the high-risk group result
ﬁ'om this group’s “greater range of risk factors,” as there existed greater outcome
predictability for those children on the extreme ends of risk, with those children
categorized as high-risk who had fewer risk factors than the comparison group exhibiting
predictable resilient outcomes while those who had more risk factors than any comparison
child consistently displaying very poor outcomes at follow-up (Bernstein & Hans, 1994, p.
355)

In a study utilizing a high risk sample of children of alcoholics and a control
sample, Moses (1992) investigated the relationship between paternal alcohol problems and
child behavior problems, as well as the moderating eﬁ‘ects of contextual risk variables,
such as parental antisocial behavior and. depression, and contextual protective factors,
such as SES and parental intelligence. This study found that while these contextual risk
and protective variables all demonstrated signiﬁcant main eﬂ‘ects on child behavior
problems, contextual protective factors did not moderate the relationship between paternal
. alcohol problems and child outcome. However, this study did not include many of the
protective factors identiﬁed by previous research (i.e. Garmezy, 1983), such as child
characteristics or aspects of the family environment other than SES or parental IQ.

A signiﬁcant interaction was found between parental psychopathology and paternal
alcohol problems, with alcohol problems demonstrating greater predictability of child
outcome when accompanied by low parent psychopathology. Moses (1992) argues that
this ﬁnding of a decreased predictability of child behavior problems from paternal alcohol

problems as parental psychopathology increases may result from the increased levels of

32
family problems attributable to the presence of parental psychopathology. In this way, the

impact on the family environment and child outcome which the presence of paternal
alcohol problems can contribute independently is attenuated by the effect of additional
psychopathology on these same variables. Although Moses (1992) found statistically
signiﬁcant, low-order correlations between risk and protective factors in this high-risk
population, this study did not investigate the relationship between these factors over
varying levels of parental psychopathology.

Studies utilizing subjects representing extreme risk or extreme protection may
reveal that there exists a differential relationship between risk and protection over the
continuum of risk, with these constructs beconring decreasingly independent as risk
increases. Additionally, it may be that outcomes are more predictable ﬁcm measures of
risk and protection in those subpopulations on the extremes of the continuum of risk than

for those subjects in the middle range.

Hypotheses

The literature review above indicates that a variety of factors may place children at
risk for externalizing behavior problems, increase the likelihood of positive outcomes, or
diminish the likelihood of externalizing behavior problems in individuals who are at risk.
Whereas researchers (e.g., Sameroﬁ’ et al., 1993; Jessor et al., 1995) have found that the
amount of risk and protection to which children are exposed is signiﬁcantly related to
outcomes, prior research typically has not investigated the relationship between risk and
protection or examined diﬁ'erences in outcome predictability over a range of risk statuses.
Additionally, research generally has not adequately addressed the roles of mediating and
moderating variables. As a result, an understanding of the relationship between risk and
protective factors and children’s outcomes remains unclear.

The current investigation will examine the relationship between risk and protective
variables, potential mediating and moderating eﬂ’ects of these variables on child
externalizing behavior problems, and variability in the proportion of outcome variance
accounted for over a range of risk statuses in a longitudinal, community-based sample of
male children of alcoholics and demographically-matched male children of non-alcoholics.
Several characteristics of the family and the environment when the targeted children are
between the ages of three and ﬁve (time 1), which the literature suggests are associated
with parental alcoholism and poor child outcomes, are expected to increase the likelihood

of child externalizing behavior problems three years later when the children are between

33

34
the ages of six and eight (time 2) and will be referred to as risk factors. Additionally,

several characteristics of the child at time 1 which have been associated with resiliency in
children are expected to diminish the likelihood of child externalizing behavior problems at
time 2 independent of the presence of risk factors as well as to moderate the relationship
between risk and negative outcomes. Similarly, characteristics of the family environment
at time 1 which have been associated with resiliency in children are expected to be
associated with more positive child outcomes and to account for the effects of risk on
child behavior problems. These characteristics of the child and the family environment
which are expected to be associated with lower levels of child externalizing behavior
problems will be referred to as protective factors. A cumulative risk index will be
calculated by summing the number of risk factors impacting each child, as determined by
criteria speciﬁed for each risk variable in this study’s description of measures. A model of
the hypothesized relationships among variables is provided in Figure 1. Several speciﬁc
hypotheses have been developed and are described below.
Predicted Main Eﬁ‘ects

1) Numerous studies have found paternal psychopathology, maternal
psychopathology, low SES, and marital conﬂict to place a child at risk for negative
outcomes (e.g., Hill & Muka, 1996; Jansen et al., 1995; Werner, 1986). The current
investigation expects to ﬁnd positive main effects between risk factors at time 1 and
externalizing behavior problems at time 2.

2) It is expected that risk factors will be negatively related to family environment
protective factors, with the strength of this relationship increasing with increased levels of

risk. For example, SES is expected to be negatively related to a stimulating home

35
environment, although it is anticipated that the strength of the relationship between these

variables will decrease with the level of risk.

3) Numerous studies have found high child IQ and positive child temperament to
be associated with positive outcomes (e. g., Garmezy et al., 1984; Fergusson & Lynskey,
1996; Smith & Prior, 1995; Werner, 1986). The current investigation expects to ﬁnd that
child protective factors at time 1 have a direct inverse effect on child externalizing
behavior problems at time 2.

4) Previous research has found family cohesion, a stimulating home environment,
and maternal warmth to be associated with positive outcomes (e.g., Phipps & Mulhern,
1995; Bradley et al., 1995; Smith & Prior, 1995). The current investigation expects to
ﬁnd that family environment protective factors at time 1 are negatively related to child
externalizing behavior problems at tirneg2.

Predictg Modgatm’ g Eﬂ‘ects

5) Researchers have found that protective child characteristics, such as positive
temperament, moderate rather than mediate the relationship between risk and outcome
(e.g., Rogosch et al., 1990). This investigation hypothesizes that the statistical interaction
of time 1 child protective factors with risk factors will moderate the main eﬁ‘ects of risk on
child externalizing behavior problems at time 2. The relationship between risk factors and
child externalizing behavior problems is expected to be greater for children low on a child
protective characteristic than for children high on a child protective characteristic.
Prﬂictﬂ Mediating Eﬁ‘ects

6) As outlined above, family environment protective factors are hypothesized to

be inversely related to child externalizing behavior problems, and risk factors are expected

36
to predict both child externalizing behavior problems and family environment protective

factors. It is anticipated that the strength of the relationship between risk factors and child
externalizing behavior problems will be signiﬁcantly weakened when the eﬁ’ects of family
environment protective factors are considered. Therefore, time 1 protective family
environment factors are hypothesized to mediate the main effect of risk factors on child
externalizing behavior problems at time 2 (Baron & Kenny, 1986; Rogosch et al., 1990).
Predicted Effects of Risk Level

7) It is hypothesized that the strength of the relationship between child
externalizing behavior problems and the proposed risk variables will diﬁ‘er over the
continuum of risk, with the predictability of child externalizing behavior problems
increasing at the extremes on the continuum of risk. A graph of the hypothesized
predictability of child externalizing behavior problems over the continuum of risk is
provided in Figure 2.

8) The current investigation expects to ﬁnd differential relationships between risk
and both child and family environment protection along the continuum of risk, with these
constructs becoming decreasingly independent as risk increases. More speciﬁcally,
subpopulations at the high end of risk are expected to have greater inverse correlations of
risk and protection than subpopulations at the low end of risk. A graph of the
hypothesized relationship between risk and protection over the continuum of risk is

provided in Figure 3.

Methods

$1M

Subjects in the current investigation are 220 families participating in the Michigan
State University-University of Michigan Longitudinal Study (MSU-UM; Fitzgerald et al.,
1993; Zucker & Fitzgerald, 1991). This ongoing study is comprised Of two population-
based samples of alcoholic families and a contrast sample of non-alcoholic families, all of
whom had a biological son between the ages of 3 and 6 at the beginning of this study. The
samme of children utilized in this study is comprised entirely of Caucasian males. As the
original focus of this longitudinal project was on the development of alcohol problems,
behavior problems, and related symptomatology in a high risk sample of children of
alcoholics, this study concentrated on males due to their increased risk for the
development of these diﬁculties (Zucker, Fitzgerald, & Moses, 1995). Given literature
supporting the relationship between patterns of alcohol involvement and ethnic/racial
status, the limited ethnic/racial composition of this sample was determined by this study’s
locale and the resulting inability to eﬂ’ectively consider this covariate, as census data
revealed that less than 10% of the population in the area where data collection took place
consisted of other ethnic and racial groups. All recruited families had been invited to
participate in a long term study of family health and child development.

The ﬁrst group of alcoholic families was recruited from the population of all

convicted male drunk drivers in four counties in the Mid-Michigan area having a blood

37

38
alcohol concentration of at least 0.15% or a blood alcohol concentration ofO. 12% with a

history of prior alcohol-related arrests. Additional inclusionary criteria for this study
were: the men were living with a biological son between 3 and 5 years of age (the target
child); the men were living with the target child’s mother; and the men met Feighner et al.
(1972) criteria for a probable or deﬁnite diagnosis of alcoholism both during recruitment,
based on the Short Michigan Alcohol Screening Test (SMAST, Selzer, 1975), and during
data collection, based on the Diagnostic Interview Schedule, Version 111 (DIS; Robins,
Helzer, Crougharr, & Ratcliﬁ‘e, 1980).

The second group of alcoholics were recruited during efforts to identify a
community contrast sample for this study. Door-to-door community canvassing of the
neighborhoods in which the court-recruited alcoholics lived was used to recruit contrast
families within the same census tract who also had two parents and a biological son
between the ages of 3 and 6 residing within the home (Fitzgerald et al., 1995). Those
families identiﬁed through the door-to-door canvassing who met study criteria and had a
father who met criteria for alcoholism but did not have a history of drunk driving
convictions were placed in a second group of alcoholic families, the community alcoholic
group. Inclusion in the contrast sample was contingent upon neither of the parents
meeting F eighner criteria for alcoholism or other drug abuse/dependence. Although
mother’s alcoholism status was assessed for all groups, it was neither a requirement or a
basis for exclusion except for in the control sample, in which neither parent could meet
criteria for a substance problem. As the original sample was recruited from the population
of all convicted male drunk drivers in four counties and the contrast samples were

recruited ﬁ'om the neighborhoods of the court-recruited alcoholics, this sample is

 

39

comprised of families who represent various points along continuums of both individual
risk factors, such as SES, as well as overall risk.
Mars

The current investigation utilizes two waves of data ﬁ'om the MSU-UM
Longitudinal Study. The ﬁrst wave of data was collected ﬁ'om the families when the
biological sons targeted for this study were between the ages of 3 and 5 years. The
second wave of data was obtained approximately three years later when the targeted
children were between 6 and 8 years of age.

Each family participating in this study completes numerous questionnaires, direct
observation sessions, and interviews during each wave of data collection. The data are
collected by trained project staﬂ‘ who are blind to family risk status over the course of nine
contacts, involving approximately 15 hOurs of time. Seven of the data collection contacts
occur in the family home, and two occur on a university campus. F arnilies receive
ﬁnancial compensation for their participation in the longitudinal project.
eres

The current investigation utilizes a subset of the battery of questionnaires collected
during the ﬁrst two waves of the MSU-UM Longitudinal Study. Risk factors examined
were assessed at time 1 and include the following: paternal psychopathology, as measured
by alcoholism, depression, and antisociality; maternal psychopathology, as measured
alcoholism, depression, and antisociality; low socioeconomic status; and marital conﬂict.
Protective factors examined were collected during time 1 and include positive child
characteristics, as measured by high child intelligence and positive child temperament, and

positive family environment characteristics, as measured by family cohesion, maternal

40
warmth, and stimulation provided by the physical and psychologic a1 home environment.

The dependent variable, child externalizing behavior problems, was measured at time 2.
The speciﬁc measures used in the current investigation are described below. Additionally,
a cumulative index of risk was created for each subject by counting the total number of
risk factors which were present (Sameroﬁ‘ et al., 1987; Bernstein & Hans, 1994). Each
risk variable was dichotomized according to criteria speciﬁed below in order to determine
the presence or absence of that risk factor for each subject. As eight risk factors were
investigated in this study, the cumulative risk index ranged from zero to eight. (For
example, if one subject’s maternal depression, paternal alcoholism, and family conﬂict
scores were all above the cut-off values outlined in the discussion of measures below
and the levels of the other risk factors were below their respective cut-offs, the
cumulative risk index for this subject would equal three). Similarly, a cumulative index
of protection was created for each subject by counting the total number of protective
factors which were present for each subject based upon the cut-offs for each protective
variable outlined below (F ergusson & Lynskey, 1996). As ﬁve protective factors were
investigated in this study, the cumulative protection index ranged from zero to ﬁve.

Measures of Risk

 

Short Michigan Alcohol Screening Test (SMAST, Selzer, 1975). As outlined in
the discussion of subjects, the SMAST is used to classify fathers and mothers as alcoholic
or nonalcoholic prior to their participation in this study based upon Feighner et al. (1972)
criteria for a probable or deﬁnite diagnosis of alcoholism. Questions such as “Do you feel
you are a normal drinker” and “Are you able to stop drinking when you want to” are

answered in a yes/no format. This instrument is a shortened l3-item version of the

41
Michigan Alcoholism Screening Test (MAST) and has been found to be a reliable and

valid screening test for alcoholism (Selzer, 1975). A reliability coeﬁcient alpha of .93 has
been reported, and studies have found scores on this measure to discriminate between
groups of alcoholics and nonalcoholics (Selzer, 197 5).

Disgpestie Intﬂ'ew Schﬂple, Versien III (DIS) (Robins, Helzer, Croughan, &
Ratcliﬁ‘e, 1980). As outlined in the discussion of subjects, father’s and mother’s responses
on the DIS are used to verify the initial classiﬁcation of alcoholic or nonalcoholic based
upon Feighner et al. (1972) criteria fora probable or deﬁnite diagnosis of alcoholism.

This instrument is a highly structured interview which is designed to assess
symptomatology in a range of areas and to make diagnoses based upon DSM-IIIR,
Feighner criteria, and Research Diagnostic Criteria. The DIS has been found to have
adequate reliability and validity, with a reported mean kappa of .69, a mean sensitivity of
75%, and a mean speciﬁcity of 94 % across all DSM-IIIR diagnoses (Robins et al., 1980).
Paternal and/or maternal alcoholism were included in the cumulative risk index based upon
each parent’s classiﬁcation as alcoholic or nonalcoholic.

