 

 

Z LIBRARY
Michigan State
'9‘ ‘ :7 University

 

 

 

This is to certify that the
dissertation entitled

DOES THE MOTHER-CHILD RELATIONSHIP
MODERATE THE EFFECTS OF DOMESTIC VIOLENCE
ON PRESCHOOL BEHAVIOR PROBLEMS AND SOCIAL

COMPETENCE?

presented by

ROBIN PIERCE WEATHERILL

has been accepted towards fulﬁllment
of the requirements for the

PhD. degree in Psychology

 

 

/ /
I I 114 Mm

l . Professor’s Signatua
11! V5 Di,

Date

 

 

MSU is an Afﬁrmative Action/Equal Opportunity Institution

un-c-n-n-n-n-u--‘-v-u-u-~o

_-.___.._-.--.-..- -’.-.L.-._J_-.L.’.‘_—E.‘.-’-'-'-'-'- 9'- -'-‘-'-»L—~"_L.-._‘_-—o_- -

 

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATEDUE

DATEDUE

DAIEDUE

 

OCT
1901'?! 290%

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/07 p:/ClRC/DaleDue.indd-p 1

DOES THE MOTHER-CHILD RELATIONSHIP MODERATE THE EFFECTS OF
DOMESTIC VIOLENCE ON PRESCHOOL BEHAVIOR PROBLEMS AND SOCIAL
COMPETENCE?

By

Robin Pierce Weatherill

A DISSERTATION

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

DOCTOR OF PHILOSOPHY
Department of Psychology

2007

ABSTRACT
DOES THE MOTHER-CHILD RELATIONSHIP MODERATE THE EFFECTS OF
DOMESTIC VIOLENCE ON PRESCHOOL BEHAVIOR PROBLEMS AND SOCIAL
COMPETENCE?
By
Robin Pierce Weatherill
Children exposed to domestic violence (DV) are at risk for a variety of negative
outcomes in their social and emotional development, including depression, PTSD,
behavior problems, and poor social skills. However, not all children exposed to DV show
these problems, making it important to identify risk and resiliency factors that might
explain the variation in outcomes. The present longitudinal study examined whether
duration of exposure over the ﬁrst four years of life and two aspects of the mother-child
relationship, maternal representations of the child and mother-child attachment,
moderated the effects of DV on behavior problems and social competence in a
community sample of 174 mother-child pairs. Child behavior was rated by both mothers
and preschool teachers. Results indicated that exposure to DV predicted more behavioral
problems and less social competence, but more exposure did not necessarily predict more
severe problems. Contrary to expectations, neither attachment nor maternal
representations were related to child behavior problems in this high-risk sample. Maternal
representations were related to child social strengths, having a main effect on social

competence at the level of a trend, but the expected interaction effect with DV was not

found.

DEDICATION

To my parents, Warren and Francie Weatherill, and my brother, Philip H. Weatherill.

iii

ACKNOWLEDGMENTS

I am extremely grateful to my mentor and dissertation chair, Alytia A.
Levendosky, for her steady support and encouragement through many challenges, her
commitment to my personal and professional development, and for her infectious
enthusiasm for research and clinical work.

I am thankful to my parents for their continual support and encouragement. I have
been sustained throughout the process by my friendships with other students who worked
on the Mother-Infant Study: Alissa Huth-Bocks, Sally Theran, Erika DeJonghe, Kerry
Leahy, Carolyn Dayton, and Archana Basu. Finally, I have to thank Jeni and Brian for
putting a roof over my head during a crucial time, Alison for her good counsel, Gita for
her inspiration, and Rupert for good company duringthe ﬁnal chapters.

Funding for this study was provided by Student Award Program grant #661 from
the Blue Cross and Blue Shield of Michigan Foundation. Additional funding was
provided by Dissertation Enhancement and Dissertation Completion Awards from the
MSU Graduate School. The larger study from which this sample was drawn was
supported by grants from the National Institute of Justice (#8-7958-MI-IJ) and the
Centers for Disease Control (ROI/CCR518519-01) to Alytia A. Levendosky, G. Anne

Bogat, William S. Davidson III, and Alexander von Eye.

iv

TABLE OF CONTENTS

List of Tables ...................................................................................... vii
Introduction ......................................................................................... 1
Attachment and Emotional Development ............................................. 4
Parenting Behavior and Infant Attachment .................................. 5

Working Models ................................................................. 6

Working Models and Parenting Behavior:

Transmission of Working Models from Parent to Child ................... 8

Attachment and Emotion Regulation ........................................ ll

Implications of Attachment for Developmental Outcomes ............... 13

Domestic Violence as a Risk Factor .................................................. 14

Use of Multiple Informants of Child Behavior ............................. 20

Two Potential Moderators of Domestic Violence ................................... 21
Attachment as a Moderator .................................................... 22

Maternal Caregiving Representations as a Moderator ..................... 24

Hypotheses and Rationale ............................................................... 26
Patterns of Exposure to Domestic Violence .................................. 27

Attachment ....................................................................... 28

Maternal Caregiving Representations ......................................... 29

Method .............................................................................................. 3O
Participants ................................................................................ 30
Procedures ................................................................................. 3O

Measures .................................................................................. 33

Domestic violence .............................................................. 33
Maternal caregiving representations ......................................... 34
Child attachment .................................................................................... 35
Preschool behavior problems ................................................. 36
Preschool social competence ................................................. 37

Use of categorical vs. dimensional data for attachment

and maternal representations ................................................ 38

Results ............................................................................................. 41
Missing Data ............................................................................. 41
Data Reduction ........................................................................... 42
Analyses .................................................................................. 44

Direct Effects of Domestic Violence. . ....................................... 44

Moderating Effects of Child Attachment Classiﬁcation

and Maternal Representations ................................................ 45

Discussion ......................................................................................... 48

Limitations ............................................................................... 54

Implications for Future Research ...................................................... 55
Appendices

A: Working Model of the Child Interview ........................................... 60

B: Strange Situation Protocol ......................................................... 65

References ........................................................................................ 76

vi

LIST OF TABLES

Table 1. Demographic Characteristics of the Sample .......................................... 67
Table 2. Sample Size of DV Groups ............................................................. 69

Table 3. Bi-variate Correlations and Descriptives for Continuous
Variables ..................................................................................... 70

Table 4. Size of Attachment and Maternal Representation Groups
for each DV Group7l

Table 5. Results of Planned Comparisons for Child Outcomes and DV Groups ........... 72

Table 6. Child Outcomes for Secure vs. Insecure Attachment Groups,
by DV Group .............................................................................. 73

Table 7. Results of Two-way ANOVA for WMCI and DV
Group ......................................................................................... 74

Table 8. Child Outcomes for Balanced vs. Non-balanced WMCI Groups,
by DV Group ............................................................................... 75

vii

INTRODUCTION

Studies have found that children exposed to domestic violence (DV) are at risk for
a number of negative outcomes in their social and emotional development, including
internalizing and externalizing problems, post-traumatic stress disorder, poor social
competence, and aggression (Edleson, 1999; Margolin, 2005). An estimated 33-10
million children are exposed to DV during childhood (B. E. Carlson, 1985; Fantuzzo,
Boruch, Beriarna, Atkins, & Marcus, 1997), and research has identiﬁed the negative
impact of DV on child development as a signiﬁcant public health problem (Margolin &
Gordis, 2000). However, not all children exposed to domestic violence show these
negative effects, making it important to identify the risk and resilience factors that might
explain why DV predicts poor outcomes for some children but not others. Potential
moderating variables include aspects of the violence itself, such as severity, contextual
variables such as parenting, and child characteristics such as gender or temperament
(Levendosky, Huth-Bocks, Shapiro, & Semel, 2003; Lieberman, Van Horn, & Ozer,
2005)

For example, the effects of DV on children may vary depending on its chronicity
over time, yet the few longitudinal studies to date that include childhood exposure to DV
tend to dichotomously group children into DV exposed vs. control (e.g., Ehrensaﬁ et al.,
2003), making it difﬁcult to identify the effects of exposure over time. Contextual factors
that affect child emotion regulation and social skills in the normal course of development
may interact with the traumatic experience of DV, moderating its effects. For example,
parental warmth and sensitivity, style of discipline, and modeling of emotion expression

are all associated with child emotion regulation and social competence (Denham, Zoller,

& Couchoud, 1994). These factors may help buffer the effect of DV, or be overridden in
the face of trauma. Similarly, child attachment, which is related to preschool peer
problems and social competence (Denham et al., 2002b), might have a protective effect
on development in the context of DV. Thus, the current study sought to test whether the
duration of exposure to DV over the ﬁrst four years of life would differentially affect
preschool emotional and behavioral outcomes. In addition, two aspects of the mother-
child relationship, maternal representations of the child and mother-child attachment,
were tested as potential moderators of the effects of DV on preschool outcomes.

The goal was to understand how DV and the mother-child relationship might
interact to inﬂuence the development of the child’s early preschool relationships with
teachers and peers. Therefore, the preschool outcomes examined in this study were
factors associated with success and difﬁculties in developing peer relationships:
internalizing and externalizing behaviors, and social competence (Casey, 1996).
Problems with internalizing and externalizing behaviors may be an early sign of difﬁculty
with emotional and behavioral regulation (N ICHD-ECCRN, 2004). On the other hand,
social competence reﬂects the age-appropriate ability to regulate emotion and express it
appropriately in order to successfully interact with others (Halberstadt, Denham, &
Dunsmore, 2001). One limitation noted in many studies of DV and children is that both
the DV data and child outcomes are rated by mothers, resulting in possible rater bias
(Edleson, 1999). For this study, behavior ratings were completed by preschool teachers as
well as mothers, thus providing cross-situational data as recommended in a recent review

of studies of domestic violence and children (Margolin & Gordis, 2000).

In order to design effective interventions for children exposed to domestic
violence, it is important to understand what factors inﬂuence the diverse outcomes seen
in prior studies. In the case of preschool children, it is likely that the mother-child
relationship may be a source of either increased vulnerability to or protection from the
effects of DV. The present study examined whether the chronicity of violence over the
ﬁrst four years of a child’s life moderated the effects of DV, and whether the internal
working models that the mother and child had of their relationship, also developed during
that time, acted as moderators. Results are interpreted in light of attachment theory,
which proposes that internal working models developed in the early caregiving
relationship are carried forward as the child develops. These working models serve as a
template for interpreting and guiding behavior in close relationships. Attachment theory
is thus used as a framework for understanding how the mother-child relationship might
moderate the effects of DV on young children’s ﬁrst relationships outside the home, with
teachers and peers.

Attachment researchers have long argued that security of attachment in infancy is
important for optimal development, in particular, for the development of social strengths
and deﬁcits. However, not all studies ﬁnd signiﬁcant consequences for insecure
attachment, and it has been argued recently that insecure attachment may be better
viewed as a risk factor in the context of other developmental liabilities (such as poverty)
than as a sign of psychopathology per se (Belsky & Fearon, 2002). Thus, this study
addresses two factors, exposure to DV and mother-child attachment, that have been
identiﬁed as having a signiﬁcant impact on child social development, but that have

proved ambiguous predictors of serious social problems. Strengths of this study include

the longitudinal measurement of DV during early childhood, observational measures of
the mother-child relationship, and cross—situational ratings of child behavior.
Attachment and Emotional Development

One of the ways young children learn to regulate their behavior in relationships is
through their early emotional experiences with their caregivers (Magai, 1999). During
their ﬁrst months of life, infants rely almost entirely on external sources, their parents, to
interpret and modulate their emotion for them (Cassidy, 1994). Infant’s ﬁrst experiences
of emotional arousal and recovery revolve around the caregiver’s presence or absence.
Bowlby (1988) argued that infants have an innate need for caregiver proximity, and that
this made evolutionary sense in that caregiver proximity was most likely to ensure the
survival of the infant. In order to maintain this proximity, infants engage in attachment
behaviors when distressed, such as approaching, reaching for the mother, crying, and
seeking to maintain contact. Once reunited, if the caregiver soothes the child, the
caregiver provides the infant with an experience of expressing and regulating negative
emotion that will inﬂuence the child’s future self-regulation. This makes separation and
reunion interactions between caregivers and children a rich source of information about
how well the dyad is able to regulate emotion (Field, 1994; Rosenblum, McDonough,
Muzik, Miller, & Sameroff, 2002).

Ainsworth (1978) found that children show individual differences in their patterns
of behavior around these separations and reunions. She identiﬁed three distinct patterns
which she hypothesized were the manifestation of the internal working model the child
has of themselves in relation to their caregiver. Secure children show distress at their

mother’s departure, are comforted by her return, and explore in her presence,

theoretically reﬂecting their conﬁdence in a benign and trusted caregiver. Avoidant
children appear indifferent to their mother’s departure and return. Ambivalent children
are upset when their mother leaves, and although they seek comfort when she returns,
they are not soothed by her presence the way secure children are. A fourth category,
Disorganized/Disoriented, was added later after it was noted that some children do not ﬁt
into one of the three standard classiﬁcations. These children appear not to have a
consistent strategy, but show contradictory behaviors, such as approaching the caregiver
with their back turned, or freezing in the middle of a gesture toward caregiver proximity
(Main & Solomon, 1990). These latter three categories are considered “insecure”, and are
thought to develop in reaction to inconsistent, intrusive, or otherwise insensitive
caregiving. The corresponding attachment behaviors reﬂect the child’s best effort to
ensure the proximity of such a caregiver.
Parenting Behavior and Infant Attachment

Ainsworth (1978) found that these infant attachment classiﬁcations are associated
with speciﬁc patterns of parenting behavior. In mother-child interactions, mothers of
secure children are noted to be responsive to their infants’ needs and appear to sensitively
read their infants’ cues signaling hunger, sleepiness, etc. Mothers of avoidant children
are less responsive and more rejecting of their children, while mothers of ambivalent
children are inconsistent, responding to their infants’ needs at some times, but not others
(Ainsworth, 1978; Van IJzendoorn, 1995). Thus, infants’ attachment behavior under
stress is believed to be adapted to accommodate what they learned to expect from their
mother (Ainsworth, 1978). Secure children are conﬁdent that their mother will be there

to meet their needs, so they can seek comfort from her when upset, and explore once they

are assured of her presence. Insecure children are more likely to experience
unpredictable parenting in which their needs must be adjusted to accommodate the
mother’s emotional state. Avoidant children learn that demanding comfort from their
mothers often brings rejection, so they adapt by minimizing their demands on their
mother. Ambivalent children appear to expect that the only thing that will ensure their
mother’s proximity is an angry demand for attention. Finally, the disorganized
classiﬁcation is found to be associated with parental maltreatment of the child (Lyons-
Ruth & Jacobvitz, 1999). Main and Hesse (1990) proposed that the disorganized,
disoriented behavior seen in this subgroup is due to what is described as “frightened and
frightening behavior” on the part of the mother. In addition to the higher incidence of
abusive behaviors, mothers in this group are also observed to engage in frightening
behaviors such as baring their teeth or sneaking up and startling the child. At the same
time, some mothers appear frightened of their child, for example, backing away from
their child or making a startled gasp during a play session. Main and Hesse argued that,
for these children, the mother is both a source of fear and a source of comfort, placing the
child in the confusing bind of seeking protection from the very person who ﬁ-ightens her.
Working Models

Insecure infant attachment behavior describes a variety of strategies for adapting
to insensitive parenting, but what underlies them all is the idea that the insecure child is
unsure that the mother will be there as a source of safety. According to the theory, this
sense of insecurity is then internalized as a belief that others in close relationships will
not reliably respond in a positive or predictable way to one’s need for closeness. In seeing

the infant’s attachment behavior as an adaptive response, Ainsworth (1978) interpreted it

as the beginnings of what Bowlby had called an “internal working model” of the world,
their expectancy of relationships. According to Bowlby, these internal working models
are internalized representations of the self and others in relationships, one’s beliefs, for
example, as to whether others can be trusted to respond to one’s needs, and whether one
is worthy of others’ care and attention. The infant carries this internal working model
forward as it grows and develops other relationships. The beliefs and expectations
described by this template are related to the child’s sense of self-worth, their belief that
they deserve to have their needs met, and colors expectations about future relationships.
For example, if they believe people are unreliable they may compensate by making few
demands on others so as not be disappointed, and instead strive for self-reliance.
Alternatively, they may interpret the caregiver’s inconsistency as an indication of their
lack of worth, and thus fail to internalize a belief that others will love them. In this case,
they may compensate by anxiously striving for evidence that others care about them,
leading to anxiety and hypervigilence in relationships. These internalized representations,
relatively stable within the individual, are what endure within the individual over time,
linking infant attachment behavior to preschool behavior and later attitudes about
attachment that can be assessed in the child, adolescent, and adult.

Main, Kaplan, and Cassidy (1985) deﬁne working models as both conscious and
unconscious rules for organizing and interpreting information about attachment
relationships, rules that also inﬂuence the emotions one is able to perceive and express in
relationships. Although internalized, these rules are not ﬁxed, but rather working models
in process, and can be reconstructed given new experiences and information. However,

the unconscious aspects of working models are difﬁcult to change, because they operate

outside of consciousness, and act as ﬁlters on the very information that might change
them. Nonetheless, studies have found that infant classiﬁcation is more likely to change
in response to major life changes in the family (Weinfeld, Sroufe, & Egeland, 2000), and
theorists argue that working models are subject to change in the context of new
environments throughout the lifespan (Bretherton & Munholland, 1999).
Working Models and Parenting Behavior: Transmission of Working Model from
Parent to Child

Much of the evidence that internal working models exert inﬂuence over
relationships, particularly parent-child relationships, comes from studies of adult
attachment representations of early caregiving relationships (Van IJzendoom, 1995). One
measure of these representations is the Adult Attachment Interview (AAI), a narrative
interview thought to tap early attachment experiences and how they have been
internalized over time, which is reﬂected both in the content and the coherency of the
narrative (Main & Goldwyn, 1984). Thus, security on the AAI is coded not just from
memories of a responsive caregiver but in the ability to integrate positive and negative
aspects of the experience. A number of studies have found signiﬁcant concordance
between adult attachment representations of early caregivers (as measured by the AAI)
and infant attachment classiﬁcation (as measured by the Strange Situation), suggesting
that parents transmit their working models of relationships to their children (Benoit &
Parker, 1994; Van IJzendoorn, 1992; Ward & Carlson, 1995). It has been hypothesized
that adult attachment representations inﬂuence sensitivity of parenting, and that this is the

link of transmission between representations and infant attachment status; a meta-analysis

found partial support for this, although parenting sensitivity did not account for all of the
concordance (Van IJzendoorn, 1995).

A similar thread of research has examined maternal caregiving representations of
the child and their predictive relationship with infant attachment. Conceptualizing the
development of working models across the lifespan, George and Solomon (1999)
hypothesized that, in the course of becoming parents, adults transform their working
models of self in relationship to a caregiver to a representation of self as caregiver.
These representations are speciﬁc to the mother’s view of herself and her child, and
reﬂect her ability to perceive her child as both an autonomous and dependent individual.

Researchers have developed interviews to assess these caregiving representations;
one such interview is the Working Model of the Child (Zeanah & Benoit, 1995). Using a
narrative interview coded for process as well as context, narratives are classiﬁed as
balanced (analogous to secure or autonomous), distorted (preoccupied or anxious) and
disengaged (dismissing or avoidant). Like other types of working models, these
representations are hypothesized to reﬂect a template of beliefs about relationships that
provide a way of interpreting others’ behavior (in this case, the child’s) and organizing an
appropriate response/action based on the interpretation. A balanced working model is
characterized by access to a wide range of affect, dominated by warmth and positive
emotion, but also the ﬂexible integration of negative feelings such as worry, frustration,
and anger. In contrast, a disengaged proﬁle is characterized by a defensive exclusion of
negative affect and denial of negative feelings or thoughts about the child. The
disengaged narrative reﬂects a pervasive impoverishment of detail about the child’s

personality or emotions, suggesting that the mother copes by repressing her awareness of

the child’s needs. This repression of awareness reﬂects an overall deactivation of the
caregiving system. On the other hand, a distorted WMCI narrative is characterized by
heightened emotion and anxiety that obscures a coherent picture of the child and the
parent’s relationship with the child. The narrative may be tangential and difﬁcult to
follow, as the parent is ﬂooded by strong emotion and unable to regulate the discourse.

These different caregiving representations have been found to be predictive of
concurrent and future infant attachment, with balanced representations predicting secure
infant attachment (Benoit, Parker, & Zeanah, 1997; Zeanah, Benoit, Hirshberg, Barton, &
Regan, 1994). In addition, a recent study found that maternal caregiving representations
were predictive of infants’ self-regulation of emotion in response to the Still Face
procedure (Rosenblum et al., 2002). Mothers with balanced representations had infants
who more quickly returned to a positive affective state following the distress exhibited
during the Still Face procedure. The authors concluded that maternal caregiving
representations reﬂect in part the mother’s ability to regulate her own affect, which, in
turn, is related to her ability to modulate her infant’s emotional arousal. The implication
is that the interaction between balanced mothers and their infants facilitates more
effective emotion regulation in the infant.

Taken together, these ﬁndings support the argument that internal working models
provide cognitive and affective maps of the relational world, shaping adaptive or
maladaptive strategies for managing strong emotions. Different attachment and
representation classiﬁcations reﬂect different patterns of coping strategies, which have
implications for the emotional development of the individual and how they manage

positive and negative affect. For parents, their own internal working

10

models/representations and affect regulation inﬂuence how they help their children learn
to manage these emotions.
Attachment and Emotion Regulation

One of the primary functions of mother-infant interaction is to provide affect
regulation for the infant, who initially relies almost completely on external regulation. A
mother modulates infant affect, arousing her infant by smiling and engaging with him,
and damping down stimulation by quieting or looking away when the child is
overstimulated. Sensitive parenting, of the kind associated with infant attachment,
requires that the mother adjust her behavior in response to the infant’s needs, maximizing
positive emotion by smiling and laughing in response to the infant’s expressions of
pleasure, and minimizing negative emotion by soothing a distressed child (Ainsworth,
1978; Cassidy, 1994). Ainsworth’s observers noticed that mothers of avoidant children
held their infants the same amount as mothers of secure children, but that they did it
when child wanted to explore, not when the child wanted to be held, suggesting that it
was the contingency of the behavior, not the act itself, that made the difference between
secure and insecure attachment (Ainsworth, 1978).

Secure children can rely on their mother to provide this kind of sensitive
responsiveness: however, insecure children have learned that their mothers may not
respond to their cues. Mothers of avoidant children are more likely to reject a distressed
child, failing to help minimize their distress, while mothers of ambivalent children are
more intrusive and, by ignoring the child’s cues, are likely to amplify the child’s distress
rather than relieve it. Attachment theorists argue that differences in child attachment

behavior represent different strategies for the regulation of negative affect associated with

separation from the caregiver (Cassidy, 1994; Weinfeld, Sroufe, Egeland, & Carlson,
1999). They hypothesize that avoidant behavior represents a deactivation of attachment
drives: the child avoids the mother, even averting their gaze, so as not to be rebuffed
(Schore, 1994). In contrast to this numbing strategy, ambivalent infants have adopted a
strategy of hyper-arousal to constantly monitor for potential abandonment on order to
protect against it. Although these strategies may be adaptive in the short term in the
context of their relationship with their mother, they may have long-term costs. For
example, ambivalent children may be so absorbed in monitoring their mothers’ faces that
they may restrict their exploration of the world (Weinfeld et al., 1999).

Maternal affective style may inﬂuence child emotion regulation as the mother
models the expression of emotion for the child. Maternal affect has been found to be
predictive of mother-child interaction and the child attachment (Izard, Haynes, Chisholm,
& Baak, 1991). In this study, Izard and colleagues found that mothers of secure infants
reported more positive emotion and less negative emotion than mothers of insecure
children, but were more open in expressing negative emotion in front of their infants.
Mothers of insecure children experienced more negative emotion but were less likely to
show it around their children. In addition, mothers of insecure children reported
expressing more positive emotion around their infants (than did mothers of secure
children), but they reported experiencing more negative emotion overall, suggesting they
were inhibiting the expression of negative emotion. These ﬁndings suggest that parental
expression of emotion inﬂuences the infant’s development of affect regulation, which, in

turn, is a factor in the child attachment.

12

Implications of A ttachment for Later Developmental Outcomes

Attachment is of interest to developmental researchers because studies have found
that infant attachment classiﬁcation is often predictive of important developmental
outcomes such as behavior problems, school functioning, and social competence.
Attachment classiﬁcation has been found to be relatively stable over time within
individuals, with classiﬁcation in infancy concordant with classiﬁcation at age six (Main
& Cassidy, 1988). Infant attachment classiﬁcation is also predictive of later behavior
problems. One study found that children who were classiﬁed as insecure as infants are
more likely to be socially withdrawn and aggressive with peers at age six (Renken,
Egeland, Marvinney, Mangelsdorf, & Sroufe, 1989). Children referred to a clinic for
disruptive behavior problems were four times more likely to classiﬁed as having an
insecure attachment than were children in a non-referred comparison group (Greenberg,
Speltz, Deklyen, & Endriga, 1992). A study of preschool social competence found that
insecure children were more likely than secure children to show unregulated anger and
were rated as lower in social competence by their teachers (Denham, Blair, Schmidt, &
DeMulder, 2002a). Disorganized attachment is associated'with higher rates of aggression
in preschool (Lyons-Ruth, 1996; Lyons-Ruth, Alpem, & Repacholi, 1993), child
behavior problems in elementary school and high school, and psychopathology and
dissociation in adolescence (E. A. Carlson, 1998).

However, not all insecure children show these developmental disadvantages
(Greenberg, 1999), nor do all insecure infants become insecure children: in general, the
insecure category tends to be less stable than the secure category (Weinfeld et al., 2000).

Belsky and Fearon (2002) note that the predictive relationship between early attachment

13

and later outcomes varies depending on the level of contextual risk in child’s
environment. Infant attachment tends to be more stable, in other words, more predictive
of later attachment classiﬁcation, in middle-class, low-risk populations. It is less stable in
high-risk samples in which families are poor, headed by single parents, and report more
major life changes (Egeland & Sroufe, 1981). In contrast, infant attachment is more
predictive of later developmental outcomes besides attachment, such as behavior
problems, social competence, and school performance, in at-risk samples. Studies using
middle-class, low risk samples ﬁnd fewer differences between secure and insecure
children (Belsky & F earon, 2002). Therefore, insecure attachment is not considered to be
equivalent to psychopathology, but rather acts as a vulnerability factor for the
development of later psychopathology in interaction with other contextual risk factors
(Weinfeld et al., 1999).
Domestic Violence as a Risk Factor

Studies show that children exposed to domestic violence (DV) tend to have more
externalizing behavior problems, such as aggression, as well as internalizing problems,
such as anxiety, withdrawal and depression (Grych, Jouriles, Swank, McDonald, &
Norwood, 2000; Kitzmann, Gaylord, Holt, & Kenny, 2003). Intemalizing problems
represent the child’s tendency to appear sad, withdrawn, or anxious, while externalizing
problems reﬂect oppositional or aggressive behavior toward others. A number of studies
have found that children exposed to DV show both types of problems at a higher rate than
their peers, although results may vary depending on gender, age, and contextual factors
(Edleson, 1999; Grych et al., 2000). Children exposed to DV may develop post-traumatic

stress disorder (PTSD) (Kilpatrick & Williams, 1998; Rossman, 1998). These and other

14

problems associated with DV have been noted as early as preschool and in children as
young as one year old, suggesting the impact of DV begins early (Levendosky et al.,
2003). In studies, including one using the current sample, trauma symptoms have been
noted as early as age one (Bogat, DeJonghe, Levendosky, Davidson, & von Eye, 2006;
Levendosky, Huth-Bocks, Semel, & Shapiro, 2002). In the present sample as well as
others, maternal experience of DV has been found to be associated with insecure and
disorganized infant attachment (Huth-Bocks, Levendosky, Bogat, & von Eye, 2004;
Zeanah et al., 1999).

Childhood exposure to DV has been estimated to affect 3.3 to 10 million children
(B. E. Carlson, 1985; Fantuzzo et al., 1997). Very young children may be
disproportionally at risk for exposure to DV as compared to older children: a national
survey found that households where domestic violence occurred were more likely to be
households with children, particularly children under the age of ﬁve (F antuzzo et al.,
1997). A recent random-sample telephone survey of 453 couples with children between
the ages of four and seven found that 30% of households reported violence occurring
between family members (Smith, Thompson, DeVellis, Earp, & Coker, 2002).
Levendosky, Huth-Bocks, Shapiro, and Semel (2003) found that exposure to domestic
violence had a negative effect on preschool child functioning as observed in a videotaped
interaction with the mother, although there was no direct effect on mother-reported
behavior problems. The authors suggested that because they found that DV negatively
affected parenting and, subsequently, attachment, that perhaps the child difﬁculties
associated with DV were only apparent within the mother-child interaction, but not

observable by the mother across different situations. A study of preschool children by

15

Lieberman, Van Horn, and Ozer (2005) found no direct effects of child exposure to DV,
but did ﬁnd that maternal PTSD mediated the relationship between overall exposure to
violence and child functioning. Similarly, using the current sample at age one, Bogat et
al. (2006) found that for mothers and infants who had been exposed to DV, maternal
trauma symptoms were predictive of infant trauma symptoms if the DV was severe
(threats of or actual physical violence). In a high-risk sample of 4-6 year olds whose
families had been referred for neglect or abuse, it was found that DV had a strong but
indirect impact on child functioning, mediated by the effects of DV on caregiver
functioning and caregiver interactions with the child (English, Marshall, & Stewart,
2003). Thus, the research on infant and preschool outcomes suggests that the mother-
child relationship mediates or moderates the impact of DV.