Beek Depression Invmteg, Short Form (B_DI) (Beck, Steer, & Garbin, 1988).
This self-report questionnaire consists of 13 items that ask respondents to indicate the
degree to which they are currently experiencing depressive symptomatology, such as
crying easily, appetite disturbances, and sleep disturbances. The BDI has been found to be
a valid and reliable test for depression, with a reported mean coeﬁcient alpha of .81 for
nonpsychiatric samples (Beck, Steer, & Garbin, 1988). Studies have found scores on the
short form of this instrument to correlate between .89 and .97 with the long form (Beck,

Steer, Garbin, 1988). Paternal and maternal depression were included in the cumulative

42
risk index based upon a score of four or greater, as this cut-oﬁ‘ has been established for

this instrument as a classiﬁcation of mild to more severe depression.

Antiseg'ﬂ Behavier Chgklig (Zucker & Noll, 1980). This self-report
questionnaire consists of 46 items that ask respondents to indicate the ﬁ'equency with
which they have participated in delinquent, criminal, and antisocial activities throughout
their lifetime. This instrument has been found to have adequate test-retest reliability (.91
over four weeks) and internal consistency (reliability coeﬁcient alpha = .93) in a series of
studies with populations ranging ﬁom male and female college students to male and
female inmates (Zucker & Noll, 1980; Zucker, Noll, Ham, Fitzgerald, & Sullivan, 1994).
Additionally, these studies have found this instrument to differentiate between individuals
with long histories of antisocial behavior (prisoners), individuals with minor oﬂ’enses in
district court, and university students (Zucker et al., 1994). Utilizing previously
established criteria for this instrument, paternal antisociality was included in the
cumulative risk index based upon a cut-oﬁ‘ score of 24, and maternal antisociality was
included based upon a cut-off score of 18 (Bingham, Zucker, & Fitzgerald, 1996; Caplan,
1996).

Revised Duncan Seciogonomic [neg (TSE12) (Stevens & Featherman, 1981;
Mueller & Parcel, 1981). This instrument is an index of occupational attainment and was
used to measure father’s and mother’s Socioeconomic Status (SES), with family SES
being deﬁned in the current study as the average of father’s and mother’s TSE12 scores or
the principle earner’s SES when there exists only one principle earner in the family
(Hauser, 1994). Based upon research which has indicated that occupation-based measures

of SES are the most reliable and valid single measures of SES, the TSE12 was utilized in

43
this study to determine family SES rather than a measure of income (Mueller & Parcel,

1981). SES was included in the cumulative risk index based upon a score below the
median split of the distribution of this variable in this sample (Bingham, Zucker, &
Fitzgerald, 1996). Bingharn et a1. (1996) argue that the use of the median as a breakpoint
for the determination of risk versus lack of risk produces a measure of cumulative risk
which more accurately represents the total range of risk experienced by the sample.
Additionally, the researchers state that this procedure is less likely than extreme groups
approaches to result in heteroscadastic group variances or to truncate variance (Bingharn
et al., 1996).

Fm’ly Envirenmmt Selle (ES) (Moos & Moos, 1986). The FES is a self-report
questionnaire that consists of 90 true-false items designed to measure the social-
environmental attributes of the family. This instrument has ten subscales which assess the
family on three broader dimensions: relationship, personal growth, and system
maintenance. The FES subscale, conﬂict, from the relationship dimension of this
instrument measures the amount of anger, aggression, and conﬂict openly expressed
- among family members. Items from the conﬂict subscale include “We ﬁght alot in our
family” and “Family members sometimes hit each other.” Family conﬂict is deﬁned in the
current study as mother’s FES conﬂict subscale score. This measure has been found to
have acceptable test-retest reliability, ranging from .68 to .86 for the various subscales, as
well as adequate internal consistency (Moos & Moos, 1986). Additionally, the FES has
been found to successfully discriminate between distressed and normal families (Moos &
Moos, 1986). Conﬂict was included in the cumulative risk index based upon a score

above the median split of the distribution of this variable in this sample (Bingham, Zucker,

& Fitzgerald, 1996).
Mees_ures of Child Protection.
Stanford-Bing Intelligence Sge, Third Revision (S-BIS; Tennan & Merrill,

1973). The S-BIS is an individually administered instrument which consists of subtests
designed to measure overall level of intellectual functioning. The child’s mental age is
compared with his or her chronological age, and a normalized standard score with a mean
of 100 and a standard deviation of 16 is obtained to represent an overall intelligence level
(IQ). ' The S-BIS has been shown to be reliable and valid and was included in the
cumulative risk index based upon a score below the standardized mean for this instrument
(I‘erman & Merrill, 1973).

The Dimensions of Temperament Survey (DOTS-Child; Lerner, Palermo, Spiro, &
Nesselroade, 1982). The DOTS-Child is a 34-item instrument which assesses child
temperament on ﬁve dimensions: activity level, attention span/distractibility, approach-
withdrawal/adaptability, rhythmicity, and reactivity. Items on the DOTS are answered as
either “more true than false” or “more false than true” and include items such as the
following: “My child moves a great deal in his/her sleep” (activity level), “Once my child is
involved in a task, he/she can’t be distracted away from it” (attention span/distractibility),
“It takes my child a long time to get used to new people” (approach-
withdrawal/adaptability), ”My child seems to get sleepy just about the same time every
ni t” (rhythmicity), and “When my child reacts to something, his/her reaction is intense”
(reactivity).

The dimensions of temperament contained within the DOTS-Child have been

found to be reliable and valid (Lerner et al., 1982). Samples of infants, preschoolers,

 

45
school-aged children, and young adults have revealed reliability coeﬂicients on the various

subscales ranging ﬁom .31 to .96, with reactivity being the only factor found to be
consistently below .60 (Lerner et al., 1982). In studies utilizing samples of preschool-age
children, Lerner et al. (1982) report that the alpha coefﬁcients for each of the ﬁve DOTS
scales are as follows: .96 for activity level, .86 for attention span/distractibility, .81 for
approach-withdrawal/adaptability, .76 for rhythmicity, and .57 for reactivity. Additionally,
studies have found temperament ratings on the DOTS-Child to be associated with grades,
self-esteem, self-concept, depression, and peer relations in samples of children and
adolescents (Windle et al., 1986). Positive temperament characteristics include low
activity level, low reactivity, high attention span, high rhythmicity, and high
approach/adaptability. Child temperament is deﬁned in the current study as mother’s
DOTS-Child scores. Subjects characterized by scores indicative of a positive ’
temperament (median split) on three of the ﬁve dimensions were included in the

cumulative protection index (Binghanr, Zucker, & Fitzgerald, 1996).

Measures of Family Environment Protection.
Family Environment che (F_ES) (Moos & Moos, 1986). Family cohesion is

measured using the same instrument described above for the measurement of conﬂict. The
FES subscale, cohesion, from the relationship dimension of this instrument measures a
family’s mutual supportiveness and commitment to each other. Items included in the
family cohesion subscale of the FES include “Family members really help and support one
another” and “There is a feeling of togetherness in our family.” This measure has been
found to have acceptable test-retest reliability, ranging ﬁom .68 to .86 for the various

subscales, as well as adequate internal consistency (Moos & Moos, 1986). Additionally,

 

46
the FES has been found to successﬁrlly discrirrrinate between distressed and normal

families (Moos & Moos, 1986). Family cohesion is deﬁned in the current study as
mother’s cohesion subscale score, with inclusion in the cumulative protection index based
upon a median split of this variable’s distribution in this sample (Bingham, Zucker, &
Fitzgerald, 1996).

TheHm 'onforM r oftheEnvir nmen-Pr h [V in
(HOME; Bradley & Caldwell, 1979). The HOME is a 54-item inventory which assesses
eight aspects of the social, emotional, and cognitive support available in the home
environment. Two subscales of the HOME are utilized in this study to assess the quality
of the home environment and maternal warmth: a safe, clean physical environment
conducive to development and maternal pride, aﬂ’ection, and warmth. Items ﬁ'om these
subscales include “Building has no potentially dangerous structural or health defect” and
“Parent holds child close ten to ﬁfteen minutes per day” and are assessed as either true of
false of the home environment through direct observations and parent interviews. This
instrument has been found to be a reliable and valid measure of the home environment
(Bradley & Caldwell, 1979). Internal consistency for the subscales reportedly range ﬁom
.53 to .83, with an estimate of .93 for the total scale, and concurrent and predictive
validity studies revealed a signiﬁcant association between scores on the HOME and child
IQ (Bradley & Caldwell, 1979). Maternal warmth and a stimulating home environment
were included in the cumulative protection index based upon a score above the median
split of the distributions of these variables in this sample (Bingham, Zucker, & Fitzgerald,

1996)

47
Measure of Child Outcome.

Child Behavior Checklist (CBCL; Achenbach & Edelbrock, 1986). The CBCL is a
standardized instrument which obtains parent’s reports of a range of child behavior
problems and competencies. The scoring of this instrument yields a total behavior
problems scale, two broad-band subscales, and eight narrow-band subscales, as well as
information regarding social competence. This study utilized the broad-band subscale,
externalizing behavior problems, which are characterized by aggression and/or delinquent
behaviors. Mother’s CBCL scores were used to assess child behavior problems. External
validity of the CBCL has been found to be adequate, with studies ﬁnding parental CBCL
ratings to be associated with independent raters’ perceptions of the child (Bird, Gould,
Rubio-Stipec, Staghezza, & Canino 1991). Additionally, test-retest reliability of the
CBCL has been found to range ﬁom .95 over a two-week period to .84 over a three-week

period (Achenbach, 1991).

Results

Missing Date Ed Qyjliers

Prior to conducting the analyses, all variables were screened for outliers and
missing data. Less than two percent of the values for any variable were initially missing,
and regression analyses were run on the available data based upon recruitment risk group
to estimate missing data. In screening for outliers, a normal curve was superimposed on a
histogram of the frequency distribution for each continuous variable, and those
nonadjacent values falling outside of the normal distribution were considered to be
outliers. Values adjacent to the closest. non-outlier value were assigned for each outlier in
the order of original ranking, therefore maintaining the rank order of subjects for each
variable. A maximum of two outlying values were altered for any variable.
Descriptive Statistics

The descriptive statistics for the hypothesized risk factors, child protective factors,
and family protective factors are presented in Table 1, as well as the descriptive statistics
for the dependent child outcome variable, mother’s report of child externalizing behavior
problems.
Multiple Regzession Anelyses

Mein Eﬂ‘ects

Hypothesis one predicted that the risk variables would be signiﬁcantly related to

child externalizing behavior problems. To test this hypothesis, risk variables were entered

48

49
into a regression equation as independent variables to predict child externalizing behavior

problems. The results of this regression, shown in Table 2, indicate that the overall
equation was signiﬁcant and explained 18% (15% adjusted R2) of the variance in the
dependent variable, child externalizing behavior problems. Although only mother’s
antisocial behavior checklist score contributed signiﬁcantly to prediction of child
externalizing behavior problems ([3=.18, senripartial r2(unique)=.023, p<.02), the
hypothesized independent risk variables in combination contributed another 16% in shared
variability. As many of the independent risk variables in this analysis are signiﬁcantly
correlated with each other, the importance of individual risk factors in the prediction of
the dependent variable is diﬂicult to discern due to the amount of shared variance

(T abachnick & Fidel], 1996). F urtherrnore, the cumulative risk index signiﬁcantly
predicted child externalizing behavior problems, accounting for 13% of the variance in
outcome (Table 2).

Hypothesis two predicted that the risk factors would be negatively related to the
family environment protective factors. To test this hypothesis, separate regression
analyses were conducted for each of the family environment protective factors. T1
measures of paternal alcoholism, paternal depression, paternal antisociality, maternal
alcoholism, maternal depression, maternal antisociality, low SES, and marital conﬂict were
simultaneously entered into the regression equation to predict Tl measures of family
cohesion. The results of this regression, shown in Table 3, indicate that the overall
equation was signiﬁcant and explained 30% (28% adjusted R2) of the variance in the
dependent variable, family cohesion. Although only three of the independent variables,

family conﬂict and mother and father’s depression scores, contributed signiﬁcantly to

50
prediction of family cohesion (family conﬂict: B=-.29, semipartial r2(unique)=.069,

p<.0001; maternal depression: [3=-.37, semipartial r2(unique)=.1 l3, p<.0001; paternal
depression: B=. 15, semipartial r2(unique)=.016, p<.029), the hypothesized independent
risk variables in combination contributed another 10% in shared variability. Similar to the
testing of hypothesis one, the importance of individual risk factors in the prediction of the
dependent variable is diﬁcult to discern due to the amount of shared variance among the
independent variables.

To test the relationship between the risk factors and the remaining family
environment protective factors, two separate analyses utilizing the same independent
variables were then conducted to predict stimulating home environment and maternal
warmth as dependent variables. As indicated in Table 4, the overall equation signiﬁcantly
predicted stimulating home environment and explained 9% (5% adjustai R2) of the
variance in the dependent variable. Paternal antisocial behavior score was the only
variable that contributed signiﬁcantly to the prediction of stimulating home environment
(B=-.17, semipartial r2(unique)=.019, p<.039), although the independent risk variables in
combination contributed another 7% in shared variability. Although the overall regression
equation predicting maternal warmth was not signiﬁcant, maternal depression was a
signiﬁcant predictor of maternal warmth (Table 5).

Hypothesis three predicted a direct inverse relationship between the child
protective factor variables and child outcome. To test this hypothesis, a regression
analysis was conducted with child externalizing behavior problems as the dependent
variable. T1 measures of child IQ and the ﬁve child temperament scales (attention,

approach~withdrawal/adaptability, activity, rhythmicity, and reactivity) were

51

simultaneously entered into the regression equation to predict T2 measures of
externalizing behavior problems. The results of this analysis are shown in Table 2. The
overall equation was signiﬁcant and explained 6% (3% adjusted R2) of the variance in the
dependent variable, child externalizing behavior problems. However, none of the
independent variables contributed signiﬁcant unique variance to the prediction of child
externalizing behavior problems.