Studies of the effects of DV on children have primarily focused on behavior
problems, self-esteem, and depression (e.g., Grych et al., 2000; Stemberg, Lamb,
Guterman, & Abbott, 2006), and have not speciﬁcally examined the effects of DV on
children’s social skills in peer relationships outside the home. Longitudinal studies offer
some evidence that childhood exposure to DV may contribute to being in a violent dating
relationship in young adulthood (Ehrensaft et al., 2003), but less is known about DV and
early peer relationships. One study found that children exposed to DV had more peer
difﬁculties (McCloskey & Stuewig, 2001). Interestingly, in this study, children of
battered women had about the same number of ﬁiends as the comparison group, but their
mothers reported more conﬂict with peers and the children themselves reported more
loneliness and more conﬂict with a close friend. The authors hypothesized that the

relational nature of DV had a negative effect on the children’s capacity for close

16

relationships, although the children were still able to maintain superﬁcial relationships
with their peers. This suggests it is important to examine children’s ability to engage with
others in a positive way, in addition to any clinical or behavioral problems they may
have. Therefore, in addition to examining behavior problems, this study will also assess
social competence as rated by the classroom teacher. Social competence has been found
to be related to emotion regulation, problem solving, and abilities to interact positively
with peers (Denham, 2002).

Domestic violence may affect children’s capacity for relationships because they
are likely to be exposed when very young, when the working models of self and others in
relationships are ﬁrst established (Bowlby, 1988). Furthermore, it is likely that the
experience of witnessing DV is a chronic and highly interpersonal trauma, the kind of
trauma that Herman (1992) argues has persistent and devastating effects on a person’s
capacity for relatedness. Thus, domestic violence represents a combination of a number
of risk factors known to be associated with peer difﬁculties and aggression in children,
although the mechanisms for the association between exposure and child outcome are
unclear. It is also possible that the link between exposure to DV and child aggression
may be due a genetic risk for conduct problems in children of parents who show
antisocial behavior (J affee et al., 2005). Extemalizing problems in preschool have been
associated with a higher risk for continued aggression during school years and conduct
disorder in adolescence (NICHD-ECCRN, 2004). There is evidence that the difﬁculties
with interpersonal aggression may place children at risk for the development of later anti-

social behavior and for being in a violent or abusive romantic relationship themselves,

17

either as the recipient or perpetrator of abuse, when they become young adults (Ehrensaft
et al., 2003; Magdol, Mofﬁtt, Caspi, & Silva, 1998).

However, not all children show these problems (Edleson, 1999; Hughes & Luke,
1998; Stemberg et al., 1993); therefore researchers have sought to ﬁnd mediating or
moderating inﬂuences, such as parenting (Grych & F incham, 1990; Levendosky &
Graham-Bermann, 1998), and maternal mental health (Kilpatrick & Williams, 1998;
McCloskey, Figueredo, & Koss, 1995). Researchers have explored both direct and
indirect effects of domestic violence. In theory, DV may constitute a direct trauma, a
traumatic event in which the child fears for their life or that of their caretaker (Scheeringa
& Zeanah, 2001). It may also affect them indirectly, through the developing mother-
child attachment relationship or its effects on maternal mental health (Huth-Bocks et al.,
2004). Levendosky and Graham-Bermann (1998) found that parenting stress moderated
the effects of DV on child ﬁrnctioning, but Kilpatrick did not ﬁnd a mediating role for
maternal mental health. Thus, although there is some evidence that mother-child
relationship factors mediate or moderate child outcomes to a certain extent, the pathways
between DV, family variables, and the severity of child outcomes remain unclear.

Outcomes for children exposed to DV may also vary with the timing and duration
of the exposure. Herman (1992) argues that chronic trauma, especially trauma that is
experienced in the context of an intimate or care-giving relationship, has particularly
devastating effects on personality and the capacity for relationships. Thus, chronic
exposure to DV may have different effects than intermittent exposure. Studies that assess
degrees of DV typically compare severity but not duration of exposure. A recent study of

the current sample found that women with a history of domestic violence across several

18

romantic relationships had worse psychosocial outcomes than those who never
experienced partner violence or experienced it in only one relationship (Bogat,
Levendosky, Theran, von Eye, & Davidson, 2003). However, no studies to date have
examined the effects of varying degrees of chronicity on child development. Terr (1991)
described a different presentation of PTSD symptoms in children depending on whether
the trauma was an acute one-time event versus chronic exposure. Thus, chronicity may
be one factor that determines the outcome of exposure to DV.

It has been hypothesized that some child characteristics may moderate the effects
of DV, for example, some studies have found an interaction between DV exposure and
gender, but some studies ﬁnd effects primarily in girls (Stemberg et al., 2006), but this is
not always the case. Some have argued that girls show primarily internalizing problems
while boys show externalizing problems (Edleson, 1999; Yates, Dodds, Sroufe, &
Egeland, 2003), but in the Stemberg et al. study, girls showed more difﬁculty with both
types of behaviors.

The developmental status of the child at time of exposure may also play a role,
and continuous exposure over childhood may result in different symptoms emerging at
different times. For example, Yates et al. (2003) found that concurrent DV predicted
behavior problems in middle-school boys but adolescent dating violence was predicted
by preschool exposure. Studies that include children in a wide range of ages have found
some age interactions, but results vary: in a study of children ages 4-12, Hughes (1988)
found that preschoolers were reported to have the most problems, but a later study found
middle-school age children to show the most symptoms (Hughes, Parkinson, & Vargo,

1989)

19

Another study using the larger sample from which this study was drawn also
found the effects of DV overrode individual child characteristics. DeJonghe, Bogat,
Levendosky, von Eye, and Davidson (2005) found that while temperament moderated
infants’ reaction to simulated adult verbal conﬂict, infants who had been exposed to DV
were more distressed by conﬂict regardless of temperament. The authors concluded that
the effects of DV overwhelmed temperament.

Use of Multiple Informants of Child Behavior

The current study proposes to use both maternal and teacher ratings of child
behavior problems. Studies of interrater agreement on child behavior have found that it
may be affected by rater characteristics such as depression and culture, but also may
reﬂect true situational differences (Youngstrom, Loeber, & Stouthamer-Loeber, 2000).
For example, teachers may be more sensitive to acting-out behavior and parents more
sensitive to internalizing behaviors. However, one study found that teachers were more
likely to agree with other teachers on depressive symptoms, suggesting some situational
continuity in these behaviors (Epkins, 1996). Cross-informant ratings have been
recommended both in the DV literature and other studies where parents are both a
variable of interest as well as the reporters of child outcomes. For example, in a study of
the effects of corporal punishment on child behavior, Deater-Deckard et al., (1996) found
signiﬁcant differences between mother and teacher ratings of child behavior problems.
Suggested reasons for the differences were that mothers may have a biased view of their
children, either under- or over-reporting, and mothers may have a smaller comparison
group to refer to when rating their child than do teachers, who work with many children.

DV studies that have used multiple reporters typically ﬁnd low levels of agreement on

20

child behavior (Levendosky et al., 2003; Stemberg et al., 2006), with contradictory
results. Margolin (2005) argues that using multiple informants may be particularly
important when assessing aggression in children of battered women, as they may under-
report in an effort to deny the effects of domestic violence, or over-report if they
associate child aggression with the abuse they have received from the their partner.
Studies of domestic violence and child outcomes often use only one reporter, the mother,
of both the violence and the child’s behavior, especially when using infant or preschool
samples where the children are too young to use self-report measures (e.g., Bogat et al.,
2006; Lieberman et al., 2005). Distortion of the relationship between parent report and
child outcomes may also occur as a statistical artifact entirely outside the mother’s
intention or control, simply because of single reporter bias.

An additional advantage of obtaining teacher report is that the child’s behavior is
assessed across different situations. Cross-situational assessment, with more than one
reporter, may detect behavioral responses to DV that are adaptive at home, but
maladaptive at school (Margolin, 2005). For example, withdrawn or avoidant behavior
may be adaptive in home where conﬂict escalates quickly and unpredictably, but it
interferes with making friends and the ability to explore and learn at school. Finally,
school data may provide information as to how school programs may support healthy
social and emotional development in children at risk for aggressive behavior.

Two Potential Moderators of Domestic Violence

Domestic violence can interfere with the acquisition of developmentally

appropriate emotion regulation and social competence, leading to behavior problems. The

mother-child relationship may moderate these effects. Studies have found that various

21

components of the mother-child relationship, including both attachment behavior and
internal representations, reﬂect the child’s developing capacity for emotion regulation
and relationships. In addition to parenting, the mother-child attachment relationship has
been proposed as one of the underpinnings of differences in parental socialization of
emotions (Magai, 1999). In support of this, Denharn (2002) found that child attachment
classiﬁcation predicts social competence in preschoolers. Thus, in the normal course of
development, children learn to regulate their emotions and behavior in the context of their
relationships with their parents. Children in families where domestic violence occurs
may face additional challenges. In addition to mastering developmental tasks faced by
children in low-stress environments, they must cope with the potentially traumatic
experience of witnessing violence between their parents. Their difﬁculties may be
exacerbated because the person that would normally be the source of comfort and the
model for coping — the mother — is herself traumatized. Thus, both the mother’s
characteristics as a parent, in this case, her caregiving representation, and the attachment
relationship itself, may play an important role in the child’s ability to cope, and thus
moderate the child’s response to domestic violence.

The current study will examine two potential moderating factors of the mother-
child relationship: child attachment and maternal caregiving representations. It is
hypothesized that children with secure attachment and children of mothers with balanced
maternal representations of their child will experience fewer negative consequences to
their social development in the context of domestic violence.

Attachment as a Moderator

22

Early researchers of attachment in young children hypothesized that insecure
infants would become insecure preschoolers and demonstrate continuing problems with
emotion regulation, disruptive behavior, and related peer problems, and this was partially
conﬁrmed by empirical studies (Greenberg, Speltz, DeKlyen, & Endriga, 1991).
However, the ﬁndings were not consistent: often insecure children did not demonstrate
signiﬁcantly more problems than their secure peers. In addition, there was discontinuity
in attachment classiﬁcation within the individual over time: insecure infants became
secure toddlers and vice versa. Gradually, with the accumulation of longitudinal studies,
a pattern of interaction with environmental context and maternal experience emerged
(Belsky & Fearon, 2002). In middle class samples, attachment classiﬁcation was more
likely to be consistent over time, while in high-risk samples where the mothers’
circumstances might change (in terms of employment, (housing, and marital status), infant
attachment classiﬁcation was more likely to change (Sroufe, 2005). However, the reverse
was true for high risk samples: while early attachment classiﬁcation did not necessarily
predict later attachment classiﬁcation, it did predict later behavior and peer problems.
Belsky and Fearon (2002) proposed that attachment might be better understood as a
moderator of risk, acting as a protective or vulnerability factor in the context of other
risks to development. Thus, recent attachment research has shifted its focus to examining
attachment in context of risk factors such as parental depression (Adam, Gunnar, &
Tanaka, 2004).

A few studies have examined the potential moderating effects of parenting for it
child outcomes in the context of domestic violence on children, with mixed results.

Levendosky and Graham-Bermann (1998) found that parenting stress partially mediated

23

child problems in school-age children, while McCloskey, Figuerdo, and K033 (1995)
found that parenting support in families where domestic violence occurred did not buffer
the effects on school-age children. Only two studies have looked at the effects of
domestic violence on the attachment relationship in preschool children (Huth-Bocks et
al., 2004; Zeanah et al., 1999), and no studies have yet examined whether attachment
may moderate the effects of domestic violence on later childhood outcomes. Given the
importance of mother-child interactions and attachment for social and emotional
development, these may be important potential protective factors to examine in the
relationship between domestic violence and child outcomes.

Attachment appears to be a better predictor of behavior problems in high-risk
samples, suggesting that it may act as moderator of other risk factors (Belsky & Fearon,
2002). In this study, it was hypothesized that secure attachment will buffer the effects of
DV, while insecure attachment will increase vulnerability to poor preschool outcomes.
Maternal Caregiving Representations as a Moderator

The representations of self and other tapped by the Working Model of the Child
Interview (WMCI) are thought to reﬂect how one regulates affect, particularly negative
affect (Cassidy, 1994; Rosenblum et al., 2002). Most studies of maternal representations
of the child to date have focused on the predictive relationship between maternal
representations and child attachment, ﬁnding that balanced maternal representations are
associated with secure child attachment (e.g., Benoit et al., 1997; Huth-Bocks et al.,
2004). In mothers, these working models may affect they way they socialize their
children. However, little is known about the relationship between maternal

representations and child behavior outcomes other than attachment. In one of the few

24

studies on other outcomes, Rosenblum et al., (2002) found that maternal representations
predicted the infant’s ability to regain positive affect following the Still Face Procedure, a
laboratory-induced stressful situation for infants. Only infants of mothers classiﬁed as
having balanced representations returned to high levels of positive affect once the
stressful portion of the procedure had ended.

This suggests that maternal representations of the child, reﬂecting the mother’s
beliefs and attitudes, may enhance or diminish a child’s social development in the context
of domestic violence. For example, a child who is traumatized may turn to the mother for
comfort and support, and if she is able to maintain a balanced representation of the child
which is sensitive and accepting, the child may be able to recover. In support of this,
Lieberman (1999, 2004) found that therapy that changes a mother’s beliefs and
attributions about her child also helps the child’s recovery from trauma. Conversely, the
child of a mother who is classiﬁed as having non-balanced (i.e., distorted or disengaged)
representations may be particularly vulnerable to developing behavior problems and
deﬁcits in social competence in the context of domestic violence.

Inasmuch as DV is associated with negative working models, it may disrupt the
mother’s own regulation of affect and, in turn, her ability to help her child learn to self-
regulate. If the mother’s working model of her caregiving relationship with the child is
damaged by DV, it may interfere with the kind of sensitive parenting necessary to
provide the scaffolding for children to learn about the expression of emotions. Because
children’s ability to self-regulate is associated with social competence (Cassidy, 1994;
Denham et al., 2003), this may explain why children exposed to DV have been found to

have deﬁcits in this area.

25

Previous analyses on the current sample found that a self-report measure of
prenatal parenting behaviors showed no differences between battered and non-battered
groups, while an interview tapping maternal prenatal representations of the child found
signiﬁcant differences between the two groups, with battered women being more likely to
have negative representations of their child (Weatherill et al., unpublished manuscript).
This study found that battered women also had more insecure adult romantic attachment
style. These ﬁndings suggest that battered women have more negative working models
of themselves and others, including their working model of their child. Maternal
caregiving representions have been found to predict both subsequent and concurrent
infant attachment (Benoit et al., 1997). Another study using the current sample found that
security of maternal prenatal representations was negatively associated with DV:
speciﬁcally, women who experienced DV during pregnancy were more likely to be
classiﬁed as having distorted or disengaged (analogous to the insecure categories for
children) rather than balanced (secure) representations (Huth-Bocks et al., 2004). This
ﬁnding has important implications for child social development because mothers with
narratives classiﬁed as distorted or disengaged are more likely to have children with
insecure attachments (Benoit et al., 1997; Fonagy, Steele, & Steele, 1991). Therefore, the
effects of DV on children’s social and emotional difﬁculties may be moderated by the
mother’s working model of the caregiving relationship and by the child’s attachment
relationship with the mother.