Hypothesis four predicted a direct inverse relationship between the family
protective variables and child externalizing behavior problems. To test this hypothesis, Tl
' measures of family cohesion, stimulating home environment, and maternal warmth were
simultaneously entered into a regression equation to predict T2 measures of externalizing
behavior problems. As indicated in Table 2, the overall equation signiﬁcantly predicted
child externalizing behavior problems and explained 5% (4% adjusted R2) of the variance
in the dependent variable. Both maternal warmth (B=-.14, semipartial r2(unique)=.012,
p<.049) and family cohesion (B=~.14, semipartial r2(uniqueF.02, p<.037) contributed
signiﬁcantly to the prediction of child externalizing behavior problems, with the three
family protective variables in combination contributing approximately another 2% in
shared variability. Furthermore, the cumulative protection index, which was a summative
index of both child and family environment protection, signiﬁcantly predicted child
externalizing behavior problems, accounting for 6% of the variance in child outcome
(Table 2).

W

Hypothesis ﬁve predicted that child protective factors, IQ and temperament, would

moderate the main effects of risk on child outcome. To assess this hypothesis, a two-step

52

regression analysis was conducted for each child protective factor. First, T1 measures of
the hypothesized risk factors (paternal alcoholism, paternal depression, paternal
antisociality, maternal alcoholism, maternal depression, maternal antisociality, low SES,
conﬂict) and child IQ were simultaneously entered in step 1 into the regression equation to
predict T2 measures of externalizing behavior problems. Next, the interaction terms for
child IQ and each of the risk variables were entered into the regression equation (i.e., child
IQ X paternal alcoholism, child IQ X paternal depression... child IQ X SES, and child IQ
X conﬂict) as the second step. The results of this analysis are summarized in Table 6 and
revealed that while there were signiﬁcant main eﬂ‘ects of the risk variables and child IQ on
child externalizing behavior problems, the interaction terms did not explain a signiﬁcant
amount of additional variance in child outcome beyond that accounted for by the main
effects (AR2=.02, AF(8,196)=.45, n<.s§).

Separate two-step analyses were run testing for interactions between the risk
factors and each of the ﬁve child temperament scales (attention, approach-
withdrawal/adaptability, activity, rhythmicity, and reactivity). The results of these'analyses
are summarized in Table 6 and reveal that while there were signiﬁcant main eﬁ‘ects of the
risk variables and each dimension of child temperament on child externalizing behavior
problems, the interaction terms between the child temperament dimensions and the risk
variables did not explain a signiﬁcant amount of additional variance in child outcome

beyond that accounted for by the main effects of these variables (attention: AR2=.05,
AF(8,196)= 1.73, p<.0934; approach-withdrawal/adaptability: AR2=.03, AF(8,196)= .83,

p<.5779; activity: AR2=.02, AF(8,196)= .56, p<.8111; rhythrnicity: AR2=.04, AF= 1.34,

53
p<.2256; and reactivity: AR2=.04, AF(8,196)= 1.14, p<.3397).

Regression analyses were then run for each risk factor individually, with a risk
factor and a child protective factor in step one (i.e., paternal depression and child IQ) and
the interaction term for those two variables in step two (i.e., paternal depression X IQ).
The analyses indicated that child reactivity moderated the eﬂ‘ect of maternal alcoholism on
child externalizing behavior problems, in that the interaction term (child reactivity X
maternal alcoholism diagnosis) explained signiﬁcant additional variance in externalizing
behavior problems over the main eﬂ'ects of maternal alcoholism and child reactivity

(AR2=DZ, AF(1, 210)= 5.50, p<.02), with the overall equation explaining 11% of the

variance (9% adjusted R2). In order to explore the nature of the moderating eﬂ‘ect of child
reactivity, separate regression analyses investigating the relationship between maternal
alcoholism (presence or absence) and child externalizing behavior problems were run for
subgroups based on high and low child reactivity (median split). The results of these
analyses revealed that for those children high on reactivity, maternal alcoholism was
signiﬁcantly related to child externalizing behavior problems (B=.28, p<.0510). However,
for thoSe children characterized by low reactivity, maternal alcoholism was not
signiﬁcantly related to externalizing behavior problems (B=.24, p<.0027) [See Figure 4].

It was also found that child attention moderated the eﬁ‘ect of paternal antisocial
behavior on child externalizing behavior problems. The interaction term, child attention X
paternal antisocial behavior, explained signiﬁcant additional variance in the outcome

variable over the main eﬂ’ects of paternal antisociality and child attention (AR2=.03,

AF(1,210)= 6.93, p<.009). The overall equation explained 13% of the variance in child

54
externalizing behavior problems (11% adjusted R2). Separate regression analyses

exploring the relationship between paternal antisociality and child externalizing behavior
problems for subgroups based on high and low child attention (median split) were then
conducted. The results of these analyses revealed that for those children low on attention,
paternal antisociality was signiﬁcantly related to child externalizing behavior problems
(B=.43, p<.0028). Although paternal antisociality was also signiﬁcantly related to
externalizing behavior problems for children characterized by high attention (B=.18,

p<. 0221), the eﬂ‘ect of this risk variable was signiﬁcantly attenuated for this subgroup of
children, as shown in Figure 5.

Additionally, it was found that child approach-withdrawal/adaptability moderated
the effects of maternal alcoholism on child externalizing behavior problems. The
interaction temr, child approach-writhdrawal/adaptability X maternal alcoholism, explained
signiﬁcant additional variance in the outcome variable over the main eﬁ‘ects of maternal

alcoholism and child approach-withdrawal/adaptability (AR2=.03, AF(1,210)= 7.14,

p<.008). The overall equation explained 8% of the variance in child externalizing behavior
problems (8% adjusted R2). In order to explore the nature of the moderating effect of
child approach-withdrawal/adaptability, separate regression analyses investigating the
relationship between maternal alcoholism and child externalizing behavior problems were
performed for subgroups based on high and low child approach-withdrawal/adaptability
(median split). The results of these analyses revealed that for children low on approach-
withdraw/adaptability, maternal alcoholism was not signiﬁcantly related to child

externalizing behavior problems (B=-.02, p<.8493). However, for children characterized

by high approach-withdrawal/adaptability, maternal alcoholism was signiﬁcantly related to

55
externalizing behavior problems (B=.40, p<.00001) [See Figure 6].

.Mrtﬁatiasﬂiws

Hypothesis six predicted that the protective family environment factors would
mediate the main eﬂ‘ects of the risk factors on child outcome. In order to test this
hypothesis, the following regression analyses were conducted in accordance with the
requirements outlined by Baron and Kenny (1986) regarding the determination of
mediation: l) The ﬁrst set of equations regressed the mediators (family cohesion,
maternal warmth, and stimulating home environment, on the risk factors (hypothesis 2),
revealing thatthe overall equation signiﬁcantly predicted family cohesion (R2=.30,
adjusted R2=.28, F(8, 205)=l 1.18, p<.0001) and stimulating home environment (R2=.09,
F(8, 205)=2.51, p<.013), but not maternal warmth (R2=.05, F(8, 205)=1.27, p<.261); 2)
The second equation regressed externalizing behavior problems on the risk factors
(hypothesis 1), revealing that the overall equation signiﬁcantly predicted externalizing
behavior problems (R2=.18, F(8, 205)=5.60, p<.0001); and 3) The third equation
predicted externalizing behavior problems and was conducted in two steps. The two
potential mediators, family cohesion and stimulating home environment were entered in
step one and the risk factors were entered in step two of the regression equation. The
results of this equation revealed that the mediators (R2=.04, F(2, 211)=3.93, p<.02) and
the risk factors (AR2=.15, AF(8, 203) = 4.5, p<.0001) signiﬁcantly predicted externalizing
behavior problems.

The following conditions must be met in order to establish mediation: 1) equation
1 must reveal that the risk factors aﬁ‘ect the mediators; 2) equation 2 must reveal that the

risk factors aﬂ‘ect externalizing behavior problems; 3) equation 3 must reveal that the

56
mediators aﬂ‘ect externalizing behavior problems; and 4) the effects of the risk factors on

externalizing behavior problems are less in the third regression equation than in the second
regression equation. The ﬁrst condition was not met for maternal warmth. While the ﬁrst
three conditions for mediation were met for both family cohesion and stimulating home
environment, the fourth condition was not, as the effects of the risk factors on
externalizing behavior problems were not signiﬁcantly less in the third regression equation
than in the second regression equation. The hypothesized mediating variables did not
substantially reduce the relationship between the risk factors and child outcome, as the
relationship between risk and externalizing behavior problems remained signiﬁcant and

was not signiﬁcantly reduced by considering the effect of the hypothesized mediators,
family cohesion and stimulating home environment (Rza=.15, FA(8,203)= 4.5, p<.0001
vs. R2A=.18 when the mediators were not considered). See Table 7 for the results of these
analyses.

Individual mediating pathways were then investigated. Regression analyses were
performed for each risk factor and mediator individually (i.e., paternal antisociality and
family cohesion). Two signiﬁcant pathways were found. Maternal warmth and family
cohesion were both found to mediate the relationship between maternal depression and
child externalizing behavior problems. The results of these analyses were as follows
below: 1) Family cohesion and maternal warmth were regressed on the risk factor,
maternal depression, revealing that maternal depression signiﬁcantly predicted family
cohesion (R2=.17, adjusted R2=.17, F(l, 212)=44.64, p<.0001) and maternal warmth
(R2=.02, F(l, 212)=5.16, p<.024); 2) The second equation regressed externalizing

behavior problems on the risk factors, revealing that maternal depression signiﬁcantly

57
predicted externalizing behavior problems (R2=.02, F(l, 212)=4.76, p<.03); and 3) Next,

extemalizing behavior problems was regressed on family cohesion in step one (Rz=.02,
F(l, 212)=5.3 l, p<.022) and maternal depression in step two of the regression equation
(aR’=.008, AF(1,211)= 1.82, p<. 1789). Another regression analysis was then run,
regressing externalizing behavior problems on maternal warmth in step one (R’=.03, F(l ,
212)=5.87, p<.0162) and maternal depression in step two of the regression equation
(AR2=.015, AF(1,211)= 3.41, p<.0663). Both maternal warmth and family cohesion met
all conditions for mediation of maternal depression, as outlined by Baron and Kenny
(1986), as the eﬁ‘ect of maternal depression on child externalizing behavior problems was
Signiﬁcantly reduced when maternal warmth or family cohesion were considered. When
externalizing behavior problems was regressed on family cohesion and maternal warmth in
step one (Rz=.05, F(2, 211)=5.24, p<.006) and maternal depression in step two of the
regression equation (AR2=.005, AF(1,210)= 1.15, p<.2840), these two variables mediated

the relationship between maternal depression and child behavior problems.

W

According to hypothesis seven, the strength of the relationship between child
externalizing behavior problems and risk variables should differ over the continuum of
risk level. A cumulative risk index was computed for each participant based upon the
cut-offs for each risk variable outlined in the Methods section of this paper. Neter,
Wasserrnan, and Kutrler (1990) state that sometimes the regression of Y (in this case
child outcome) on X (in this case cumulative risk) follows a particular linear relation in

one range of X (cumulative risk) while following a different linear relation in another

58
range of X (cumulative risk). This relationship among variables earl be addressed

statistieally through piecewise regression (Neter, Wasserman, & Kutner, 1990). In the
current investigation, the breakpoints along the continuum of risk level at which the
relationship between risk and child externalizing behavior problems changes was
determined through the use of the nonlinear regression function in Systat, which allows
piecewise nonlinear regression equations with unknown breakpoints to be calculated
(SPSS Inc., 1996). In this way, different regression functions earl be ﬁtted to the same
data, and the piecewise nonlinear regression function in Systat estimates the point at
which the regression function changes. The results of this analysis reveal that while the
regression slopes calculated through this procedure were not signiﬁcant, the break point
estimate was signiﬁeant (p .0494), with the slope of the relationship between risk and
child externalizing behavior problems “signiﬁeantly changing at a cumulative risk level
of two [See Figure 7].

The break point identiﬁed by the piecewise nonlinear regression analysis was
then utilized to divide the subjects into subpopulations based on the cumulative risk
index cut-off of two. Chi-square tests for independence were utilized to explore the
relationships among the levels of child externalizing behavior problems and both
cumulative risk and individual risk factors among the two subpopulations.
Additionally, by computing a cross-product ratio from the 2 by 2 table of values
obtained from the chi-square analysis, odds ratios were calculated to describe the
association between the 2 levels of each risk variable and the level of child
externalizing behavior problems (Dillon & Goldstein, 1996; Tabachnick & Fidell,

1996). In this way, the increase (or decrease for ratios less than one) in the likelihood

59
of being in an outcome category given inclusion in another category (such as risk) can

be computed (Dillon & Goldstein, 1996; Tabachnick & Fidell, 1996). The results of
the analysis which investigated the relationship between level of externalizing behavior
problems and level of risk revealed that the level of child behavior problems was
signiﬁcantly related to the level of risk (Xﬂl «F675 $309), with the high risk
children being over two times as likely to have a high level of externalizing behavior
problems compared with those from the low risk group (odds ratio = 2.05). Also,
while high versus low levels of externalizing behavior problems were almost equally
likely for the low risk group (46% high externalizing behavior problems versus 54 96
low externalizing behavior problems), high externalizing behavior problems were found
in 64% of the high risk group.

The relationships between levels of individual risk factors (based upon those
criteria utilized for the calculation of the cumulative risk index) and the level of
externalizing behavior problems were then investigated among the break-point
subpopulations. Table 8 displays the results of these analyses. As can be seen,
differential relationships between individual risk variables and level of child
externalizing behavior problems did exist among the two subpopulations of this sample.
Additionally, it was found that the presence of maternal alcoholism was signiﬁcantly
related to high externalizing behavior problems in the high risk group, with those
children with an alcoholic mother being over two times as likely to have a high level of
behavior problems than those without maternal alcoholism (odds ratio =2.28).

Children in the low risk group with an alcoholic mother were approximately half as

60
likely as those without an alcoholic mother to have behavior problems (odds

ratio= .47). Also, while high levels of externalizing behavior problems were only
present in 30% of the low risk children with alcoholic mothers, behavior problems
were found in 72 96 of the high risk group children with alcoholic mothers.

Additionally, overall risk level was associated with a differential relationship
between parental depression and child externalizing behavior problems. Maternal
depression was associated with a one and a half times greater chance of a high level of
child externalizing behavior problems in the high risk subgroup (odds ratio= 1.48),
while those children of depressed mothers in the low risk group were half as likely as
those whose mother was not depressed to have behavior problems (odds ratio= .48).
While high levels of externalizing behavior problems were only present in 31 % of the
low risk children with depressed mothers, behavior problems were found in 69% of the
high risk group children with depressed mothers. Similarly, child behavior problems
were over two times as likely in high risk children with depressed fathers (odds
ratio=2.06), whereas paternal depression was associated with an almost equal chance
of exhibiting child behavior problems in the low risk group.

Furthermore, the differential relationship between the risk variables and child
outcome can also be seen when separate multiple regression analyses were run for
subsections of the sample, divided according to the risk level cut-off previously
identiﬁed by the break point analysis. The hypothesized risk variables were entered
into a regression equation as independent variables to predict child externalizing
behavior problems for these two subsamples of subjects. The results of the regression

analyses, shown in Table 9, indicate that while the overall equation was signiﬁcant and

61
explained 17% (10% adjusted R2) of the variance in child externalizing behavior

problems for those subjects with a cumulative risk index greater than two (R2 = .17,
F(8, 99)=2.46, p< .0178), the risk factors do not signiﬁcantly predict child outcome
for those subjects with a cumulative risk index less than or equal to two (R’=.06, F(8,
97)=.76, p< .6395).

Hypothesis eight predicted differential relationships between risk and protection
along the continuum of risk, with these constructs becoming decreasingly independent
as risk increases. Piecewise nonlinear regression was utilized to investigate the
relationship between the cumulative risk index and a cumulative protection index (see
Methods section for discussion of cumulative protection index). The results of this
analysis did not reveal a point along the continuum of risk at which the slope of the
regression function representing the relationship between risk and protection changed
signiﬁcantly.

The correlations between the individual risk and protective variables were then
investigated within the two subgroups divided according to the cumulative risk index
cut-off of two. Signiﬁcant differences in the relationship between several of the risk
and protective variables earl be found among the two subpopulations [See Table 10].
For example, while maternal depression was signiﬁcantly negatively correlated with
family cohesion in the high risk group (r=-.46, 125.0001), these variables were not
signiﬁcantly related in the low risk group (r=-.01, 25.905). Similarly, paternal
antisociality was signiﬁcantly negatively correlated with a stimulating home

environment in the high risk group (r=-.24, ps.014) but not in the low risk group (r=-

62
.02, 25.822). Additionally, it can be seen that the correlations between risk and

protection found for the total sample are not representative of the relationship between
several of these variables across the continuum of total risk [i.e. , total sample:
maternal depression and family cohesion (r=-.42, 23.0001); paternal antisociality and
stimulating home environment ( =-.26, 1230001)].

Chi-square analyses and odds ratios were then utilized to explore the different
relationships between the protective factors and both the cumulative risk index and
individual risk factors among the two subgroups. Table 11 displays the results of these
analyses. High levels of cumulative protection (median split of cumulative protection
index) were one quarter as likely in children from the high risk group than in children
from the low risk group (Xﬂl m=21.4, pg .0001; odds ratio = .26).

Additionally, it can be seen that there exist many differential relationships
between individual risk and protective variables among the two subgroups. For
example, in the high risk group, the presence of high family cohesion is less than half
as likely when maternal depression is present (Xﬂ1 ”=5 .51, pg .018; odds ratio =
.40), whereas maternal depression is almost unrelated to the presence of family
cohesion in the low risk group (odds ratio = .76). Although high family cohesion can
be found in 69% of the low risk group with maternal depression, only 35 % of the high
risk group with maternal depression has high family cohesion present. Similarly,
paternal alcoholism is almost equally likely to occur with a stimulating home
environment in the low risk group (odds ratio= .73), whereas in the high risk group,

the presence of paternal alcoholism makes it one quarter as likely that the home

63
environment is stimulating (odds ratio=.27).

There do exist several relationships between risk and protective variables which
differ across subgroups in a direction which appears contradictory to that anticipated in
hypothesis eight. For example, while paternal alcoholism is equally likely to occur
with high family cohesion in the low risk group, high risk group families with an
alcoholic father are over three times as likely to have high levels of family cohesion
(odds ratio = 3.43). However, it appears that this ﬁnding may have resulted from the
signiﬁcantly unequal distribution of paternal alcoholism in the high risk group due to
sample recruitment procedures (high risk group: n=9 with no paternal alcoholism,
n=99 with paternal alcoholism). In fact, while 90% of the low risk group with
paternal alcoholism is characterized by high family cohesion, only 41% of the high risk
group with alcoholic fathers has high family cohesion.

Furthermore, two differential relationships based upon risk level subgroup were
found in the ability of the risk factors to predict protective factors. Although the risk
factors entered together into a regression equation signiﬁcantly accounted for 16% of
the variance in child reactivity for the high risk subgroup (R2=.16, F(8,99)=2.34,
p<.023 9), these variables did not signiﬁcantly predict child reactivity in the low risk
subgroup (Rz=.07, F(8,97)==.9040, p<.5165). Similarly, the risk factors signiﬁcantly
accounted for 36% of the variance in family cohesion for the high risk group (R2=.36,
F(8,99)=6.80, p<.0001) but not for the low risk group (R2=.11, F(8,97)=l.49, p<.1705).

(See Tables 12 and 13).

Discussion

The purpose of this study was to examine the relationship between multiple risk
and protective factors and the role of these variables over time in the development of
externalizing behavior problems in children. Based on the literature, it was hypothesized
that such child characteristics as intelligence and temperament would moderate the
relationship between risk factors and externalizing behavior problems. It was also
hypothesized that aspects of the family environment, such as family cohesion and maternal
warmth, would mediate the relationship between risk and externalizing behavior problems.

Additionally, as prior research has not adequately addressed the relationship
between risk and protection, particularly in high risk samples, an important goal of the
current investigation was to investigate the orthogonality of risk and protection at
different points along a continuum of risk. This investigation hypothesized that there
exists a differential relationship between risk and protection among risk status
subpopulations, with these constructs becoming decreasingly independent as risk
increases. It was therefore expected that although risk and protection would occur
together on the lower end of the continuum of risk, high protection would be less likely on
the high end of the continuum of risk. Additionally, it was anticipated that the
predictability of child externalizing behavior problems would increase at the extremes on

the continuum of risk (i.e., Bernstein & Hans, 1994). Therefore, this study investigated

65
the relationship between multiple risk and protective factors and the effects of these

factors on children’s development of externalizing behavior problems over time in order to
begin to explain the variations in outcome which are observed in children of alcoholics.
Marn' Eggs of Risk Fgors

The results of this study indicate that the hypothesized risk factors at time 1
signiﬁcantly predicted child externalizing behavior problems three years later at time 2.
This ﬁnding is consistent with the literature regarding the eﬁ‘ect of parental
psychopathology and environmental risk factors on children’s development of behavior
problems (e.g. Fitzgerald et al., 1993; Zucker & Fitzgerald, 1991; Jansen, Fitzgerald,
Ham, & Zucker, 1995; Hill & Muka, 1996; West & Prinz, 1987; Werner, 1986). While
the risk factors were signiﬁcantly related to child externalizing behavior problems, only
18% of the variability in child outcome was accounted for by these variables. It is
therefore apparent that the incidence of child behavior problems could not be fully
explained by the risk factors in this model. However, the proportion of outcome variance
accounted for by the risk factors in the current investigation was fairly consistent with the
literature on behavior problems in children. For example, Chassin et al. (1991) and Rubio-
Stipec et al. (1991) found that risk variables explained 18% (adjusted R2) and 29% of the
variability in child behavior problems, respectively. Furthermore, given the present study’s
ﬁndings regarding both the direct and moderating effects of protective factors on child
outcome, it is not surprising that risk alone does not account ﬁllly for the variance in child
externalizing behavior problems.

The results of the present investigation are also consistent with Walker et al. ’5

(1989) conclusion that the speciﬁc eﬁ‘ects of individual risk factors on child development

66
are often confounded due to the nonorthogonality of risk factors. Although the risk

factors together signiﬁcantly predicted child externalizing behavior problems, only one risk
factor, maternal antisociality, contributed signiﬁcant unique variance to the prediction of
child behavior problems when all risk factors were considered simultaneously. However,
the other risk factors together explained an additional 16% in child outcome variability.
The variance shared by these highly correlated risk factors made it diﬁcult to discern the
importance of individual risk factors. This ﬁnding is not surprising in light of Kazdin and
Kagan’s (1994) argument that risk factors often do not exist in isolation but rather are
interrelated or appear in “packages” (p. 39). Other researchers have argued that the
inability to ﬁnd speciﬁc child outcomes associated with speciﬁc risk factors often results
ﬁ'om the fact that risk factors co-occur (e.g., Frick et al., 1992). Therefore, it is diﬁcult,
and perhaps artiﬁcial, to investigate the isolated eﬁ‘ects of individual risk factors on child
outcome.

As a result, the cumulative risk index utilized in the current study attempted to
capture the additive eﬁ'ects of co-occurring risk factors suggested by the literature (e.g.,
Frick et al., 1992; Blackson & Tarter, 1994). In fact, the results of this study are
consistent with the literature regarding the explanatory power of summative indexes of
risk (e.g., Sameroﬁ‘ et al., 1993; Jessor et al., 1995; Williams, Anderson, McGee, & Silva,
1990), as the cumulative risk index alone explained 13% of the variance in child
externalizing behavior problems (compared to the 18% accounted for by all the risk
factors considered together). However, the cumulative index of risk in this investigation
did not account for a proportion of variance in outcome which is as substantial as has been

evidenced in the literature. For example, Sameroﬁ‘ et al. (1993) found that a summative

67
risk index accounted for 37% of the variance in child IQ, leading these researchers to

conclude that it was not the type but rather the number of risk factors which a child
encountered that determined outcome. Although the results of the present study indicate
that the amount of risk encountered does impact child development, a signiﬁcant
proportion of variance in child outcome remained unaccounted for by the cumulative risk
index. When this ﬁnding is considered in light of the moderators of risk also identiﬁed in
this study, the assumption that the relationships between risk and outcome lack speciﬁcity
appears not to be fully supported.

As has been suggested by the literature, the risk factors in this study were also
predictive of an absence of family environment protective factors. Again, the importance
of individual risk factors in the prediction of the family environment protective factors
remained diﬁcult to discern due to the amount of variance shared by the risk factors (10%
for family cohesion and 7% for stimulating home environment). Additionally, while the
risk factors together did not signiﬁcantly predict maternal warmth, maternal depression
was predictive of a lack of maternal warmth. As the risk factors together were not
predictive of maternal warmth, it seems possible that this variable may be a particularly
important protective factor in high risk environments. This is consistent with prior
research which suggests that maternal warmth is associated with better outcomes and is
protective in the development of children exposed to risk (e. g., Smith & Prior, 1995;
Bradley et al., 1991; Pettit & Bates, 1989; Bradley & Caldwell, 1981). However, an
important implication of these ﬁndings is that the presence of maternal depression may
signiﬁcantly interfere with the development of mother-child warmth, therefore placing

children at an increased risk for poor outcomes.

68
ME Eﬁ‘m of Prgtﬁive Fagors

The results of this study are also consistent with previous studies that have found
high intellectual ability, positive temperament characteristics, and family environment
factors to play a protective role in the development of high-risk children (e. g., Werner,
1986; Rogosh et al., 1990; Corbo-Richert, 1994; Zucker & Gomberg, 1986; Phipps &
Mulhern, 1995; Varni et al., 1990). Similar to the results found for the risk factors, most
of the-protective factors did not contribute signiﬁcant unique variance to the prediction of
child externalizing behavior problems when considered together. Additionally, as is
suggested by the literature, the cumulative protection index was predictive of child
behavior problems, although it did not account for a substantial proportion of variance in
outcome (Fergusson & Lynskey, 1996; Jessor et al., 1995).

Overall, the ﬁndings of the current investigation regarding the relationship between
protective factors and child behavior problems may suggest that protective factors are
similar to risk factors in that they are interrelated or appear in “packages,” making it
diﬁcult to determine the importance of individual factors (Kazdin & Kagan, 1994, p. 39).
The results of this study also appear to highlight the importance of investigating both risk
and protective factors in the development of child behavior problems, as neither the risk
nor the protective variables were able to fully account for the variance in child outcome.
Furthermore, given the moderators eﬁ‘ects which were also identiﬁed in this study, it may
be that models which merely investigate main eﬁ‘ects can not suﬁciently capture the
complex relationships among risk, protection, and the development of child behavior

problems.

69

Modmting Eﬁ‘e_ct§ of thld Characteristics

This study also investigated the question of whether child characteristics serve as
moderators of risk, by affecting the strength and direction of the relationship between risk
and outcome. The results of this study revealed that neither the interaction of the risk
variables with child IQ nor the interaction of the risk factors with any of the child
temperament dimensions accounted for a signiﬁcant portion of variance above that
accounted for by the main eﬁ‘ects of these variables. This ﬁnding indicates that the effect
of the risk factors together on child outcome is not signiﬁcantly dependent upon either IQ
level or child temperament status. It seems that although a good child temperament and a
high IQ directly protect a child ﬁom poor outcomes, the adverse eﬁ‘ect of risk factors on
child behavior was not signiﬁcantly attenuated by the presence of these protective child
characteristics, perhaps limiting the amount of protection these characteristics can provide
for children exposed to high risk.

Overall, the results of this study did not support those of prior research regarding
the role of child personality variables as moderators of the relationship between risk and
outcome in children of alcoholics (Rogosch et al., 1990; Masten et al., 1988). Masten et
al. (1988) found that child IQ moderated the relationship between stress and children’s
competence at school, although child IQ was not found to moderate the eﬁ‘ects of risk on
child behavior problems in the current investigation. However, Masten et al. (1988) also
reported that the role of child IQ was dependent upon the competence criterion utilized.
For example, while high child IQ was found to be protective against the adverse impact of
stress on children’s classroom disruptiveness, low child IQ was protective when

considering classroom engagement (Masten et al., 1988). It seems that the eﬁ‘ects of risk

70
on the outcome criterion utilized by the present study, child externalizing behavior

problems, are not moderated by child intelligence.