Hypotheses and Rationale
Although numerous studies have documented the negative effects of exposure to

DV on young children, the mechanisms of effect and reasons for the wide range of

26

outcomes are not yet understood (Edleson, 1999; Margolin, 2005). The present study
sought to examine characteristics of the experience of DV (e.g., chronicity) as well as
potential protective factors in the mother-child relationship. Attachment theory was used
as a framework to understand the interaction of trauma and the caregiving relationship
and potential effects on social and emotional behavior in preschool children.

Patterns of Exposure to D V

It was hypothesized that domestic violence would negatively affect preschool
children’s outcomes, but that this effect would vary depending on the pattern of violence
exposure over time. A longitudinal design was used to assess violence over time,
speciﬁcally the ﬁrst four years of the child’s life, a time when the working models of self
and others are developed in the context of the mother-child relationship and then applied
to new relationships with peers and teachers. The range of exposure that young children
may experience, from short-term to chronic violence, may have very different effects on
their developing working models of relationships.

Exposure to violence was assessed at 4 time points: ages 1, 2, 3 and 4 (preschool).
Children were categorized into three groups: no exposure, exposure at one or two time
points (intermittent exposure), and exposure at 3 or 4 time points (chronic exposure). It
was predicted that children exposed to chronic violence would have the worst preschool
outcomes, non-exposed children would have the best outcomes, and children exposed
intermittently would fall in between in their degree of preschool behavior difﬁculties.
This was tested using two-way ANOVAs and planned comparisons. In addition, to
determine whether any effect seen in the chronic group was due to the fact that many in

the chronic group also experienced recent or concurrent DV during the past year, a two-

27

way ANOVA was used to compare children who had experienced DV at age four and
those who had been exposed prior to age four but not in the past year. This study used
cross-situational assessment to determine if exposure to domestic violence was associated
with difﬁculties in the home setting, at school, or both. Mother and teacher ratings were
assessed as separate outcomes.
Attachment

In light of recent research that suggests that attachment may affect child outcomes
in interaction with the context of development — high or low risk environments — a
moderating hypothesis was expected, such that the relationship between attachment and
child outcomes was predicted to be the most robust in the context of the risk factor of
DV. Child attachment behavior is theoretically an indicator of internalized
representations of self and relationships that the child carries forward into new
relationships. Therefore, in the current study, it was expected that attachment would
moderate the effects of domestic violence on child behavior problems and social
competence at school, as reported by the teacher. Speciﬁcally, a secure attachment
relationship was predicted to enhance a child’s ability to cope with the trauma and
maintain a normal developmental trajectory, while insecure attachment was expected to
exacerbate the negative effects of domestic violence on child behavior problems.
Previous studies have found that preschoolers with externalizing problems are more
likely to be insecurely attached (Greenberg et al., 1991), and that insecure attachment
may function as a risk factor for behavior problems in the context of other vulnerabilities
(Belsky & Fearon, 2002). Therefore it was predicted that there would be a signiﬁcant

interaction effect between attachment and DV group on child outcomes.

Maternal Caregiving Representations

In the present study it was hypothesized that balanced maternal representations would
buffer the negative effects of domestic violence on behavior problems in the home
context for children, for both chronic and intermittent exposure, and that non-balanced
representations would function as a vulnerability factor in these groups. As with
attachment security, it was expected that there would be an interaction between WMCI
classiﬁcation and DV group. Thus, it was predicted that children of mothers classiﬁed as
balanced and exposed to domestic violence would have fewer behavior problems than

children of non-balanced mothers who were exposed to domestic violence.

29

METHOD

Participants

The sample of 174 mother-child pairs used for this study was drawn from a larger
longitudinal study of 206 pairs (Levendosky, Bogat, Davidson, & Von Eye, 2000).
Subjects were included in this study if both mother and child completed the age four
interview and mothers had custody of their child. Demographically, the subset used for
this study did not differ signiﬁcantly from the total sample. For the larger study, women
were initially recruited during pregnancy to participate in a longitudinal study of the
effects of DV on women and their children. Women were enrolled in the study and ﬁrst
interviewed during their third trimester of pregnancy, and then each year following at the
time of their child’s birthday. The present study used data from the age one through age
four interviews (Time 1, 2, 3, and 4).
Procedures

Recruitment

Participants were recruited by posting ﬂiers in public areas and obtaining referrals
from hospitals and other programs that provide ﬁnancial, legal, and health services to
pregnant women or victims of DV throughout the tri-county Lansing area, including
rural, urban, and suburban areas. Recruitment efforts were made to include women who
were ethnically and socio-economically diverse by using recruitment sites that served
women with a variety of backgrounds and experiences. See Table 1 for participant
demographics.

Women contacted the study ofﬁce if they were interested in participating, at

which time a research assistant conducted a brief screening to determine eligibility.

30

Women who had not been involved in a romantic relationship for at least six weeks
during their pregnancy or who were under the age of 18 or over the age of 40 were
excluded. Women were excluded from the study if it was believed they would have
difﬁculty understanding questionnaires due to limited knowledge of the English
language.

The women were told the study was about women’s relationships with the
important people in their life, including partners, family members, and children, and that
if they participated in the study they would be asked about their thoughts and feelings
about relationships and recent life events, including DV. They were also told that they
did not necessarily need to have experienced DV in order to be eligible for the study, nor
did they have to be in an ongoing relationship at the time of the interview: “current
partner” at each interview was deﬁned as their current or most recent romantic
relationship that had lasted at least six weeks. ‘

The focus of the larger study was the effect of DV during pregnancy on infant
development, therefore, in addition to speciﬁcally recruiting women who had
experienced DV, the telephone screen was used to ensure that approximately half the
participants had experienced DV during pregnancy, and the other half had not. Women
were told that they would be asked some questions about themselves and their
relationships over the telephone in a ﬁve-minute interview. After approximately half the
sample had been recruited and interviewed (g=96), the Conﬂict Tactics Scale (Straus,
1979) began to be administered over the telephone during the initial screening procedures
in order to keep track of the number of battered and non-battered women. For the purpose

of recruitment, women were categorized as “battered” if they had experienced physical

31

violence (items 6-14) during pregnancy according to this measure. This screen was used
to exclude women who had not experienced DV during pregnancy once it was
determined that there were enough non-battered women to make up half of the target
sample of 2 1 0.

After 137 participants had been recruited and interviewed, it was discovered that
many of the “non-battered” women had experienced DV in a prior relationship. Thus, the
telephone screen was then also used to enroll women who had never experienced DV, in
addition to those who experienced violence in the current pregnancy.

Time 1/ age one interview. Research assistants contacted mothers by phone to
schedule an interview on or about the child’s ﬁrst birthday. Subjects were then informed
that portions of the 3-hour interview would be videotaped, and that they would receive
$75 and a giﬂ for their infant as payment for participating. Mothers and infants were
interviewed concurrently in most cases during a 3-hour interview at the project’s ofﬁces
on campus.

Time 2 and 3/ ages 2 and 3 interviews. Administration procedures were similar to
those described for Time 1. Women were contacted and invited to participate in an
interview to be held either at their homes or the project’s ofﬁces, and paid for their
participation.

Time 4 / age 4 interview. Administration procedures were similar to those
described for T1. Women were contacted and invited to participate with their child in an
interview to be held at the project’s offices on campus. Women were given $150.00 and
a gift for their child upon completion of the interview. In addition, a release of

information for the child’s school or daycare was requested: with permission, the child’s

32

teacher/daycare provider was contacted and asked to complete a behavioral checklist
about the child. Teachers were told only that the child’s family was participating in a
study of child development, but nothing else about the participant or the study. Teachers
were paid $10.00 cash when they returned the completed the form.

T raining for the interviews

Graduate and undergraduate research assistants received extensive training in
conducting the interviews. Interviewers attended a weekly training meeting throughout
their time on the project, and did two to ﬁve interviews under supervision until they
reached 95% reliability for standard administration of measures. At T1, research
assistants were trained to administer either the mother interviewing protocol or the infant
tasks protocol. Mother interviewers were trained to administer the questionnaire packet,
the Mother-Infant Interaction, the Working Model of the Child Interview (WMCI;
Zeanah, Hirshberg, Danis, & Brennan, 1994), and to perform the role of experimenters
and videotapers in the Strange Situation; infant interviewers performed the role of the
stranger fer the Strange Situation. To ensure standardization of administration tape-
recorded role-plays of the WMCI and videotapes of the Strange Situation were rated
according to a protocol checklist and all interviewers met or exceeded 85% reliability;
during data collection, 25% of each interviewer’s WMCI and SS tapes were reviewed to
prevent drift.
Measures

Domestic violence. This was collected at every Wave, T1 through T4. ﬁe
Severity of Violence Aga_ir_r§t Women Scales (SVAWS: Marshall, 1992) is a 46-item

questionnaire that assesses violent behaviors and threats the woman may have

33

experienced from her romantic partner, including physical and sexual violence.
Respondents rated their experiences of abuse on a 4-point scale ranging from “Never” (0)

99 ‘6

to “Many Times” (3). Examples of items include “threatened to hurt you, choked you,”
and “used a knife or a gun on you.” Each item has a physical harm impact weight
according to the level of seriousness, abusiveness, and aggressiveness of the act. Severity
of violence was calculated by multiplying each item’s frequency score by its physical
harm impact weight and summing the products (Marshall, 1992). Marshall reported high
internal consistency for a community sample; alphas in this study ranged from 92-95 for
the four time periods.

Exposure to violence was assessed at 4 time points: ages 1, 2, 3 and 4 (preschool)
by mother self-report. Women were asked to complete the scale for each partner they had
been involved with for more than 6 weeks in the previous year. This yielded a weighted
sum score for each wave, which was then converted into a dichotomous violence or no-
violence score for each wave. For this study, items describing physical violence or threats
of physical violence were used to determine DV. Based on the number of waves at which
mothers reported experiencing DV, children were categorized into three groups: No DV,
Intermittent DV (at 1 or 2 time points), and Chronic DV (at 3 or 4 time points).

Maternal caregiving representations. The Working Model of the Child Interview
(WMCI: Zeanah et al., 1994) is a semi-structured, hour-long interview that elicits
mothers’ perceptions of and thoughts and feelings about their infant and their relationship
with their infant. This was administered at T1 when the infant was one year old.

Interviews were audio-taped, transcribed, and coded by trained coders for both process

and content of the participants’ narratives (cf the AAI, George, Kaplan, & Main, 1985)

34

 

using 5-point Likert scales that assess the quality of the narrative (e. g. richness), content
(e. g. infant difﬁculty), and the affective tone of the narrative. Based on the pattern of
scores on these subscales, an overall classiﬁcation of the narrative as balanced,
disengaged, or distorted was determined. These subscales may also be summed to create
a continuous score. This has been used in one previous study (Huth-Bocks et al., 2004).
For the purpose of this study, disengaged and distorted groups were collapsed to form a
non-balanced group.

The author and another graduate-level research assistant were trained by a third
graduate assistant to code the scales and determine classiﬁcations using the system
developed by Zeanah, Benoit, Barton, and Hirshberg (1996). Coders were blind to the
violence status of the participants. Occasionally discussion of DV occurred in the
interview, but this was unusual. Coders received extensive training over a period of 6
months, during which they read relevant literature and attended a 90-minute weekly
meeting to review and discuss transcripts. They coded 18 interviews with the trainer
during this period, and were required to achieve 80% agreement with the trainer for the
three-way classiﬁcation. Disagreements were settled by consensus after the initial
training period. Inter-rater reliability was calculated using percent agreement and
Cohen’s kappa for overall classiﬁcations, based on 25% of the 201 transcripts. Final
inter-rater agreement averaged 80%, with a kappa of .62 (p < .001).

Child attachment. The Strange Situation procedure (Ainsworth, Blehar, Waters,
& Wall, 1978) is a 22-minute observational procedure in which the infant and mother
participate in eight episodes of play, separation and reunion. This was administered at T1

when the infant was one year old. This series of episodes is considered to be mildly to

35

moderately stressful for the baby and thus induces particular attachment behaviors, such
as crying, clinging, approach and withdrawal. The behavior of the infant as videotaped is
coded into one of four general types of attachment classiﬁcations: secure, insecure
avoidant, insecure ambivalent, (Ainsworth et al., 1978) and disorganized (Main &
Solomon, 1990). Coding was done by trained, reliable coders supervised by Susan Paris
of the NICHD project at the University of Washington, who had extensive experience
coding the Strange Situation. Coders achieved a 90% agreement rate on overall
classiﬁcations, with a kappa of .84 (p < .001 ).

Preschool behavior problems. The Child Behgior Checklist — Parent Report.
ages 4-18 (CBCL: Achenbach, 1991) is a 112-item instrument measures the child’s social
and emotional functioning. The mother completed this measure at T4 when her child was
four years old. The instrument yields eight subscales, two broadband subscales
concerning internalizing and externalizing behavior, and a total problem behavior score.
These analyses used the total score. The CBCL is a well-standardized measure with
extensive reliability and validity data and alphas in the .905 for internal consistency
(Achenbach, 1991). The internalizing cluster is made up of the Withdrawn/Depressed,
Somatic Complaints, Anxious/Depressed, Thought Problems subscales, and the
externalizing cluster consists of the Attention Problems, Social Problems, Delinquent
Behavior/Rule breaking, and Aggression subscales. Participants were given a list of
symptoms and asked to indicate how true the statement is for their child within the last
six months. Participants rated the symptoms on a 3-point scale from “Not True” to “Very
True or Often True.” Alphas for this study were .69 for externalizing, .86 for

internalizing, and .87 for the total score.

36

The Cﬁegiver-Teacher Report Form, ages 1 1/2 - 5. (C-TRF: Achenbach, 1997) is
a lOO-item measure is similar to the CBCL, yielding two broadband subscales concerning
internalizing and externalizing behavior, and a total problem behavior score. The child’s
preschool teacher completed this measure at Time 4 when the child was four years old.
Like the CBCL, the C-TRF is well-standardized, with alphas reported in the .905
(Achenbach & Rescorla, 2001). There are six cross-informant subscales shared with the
CBCL: Emotionally Reactive, Anxious-Depressed, Somatic Complaints, Withdrawn
(these make up the internalizing cluster), Attention Problems, and Aggressive Behavior
(these make up the externalizing cluster). There is also a Sleep Problems syndrome.
Alphas in this study were alphas in this study were .96 for externalizing, .84 for
internalizing, and .96 for the total score. These analyses used the total score, a sum of the
6 major subscales (excluding Sleep Problems).