When the moderating ‘eﬁ‘ects of individual temperament characteristics were
examined in relation to the impact of a single risk factor on child outcome, three
signiﬁcant interaction effects were found in the current investigation. The results of these
analyses revealed that for those children low on reactivity and low on approach-
withdrawal/adaptability, maternal alcoholism was not signiﬁcantly related to child
externalizing behavior problems, whereas for those children high on reactivity and high on
approach-withdrawal/adaptability, maternal alcoholism did signiﬁcantly predict poor child
outcome. Additionally, child attention was found to moderate the eﬁ’ects of paternal
antisociality, with the eﬁ‘ect of this risk factor on child outcome being signiﬁcantly more
pronounced for those children characteiized by a low level of attention than for those
children with a high level of attention.

The results found regarding the moderating eﬁ’ects of both child attention and child
reactivity are consistent with the literature, which has found that an easy temperament,
including the characteristics of high attentiveness and low reactivity, attenuates the
relationship between risk and poor outcomes (e.g., Kyrios & Prior, 1991). The
moderating eﬁ‘ect of child approach-withdrawal/adaptability was not in the direction
predicted by much of the prior research. Indeed, the hypothesized protective
characteristic of high approach/adaptability (or approach behavior to novel stimuli) did not
attenuate the effect of maternal alcoholism, but rather the absence of this child
characteristic in the presence of maternal alcoholism resulted in more positive child

outcomes.

71
Although the direction of this ﬁnding was not anticipated, it seems possible that a

child temperament characterized by a tendency to approach novel situations and people
may be a positive attribute in an average environment, while a temperament characterized
by a tendency to withdraw ﬁ'om unfamiliar and uncomfortable situations or people may be
a more adaptive personality characteristic in the presence of maternal alcoholism. In fact,
Tschann et al. (1996) found that child approachability played a similar moderating role in
the relationship between high family conﬂict and child social withdrawal, as low approach
children from high conﬂict families were less withdrawn than were similar children ﬁ'om
families characterized by low conﬂict. The opposite was found for high approach
children, who were the least withdrawn in low conﬂict families and were more withdrawn
in the presence of high conﬂict (T Schann et al., 1996). These researchers concluded that
child approachability functions as either. a protective or a vulnerability factor, depending
upon the level of family conﬂict to which the child is exposed (T schann et al., 1996).
I When considered in light of the goodness of ﬁt model proposed by Thomas and Chess
(1981), which conceptualizes temperament not as simply good or bad but rather examines
it in regard to its ﬁt with a particular environment, it may have been ambitious to assume
that speciﬁc temperament characteristics would be protective in all environments.
Mgiating Effects of Fm’ly Environment Characteristics

Based on prior research which has found characteristics of the family environment
to be associated with positive outcomes in children who are exposed to risk factors (e. g.,
Smith & Prior, 1995; Seifer et al., 1992; Bradley et al., 1995), the current study not only
hypothesized that protective factors would aﬁ‘ect outcomes directly but also anticipated

that family environment protective factors would account for the “how” or “why” of the

72
relationship between risk and outcome (Rogosch et al., 1990). However, the results of

this study indicated that the eﬁ‘ects of the risk factors on child behavior problems were not
signiﬁcantly reduced when the hypothesized mediating factors were considered. One
possible explanation for this ﬁnding may be that the risk factors affect numerous aspects of
the family environment and, therefore, the two mediating factors tested in this model could
not suﬁciently account for the relationship between risk and child outcome. Perhaps a
model which considered additional aspects of the family enviromnent, such as
expressiveness, control, and communication among family members (e.g., Bischoﬁ Stith,
& Whitney, 1995; Grossman et al., 1992), could account for the eﬁ‘ects of risk on child
extemalizing behavior problems. However, the effects of the eight risk factors considered
together in the current model account for such a substantial proportion of variance (1 8%)
in child externalizing behavior problems that the signiﬁcant, yet more modest, relationship
between the two family environment protective factors and outcome (4%) can not
suﬁciently account for the eﬁ‘ects of risk.

When the potential mediating eﬁ‘ects of family environment characteristics were
considered individually in relation to the impact of a single risk factor on child outcome,
maternal warmth and family cohesion both were found to mediate the relationship between
maternal depression and child externalizing behavior problems. In this way, maternal
depression aﬂ‘ects children indirectly, by decreasing the maternal warmth and the family
cohesion present in the child’s environment. The two independent mediating pathways
identiﬁed for the eﬁ‘ects of maternal depression indicate that one risk factor can adversely
impact multiple aspects of the family environment. It therefore seems that a model which

does not account for numerous family environment characteristics, such as the one

73
proposed in the current investigation, will not be capable of sumciently capturing the

“how” or “why” of a risk factor’s eﬁ‘ects on child outcome. A model that accounts for a
diversity of family environment characteristics may be more equipped to account for the
effects of risk on child externalizing behavior problems.

Eﬁ‘gs of Risk Lag]

The results of this study indicate that the slope of the regression ﬁlnction
representing the relationship between risk level and child externalizing behavior problems
changes at a cumulative risk level of two (see Figure 7). This study also found that the
level of risk and the level of externalizing behavior problems were signiﬁcantly related and
that there was a greater variability in child outcome in the low risk group than in the high
risk group. These ﬁndings are similar to the diﬂ‘erential relationships between cumulative
risk and child outcome among high- and low-risk subgroups identiﬁed by Bernstein and
Hans (1994). Additionally, the results of this analysis replicate the ﬁndings of both
Bernstein and Hans (1994) and Sameroﬂ‘ et al. (1987) in which there were changes in the
relationship between cumulative risk and child outcome following the point where there
were two risk factors present.

One possible way to conceptualize the change in the slope of the regression
function along the continuum of risk is as a threshold of risk level, after which the
accumulation of additional risk factors impacts child outcome to a greater extent than it
did prior to the breakpoint in the regression function. However, Bernstein and Hans
(1994) argue that while the eﬁ‘ects of some risk factors may simply be cumulative,
combinations of some speciﬁc risk factors may be especially damaging to children’s

development. A particular combination of risk factors which the literature associates with

74
a signiﬁcantly increased rate of behavior problems in children is parental comorbidity for

psychopathology (Steinhausen et al., 1984; Hill & Muka, 1996; Chassin et al., 1991;
Warner et al., 1995; Merikangas et al., 1988; Frick et al., 1991). For example, Hill and
Muka (1996) found that high. risk children’s odds of having a diagnosed disorder were
increased when the child lived with two alcoholic parents. Similarly, Chassin et al. (1991)
found that the risk associated with parental alcoholism was mediated by co-occurring
parental psychopathology and environmental stress.

In the current study, it is possible that the presence of psychopathology,
particularly alcoholism, in both parents may be driving the change in the slope of the
regression function, as a substantial increase in the number of children with two alcoholic
parents occurred following a cumulative risk index value of two. In addition, Fitzgerald et
al. (1993) previously found that maternal psychopathology not only was more likely to be
found in the high risk families of this sample but also was predictive of child outcome.
Ellis et al. (1996) also found that the families of antisocial alcoholics in this sample had
mothers with the highest levels of psychopathology and children with the greatest degree
of behavior problems. Furthermore, the current investigation found a signiﬁcant
relationship between maternal alcoholism and child behavior problems in the high risk
group, with those children exposed to maternal alcoholism being more than two times as
likely as children without alcoholic mothers to exhibit externalizing behavior problems.
This ﬁnding is consistent with the literature regarding the adverse eﬁ‘ects of maternal
psychopathology on child outcome (i.e., Hill & Muka, 1996; Fitzgerald et al., 1993;
Werner, 1986).

In fact, post-hoe X2 analyses conducted in the present investigation revealed that

75
within the total sample, parental comorbidity for alcoholism was associated with a

signiﬁeantly increased chance of a high level of externalizing behavior problems in
children (p< .02; odds ratio=2.05). Similarly, within the high risk subgroup, parental
comorbidity for alcoholism remained signiﬁcantly related to an increased chance of a
high level of externalizing behavior problems in children (n < .04; odds ratio=2.24).
However, merely the presence of psychopathology in both parents was not signiﬁcantly
related to an increased chance of a high level of externalizing behavior problems in
children. These post-hoc ﬁndings seem to provide support for the argument made by
Bernstein and Hans (1994) regarding the particularly damaging eﬁ‘ect of some
combinations of speciﬁc risk factors on children’s development.

Although other signiﬁcant relationships between the level of individual risk factors
and child behavior problems were not identiﬁed within the subgroups, several relationships
between risk factors and externalizing behavior problems were found to diﬁ'er between the
risk level groups, providing further support for the importance of investigating the
cumulative effects of risk in addition to combinations of speciﬁc risk factors (Bernstein &
Hans, 1996). For example, parental depression was more strongly related to externalizing
behavior problems in the high risk group than in the low risk group.

While the results of this study highlight the importance of investigating the role of
cumulative risk level in the relationship between individual risk factors and child outcome,
a potential problem with comparisons across risk subgroups was also revealed, as it was
found that the number of subjects characterized by a risk factor within a risk subgroup
often was very small. When this occurred, it was less likely that a signiﬁcant relationship

between risk and outcome could be detected due to insuﬁcient power, although important

 

 

76
information was still provided regarding the likelihood that a particular risk factor exists

within a context of high or low overall risk. For example, the results of these analyses
reveal that maternal antisociality did not occur within the low risk group and that all low
risk children with an antisocial father exhibited a high level of externalizing behavior
problems.

Additionally, the results of the regression analyses for the risk level subgroups
revealed that the individual risk factors were diﬁ‘erentially predictive of child outcome
contingent upon risk level. This ﬁnding is consistent with that of Seifer et al. (1992), who
found diﬁ'erential predictability of outcome based upon risk group status. It can therefore
be seen that studies which do not involve high risk samples or do not account for multiple
risk factors may be unable to successfully address the relationship between risk and
outcome. .

The slope of the regression function representing the relationship between
cumulative risk level and cumulative protection level did not change at any point along
the continuum of risk level, as the relationship between these two constructs was best
represented by an inverse, liner regression line. It therefore appears that there did not
exist a threshold of risk after which the accumulation of additional risk factors precluded
the accumulation of additional protective factors to a greater extent than it did prior to the
threshold. However, risk and protection were found to be negatively related, and the
relationship between risk and protection differed at points along the continuum of risk for
some individual variables, a ﬁnding similar to the relationships between risk and outcome
found by both Bernstein and Hans (1994) and the present investigation. Signiﬁcant

diﬁ‘erences between the correlations of individual risk and protective factors were found

77

between the two risk level subgroups, as many risk and protective factors were more
highly inversely correlated in the high risk group than in the low risk group. While the
relationship between risk and protection identiﬁed in the low risk group is consistent with
Jessor et al.’s (1995) argument that these are orthogonal constructs, the increased inverse
correlation between risk and protection in the high risk group suggests that risk and
protection may become decreasingly orthogonal as cumulative risk increases.

The eﬂ'ect of overall risk level on the relationship between risk and protection
could also be seen in diﬁ‘erential relationships which were found involving protective
variables which previously had been identiﬁed as mediators or moderators of the eﬂ‘ects of
risk on child behavior problems. For example, maternal depression was found to be
signiﬁcantly related to family cohesion in the high risk group (r=-.46) but not in the low
risk group (r=-.01). Findings such as this involving previously identiﬁed mediating
pathways and moderating eﬁ‘ects raise the question of whether these relationships exist
along the continuum of risk or only within risk level subgroups.

Additionally, it was found that the ability of the risk factors taken together to
predict two protective factors, child reactivity and family cohesion, signiﬁcantly
differed between the two risk level subgroups, as the risk factors accounted for a
signiﬁcant amount of the variance in these protective variables in the high risk group
but not in the low risk group. This ﬁnding provides additional information which
augments the results of Jessor et al. ’s (1995) study regarding the relationship between
risk and protection in a sample which did not include extremes on a continuum of risk.
In the current investigation, it appears that while both high and low levels of family

cohesion and child reactivity coexisted with risk within an environment characterized

78

by a low level of overall risk, low levels of these protective factors were found within a
context of high risk. It can be seen that the orthogonality of these constructs differed at
different points along the continuum of risk, a ﬁnding which is inconsistent with Jessor
et a1. ’s (1995) conclusion that high risk does not necessarily imply low protection.

Additionally, the relationships between several individual risk and protective
variables within the subgroups revealed that it was signiﬁcantly less likely to have high
levels of individual protective variables in the presence of particular risk factors in the high
risk subgroup than it was in the low risk subgroup. Furthermore, the results of the
analyses investigating the relationship between cumulative risk and protection level
revealed that the odds were four to one against ﬁnding a high level of protection with a
high level of risk. It appears that studies which do not include high risk samples (e.g.,
Jessor et al., 1995) may be unable to accurately determine the relationship between risk
and protection, as the results of the current investigation suggest that high levels of
protection are unlikely to exist within an environment characterized by high risk.

There also existed several relationships between risk and protection which were
statistically signiﬁcant in a direction which was not anticipated. For example, it was found
that maternal antisociality was inversely related to child rhythmicity in the low risk group
but not in the high risk group. As prior research has found that the impact of a speciﬁc
parental psychopathology can be attenuated by the presence of additional
psychopathology (Moses, 1992), it may be that even low levels of maternal antisocial
behavior in a context of low risk can aﬁ’ect a child’s rhythmicity, while within a high risk
context there exist numerous factors which could adversely aﬁ‘ect a child’s rhythmicity.

Similarly, low child reactivity was associated with high family conﬂict, particularly in the

79
high risk subgroup. Perhaps the presence of high conﬂict decreases the likelihood that

children will react intensely to stimuli, as children may learn that an intense reaction on
their part exacerbates an already conﬂictual environment. In this way, low child reactivity
may serve an important protective function in high conﬂict families, a hypothesis which is
supported by the literature (e.g., Tschann et al., 1995). In fact, these results are
consistent with Smith and Prior’s (1995) hypothesis, based upon their discovery of
independent factors for teacher and mother ratings of child temperament, that “contextual
inﬂuences on expressive behaviors” may exist (p. 173).

In addition, the results of these analyses again raised questions regarding mediating
and moderating relationships across risk levels, as diﬂ‘erential relationships between risk
and previously identiﬁed mediating and moderating protective factors were found.
Perhaps additional mediating pathways and moderating effects would have been identiﬁed
in the present investigation if these relationships had been examined within risk level
subgroups.

Limitations of the Current Study and Directions for Future Research

While it was important to control for the effects of gender and ethnicity, the
entirely male, Caucasian sample utilized in this study necessarily limited the generalizability
of these ﬁndings to both females and individuals from other ethnic/racial backgrounds.
Future studies which investigate these relationships in more diverse populations will
provide further information regarding the eﬁ‘ect of risk and protection in the general
population and may reveal differential relationships among these variables in various

populations.