Preschool social competence. The Social Cormretence and Behavioral Evaluation
— Short Form (SCBE-30: LaFreniere & Dumas, 1996) is a 30-item measure assesses the
child’s strengths and weaknesses in social interactions with peers and adults in a school
or daycare setting. It was completed by the child’s teacher or daycare provider at T4
when the child was four years old. The SCBE-30 is a shortened form of the SCBE-80 and
yields 3 subscales: an anxious-withdrawal scale, an angry-aggression scale, and a
sensitive-cooperative scale. It is designed speciﬁcally to assess preschool behaviors and
taps both positive and negative behaviors (in contrast to the CBCL, which indexes
problems only). It has been shown to have good inter-rater reliability (78-91) as well as
construct validity when compared to other conduct disorder and anxiety measures for

young children; internal consistency is also good, with alphas in previous studies ranging

37

 

from .80 - .92 for all three scales (Denham et al., 2003; LaFreniere & Dumas, 1996).
Because the anxious-withdrawal and angry-aggressive scales tap similar dimensions to
those measured by the C-TRF, the SCBE was included primarily as a measure of
strengths in social competence, therefore these analyses used only the sensitive-
cooperative scale. The alpha for this scale was .88.
Use of Attachment and Representation Categories vs. Dimensional Ratings
Ainsworth’s (Ainsworth, 1978) original studies of infant attachment behavior and
corollary parent behavior used a categorical system rather than dimensional scores, and
subsequent research on adult representations of early attachment memories, such as the
AAI, developed categories that parallel the infant classiﬁcation system (Main &
Goldwyn, 1984), in part because of an interest in studying the intergenerational
transmission of working models (e. g., Benoit & Parker, 1994). The authors of the WMCI
and other measures of caregiving representations followed suit. This led to a convention
among developmental researchers to use categorical classiﬁcations of attachment
behavior or representations in studies of parents and children.(Van IJzendoorn, 1992).
However, in coding behavior or narratives to arrive at these categories, these
systems include numerical ratings on a number of scales thought to be important to the
overall classiﬁcation, although this classiﬁcation is typically reached by making a gestalt-
type decision rather than by summing the scores. For example, behavior during the
Strange Situations is rated for degree of avoidance, contact maintenance, proximity
seeking, resistance, and disorganization. Although Ainsworth’s group found these scores
were strongly related to the category, she chose to preserve the categories because she

thought they conveyed a better picture of the overall patterns, that a dimensional system

38

 

could not capture all the behaviors considered in making the ﬁnal decision, and the
pattern categories could be easily related to their origins in parenting differences
(Ainsworth, 1978).

Researchers of adult attachment style also initially used categories, but then
shifted to considering dimensions of attachment (Fraley & Shaver, 2000). Social
psychology researchers now primarily consider dimensions of attachment along two axes,
for example, positive and negative models of self and others (F eeney, Noller, &
Hanrahan, 1994), or, more recently, 1) anxiety about rejection and 2) how to cope with
this anxiety, using comfort seeking or avoidance (Fraley & Shaver, 2000). Recently,
developmental researchers have begun to explore a dimensional approach to infant
attachment behavior, using new statistically techniques to ﬁnd that infant attachment may
be more dimensional, along axes that tap anxiety and comfort with proximity seeking
(Fraley & Spieker, 2003). One of the arguments for using dimensional data is that it
allows for more statistical power in analyses and increases the likelihood of ﬁnding
results (Fraley & Spieker, 2003).

For the current study, it was decided to focus on the categorical data on
attachment behavior and maternal representations for several reasons. The ﬁrst was to be
able to compare the results to previous studies that used categorical classiﬁcations. One
of the questions to be explored was whether insecurity of attachment or non-balanced
representations matter only in the context of risk, rather than being a general marker of
psychopathology, and this could only be answered by comparing categories as other
studies have done. Secondly, I hypothesized that the two non-balanced categories of

maternal representations reﬂect alternative ways of coping with negative emotion, either

39

by deactivating or over-arousal, and thus needed to keep these categories distinct. Finally,
less is known about the reliability and validity of the sum scores of the WMCI, and to a
lesser extent, the rating subscales used for the Strange Situation, so the more conservative
categorical approach was chosen. Although the primary analyses were done with
categorical data, some exploratory analyses were done using by summing of the WMCI

subscales and using this continuous score of representations in analyses.

40

RESULTS
Missing Data

Missing scale items were imputed using series mean in SPSS. When case
imputation was necessary, continuous variables were imputed using the regression
method (ML) in SYSTAT, and categorical variables were imputed using the Hotdeck
method in LISREL. Nine teacher SCBE’s were missing because this measure was not
included in the initial interviews for the age 4 data collection. For these cases, SCBE sum
scores were imputed from CBCL and C-TRF scores using the regression method (ML) in
SYSTAT .

For the Working Model of the Child Interview (WMCI) data, one case was
missing due to tape problems and 2 women missed the age one interview. WMCI
variables were imputed using the Hotdeck method in PRELIS based on variables found to
be signiﬁcantly related to the WMCI in previous analyses of this sample (Huth-Bocks et
al., 2004; Theran, Levendosky, Bogat, & Huth—Bocks, 2005). For the age one WMCI,
the variables used for imputation were: prenatal WMCI category, mother’s education,
household income, and depression scores.

For the Strange Situation (SS) data, ten cases were missing either because the
interview was done by phone because the participant lived out of state (n=6), the mother
refused to be videotaped (n=1), there was a tape malfunction (1), or the mother missed
the age one interview (2). It was decided not to impute this data due to the large number
of missing cases.

113 of the 174 children were enrolled in preschool at Time 4. Twenty-one of

these 113 children were missing all teacher data (C-TRF’s and SCBE’s), either because

41

 

the mothers refused to allow us to contact the school (n=6), or the teachers did not return
the forms (n=15). Therefore, for the analyses of teacher-rated data, the total sample was
92.

Data Reduction
Domestic Violence

Exposure to violence was assessed at all four waves: ages 1, 2, 3 and 4
(preschool) by mother self-report, yielding a weighted sum score for each wave, which
was then converted into a dichotomous violence or no-violence score for each wave.
Criteria of inclusion in the DV groups was endorsement of physical violence or threat of
physical violence on the SVAWS. Based on the number of waves at which mothers
reported experiencing violence, children were categorized into three groups reﬂecting the
duration of domestic violence during the ﬁrst four years of life: No exposure, Intermittent
exposure (at 1 or 2 time points), and Chronic exposure (at 3 or 4 time points) (see Table
2).

To control for severity of violence, a peak score was calculated by taking the
highest severity score of the four time periods (see Table 3). In addition, in order to
conduct a secondary analysis comparing recent vs. distal exposure, another three
categories were created to reﬂect recent (past year) exposure, prior but not recent
exposure, and no exposure.

Attachment and Maternal Representation Categories

The infant attachment coding system used for the Strange Situation procedure

yields four categories: secure, avoidant, ambivalent, and disorganized. For the purpose of

this study, the latter three were combined to form an insecure group. Analyses for this

42

study compared the secure vs. the insecure attachment groups. The WMCI coding yields
3 categories: balanced, distorted, and disengaged; the latter two were combined to form a
non-balanced group for these analyses, in order to compare children of balanced vs. non-
balanced mothers (see Table 4).
Child Behavior Outcomes

Mothers and teachers completed parallel behavior rating scales for the children,
with teachers using the C-TRF, and mothers using the CBCL. Both scales yield
internalizing and externalizing behavior subscales. Teachers also completed a measure of
social competence. There were initially ﬁve child outcomes to consider: mother-rated
internalizing behavior and externalizing behavior, teacher-rated internalizing behavior
and externalizing behavior, and teacher-rated social competency. See Table 3 for the
bivariate correlations and descriptive statistics on these variables. Within raters, the
internalizing and externalizing subscales were so highly correlated it was decided to use
the total score for each rater instead of the two subscales (see Table 3). However,
correlations between raters were smaller, and in the case of teacher-rated internalizing,
non-signiﬁcant, so it was decided to analyze mother and teacher ratings separately.
Although social competence was signiﬁcantly correlated with the other 4 outcomes, it
was decided to preserve this variable because it is a measure of strengths rather than
problems, and was therefore provided unique information. Thus, three child outcomes
were used for these analyses: mother-rated total problems (CBCL), teacher-rated total

problems (C-TRF), and teacher-rated social competence (SCBE).

43

Analyses
Direct Effects of Domestic Violence

It was hypothesized that children who were exposed to DV would have more
behavior problems and lower social competence than children in the no-violence group. It
was further hypothesized that problems would increase as the duration of exposure
increased. First, a one-way AN OVA was used to see if DV had a signiﬁcant effect on
child outcomes, entering DV group as the independent variable and child outcomes as the
dependent variable. Next, planned comparisons were used to compare the DV groups,
anticipating that the Chronic group would have more severe problems than the
Intermittent group.

The initial one-way ANOVA supported the hypothesis that children exposed to
domestic violence would have signiﬁcantly more behavior problems and lower social
competence than their non-exposed peers (F (2, 171) = 5.212, p < .01). However, results
of the planned comparisons only partially supported the hypothesis that problems would
increase as exposure to DV increased. For social competence, the Chronic group looked
signiﬁcantly worse than the no-violence group, but the Intermittent group was not
signiﬁcantly different than either of the other groups (see Table 5). Both the Chronic and
Intermittent groups had signiﬁcantly more teacher-rated behavior problems than the no-
violence group, but did not differ signiﬁcantly from each other. For mother-rated
behavior problems, the Chronic group was signiﬁcantly worse than both the Intermittent
and no-violence groups, but the Intermittent group did not differ from the no-violence
group (see Table 5). Thus, for mother-rated problems, longer duration was associated

with more severe problems, but for the other outcomes the Intermittent group either did

44

not differ from the two other groups or looked similar to the no-DV group. Finally, a one-
way ANOVA was used to compare children who had been exposed to DV in the past
year with those who had been exposed prior to but not during the past year: there were no
differences between these two groups for any of the three outcomes. Results for teacher-
rated problems were F (1 , 58) == .002, p = .968, and results for the other child outcomes
were similar.

In summary, children exposed to DV did show more behavior problems at home
and school at age four, but greater exposure did not necessarily mean more severe
problems. There was no effect related to having experienced DV in the past year vs. only
at an earlier age. The most signiﬁcant factor seemed to be whether or not the child was
exposed at all.

Moderating Effects of A ttachment and Maternal Representations

Although differences in preschool outcomes between secure and insecure children
were not expected for the group as whole, in the DV-exposed groups, secure children
were expected to have better outcomes than insecure children. Thus, an interaction was
predicted between attachment and DV status, with attachment having more effect on
preschool behavior in the violence-exposed groups. A two-way ANOVA was used to test
for main and interaction effects. For social competence, there was no main effect for
attachment category [F (1, 82) = .044, p =.84] or for the covariate DV score [F (1, 82) =
1.98, p =.16], nor was there a signiﬁcant interaction effect for DV X attachment [F (2,
82) = .458, p =.63]. For teacher-rated behavior problems, there was no main effect for
attachment [F (1, 82) = .719, p = .40] or peak DV score [F (l, 82) = 1.29, p =.26], nor

was there a signiﬁcant effect of the interaction variable [F (2, 82) = .831, p =.439].

45

Similarly, for mother-rated behavior problems, there was no main effect for attachment
[F (1, 168) = l.04,p =.31] or for the interaction variable [F (2, 88) = 1.33,p =.29].

Although no signiﬁcant effects were found for attachment, the group means were
generally in the expected direction, with secure groups having lower problem scores and
higher competence scores (see Table 6). For these analyses, the Eta value from the
AN OVA was used as an estimate of the effect size. The Eta value here was negligible,
ranging from 0.00 to 0.06.

A similar moderating hypothesis was proposed for maternal representations in the
high-risk, DV-exposed groups, but this hypothesis also was contradicted by the data. For
the WMCI, it was predicted that there would be an interaction between WMCI category
and DV group. This was tested using a two-way ANOVA with DV group and WMCI
classiﬁcation as the independent variables, peak DV score as a covariate, and child
outcome as the dependent variable. Univariate ANOVA’s were run separately for each
child outcome. There were no main or interaction effects on teacher or mother-rated
behavior problems, but there was a main effect of WMCI classiﬁcation on child social
competence at the level of a trend (see Table 7). There was no signiﬁcant interaction
effect of DV and WMCI for social competence. Although there were signiﬁcant
differences for behavior problems, means were generally in the expected direction, with
children in the balanced group having fewer problems in most cases (see Table 8).

Finally, in order to test whether these ﬁndings might be limited by the use of
categorical data, an exploratory analysis was done using a continuous score of maternal
representations created by summing the six qualitative subscales. Linear regressions

were performed using this score as the independent variable and each of the preschool

46

outcomes as the dependent variable, running separate regressions for each DV group. The
WMCI sum score was not related to child outcomes in any of the groups. For example,
although WMCI category was related to social competence in the two-way ANOVA
analyses, it was not signiﬁcant in the regression analysis with social competence (B=
.406, F (2, 168) = 1.083, p = .306). Results for the other child outcomes were similar.

In summary, the expected interaction of attachment and high-risk (DV exposure)
was not found in this sample, nor did maternal representations appear to moderate the
effects of DV of behavior problems. There was, however, a main effect of maternal

representations on preschool social competence.

47

DISCUSSION

The goal of the present study was to understand how DV and the mother-child
relationship might interact to inﬂuence the development of the child’s early preschool
relationships with teachers and peers. It was hypothesized that DV during the ﬁrst four
years of life would predict preschool behavior problems, and that severity of problems
would increase in proportion to the extent of the exposure. It was also hypothesized that
there would be an interaction effect for DV and attachment and DV and maternal
representations.

The ﬁrst hypothesis was partially supported, in that DV predicted preschool
behavior problems, but severity of behavior problems did not increase in proportion to
exposure. These results are consistent with ﬁndings from previous studies indicating that
exposure to domestic violence is associated with increased behavior problems
(Levendosky et al., 2003; McCloskey & Stuewig, 200.1; Stemberg et al., 1993). The fact
that higher levels of behavior were seen in the DV-exposed groups suggests that DV
affects children’s emotion regulation (as seen in increased anxiety, sadness, and anxiety)
and impairs their ability to interact appropriately with others (as seen in increased
aggression). It could be that children exposed to DV are imitating behavior seen at home
(unregulated anger) and/or that they are distressed by frightening behavior seen at home
and express this behaviorally.

Results did not support the hypothesis that attachment and maternal
representations would moderate the effects of DV. Given that secure attachment in
infancy is theoretically associated with the development of effective emotion regulation

(Magai, 1999), it is surprising that attachment security did not predict lower levels of

48

 

problem behaviors and better social competence. Previous studies examining whether
maternal characteristics or the mother-child relationship mediate or moderate the effects
of DV on children show mixed results. Some studies, including one of the present
sample, found that the mother’s life stress and symptoms of PTSD mediated the
relationship between DV and child symptoms (Bogat et al., 2006; Lieberman et al.,
2005). Another study failed to ﬁnd a mediating effect for maternal mental health on child
PTSD symptoms (Kilpatrick & Williams, 1998). For the current sample, the direct effect
of domestic violence appears to override the potential protective effect that might be
provided by secure attachment or balanced maternal caregiving representations.