80
Another potential limitation of the current study was that the determination of

child behavior problems relied exclusively on parent report. This is particularly
problematic if diﬁ‘erences exist between the risk groups in parental perceptions of child
behavior. As previous studies have suggested that psychologically distressed parents
provide ratings of their children comparable to independent sources, the extent to which
this was a problem in the current investigation remains unclear (Richters & Pellegrinni,
1989). Similarly, this study also relied exclusively on self-report measures for the
assessment of risk and protective factors. Further investigation of the relationship
between risk, protection, and child behavior ratings obtained ﬁom independent sources,
such as teachers, will provide information regarding the limitations of relying on parent
ratings and self report for the assessment of these variables.

As Masten et al. (1988) found that the relationship between stress and children’s
competence was dependent upon individual characteristics as well as the competence
criterion utilized, the current investigation’s use of only one area of outcome, externalizing
behavior problems, may have limited the ﬁndings of this study. Perhaps studies which
investigate the relationship between risk, protection, and other dimensions of outcome,
such as depression and anxiety disorders, will ﬁnd differential relationships among these
variables. Furthermore, as mentioned previously, this study only investigated a limited
number of family environment characteristics as potential mediating variables. As a result,
it is likely that numerous effects of risk on the family environment, and ultimately on child
outcome, were not accounted for by the present study. The inclusion of additional family
environment characteristics in future studies may reveal important mediating pathways of

the eﬁ‘ects of risk on child outcome.

81
The potential protective function of external support systems, such as peers and

teachers, remains another important area for future research. Although child
characteristics and aspects of the family environment were investigated in this study, the
third general category of protective factors speciﬁed by Garmezy (1983), external support
systems, generally is not available to very young children and therefore could not be
examined in the current investigation. Future studies which examine the role of external
support systems in children’s lives will provide important information regarding this
potential source of protection for high risk children.

Based upon the results of the current investigation, it appears that future studies
which investigate both mediating and moderating relationships between risk and protective
variables within risk level subgroups may reveal signiﬁcant protective pathways.
Additionally, it would be interesting to investigate the relationship between risk and
protection in the lives of older high risk children. Such an investigation could include both
following the present sample over time as well as exploring the relationship between risk
and protection in other samples. Finally, future studies should attempt to identify
particularly damaging combinations of risk factors and continue to examine the
nonspeciﬁc, cumulative effects of multiple risk factors on the development of externalizing

behavior problems in children.

APPENDICES

APPENDIX A

TABLES

82
Table 1

".1 'ive iii or"kVari.--l Pro-.at-t'veV't-l ad-rr' {-41

Behavier Problems (n=214)

 

 

Variable M D
Risk
Paternal Alcoholism 1.68 .47
Paternal Antisocial Behavior Score 16.28 9.95
Paternal Depression Score 2.44 2.61
Maternal Alcoholism 1.33 .47
Maternal Antisocial Behavior Score 10.24 6.56
Maternal Depression Score 2.66 2.86
Low Socioeconomic Status 340.62 126.87
Family Conﬂict 3.57 2.07
Family
Environment
Protection ..
Family Cohesion 7.39 1.78
Stimulating Home Environment 6.59 .97
Maternal Warmth 5.05 1.20
Child
Protection
IQ 105.42 13.91
Child Activity 1.39 1.44
Child Approach-withdrawal/Adaptability 3.66 1.96
Child Attention 5.09 3.01
Child Reactivity 3.15 1.60
Child Rhythrrricity 5.53 2.24
Child
Outcome
Extemalizing Behavior Problems 10.89 6.52

 

 

83
Table 2

m ef Multiple Regreesion Analme Prﬁieting Child Extemalizing BeMvier
Prlmsn=214'M' ﬁ‘eszikF r nP ivF

 

 

Overall Individual Variables a} E a r sf—
Model

Risk .18 560*"
Paternal Alcoholism -.09 -1.3 .006
Paternal Antisocial Behavior .11 1.4 .008
Score
Paternal Depression Score .10 1.4 .008
Maternal Alcoholism .11 1.5 .009
Maternal Antisocial Behavior .18 2.4* .023
Score
Maternal Depression Score .04 .5 .001
Low Socioeconomic Status .07 1.0 .004
Family Conﬂict . .11 1.5 .009

Family

Environment

Protection .05 3.95 "
Family Cohesion -.14 -2.1"‘ .020
Stimulating Home -.08 -1.2 .006
Environment
Maternal Warmth -.14 -2.0* .018

Child

Protection .06 2.24“
IQ -.09 -1.25 .008
Child Activity .01 .19 .000
Child Adaptability -.04 -.60 .002
Child Attention -. 13 -1.69 .000
Child Reactivity .1 1 1.54 .011
Child Rhythnricity .07 1.04 .005
Cumulative Risk Index .13 303*"

Cumulative Protection Index .06 123*"

 

‘23 05. "pg 01. "ﬁgs 001.

Table 3

Sum of Multi le Re ession Anal sis Predictin Farnil Cohesion n=214:

 

 

Main Effects of Risk Factors
Overall Individual Variables e2 g Q r srz
Model

Risk .30 l l.2"”‘“'l
Paternal Alcoholism -.048 -.70 .002
Paternal Antisocial Behavior -.095 -1.3 .006
Paternal Depression .145 2.20“ .017
Maternal Alcoholism -.059 -.86 .003
Maternal Antisocial Behavior -.03 -.415 .000
Maternal Depression -.366 6.76"“ .1 13
Low Socioeconomic Status -.008 -. 120 .000
Family Conﬂict -2.92 -4.52**“‘ .069

 

*95 05. “pg 01. "*pg 001.

85
Table 4

Summag of Multiple Regression Analysis Predicting Stimulating Home Environment

(n=214 ): Main Eﬁ‘ects of Risk Factors

 

 

Overall Individual Variables e2 E e t i
Model
Risk .09 2.5“

Paternal Alcoholism -.099 -l.29 .008
Paternal Antisocial Behavior -.173 -2.07* .021
Paternal Depression Score -.049 -.652 .002
Maternal Alcoholism .120 .156 .000
Maternal Antisocial Behavior .056 .687 .002
Maternal Depression -.032 -.438 .000
Low Socioeconomic Status -.117 -l.56 .012
Family Conﬂict -.019 -2.65 .000

 

‘23 05. "pg 01. "*9: 001.

86
Table 5

Summeg ef Multiple Regession Anelysis Predicting Maternal Wermth (n=214):

Mm Eﬁ‘eete efRiek Fegors

 

 

Overall Individual Variables E F(dﬂ Q t i
Model
Risk .05 1.27
(8,205)
Paternal Alcoholism -.04 -.503 .001
Paternal Antisocial Behavior .00 .025 .000
Paternal Depression .06 .823 .003
Maternal Alcoholism -.04 -.544 .001
Maternal Antisocial Behavior .00 -.017 .000
Maternal Depression -. 15 -2.05 ‘ .019
Low Socioeconomic Status .12 -1.58 .011
.000

Family Conﬂict -.03 -.373

 

”as 05. “pg 01. "‘25 001.

87

 

 

 

 

 

 

 

Table 6
as fI-Iier chi .: R14 e in a 1] 1' fr V ° vl' P “1"”:
Ext malizin Behavi r Problems n=214 ' Modera in f hil Prot i n
Predictors AK AI _
Step 1
Risk Variables and Child IQ .18 5.06"”
Step 2
Interaction terms of Child IQ X Risk Variables .02 .45
Step 1
Risk Variables and Child Activity .18 596*"
Step 2
Interaction terms of Child Activity X Risk .02 .56
Variables
Step 1
Risk Variables and Child Adaptability .18 496*"
Step 2
Interaction terms of Child Adaptability X Risk .03 .83
Variables
Step 1
Risk Variables and Child Attention .20 506*“
Step 2
Interaction terms of Child Attention X Risk .05 1.73
Variables
Step 1
Risk Variables and Child Reactivity .20 567*"
Step 2
Interaction terms of Child Reactivity X Risk .04 1.14
Variables
Step 1
Risk Variables and Child Rhythmicity .20 5.71 ***
Step 2
Interaction terms of Child Rhythmicity X Risk .04 1.34

Variables

ﬁgs 05. "pg 01. “ﬁgs 001.

 

88
Table 7

Summm of Hierarchical Regression Analyses for Variables Predicting Child Extemalizing
Behavior Problems (n=214 ): Mediating Eﬁ‘ects ofFamily Environment Protection

 

 

 

 

Predictors A_R2 A15

Step 1

Family Cohesion and Stimulating Home Environment .04 3.93“
Step 2

Risk Variables .15 .4.5"'**

(Risk Variables when mediators not considered) (. 18) (5.6"*)
Step 1

Family Cohesion .02 5.3 1*
Step 2

Maternal Depression .008 1.82

(Maternal Depression when mediator not considered) (.02) (4.76“)
Step 1 '

Maternal Warmth .03 5.87*
Step 2

Maternal Depression .015 3.41

(Maternal Depression when mediator not considered) (.02) (476*)

 

*2: 05. “pg 01. "“25 001.

89
Table 8

Sum oftheRl 'nhi Be nRik daHih

khavier Preblems Agrees Risk Level Sebmpeleg‘ens

V] f hild Ex rn izin

 

High Risk (n=108)

Low Riel; (n=106)

 

 

Variables % w/ odds % w/ odds
Risk ratiol Risk ratiol
Factor Factor

Paternal Alcoholism 63 .48 45 .89
Maternal Alcoholism 72 2.28“ 30 .47
Paternal Antisocial Behavior Score 65 1.09 100 -
Maternal Antisocial Behavior Score 69 1.37 0 risk -
Paternal Depression Score 73 2.06 50 1.18
Maternal Depression Score 69 1.48 31 .48
Low Socioeconomic Status 63 .84 50 1.24
Family Conﬂict 63 .88 50 1.19

 

 

*2: 05. "pg 01. "*pg 001.

lgiven presence of risk factor, odds ratio equals the increased likelihood of high level of

externalizing behavior problems

89
Table 8

Summm of the Relag'enships Betwgn Ri§k end a High Level ef Chile Extemalizing

ﬁehevier Preelems Aemee Risk Qvel Sebmpelatjens

 

High Risk (n=108)

Low Risk (n=106)

 

 

Variables % w/ odds % w/ odds
Risk ratiol Risk ratiol
Factor Factor

Paternal Alcoholism 63 .48 45 .89
Maternal Alcoholism 72 2.28“ 30 .47
Paternal Antisocial Behavior Score 65 1.09 100 -
Maternal Antisocial Behavior Score 69 1.37 0 risk -
Paternal Depression Score 73 2.06 50 1.18
Maternal Depression Score 69 1.48 31 .48
Low Socioeconomic Status 63 .84 50 1.24
Family Conﬂict 63 .88 50 1.19

 

 

*9: 05. "pg 01. ““93 001.

lgiven presence of risk factor, odds ratio equals the increased likelihood of high level of

externalizing behavior problems

90

Table 9
Semmgy efMeltiple RegLession Analygg Predicting Child Extemalizing Behavior

Problems: Main Effects of Risk Factors Within Risk Level Subggoups

 

 

Risk Level B: _F_
High Risk (n=108)
Risk Variables .17 2.46“

Low Risk (n=106)
Risk Variables .06 .76

 

ﬁgs 05. "pg 01. “*2: 001.

91

.So mac: .8 was... .mo was

 

 

 

 

 

82 .- 88.- 22.- 82. 82.- 88.- 88.- 88.- 2.83- 288 8:5
88. 82.- 8_ _r 82.. 88.- 2.2.- 887 88.- 88. 28 33
.82. 8:. 82. 88.- 8:.- 88- 88.- 8:. 28.- 882an ease:
28. 82.- 82. 88.- 82.- 88? 88.- 88.- 88.- 88885 3538
88. .82.- 28.- 88. 88.- 88.- 88. 88.- 82.- been mm< ease:
88.- 22.- 88. 2.8.- 88.- 88.- N2 2.- 88.- E..- 288 m3. 3.53
82 7 88.- 82a.- 82. 88. 88. 2.8. 88. 82. 52282 neon:
28.- 88. e2 _. 88. 828. 2.8.- 8.8.- 82 .- 88. 582.82 3538
828 82 33

aaawﬁam.
:28.- 28.- 88..- 22. 88. 2.2. 88.- 88. - 6880 ans-.2
88. :88.- 88. 88. 82. .22.- 28. 82.- 8.8. 88 33

ccchoov.
88. 2.8.- 58.- 82 .- 8:. 88. a: _ .- 88.- - 88.-8am 353a:
88. 88.- 82. 82. 88. 2.8.- 88. 22.- 88. 888an 3828
88. :8.- 8:. 82.- 8:. 88. 88.- 88. 88.- 288 m3. nabs:
22. 82 7 88. .82. 88. 82.- 2.8. :88- 88. 288 mm< 3288
2.8.- 88.- 2.8.- .82 .- 88. 88. 88.- 88. 82 .- 582.82 ease:
28.- 88.- N2 _. 88. 88.- 88. 88. 88.- 82 _. 22.2.82 _aeaaa

8T5 8.2 82
.682 822 .883. 8853. 323. 8:53 .888 8880
220 85 220 86 220 9 15222 .8888 been
328m 028805 838m 882
3.. .fw .J. ...-.-..< - . . .ltt... . a .. .. .. v.44-“ :.....~. ﬂaw—um

S 033-

92

 

88:: 02888: 1:222: me 8:88: .8 3:8... vow-882 05 23—8 0.8. 8.28 .88.: 2...: .8 8:80.: .829
888: 88...: 9:808... 1:222: .8 25.— 3.. 2.3 .58.: 2...: 8:222: 5. .23 8803:» 805 .8 $8580....

.80 mm... .3 mm: .m: mu.