The association between DV and child behavior problems and lower social
competence found here is consistent with results from previous studies with school-age
children (McCloskey et al., 1995; Stemberg et al., 2006) and preschool children
(Levendosky et al., 2003; Lieberman et al., 2005). However, contrary to expectation, the
intermittent and chronic groups differed from each other only in one of the three
outcomes, mother-rated problems, and here only at the level of a trend. These ﬁndings
suggest that the effects of DV take place upon ﬁrst exposure, rather than following a
dose-response pattern.

It had been hypothesized that duration of DV might account for some of the
variation in child outcomes that has been found by other studies, but that was not the case
in this sample. Rather, these results suggest that any DV experienced during the ﬁrst four
years will contribute to preschool behavior problems. This is in contrast to another study
based on a subgroup of this sample, which found that witnessing severe domestic

violence predicted trauma symptoms at age one while witnessing less severe DV did not

49

(Bogat et al., 2006). However, that study was cross-sectional, using DV from one wave
(ﬁrst year of life), and focused on a concurrent outcome speciﬁc to witnessing a violent
episode (PTSD symptoms). It may be that when examining proximal violence and child
response, symptoms increase in proportion to the severity of DV, while when looking at
the effects of DV on general child problems over four years, the extent of DV matters
less than the initial exposure. Finally, the lack of differences between the Intermittent and
Chronic groups may be a function of the small sample size in each group. Future studies
using larger samples should examine whether duration is one factor that contributes to the
variability of outcomes.

In this sample of mixed low- and high risk children (based on exposure to DV),
attachment and maternal representations did not have a direct effect on preschool
behaviors for the group as a whole. In general, the effect size for attachment within each
violence group was negligible, with eta’s ranging from .00 to .09 in all three groups. The
effect size of maternal representations was similarly small, ranging from .00 to .13 (for
social competence). This suggests that other factors outside the scope of this study
outweighed the inﬂuence of attachment on preschool outcomes in this high-risk sample.

The lack of moderation ﬁndings was somewhat surprising because previous
studies have found that the relationship of early attachment to later behavior problems is
most robust in high-risk samples (Belsky & Fearon, 2002; Fish, 2004; Lyons-Ruth,
1996). It may be that the sample as a whole is sufﬁciently high-risk in Belsky and
F earon’s terms (2002) to show such attachment-related differences, but when DV is

added to the demographics risks, the effect of attachment is neutralized.

50

 

Examining the overall pattern of behavior problems, DV, and attachment in Table
6 reveals that mean problem behavior scores rose in both secure and insecure groups
while social competence fell as the duration of violence increased. Thus, the direction of
the means suggests that the exposure to violence had the same effect on both secure and
insecure children, overpowering any buffering effect security might have had for
preschool behavior and social problems. Such a threshold effect, in which one factor
ceases to have inﬂuence in the presence of another factor, was found in an earlier study
of the larger sample from which this study was drawn. In a study of children’s reaction
to a demonstration of adult anger in a laboratory situation, temperament predicted
postural distress response in children who had not been exposed to DV, but not in those
who had been exposed to DV (DeJonghe et al., 2005). This threshold effect may explain
why attachment was not related to preschool outcomes in the groups exposed to DV.

The failure to ﬁnd an interaction between attachment and risk, in this case DV, is
in contrast to other studies of attachment in high-risk samples. However, these other
studies, described in the introduction, used demographic factors such as SES, single
parenthood, and ethnicity to determine the level of risk (Fish, 2004; Huth-Bocks et al.,
2004). It may be that an interpersonally traumatic experience such as inter-parental
violence has a qualitatively different effect on child development of emotion regulation
and social skills with peers, at least in the ﬁrst few years of life. It is important to note
that a number of children exposed to violence nonetheless were categorized as having
secure attachments to their mothers. Although attachment security is not buffering their
behavioral difﬁculties at age four, there may be a later effect on peer relationships as the

child grows older, or a recuperative effect if the violence between the parents ends.

51

There are no empirical studies to date of child recovery from witnessing domestic
violence, but other studies of attachment and child behavior problems suggest that
children who are in the same attachment or problem group at a young age may follow
different trajectories. For example, a large longitudinal study of child behavior problems
found that children with early behavior problems followed different trajectories: some
children improved, some worsened, some stayed the same, and these paths reﬂected other
differences in parenting variables, demographics, and child characteristics (NICHD
ECCRN, 2004). Fish (2004) found that infant attachment was a better predictor of
preschool social and cognitive functioning than concurrent age four attachment. Although
the current study examines DV over time, the outcomes are from one time period, age
four. Re~examining the developmental paths of this sample when they reach school-age
may provide more information about which preschool children with behavior problems
continue to have such problems when they are age seven or eight. Margolin (2005)
argues that the adaptations that children make to survive in a household with DV, such as
becoming passive or withdrawn to avoid drawing parental anger may be maladaptive
outside the home, interfering with later developmental tasks including making friends at
school or asserting themselves in academics or sports. It may be that the protective
effects of attachment security or balanced maternal representations emerge at a later stage
in development, perhaps only several years after the exposure to violence ends, and that
the lack of difference seen in this study reﬂected the inﬂuence of ongoing DV.

In contrast to previous studies, no relationship between attachment and social
competence was found (Denham et al., 2003; Denham et al., 2002a; Rose-Krasnor,

1997). Effective social competence requires recognizing emotions in oneself and others,

52

effective regulation of one’s own emotion and an ability to respond appropriately to
emotions in others (Denham et al., 2002b; Rose-Krasnor, 1997; Zhou et al., 2002).
Theoretically, attachment provides a foundation for learning these things, particularly
emotion regulation (Cassidy, 1994; Magai, 1999). However, in this sample attachment
was not related to social competence. This may have been a function of small sample
size, or it may be that the positive effects of age one attachment on preschool behavior
are less discernible given the three-year period between assessments.

Maternal representations were related only to teacher-rated social competence at
the level of a trend. Although this is not a robust ﬁnding, it contributes to the limited
evidence that maternal representations reﬂect a pattern of emotion regulation that may be
transmitted to the child (Rosenblum et al., 2002). However, it remains to be seen whether
this is because maternal representations are inﬂuenced by an underlying
temperamental/ genetic factor that is shared by mother and child, or by modeling of
emotion regulation. The association between balanced representations and higher social
competence is theoretically consistent with research on parenting and the development of
social competence, although there are no prior empirical studies to compare. It is
particularly notable that a relationship between maternal representations and social
competence was found given that they were assessed three years apart. It would be
interesting to test this effect in a low-risk sample as a way of getting a baseline measure
of the strength of the effect of caregiving representations, or to measure concurrent age
three caregiving representations and child behavior. Theoretically, parents who can
integrate both positive and negative affect in a coherent way while being open and

accepting of the child’s emotional experience provide a model for emotion regulation as

53

well as for interpreting and responding appropriately to others (Denham et al., 1994;
Magai, 1999). This is a cornerstone of social competence (Halberstadt et al., 2001).
Children who can regulate their emotions and accurately read others’ emotions are more
likely to succeed in interpersonal interactions, to behave cooperatively in a peer setting
(Denham et al., 2002b).

Previous studies have found that social competence is facilitated by parents who
can express and respond to both positive and negative emotions, who respond sensitively
to their children’s expression of emotions and who help their children learn to regulate
their emotions and their behavior (Denham et al., 2000). Sensitivity, acceptance, and the
ability to balance both positive and negative emotions are all components of balanced
maternal representations (Zeanah et al., 1996), but this is the ﬁrst study to show that these
qualities coded from a narrative may be related to social competence three years later. It
may be that a balanced maternal representation reﬂects important and enduring maternal
characteristics that contribute to the development of emotional strengths as the child
develops. However, it does not seem to protect against the development of behavior
problems, because it was unrelated to those rated by either the teacher or the mother.
Limitations

Perhaps the strongest limitation of the present study is the small sample size for
each of the DV, attachment, and WMCI groups used for the analyses (see Table 4). In
addition, as noted, the moderator variables were measured three years prior to the
outcome variables. While this is typical of longitudinal attachment studies, attachment is
not necessarily stable over time, particularly in high-risk samples (Belsky & Fearon,

2002). Thus, children who were categorized as secure at age one may be insecure at age

54

 

four. The focus of this study was the longitudinal effect of infant attachment on later
preschool behavior. However, comparing the relative effects of past and concurrent
attachment on preschool behavior might provide more information about potential
moderating effects. Such a study could also consider stability of attachment over time as
an additional factor. In a study of children at risk due to poverty and rural environment,
Fish (2004) found preschool outcome differences between stable secure and children who
changed attachment classiﬁcation between age one and age four. She also found that
secure infant attachment had a protective effect, but that there was no such beneﬁt from
preschool attachment status. For that sample, risk status was generally stable over time,
while DV has been found to be a highly changeable risk factor in longitudinal studies

(e. g., Stemberg et al., 2006). As such, measuring attachment in a DV sample across time
may reveal interactions at different developmental stages.

Another potential limitation stems from the decision to group children with
similar duration of violence exposure. While this has the advantage of testing for the
effects of DV over a crucial developmental period, it comes at the expense of considering
severity of violence. Thus, children within one theoretically homogenous group may have
experienced widely different levels of severity. Finally, this is a community sample
reporting mostly mild to moderate levels of DV. Thus, results may not be generalizable to
shelter samples who experience more extreme forms of DV.

Implications for Future Research and Intervention

This study makes a unique contribution to the literature showing negative effects

of domestic violence on child development, because it is one of the few studies that has

used a very young sample and is longitudinal over the time period, thus showing that the

55

effects of DV may take hold early in development. Further research is needed to examine
when attachment is most predictive of later behavior and to what degree attachment can
facilitate resiliency. Although Belsky and Fearon (2002) suggested attachment security
may act as a protective factor in the context of demographic risks, in this study, DV
appeared to overwhelm the effects of attachment. However, this study examined only the
ﬁrst four year of the child’s life: secure attachment may provide a protective effect in
later years or in different relationship contexts (i.e. dating relationships during
adolescence).

Longitudinal studies that encompass signiﬁcant developmental milestones are
needed to understand the long-term relationship between childhood exposure to DV and
attachment. Longitudinal studies have found that both exposure to DV and the parent—
child relationship are important contributors to risk for being in a violent dating
relationship (Ehrensaft et al., 2003; Magdol et al., 1998), but these studies did not have
early attachment data, only later self-report. The working models of infancy identiﬁed in
the Strange Situation may have a different effect on relationship behavior than adolescent
self-reported attachment style. It will be important to include both in longitudinal studies.
In particular, the Disorganized subtype of infant attachment, which at this time is only
assessed using the Strange Situation, may have a unique relationship to later dating
violence, (Lyons-Ruth & Jacobvitz, 1999).

Previous studies have examined the inﬂuence of maternal representations on
attachment and child behavior only as far as infancy (Benoit et al., 1997; Huth-Bocks et
al., 2004; Rosenblum et al., 2002). In this ﬁrst study of maternal representations and

child behavior during preschool, the ﬁnding of a relationship between balanced mothers

56

and social competence in the child is consistent with theory about transmission of
working models from mother to child and the development of social competence. The
WMCI theoretically taps the extent to which a mother can integrate positive and negative
emotions about her relationship with her child and emotionally regulate herself in order
to remain coherent and aware of complex and rich detail of her feelings and experiences
(Zeanah & Benoit, 1995; Zeanah et al., 1996). Studies using the WMCI have found it to
be concordant with concurrent and future infant attachment, suggesting that the internal
working model reﬂected in the mother’s caregiving representations somehow inﬂuences
the child’s working model, as manifest in infant attachment behavior. It may be that the
qualities of emotion regulation, sensitivity, and acceptance inﬂuence the child’s working
model. The mechanism of transmission is unclear, but Rosenblum (2002) found evidence
that maternal caregiving representations affects infant emotion regulation under stress.
More studies will be needed to see if this can be replicated, and if the effect is mediated
by observed parenting or by maternal emotion regulation. Researchers have found that
attachment and parental emotional style are important building blocks for the
development of social competence (Denham et al., 1994). Thus, if maternal
representations can inﬂuence attachment and emotion regulation, theoretically they
should also affect the development of social competence.

In summary, the ﬁndings of this study suggest that the effects of DV begin early
in life and are not necessarily buffered by a secure mother-child relationship.
Furthermore, results from this study suggest that the damage done by short-term exposure
is similar to that done by longer-term exposure. Thus, while a secure parent-child

relationship is important to foster for many reasons, parents should be educated that the

57

quality of their relationship with their children may not negate the damaging effects that a

violent intimate partner relationship of any duration might have on their children.

58

mum.hua'-h- ~. A... ' a'. .

APPENDICES

59

Appendix A

2/6/00 Working Model of the Child Interview
We are interested in how parents think and feel about their young children. This interview
is a way for us to ask you about (child’s name) and your relationship to him/her. The

interview will take us about an hour to complete.

1a.

1b.

1c.

1d.

The last time we interviewed you, we began by asking all about your pregnancy. Today I’d
like to begin by asking about your labor and delivery. What was it like? (Give some time to
respond before proceeding.)

1. How did you feel and react at that time?

2. What was your first reaction when you saw the baby?

3. What was your reaction to having a boy/girl?

4. How did your family react?
(Be sure to include husband/partner, other children in the family, etc.)

Did the baby have any problems in the first few days after birth? . (Give some time to
respond before proceeding.)

1. How soon did the baby leave the hospital?

2. Did you decide to breast-feed or bottle-feed?
3. Why?

4. How did you come to that decision?

How would you describe the ﬁrst few weeks at home with the baby ? (give the subject some
time to respond, then ask about feeding, sleeping, crying, etc. if she hasn ’t mentioned them)

How old was your baby when he/she started sitting up?

1. Was this ahead, behind, or at about the same time as when you think other children might
do this for the ﬁrst time?

(F or each of the following, be sure to get a sense of the ways in which the baby was thought to be
different, ahead, or behind in motor, social and language development.)

2. How old was your baby the first time he/she

 

 

 

Crawled? ahead, behind, or about the same....
Walked? ahead, behind, or about the same....
Smiled? ahead, behind, or about the same....
Talked? ahead, behind, or about the same....

 

3. Did you have any sense of your baby’s intelligence early on? What did you think?

60

1e.

1f.

2a.

2b.

3a.

3b.

3c.

Did your baby seem to have a regular routine?

1. What happened if you or your baby couldn’t stay in the routine?

How has the baby reacted to separations from you? (Try to get a sense of the baby ’s reactions
at various ages.)

1. Were there any separations of more than a day in the ﬁrst year?
If she says “NO, ” go to 2a.

2. How did your baby react?
3. How did you feel?
4. What did you do?

Describe your impression of your child’s personality now. (Give the subject enough time to
respond to this before proceeding to speciﬁc descriptors below.)