 

 

 

 

 

0
8...: .88 8:88 .8: .88 .8: .88 8...: 88 .2: .88 .8: .88 3.: 88 28:88 58:8 2.5..
.8: .88 .8: .88 8...: .88 88: .88 8.: .88 .8: .8. .8: .88 5: .88 .8: .88 28 5..
.8: .88 .2: .82. 5:88 .2: .88 .8: .88 .8: .88 .8: .88 3 .82 8.: .88 5.5.5: .532
.8: .88 L2: .88 .8: x... .2: .88 .8: 88 .2: .88 8...: .88 .8: .88 88.: .88 558...: .25....
mm... mm... mm... :9. mm... 8.. 8a. .8... mm...
\3 0:02 \3 0:02 \3 0:07— \3 0:02 \3 0:02 \3 0:07— \3 0:02 \3 0:02 \3 0:02 0:00m mmd. 352::
3 .82 .8: .8. 3 .82 3 .82 .8: .88 3 .8 .8: .88. 3 .88. 8.: .8. 28... mm... .25....
.8: .\.8 .8: .8. :2: .88 3 .82 .8: .8... 81:82. 3 .82 .8: .88 .8: .88 55.8.... 2552
.2: .88 .2: .88 .8: .88 .8: .88 .8: .88 .8: x8 8...: .88 E: 88 .8: .\.8 5.2.8.... .85....
82".: .2... 33
£8,228. .2: .88 .8: .8. 3.: .8.- 88: .88 an: .88 .8: 88 2...: 8: .8: .82. 5850 2.5..
.8: .8: r8: .8... .8: .88 .2: .82. .8: .88 .8: .8... 8...: x8 .8: .88 ....: 8.... 88 33
.8: .8... .8: .8. .8: 8... .8: .88 .8: .88 .8: .88 88: .\.8 .8: 8:. .88: .88. 5.5.5: .2552
8...: .8... .8: .88 .2: .88 .8: x8 .8: .88 .8: .88 8...: .88 .8: .8: .8: .8. 5.5.5: .25....
.8: .88 .8: .88 .8: .88 8.: .88 .2: 88. .8: 8.... .8: .88 .8: .88 .8: 82. 88: mm... .2552
.8: .88 .8: 88 .8: .88 .8: 8:. .8: .82. .8: 8.... .8: .88 88: 8.8 .8: .2... .58 mm... .25....
.8: x... .8: .88 88.: .\.8 £8.88 .8: .88 .8: .82. .8: .88 .8: .88 88: .88 5.2.8.... 5.552
8.: .88 .8: .88 .8: 88 .8: .8: 88.: :8 .8: .88 .8: 8.8 .8: 88 8...: .8... 5.2.82 .55....
82".: .5. 8...
«28. 8.28: \ .88.: 02889.: 8:22—82 52>» 8.
52.8.... .85. 5.85.58... 5.5.2 :25... 5:53 5.2.... 5.8.8
2.8 2.8 2.8 2.8 2.8 o. .2552 5.5.8 2.5...
2. .2... - a. . ..... -s .u ._ .lu -. .vﬁJ

...-,. .a

. .«O 41.85:

_ _ 038,—.

93

Table 12
Summag of Multiple Reggession Analyses Predicting Child Reactivig:

Main Eﬁ‘ects of Risk Factors Within Risk Level Subggoups

 

 

Risk Level B: B
High Risk (n=108)
Risk Variables .16 2.34“
Low Risk (n=106)
Risk Variables .0 .90

 

‘95 05. "pg 01. "“23 001.

94

Table 13

Summg of Multiple Regression Analyses Predicting Family Cohesion:

Main Effects of Risk Factors Within Risk Level Subgroups

 

 

Risk Level 1:2 E
High Risk (n=108)
Risk Variables .36 6.80"”
Low Risk (n=106)
Risk Variables .11 1.49

 

*9: 05. "pg 01. “*23 001.

 

APPENDIX B

FIGURES

95

 

2:325 230 E... .5388; £5— 5.93qu— maEmgea—om 33853.»: a: .252 .. — 253m

 

 

 

 

«Eu—co:— 3323 mes—uﬁoﬁo

 

 

£8.33 HESS:

comma—*8 bun—é

“55835 089— mesa—=53".

 

 

 

328m 028805
“nogogam £55m

 

 

 

05830 2.5

 

 

sacs 3E8

mMm 26.

€239.93 BBS“
Sum—£0855 €39.83
awe—osaanonommq _quSE

Acommmoaoe 8:28
.b=a_oom:§ .Emzogoo—av
awe—ogaaqoaomma 3588

 

 

22:82—52 220

O— 220

 

 

 

888m o>uo£o£
EEO

€28.“ v6.3

 

.3954 “EM— ? 55:5an 2: .35

mac—a9..— ucgaaom mama—«Evin.— ES V7.2 5959m— aEmncquy— amino—=3»: - N 9...»:—

_o>3 “Em
.35 page: 33

 

 

2:03on
Branom
wEE—wEme

 

97

 

2954 «EM .3 8:55.80 2: .35

5398:...— EE :5— uooszom 9523.33— 382: conﬁne—=3»: .. n 9...»:—

Eﬁ ﬁx
:2: 5:602 33 od

 

5588.5 am am?—
5953— 5322.80

 

o._-

98

mac—noun .8339: mama—«Egan 230 E; 83—2—82 _anua—Z

5950m— nEmueua—oy— 05 .8 33:23.: EEO u: 30...:— wnuauoeo—z .v 9.5»:—

Ew__ocoo_< $5922 a0 8535

 

mo> oz
26.. m 3
a ll 4 w
c _I \\\
I \\\\\ .2 ml.
238mm 230 \ \ \ \ \ l m.”
\\\ IN? 8
3
U.
B
.3 w.
H
.8 W
mm
m
8 s

 

 

 

99

mac—acum 3330mm mus—«Eoﬁm EEO ES b=£oomua< _aESam
:ooEom augmuotﬂom 2: no 5653< 250 no 80...:— wEﬁB—SE - m unsur—

b__m_oomuc< _mEQmm ho mocommi

 

mm; 02
:9: a
26.... l \ \ l 1.9

8:5? pilo.

 

swe|q01d Jomeuag ﬁugzuewelxa

 

 

 

 

mo>
:9:

26..

2:59am? 23m

26:38? .2565. B 8585

O
(D ID V 1") N ‘- 0 032
1- 1- 1— u— 1- 1- 1-

swelqwd Jomeuea Buszneweixa

 

-e 2a»...—

 

101

mac—ac...” hogs—_om anﬁﬁﬁoim @536on

552.3— =£mmoumom 2: mo 2.2m E ESE—«ohm .«o Ewan—3mm -2 Ban:—

muouoﬁm v—mmm mo 200852”

 

 

 

 

m 0 m" D
_ m 1 -T .
o T Av
D 0 _. C
W 0 O 0 n C
m 0 MW . c
I 0 o 0 . lh
o a m w m m C:
a m w 3 w a
m w m w m" a
I a a w w 1:3
m o m M" .v
0
i 0 o 0 O O “Cacao—Ohm
his. 0 W W 0 U LWPN
O O n. 0 homxrnnmom
O WENSNGHDme
n-
I' o I.
m 9 mm
0
_ _ mm

 

 

 

REFERENCES

References

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

Achenbach, T. M., & Edelbrock, C. S. (1983). Mgual for the child behavior
checklist and revised child behavior proﬁle. Burlington, VT: T.M. Achenbach Publishing.

Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable
distinction in social psychological research: Conceptual, strategic, and statistical

considerations. Journal of Personality and Social chhology, 51(6), 1173-1182.

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 8 77-100.

 

Bernstein, V., & Hans, S. (1994). Predicting the developmental outcome of two-
year-old children born exposed to methadone: Impact of social-environmental risk

factors. Journal of Clinical Child Psychology, 23, 349-359.

Bird, H. R., Gould, M. S., Rubio-Stipec, M., Staghezza, B. M., & Canino, G.
(1991). Screening for childhood psychopathology in the community using the Child
Behavior Checklist. Journal of the American Academy of Children and Adolescents,
3011), 116-123.

Bischof, G., Stith, S., & Whitney, M. (1995). Family environments of adolescent
sex offenders and other juvenile delinquents. Adolescence, 301117), 157-170.

Black, A E., & Pedro-Carroll, J. (1993). Role of parent-child relationships in
mediating the eﬂ‘ects of marital disruption. Journal of the American Academy of Child and
Adolescent Pachiatry, 3215 1, 1019-1027.

Blackson, T. C., & Tarter, R. E. (1994). Individual, family, and peer aﬁiliation
factors predisposing to early-age onset of alcohol and drug use. Alcoholism: Clinical and

Experimental Research, 18(4), 813-821.

103

Bolger, K. E., Patterson, C. J., Thompson, W. W., & Kupersmidt, J. B. (1995).
Psychosocial adjustment among children experiencing persistent and intermittent family
economic hardship. Child Development, 66, 1107-1129.

Bond, C. R., & McMahon, R. J. (1984). Relationships between marital distress
and child behavior problems, maternal personal adjustment, maternal personality, and
maternal parenting behavior. ourn f Abn rm P ha] 93 3 348-351.

Bradley, R. H., & Caldwell, B. M. (1979). Home observation for measurement of
the environment: A revision of the preschool scale. Amerim 19mm! pf Meng

Deﬁciency, 84(3 1, 235-244.

Bradley, R. H., & Caldwell, B. M. (1984). The HOME inventory and family
demographics. D vel men P ch 1 20 2 315-320.

Bradley, R. H., & Caldwell, B. M. (1981). The HOME inventory: A validation of
the preschool scale for black children. Child Developmgm, 52, 708-710.

Bradley, R. H., Caldwell, B. M., Rock, S. L. (1988). Home environment and
school performance: A ten-year follow-up and examination of three models of
environmental action. Child Development, 59, 852-867.

Bradley, R. H., Caldwell, B. M., Rock S. L., & Harris, P. T. (1986). Early home
environment and the development of competence: Findings from the Little Rock
Longitudinal Study. hildren’ Envir nm n u r1 3 1 10-22.

Bradley, R. H., Whiteside, L., Mundfrom, D. J ., Blevins-Knabe, B., Casey, P. H.,
Caldwell, B. M., Kelleher, K. J ., Pope, S. K., & Barrett, K. (1995). Home environment
and adaptive social behavior among premature, low birth weight children: Alternative

models of environmental action. Journal pf Pediatric Psychology, 203 1, 347-362.

Bradley, R. H., Whiteside, L., Mundfrom, D. J., Casey, P. H., Kelleher, K. J ., &
Pope, S. K. (1994). Contribution of early intervention and early caregiving experiences to
resilience in low-birthweight premature children living in poverty. Journal pf Clinipal

Child Psychology, 23, 425-434.

Caplan, H. D. Moses, (1996). Subtypes of alcoholic women. Unpublished
doctoral dissertation. Department of Psychology, Michigan State University, East
Lansing, MI.

Chase-Landsdale, P. L., Wakschlag, L. S., & Brooks-Gunn, J. A. (1995). A
psychological perspective on the development of caring in children and youth: The role of

the family. Journal of Adolescence, 18(5 1, 515-556.

104

Chassin, L., Rogosch, F., & Barrera, M. (1991). Substance use and
symptomatology among adolescent children of alcoholics. Journal of Abnormal

Psychology, 100(4), 449-463.

Corbo-Richert, B. H. (1994). Coping behaviors of young children during a chest
tube procedure in the pediatric intensive care unit. Matemal-Child Nursing Journal, 22(4),
134-146.

Corrao, G., Busellu, G., Valenti, M., & Lepore, A. R. (1993). Alcohol-related
problems within the family and global functioning of the children: A population-based
study. Social P chiat & P chiatric E idemiolo 28 6 304-308.

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. Journal of Abnormal Child Pachology, 19(5), 553-567.

Dillon, W., & Goldstein, M. (1984). Multivariate Analysis: Methods and
Applications. New York: John Wiley & Sons.

Donovan, J. M. (1986). An etiologic model of alcoholism. The American Journal
ofPsychiatigy, 143(1), 1-11.

Downey, G., & Coyne, J. C. (1990). Children of depressed parents: An
integrative review. Psychological Bulletip, 108(1), 50-76.

Downey, G., & Walker, E. (1992). Distinguishing family-level and child-level
inﬂuences on the development of depression and aggression in children at risk.
Development and Psychopathology, 4, 81-95.

Ellis, D. A., Bingham, C. R., Zucker, R. A, & Fitzgerald, H. E. (1996).

Typologjcal differences in patterns of risk load and child adaptation among young
alcoholic families. Unpublished manuscript, Michigan State University, East Lansing.

Erel, 0., & Burman, B. (1995). Interrelatedness of marital relations and parent-
child relations: A meta-analytic review. Pachological Bulletip, 118(1), 108-132.

Fergusson, D. M., & Lynskey, M. T. (1996). Adolescent Resiliency to family
adversity. Journal of Child Psychology and Psychiatg, 37(3 ), 281-292.

Fitzgerald, H. E., Sullivan, L. A., Ham, H. P., Zucker, R. A., Bruckel, S., &
Schneider, A. M. (1993). Predictors of behavior problems in three-year-old sons of
alcoholics: Early evidence for the onset of risk. Child Development, 64, 110-123.

105

Fitzgerald, H. E., Zucker, R. A, Yang, H. (1995). Developmental systems theory
and alcoholism: Analyzing patterns of variation in high-risk families. P h l f

Addigp'vg Bahavipra 9(1), 8-22.

Frick, P. J., Lahey, B. B., Loeber, R., Stouthamer-Loeber, M., Christ, M. An. G.,
& Hanson, K. (1992). Familial risk factors to oppositional deﬁant disorder and conduct
disorder: Parental Psychopathology and maternal parenting. Journal pf Conaalting and

Cliniﬂ Pachplpgy, 60(1), 49-55.

Garmezy, N. (1983). Stressors of childhood. In N. Garmezy & M. Rutter (Eds)

MW (99 43-84) Baltimore MD The Johns
Hopkins University Press.

Garmezy, N., Masten, A S., & Tellegen, A (1984). The study of stress and
competence in children: A building block for developmental psychopathology. Child

Devalopment, 55, 97-111.

Grizenko, N., & Fisher, C. (1992). Review of studies of risk and protective
factors for psychopathology in children. Canadiap 1;)qu prsyahiam, 37, 711-721.

Grossman, F. K., Beinashowitz, J., Anderson, L., Sakurai, M., Firmin, L., &
Flaherty, M. (1992). Risk and resilience in young adolescents. Journal of Youth and

Adolescence, 21(5), 529-550.

Hammen, C., Burge, D., & Stansbury, K. (1990). Relationship of mother and
child variables to child outcomes in a high-risk sample: A causal modeling analysis.
Dev 10 men P ch 1 6 1 24-30.