Pick five words (adjectives) to describe your child’s personality. After you have told me
what they are, I will ask you about each one. (You may tell the subject that it is ﬁne to use any
of the descriptors they used in response to the general probe above, but do not remind them what
they said before you have given them time to recall themselves. Some subjects will have a hard

time coming up with ﬁve descriptors. If you feel that they cannot come up with five, then move on.
The numbers are less important than the descriptions.)

1. Interviewer: write the 5 words down, then go back and ask for each:

For each one, what is it about him/her that makes you say that? Tell me a speciﬁc incident
or memory to show what you mean.

At this point, whom does your child remind you of?

1. In what ways?

2. When did you first notice the similarity?

"*If only one parent is mentioned, ask. “In what ways does the child remind you of (the other
parent)? ”

(The following questions should be asked whether or not the parents have been mentioned.)

3. Which of his/her parents is your child most like now?

4. In what ways is your child’s personality like or unlike yours or his/her father’s
personality?

Are there any family characteristics on your side that you see in your child’s personality?
1. What about the baby’s father’s side of the family?
How did you decide on your child’s name?

1. Does this name have special meaning in your family or the baby’s father’s family?

61

5a.

5b.

5c.

6a.

6b.

7a.

7b.

7c.

2. How well does the name seem to fit?

In what ways is your child unique or different compared to what you know of other
children?

What about your child’s behavior now is the most difficult for you to handle? Give a typical
example. (if she has trouble thinking of one, tell her it doesn ’t have to be a Leggy difﬁcult
behavior: just whatever is the ms; difficult for her to handle now.)

How often does this occur?

1. What do you feel like doing when your child reacts this way?

2. How do you feel when your child reacts this way?

3. What do you actually do?

Do you think he/she knows that you don’t like this behavior?

1. Why do you think hc/she does it?

What do you imagine will happen to this behavior as your child grows older?

1. Why do you think so?

How would you describe your relationship to your child now? (Give time to respond.)

Pick five words (adjectives) to describe your relationship with your child now.

PM“???

Interviewer: As before, write the 5 words down, then go back and ask for each one:
1. For each word, describe an incident or memory that illustrates what you mean.

What pleases you most about your relationship with your baby?
1. What do you wish you could change about it?

How do you feel your relationship with your child has affected your child’s personality?
(Give ample time to respond to this.)

Has your relationship to your child changed at all over time?
1. In what ways?

2. What’s your own feeling about the change?

8. Which parent is your child closest to now?

1. How can you tell?
2. Has it always been that way?

3. Do you expect that to change (as the child gets older, for instance)?

62

9a.

9b.

9c.

4. How do you expect it to change?

Does your baby get upset often? (Give some time to respond before proceeding to
specific queries.)

1. What do you do at these times?
2. What do you feel like doing when this happens?

3. What are your feelings when this happens?

What about when he/she becomes emotionally upset? Can you recall a

speciﬁc example? (Indicate that you want an example by providing a reasonably long time to
think of one.)

1. What did you do when that happened?
2. What did you feel like doing?

3. What were your feelings when that happened? (If the subject becomes extremely anxious
and cannot recall an example then proceed to the next question.)

What about when he/she has been physically hurt a little bit? Can you give an example and
describe what happened?

1. What did you do when that happened?
2. What did you feel like doing?

3. What were your feelings when that happened? (If the subject becomes extremely anxious and
cannot recall an example then proceed to the next question.)

Has your child been sick at all? Can you give an example?
1. What did you do when that happened?
2. What did you feel like doing?

3. What were your feelings when that happened? (If the subject becomes extremely anxious
and cannot recall an example then proceed to the next question.)

10. Tell me a favorite story about your child, perhaps one you’ve told to family or friends. I’ll give

11.

you a minute to think about this one. (If the subject is struggling, you may tell them that this
doesn’t have to be the favorite story, only a favorite).

1. What do you like about this story?

Are there any experiences which your child has had which you feel may have been a setback
for him/her? (Setback: something that might make things harder for your baby than for other
children.)

1. Why do you think this might have been a setback? (Be sure to give time to respond before
moving on to the more direct questions which follow.)

63

12.

l3.

14.

15.

16.

17.

Knowing what you know now, if you started all over again with your child, what would you
do differently? (Give some time to respond.)

Do you ever worry about your child?

1. What do you worry about?

If your child were to be one particular age, what age would you choose?

1. Why?

As you look ahead, what will be the most difﬁcult time in your child’s development?
1. Why do you think so?

What do you expect your child to be like as a teenager?
(between the ages of 12-18)

1. What makes you feel this way?
2. What do you expect to be good and not so good about this period in your child’s life?

Think for a moment of your child as an adult. What hopes and fears do you have about that
time?

64

Appendix B

Strange Situation Procedure

 

 

 

 

Episode # Persons Present Duration Action

1 Mother, baby, 30 seconds Walk the mother and child into the room
experimenter

2 Mother & baby 3 minutes Tape

3 Stranger, mother, & 3 minutes > After 1 minute, knock once on the
baby mirror (stranger talks to mom)

> After 2 minutes, knock once on the
mirror (stranger initiates play)

> After 3 minutes, knock twice on the
mirror (mother leaves)

 

 

 

 

 

 

 

 

 

 

4 Stranger & baby 3 min. or Tape; after 3 minutes, cue the mother into
less1 the room

5 Mother & baby 3 min. or Tape; after 3 minutes, knock twice on the
more2 mirror to cue the mother to leave

6 Baby alone 3 min. or Tape
less1

7 Stranger & baby 3 min. or Tape; after 3 minutes, cue the mother into
lessl the room

8 Mother & baby 3 minutes Tape

 

' Episode might be curtailed if the baby is extremely distressed. See below
2 Episode might be extended if the baby does not seem settled enough.

Judging distress: as a rule, if child is calling out at door/intennittent crying - keep going
Full cry: send stranger in after 30 seconds. Do not allow full cry to go past 1 minute.

 

Difﬁcult separations

Episode 4: if child is extremely upset at ﬁrst separation and clings to mother in hall, have stranger lead
child away from door and have mother at least close door behind her — then she can return right away
(stranger can explain this to mom).

.If the child is hysterical after attempting episode 4 and does not settle, do not try for second separation -—
have mother reassure child she will not leave again.

Episode 5: if child has not settled, extend episode up to one minute, then see below

Episode 6: if child is extremely distressed, again, attempt to have mother at least close door behind her and
then send stranger right in, then mother in too if necessary.

If the stranger has to escort the child back into the room for episode 6, you can skip episode 6 (child alone)
and go right on to episode 7.

65

TABLES

66

Table 1. Demographic Characteristics of the Sample at Time of Recruitment, during

Third Trimester of Pregnancy
Characteristics
Mother’s racial/ethnic group
Caucasian
Afiican-American
Biracial
Latina
Asian or Native American
Child’s racial/ethnic groupa
Caucasian
African-American
Multi-racial
Asian or Latina
Child Gendera
Marital status
Single, never married
Married
Separated, divorced or widowed
Educational Status

Did not complete high school

a As reported by mother after birth

Percentage of Sample (N = 174)

64%
24%
5%
5%

2%

49%
24%
24%

3%

51% girls
50%
42%

8%

15%

67

Table 1. Continued

 

 

Characteristics Percentage of Sample (N = 174)
High school diploma 28%

Trade school, some college, or associate’s degree 42%

Bachelor’s or graduate degree 15%
Receiving public assistance 62%

Monthly household income (1999-2000)

Mother’s age

Median = $1,500
(range: 0-$6,200; SD = $1,400)
M = 25 years
(range: 18-40 years; SD = 5

years)

 

68

Table 2. Sample Size of DV Groups

 

Preschool n = 92

Total sample n = 174

 

No violence 29
Intermittent 40
Chronic 58

59

67

48

 

69

mo.vm.*.—O.V&**

 

 

a: 33 Be one was 8-: sagas a swam
953: 8.83 333 8.52 3.33 8.399. 83:82
8. 3.- a. 8. 3. sea >a case a
:8. .mm. 2. ram; assesses .26 .m
.8. 2. ram- masseuse Eu .4
ram. .18.- manages Ed .m
:3..- wasaeaacad .N
oocBoQEoo 38m ._
o m a. m N _

 

833.23 302:3:00 mo moquComoQ Ea macaw—880 039me .m 2an

70

Table 4. Size of Attachment and Maternal Representations Groups within each DV
Group

 

 

No violence Intermittent Chronic
Attachment Category (n=89)
Secure (52) 17 20 15
Insecure (37) l2 l9 6
WMCI (n=169)
Balanced (102) 42 34 26

Non-balanced (67) 15 31 21

 

71

C E .3 ammo How was 8w :8 2?» 85895“. .308 can ”Er—HO Sm :88on mo moouwuu £8?”— Bmtng Sm a

.2. Va. .8. vs... .8. var.

 

 

.m. A a .to A a team awe 8.2 698 3.2 83V 2.x Es ammo
.Q A a .85 sea as... in 3.9. 83: Ga. 85338 38m
.0 A a .......a A m 3.20.4 83:32 32: $2 $2 3 ms 32 ”Rho
A ”a Q B
oEoEU EoEEBE 8:203 02
Batman—co 3583 am A99 3on some 8.“ £832

 

3:80 >D 28 880850 230 no.“ 3033800 BEBE .«o 333% .m 2an

72

 

 

 

 

$8 8.2 :35 an: 33 2.: 2:82:
Get 3: $2: 8.2 8%: 3. .N aaosm
2850

3.3 8.2 awe e2; 63: 3.: cause:

5.2 82 get 2.: a3: 5.8 038m
EoEEBE

$4.2 a: £98 3:. 5.2V mm: @5035

see a; 5.6 an? e: c we.” seem
mono—o; OZ

6% 2
.33 Home massage snow as... Ed
powBLoﬁo—Z “688-553...

 

955 >Q .3 £985 “5858?. 9303.5 .m> 053m com 880330 EEO .o 033.

73

 

 

 

8m. can. mm. 2: 2m. 2m.
Em. 82 9;. 2a Se. 4%.
48. 83 so. 2: 2. RS
3.. N8. 2;. .m _ .m so. 08.
a «a: i m a am i a a «a. i a
nae home 032288 38m 33 go
Bamakoﬁoz woaemaonowok

5.23 x 955 >o
Boom xenon >Q
@280 >Q

52>»

 

955 >n Ea aoeaoeaaa 62? ea <>oz< $3 a sauce .5 use.

74

 

 

 

 

see 8.2 $8: :8 88.28 22 885.882
25 8.2 85 8.: 8.2V 8.: 3233
cacao

A82 :2 8.3 3.8 $3: 8.2 8853-82

$5 8.8 5.5 8.9. 2 .3 8.8 3853
«CHEER—CH

8.8 8.2 82 c 8.8 82: 8.: eooaamaoz

:8: 8s as 8.3. 38 m; Begum
8:203 02

Se 8
use Home 083888 38m 38 5:6
panamaoﬁo—Z cosmaoaoaoh

 

9:80 >9 .3 .3380 52>» wooed—3-52 .m> wooaeﬁm 8m 380830 EEO .w 03m...

75

REFERENCES

76

Achenbach, T. M. (1991). Manual for the Child Behavior Checklist / 4-18 and 1991
proﬁles. Burlington, VT: University of Vermont.

Achenbach, T. M., & Rescorla, L. A. (2001 ). Manual for the ASEBA school-age forms &
proﬁles. Burlington, VT: University of Vermont, Research Center for Children,
Youth, and Families.

Adam, E. K., Gunnar, M. R., & Tanaka, A. (2004). Adult attachment, parent emotion,
and observed parenting behavior: Mediator and moderator models. Child
Development, 75(1), 110-122.

Ainsworth, M. D. S., Blehar, M.C., Waters, E., & Wall, S. (1978). Patterns of
attachment: A psychological study of the Strange Situation. Hillsdale, NJ:
Erlbaurn.

Belsky, J., & Fearon, R. M. P. (2002). Infant-mother attachment security, contextual risk,
and early development: A moderational analysis. Development and
Psychopathology, 14, 293-310.

Benoit, D., & Parker, K. C. H. (1994). Stability and transmission of attachment across
three generations. Child Development, 65, 1444-1456.

Benoit, D., Parker, K. C. H., & Zeanah, C. H. (1997). Mothers' representations of their
infants assessed prenatally: Stability and association with infants' attachment
classiﬁcations. Journal of Child Psychology & Psychiatry & Allied Disciplines,

38(3), 307-313.

Bogat, G. A., DeJonghe, E. S., Levendosky, A. A., Davidson, W., S., & von Eye, A.
(2006). Trauma symptoms among infants who witness domestic violence toward
their mothers. Child Abuse & Neglect, 30(2), 109-125.

Bogat, G. A., Levendosky, A. A., Theran, S. A., von Eye, A., & Davidson, W. S. (2003).
Predicting the psychosocial effects of interpersonal partner violence (IPV): How

much does a woman's history of IPV matter? Journal of Interpersonal Violence,
18,1271-1291.

Bowlby, J. (1988). A secure base: Parent-child attachment and healthy human
development. New York: BasicBooks.

Bretherton, I., & Munholland, K. (1999). Internal working models in attachment
relationships. In J. Cassidy & P. R. Shaver (Eds), Handbook of attachment (pp.
89-11 1). New York: The Guildford Press.

Carlson, B. E. (1985). Children's observations of interparental violence. In A. R. Roberts
(Ed.), Battered women and their families (pp. 147-167). New York: Springer.

77

Carlson, E. A. (1998). A prospective longitudinal study of attachment
disorganization/disorientation. Child Development, 69(4), 1107-1128.

Casey, R. J. (1996). Emotional competence in children with externalizing and
internalizing disorders. In M. Lewis & M. W. Sullivan (Eds), Emotional
development in atypical children (pp. 161-183). Mahwah, New Jersey: Lawrence
Erlbaum Associates.

Cassidy, J. (1994). Emotion regulation: Inﬂuences of attachment relationships.
Monographs of the Society for Research in Child Development, 5 9(2-3, Serial No.
240), 228-249.

Deater-Deckard, K., Dodge, K. A., Bates, J. E., & Pettit, G. S. (1996). Physical discipline
among African American and European American mothers: Links to children's
externalizing behaviors. DeveIOpmental Psychology, 32(6), 1065-1072.

DeJonghe, E. S., Bogat, G. A., Levendosky, A. A., von Eye, A., & Davidson, W., S.
(2005). Infant exposure to domestic violence predicts heightened sensitivity to
adult verbal conﬂict. Infant Mental Health Journal, 26(3), 268-281.

Denham, S. A., Blair, K., DeMulder, E., Levitas, J., Sawyer, K., Auerbach-Major, S., et
al. (2003). Preschool emotional competence: Pathway to social competence. Child
Development, 74(1), 238-256.

Denham, S. A., Blair, K., Schmidt, M., & DeMulder, E. (2002a). Compromised
emotional competence: Seeds of violence sown early? American Journal of
Orthopsychiatry, 72(1), 70-82.

Denham, S. A., Caverly, S., Schmidt, M., Blair, K., DeMulder, E., Caal, S., et al.
(2002b). Preschool understanding of emotions: Contributions to classroom anger
and aggression. Journal of Child Psychology & Psychiatry, 43(7), 901 -916.