Hauser, R. M. (1994). Measuring socioeconomic status in studies of child
development. Child Development, 65, 1541-1545.

Herrenkohl, E. C., Herrenkohl, R. C., & Egolf, B. (1994). Resilient early school-
age children from maltreating homes: Outcomes in late adolescence. American Journal of
Ortho chiat 64 2 301-309.

Hill, S. Y., & Muka, D. (1996). Childhood psychopathology in children from

families of alcoholic female probands. Jpprnal pfthe American Aaadamy of Child md
Adole n P chiat 35 6 725-733.

Jacob, T., & Leonard, K. (1986). Psychosocial functioning in children of alcoholic
fathers, depressed fathers and control fathers. Jpprnal of Studies pn AlQhol, 47(5), 373-
380.

Jansen, R. E., Fitzgerald, H. E., Ham, H. P., & Zucker, R. A. (1995). Pathways
into risk: Temperament and behavior problems in three- to ﬁve-year-old sons of

106

alcoholics. Alcoholism: Clinical and Experimental Research, 19(2), 501-509.

Jessor, R., Van Den Bos, J ., Vanderryn, J., Costa, F. M., & Turbin, M. S. (1995).
Protective factors in adolescent problem behavior: Moderator eﬂ‘ects and developmental

change. Developmental Psychology, 31(6), 923-933.

Johnson, J. L., Sher, K. J., & Rolf, J. E. (1991). Models of vulnerability to

psychopathology in children of alcoholics. Alcohol Health & Research World, 15(1), 33-
42.

Johnson, 8., & Leonard, K. E., & Jacob, T. (1989). Drinking, drinking styles and
drug use in children of alcoholics, depressives and controls. Journal of Studies on

Alcohol, 50(5), 427-431.

Jouriles, E. N., Pﬁﬂiier, L. J., & O’Leary, S. G. (1988). Marital conﬂict,

parenting, and toddler conduct problems. Journal of Abnormal Child Psychology, 16(2),
197-206.

Katz, L. F., & Gottman, J. M. (1993). Patterns of marital conﬂict predict
children’s internalizing and externalizing behaviors. Developmental Psychology, 29(6),
940-950.

Kazdin, A. E., & Kagan, J. (1994). Models of Dysfunction in Developmental
Psychopathology. Clinical Psychology: Science and Practice, 1(1), 35-52.

Kyrios, M., & Prior, M. (1991). Temperament, stress and family factors in
behavioral adjustment of three- ﬁve-year-old children. In S. Chess & M. E. Hertzig

(Eds), Annual Proggess in Child Psychiatgy and Child Development (pp. 285-311). New
York: Brunner/Mazel Publishers.

Lavigne, J. V., & Faier-Routman, J. (1993). Correlates of psychological
adjustment to pediatric physical disorders: A meta-analytic review and comparison with

existing models. Journal of Developmental & Behavioral Pediatrics, 14(2), 117-123.

Lempers, J. D., Clark-Lempers, D., & Sirnons, R. L. (1989). Economic hardship,
parenting, and distress in adolescence. Child Development, 60, 25-39.

Lerner, R. M., Palermo, M., Spiro, A., & Nesselroade, J. R. (1982). Assessing the
dimensions of temperamental individuality across the life span: The Dimensions of
Temperament Survey (DOTS). Child Development, 53, 149-159.

McClelland, G., & Judd, C. (1993). Statistical diﬁiculties of detecting interactions
and moderator eﬂ‘ects. Psychological Bulletip, 114(2), 376-390.

107

Masten, A S., Garmezy, N., Tellegen, A, Pellegrini, D. S., Larkin, K., & Larsen,
A (1988). Competence and stress in school children: The moderating eﬂ‘ects of

individual and family qualities. Joamﬂ pf Child Pachplogy and Pachiatg, 29(6), 745-
764.

Merikangas, K. R., Prusoﬁ‘, B. A, & Weissman, M. M. (1988). Parental
concordance for aﬂ‘ective disorders: Psychopathology in oﬂ‘spring. Jgrnal of Aﬁ‘ective

Digrdgs, 15(3 ), 279-290.

Moos, R. H., & Moos, B. S. (1986). Famjly Enytt'p' nmant Scalp Mgpal (2nd (31. ).
Palo Alto, CA: Consulting Psychologists Press.

Moses, H. D. (1992). n m e r fth r 1 ti nshi fth
aJmhpl prpblema and ahild bahavior problema in a high-riak pppulatipn. Unpublished

master’s thesis, Michigan State University, East Lansing.

Mueller, C. W., & Parcel, T. L. (1981). Measures of socioeconomic status:
Alternatives and recommendations. Child Development, 32, 13-30.

Neter, J ., Wasserman, W., and Kutner, M. (1990). Applied Linear Statimcal
Modals. Boston, MA: Irwin Publishing Company.

Pettit, G. S., & Bates, J. E. (1989). Family interaction patterns and children’s
behavior problems from infancy to 4 years. Davelopmental Pyohplogy, 25(3), 413-420.

Phares, V., & Compas, B. E. (1992). The role of fathers in child and adolescent
psychopathology: Make room for daddy. Pachologigal Review, 111(3 ), 387-412.

Phipps, S., & Mulhem, R. K. (1995). Family cohesion and expressiveness
promote resilience to the stress of pediatric bone marrow transplant: A preliminary report.

Developmentﬂ and Behavioral Pediatriaa, 16(4), 257-263.

Rak, C. F., & Patterson, L. E. (1996). Promoting resilience in at-risk children.
qurnal pf Cppnsaling apd Development, 74, 368-373.

Richters, J ., & Pellegrinni, D. (1989). Depressed mother’s judgments about their
children: An examination of the depression-distortion hypothesis. Child Devalppment,
QC), 1068-1075.

Robins, L. N., Helzer, J. H., Croughan, J., & Ratcliﬂ K. S. (1981). National
Institute of Mental Health Diagnostic Interview Schedule: Its history, characteristics, and

validity. Archivas pf ﬁneral Psychiatty, 38, 381-389.
Rogosch, F., Chassin, L., & Sher, K. J. (1990). Personality variables as mediators

108
and moderators of family history risk for alcoholism: Conceptual and methodological

issues. Journal of Studies on Alcohol, 51(4), 310-318.

Rubenstein, J. L, Heeren, T., Housman, D., Rubin, C., & Stechler, G. (1989).
Suicidal behavior in “normal” adolescents: Risk and protective factors. American Journal
of Ortho chiat 59 1 59-71.

Rubia-Stipec, M., Bird, H., Canino, G., Bravo, M., & Alegria, M. (1991).
Children of alcoholic parents in the community. Journal of Studies on Alcohol, 52 (1), 78-
88.

Rutter, M. (1987). Psychosocial resilience and protective mechanisms. American

Journal of Orthopgzchiatg, 57(3), 316-331.

Sameroﬂ‘, A. J ., & Seifer, R. ( 1983). Familial risk and child competence. Child
Development, 54, 1254-1268.

Sarneroﬂ‘, A J., Seifer, R., Baldwin, A, & Baldwin, C. (1993). Stability of
intelligence ﬁom preschool to adolescence: The inﬂuence of social and family risk factors.

Child Development, 64, 80-97.

Sameroﬂ‘, A. J., Seifer, R., Barocas, R., Zax, M., & Greenspan, S. (1987).
Intelligence quotient scores of 4-year-old children: Social-environmental risk factors.
Pediatrics 79 343-350.

 

Seifer, R., Sameroﬁ‘, A J., Baldwin, C. P., & Baldwin, A (1992). Child and
family factors that ameliorate risk between 4 and 13 years of age. Journal of the American

Academy of Child and Adolescent Psychiatg, 31(5), 893-903.

Selzer, M. L., Vinokur, A, & van Rooijen, L. (1975). A self-administered short

lvﬁchigan Alcoholism Screening Test (SMAST). Journal of Studies on Alcohol, 36(1),
117-126.

Sher, K. (1991). Children of Alcoholics: A Critical Appraisal of Theog and

Research. Chicago: University of Chicago Press.

Sher, K., & Trull (1996). Methodological issues in psychopathology research.
Annual Review of Pachology, 47, 371-400.

Smith, J ., & Prior, M. (1995). Temperament and stress resilience in school-age
children: A within-families study. Journal of the American Academ of Child and

Adolescent Psychiatﬂ, 34(2), 168-179

 

109
SPSS, Inc. (1996). Statistics: Systat 6.0 for Windows. Chicago, IL: SPSS, Inc.

Sroufe, L. A, & Rutter, M. (1984). The domain of developmental
psychopathology. Child Development, 55, 17-29.

Stanger, C., Achenbach, T. M., & McConaughy, S. H. (1993). Three-year'course
of behavioral/emotional problems in a national sample of 4- to 16- year-olds. Predictors

of signs of disturbance. Jaumg of Cpnsplting apd Clinian Payahplpgy, 61(5), 839-848.

Stevens, 6., & Featherman, D. L. (1981). A revised socioeconomic index of
occupational status. Saga! Sciengp Rama, 10, 364-395.

Tabachnick, B. G., & Fidell, L. S. (1996). Using Multivariata Statigiaa, New
York: HarperCollins College Publishers.

Terman, L. M., & Merrill, M. A (1973). Stanfprd-Binet Intelligeng Sade:
Manual for tha Third Revision Form L-M. Chicago: The Riverside Publishing Company.

Thomas, A, & Chess, 8. (1981). The role of temperament in the contributions of
individuals to their development. In R. M. Lerner & N. A. Busch-Rossnagel (Eds),

Individuals as producers of their develppment: A life-span perspective. New York:
Academic Press. ..

Tschann, J. M., Kaiser, P., Chesney, M. A, Alkon, A, & Boyce, W. T. (1996).
Resilience and vulnerability among preschool children: F arnily functioning, temperament,

and behavior problems. Journal of the American Academy of Child and Adolescent
chhiatgt, 35(2), 184-192.

Varni, J. W., Rubenfeld, L. A, Talbot, D., Setoguchi, Y. (1990). Family
functioning, temperament, and psychologic adaptation in children with congenital or

acquired limb deﬁciencies. Annuﬂ Progtess in Child Psychiatty and Child Development,
279-291.

Walker, E., Downey, G., & Bergman, A (1989). The eﬂ‘ects of parental
psychopathology and maltreatment on child behavior: A test of the diathesis-stress model.

Child Developmant, 60, 15-24.

Warner, V., Mufson, L., & Weissman, M. (1995). Oﬁ‘spring at high and low risk
for depression and anxiety: Mechanisms of psychiatric disorder. Journal of the American

Academy of Child ES! Adolescent Psychiatgg, 34(6), 786-797.

Weist, M. D., Freedman, A. H., Paskewitz, D. A, Proescher, E. J ., & Flaherty, L.
T. (1995). Urban youth under stress: Empirical identiﬁcation of protective factors.

Journal pf Youth and Adolescence, 24(6), 705-720.

110

Werner, E. E. (1986). Resilient oﬁ'spring of alcoholics: A longitudinal study ﬁom
birth to age 18. Jotm of Studies an Alcohpl, 47, 34-40.

Werner, E. E.(l990). High-risk children in young adulthood: A longitudinal study
ﬁom birth to 32 years. In 8. Chess & M. E. Hertzig (Eds), Annpal Prpgteas in Child
chhiatly gtd Child Davelopment (pp. 180-193). New York: Brunner/Mazel Publishers.

Werner, E. E. (1992). The children of Kauai: Resiliency and recovery in
adolescence and adulthood. Journal of Adolagant Haalth, 13, 262-268.

West, M. O., & Prinz, R. J. (1987). Parental alcoholism and childhood
psychopathology. P h l ' B 11 ' 10 204-218.

Williams, S., Anderson, J., McGee, R, & Silva, P. A (1990). Risk factors for
behavioral and emotional disorder in preadolescent children. MAM

Agdm of Child gd Adplas_<‘&nt Psyghjatg, 29(3 ), 413-419.

Windle, M., Hooker, K., Lenerz, K., East, P. L., Lerner, J. V., Lerner, R. M.
(1986). Temperament, perceived competence, and depression in early and late
adolescents. Developmental Pacholdgy, 22, 384-3 92.

Wyman, P. A, Cowen, E. L., Work, W. C., & Parker, G. R. (1991).
Developmental and family milieu correlates of resilience in urban children who have

experienced major life stress. Amarim ngal pf Cpmmpnig Psychplpgy, 19(3 ), 405-
426.

Wyrnan, P. A, Cowen, E. L., Work, W. C., Raoof; A, Gribble, P. A, Parker, G.
R, & Wannon, M. (1992). Interviews with children who experienced major life stress:
Family and child attributes that predict resilient outcomes. derpaLpﬂhaApm

Academy of Child and Adolescent nghiatg, 31(5), 904-910.

Zucker, R. (1989). Is risk for alcoholism predictable? A probabilistic approach to
a developmental problem. Drugs and Spciaty, 3, 69-93.

Zucker, R. A. (1994). Pathways to alcohol problems and alcoholism: A
developmental account of the evidence for multiple alcoholisms and for contextual
contributions to risk. In R. Zucker, G. Boyd, & J. Howard (Eds), Tha Development pf
Al hol Problem : Ex lorin he Bio ch 3 ci M trix f Risk (pp. 254-289).

Zucker, R. A, & Fitzgerald, H. E. (1991). Early developmental factors and risk
for alcohol problems. A] hol Health R s h W r 15 1 18-24.

Zucker, R. A, Fitzgerald, H. E., & Moses, H. D. (1995). Emergence of alcohol

111

problems and the several alcoholisms: A developmental perspective on etiological theory
and life course trajectory. In D. Cicchetti & D.J. Cohen (Eds), Developmental

Psychopathology, Volume 2: Risk disorder and adaptation (pp. 677-711). New York:
Wiley.

Zucker, R. A, & Gomberg, E. S. L. (1984). Etiology of alcoholism reconsidered:
The case for a biopsychosocial process. American chhologist, 41(7), 783-793.

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

.Zucker, R. A, Noll, R. B. ,Ham, H. P. ,Fitzgerald, H. E., & Sullivan, L. S. (1994).
" and validi

 

studies. Unpublished instrument, Michigan State University, East Lansing, MI.

 

HIGQN STQTE UNIV. LI BRQR

llgllll 2lllgl 3l

gllll lgllgl llllL llglllglllgllll