Denham, S. A., Workman, E., Cole, P. M., Weissbrod, C., Kendziora, K. T., & Zahn-
Waxler, C. (2000). Prediction of externalizing behavior problems from early to
middle childhood: The role of parental socialization and emotion expression.
Development and Psychopathology, 12(1), 23-45.

Denham, S. A., Zoller, D., & Couchoud, E. A. (1994). Socialization of preschoolers'
emotion understanding. Developmental Psychology, 30(6), 928-93 6.

Edleson, J. (1999). Children's witnessing of adult domestic violence. Journal of
Interpersonal Violence, 14(8), 839-870.

Egeland, B., & Sroufe, L. A. (1981). Attachment and early maltreatment. Child
Development, 52, 44-52.

78

Ehrensaft, M. K., Cohen, R, Brown, J ., Smailes, B... Chen, H., & Johnson, J. G. (2003).
Intergenerational transmission of partner violence: A 20-year prospective study.
Journal of Consulting and Clinical Psychology, 71 (4), 741-753.

English, D. J ., Marshall, D. B., & Stewart, A. J. (2003). Effects of family violence on
child behavior and heath during early childhood. Journal of Family Violence,
18(1), 43-57.

Fantuzzo, J ., Boruch, R., Beriama, A., Atkins, M., & Marcus, S. (1997). Domestic
violence and children: Prevalence and risk in ﬁve major US. cities. Journal of the
American Academy of Child and Adolescent Psychiatry, 36(1), 1 16- 122.

Feeney, J. A., Noller, P., & Hanrahan, M. (1994). Assessing adult attachment. In M.
Sperling & W. Berman (Eds), Attachment in adults (pp. 128-152). New York:
The Guilford Press.

Field, T. (1994). Effects of mother's physical and emotional unavailability on emotion
regulation. Monographs of the Society for Research in Child Development, 5 9(2-
3, Serial No. 240).

Fish, M. (2004). Attachment in infancy and preschool in low socioeconomic status rural
Appalachian children: Stability and change and relations to preschool and
kindergarten competence. Development and Psychopathology, 16.

Fonagy, P., Steele, H., & Steele, M. (1991). Maternal representations of attachment
during pregnancy predict the organization of infant-mother attachment at one year
of age. Child Development, 62(5), 891-905.

F raley, R. C., & Shaver, P. R. (2000). Adult romatic attachment: theoretical
developments, emerging controversies, and unanswered questions. Review of
General Psychology, 4(2), 132-154.

Fraley, R. C., & Spieker, S. J. (2003). Are infant attachment patterns continuously or
categorically distributed? A taxometric analysis of Strange Situation behavior.
Developmental Psychology, 39(3), 387-404.

George, C., Kaplan, N., & Main, M. (1985). Adult attachment interview.Unpublished
manuscript, University of California at Berkeley.

Greenberg, M. T. (1999). Attachment and psychopathology in childhood. In J. Cassidy &
P. R. Shaver (Eds), Handbook of attachment: Theory, research, and clinical
applications (pp. 469-496). New York: Guilford Press.

Greenberg, M. T., Speltz, M. L., DeKlyen, M., & Endriga, M. (1991). Attachment

security in preschoolers with and without externalizing behavior problems: A
replication. Development and Psychopathology, 3, 413-430.

79

Greenberg, M. T., Speltz, M. L., Deklyen, M., & Endriga, M. C. (1992). Attachment
security in preschoolers with and without externalizing problems: A replication.
Development and Psychopathology, 3, 413-430.

Grych, J. H., & Fincham, F. D. (1990). Marital conﬂict and children's adjustment: A
cognitive-contextual framework. Psychological Bulletin, 108, 267-290.

Grych, J. H., Jouriles, E. N., Swank, P. R., McDonald, R., & Norwood, W. D. (2000).
Patterns of adjustment among children of battered women. Journal of Consulting
and Clinical Psychology, 68(1), 84-94.

Halberstadt, A. G., Denham, S. A., & Dunsmore, J. C. (2001). Affective social
competence. Social Development, 10, 79-119.

Herman, J. L. (1992). Trauma and recovery. New York: Basic Books.

Hughes, H. M. (1988). Psychological and behavioral correlates of family violence in
child witnesses and victims. American Journal of Orthopsychiatry, 58, 77-90.

Hughes, H. M., & Luke, D. A. (1998). Heterogeneity in adjustment among children of
battered women. In G. W. Holden, E. N. J ouriles & R. Geffner (Eds), Children
exposed to marital violence: Theory, research, and applied issues (pp. 185-221).
Washington, DC: American Psychological Association.

Hughes, H. M., Parkinson, D., & Vargo, M. (1989). Witnessing spouse abuse and
experiencing physical abuse: a "double whammy"? Journal of Family Violence, 4,
197-209.

Huth-Bocks, A. C., Levendosky, A. A., Bogat, G. A., & von Eye, A. (2004). The impact
of maternal characteristics and contextual variables on infant-mother attachment
Child Development, 75(2), 480-496.

Izard, C. E., Haynes, O. M., Chisholm, G., & Baak, K. (1991). Emotional determinants of
infant-mother attachment. Child Development, 62, 906-917.

Jaffee, S. R., Caspi, A., Mofﬁtt, T. E., Dodge, K. A., Rutter, M., Taylor, A., et al. (2005).
Nature x nurture: Genetic vulnerabilities interact with physical maltreatment to
promote conduct problems. Development and Psychopathology, 1 7(1), 67-84.

Kilpatrick, K. L., & Williams, L. M. (1998). Potential mediators of post-traumatic stress

disorder in child witnesses to domestic violence. Child Abuse and Neglect, 22(4),
319-330.

80

Kitzmann, K. K., Gaylord, N. K., Holt, A. R., & Kenny, E. D. (2003). Child witnesses to
domestic violence: A meta-analytic review. Journal of Consulting and Clinical
Psychology, 71(2), 339-3 52.

LaFreniere, P. J ., & Dumas, J. E. (1996). Social competence and behavior evaluation in
children ages 3-6 years: The short form (SCBE-30). Psychological Assessment,
8(4), 369-377.

Levendosky, A. A., Bogat, G. A., Davidson, W., S., & Von Eye, A. (2000). Risk and
protective factors for domestic violence. Centers for Disease Control, Injury
Prevention Center, ROI/CCR518519-01.

Levendosky, A. A., & Graham-Bermann, S. A. (1998). The moderating effects of
parenting stress on children's adjustment in woman-abusing families. Journal of
Interpersonal Violence, 13(3), 383-397.

Levendosky, A. A., Huth-Bocks, A. C., Semel, M. A., & Shapiro, D. L. (2002). Trauma
symptoms in preschool-age children exposed to domestic violence. Journal of
Interpersonal Violence, 17(2), 150-164.

Levendosky, A. A., Huth-Bocks, A. C., Shapiro, D. L., & Semel, M. A. (2003). The
impact of domestic violence on the matemal-child relationship and preschool-age
children's functioning. Journal of Family Psychology, 1 7(3), 275-287.

Lieberman, A. F ., Van Horn, P., & Ozer, E. J. (2005). Preschooler witnesses of marital
violence: Predictors and mediators of child behavior problems. Development and
Psychopathology, 1 7, 385-396.

Lyons-Ruth, K. (1996). Attachment relationships among children with aggressive
behavior problems, the role of disorganized early attachment patterns. Journal of
Consulting and Clinical Psychology, 64(1), 64-73.

Lyons-Ruth, K., Alpem, L., & Repacholi, B. (1993). Disorganized infant attachment
classiﬁcation and maternal psychosocial problems as predictors of hostile-
aggressive behavior in the preschool classroom. Child Development, 64, 572-585.

Lyons-Ruth, K., & Jacobvitz, D. (1999). Attachment disorganization: Unresolved loss,
relational violence, and lapses in behavioral and attention strategies. In J. Cassidy
& P. R. Shaver (Eds), Handbook of attachment (pp. 89-111). New York: The
Guildford Press.

Magai, C. (1999). Affect, imagery, and attachment. In J. Cassidy & P. R. Shaver (Eds),
Handbook of attachment (pp. 787-802). New York: The Guilford Press.

81

Magdol, L., Mofﬁtt, T. E., Caspi, A., & Silva, P. A. (1998). Developmental antecedents
of partner abuse: A prospective-longitudinal study. Journal of Abnormal
Psychology, 107(3), 375-389.

Main, M., & Cassidy, J. (1988). Catagories of response to reunion with the parent at age
six: Predicted from infant attachment classiﬁcations and stable over a one-month
period. Developmental Psychology, 24, 415-426.

Main, M., & Goldwyn, R. (1984). Adult attachment scoring and classification systems
(Second ed.): Unpublished manuscript, University of California at Berkeley.

Main, M., & Hesse, E. (1990). Parent's unresolved traumatic experiences are related to
infant disorganized attachment status: Is frightened and/or frightening parental
behavior the linking mechanism? In M. T. Greenberg, D. Cicchetti & E. M.
Cummings (Eds), Attachment in the preschool years: Theory, research, and
intervention (pp. 161-184). Chicago: University of Chicago Press.

Main, M., Kaplan, N., & Cassidy, J. (1985). Security in infancy, childhood, and
adulthood: A move to the level of representation. Monographs of the Society for
Research in Child Development, 50(1-2), 66-104.

Main, M., & Solomon, J. (1990). Procedures for identifying infants as
disorganized/disoriented during the Ainsworth Strange Situation. In M. T.
Greenberg, D. Cicchetti & E. M. Cummings (Eds), Attachment in the preschool
years (pp. 121-160). Chicago: University of Chicago Press.

Margolin, G. (2005). Children's exposure to violence: Exploring developmental pathways
to diverse outcomes. Journal of Interpersonal Violence, 20(1), 72-81.

Margolin, G., & Gordis, E. B. (2000). The effects of family and community violence on
children. Annual Review of Psychology, 51, 445-479.

Marshall, L. (1992). Development of the severity of violence against women scales.
Journal of Family Violence, 7, 103-121.

McCloskey, L. A., Figueredo, A. J ., & Koss, M. P. (1995). The effects of systemic family
violence on children's mental health. Child Development, 66(5), 1239-1261.

McCloskey, L. A., & Stuewig, J. (2001). The quality of peer relationships among
children exposed to domestic violence. Development and Psychopathology, 13,
83-96.

NICHD-ECCRN. (2004). Trajectories of physical aggression from toddlerhood to middle
childhood. Monographs of the Society for Research in Child Development, 69(4).

82

Renken, B., Egeland, B., Marvinney, D., Mangelsdorf, S., & Sroufe, A. L. (1989). Early
childhood antecedants of aggression and passive withdrawal in early elementary
school. Journal of Personality, 5 7, 257-281.

Rose-Krasnor, L. (1997). The nature of social competence: A theoretical review. Social
Development, 6, 111-135.

Rosenblum, K. L., McDonough, S., Muzik, M., Miller, A., & Sameroff, A. (2002).
Maternal representations of the infant: Associations with infant response to the
Still Face. Child Development, 73(4), 999-1015.

Rossman, B. B. R. (1998). Descartes's error and posttraumatic stress disorder: Cognition
and emotion in children who are exposed to parental violence. In G. W. Holden,
R. Geffner & E. N. Jouriles (Eds), Children exposed to marital violence (pp. 223-
256). Washington, D. C.: American Psychological Association.

Scheeringa, M. S., & Zeanah, C. H. (2001). A relational perspective on PTSD in early
childhood. Journal of Traumatic Stress, 14(4), 799-815.

Schore, A. N. (1994). Aﬂect regulation and the development of the self: The neurobiology
of emotional development. Hillsdale, New Jersey: Lawrence Erlbaum Associates.

Smith, P. H., Thompson, G. E., DeVellis, R., Earp, J ., & Coker, A. L. (2002). A
population-based study of the prevalence and distinctiveness of battering, physical
assault, and sexual assault in intimate relationships. Violence Against Women,

8(10), 1208-1232.

Sroufe, A. L. (2005). Attachment and development: A prospective, longitudinal study
from brith to adulthood. Attachment & Human Development, 7(4), 349-367.

Stemberg, K. J., Lamb, M. E., Greenbaum, C., Ciccetti, D., Dawud, S., Cortes, R. M., et
al. (1993). Effects of domestic violence on children's behavior problems and
depression. Developmental Psychology, 29(1), 44-52.

Stemberg, K. J ., Lamb, M. E., Guterrnan, E., & Abbott, C. B. (2006). Effect of early and
later family violence on children's behavior problems and depression: A

longitudinal, multi-informant perspective. Child Abuse & Neglect, 30, 283-306.
Theran, S. A., Levendosky, A. A., Bogat, G. A., & Huth-Bocks, A. C. (2005). Stability
and change in mothers' internal representations of their infants over time.

Attachment & Human Development, 7, 253-268.

Van IJzendoom, M. H. (1992). Intergenerational transmission of parenting: a review of
studies in nonclinical populations. Developmental Review, 12, 76-99.

83

Van IJzendoom, M. H. (1995). Adult attachment representations, parental responsiveness
and infant attachment: A meta-analysis on the predictive validity of the Adult
Attachment Interview. Psychological Bulletin, 11 7(3), 387-403.

Ward, M. J., & Carlson, E. A. (1995). Associations among adult attachment
representations, maternal sensitivity, and infant-mother attachment in a sample of
adolescent mothers. Child Development, 66(1), 69-79.

Weinfeld, N. S., Sroufe, A. L., & Egeland, B. (2000). Attachment from infancy to early
adulthood in a high risk sample: Continuity, discontinuity, and their correlates.
Child Development, 71(3), 695-702.

Weinfeld, N. S., Sroufe, A. L., Egeland, B., & Carlson, E. A. (1999). The nature of
individual differences in infant-caregiver attachment. In J. Cassidy & P. R. Shaver
(Eds), Handbook of attachment: Theory, research, and clinical applications (pp.
68-88). New York: The Guilford Press.

Yates, T. M., Dodds, M. F., Sroufe, A. L., & Egeland, B. (2003). Exposure to partner
violence and child behavior problems: A prospective study controlling for child
physical abuse and neglect, child cognitive ability, socioeconomic stress, and life
stress. Development and Psychopathology, 15, 199-218.

Youngstrom, E., Loeber, R., & Stoutharner-Loeber, M. (2000). Patterns and correlates of
agreement between parent, teacher, and male adolescent ratings of externalizing
and internalizing problems. Journal of Consulting and Clinical Psychology, 68(6),
103 8-105 0.

Zeanah, C. H., & Benoit, D. (1995). Clinical applications of a parent perception interview
in infant mental health. Infant Psychiatry, 4(3), 539-554.

Zeanah, C. H., Benoit, D., Barton, M. L., & Hirshberg, L. (1996). Working Model of the
Child Interview Coding Manual.Unpublished manuscript, Louisiana State
University School of Medicine.

Zeanah, C. H., Benoit, D., Hirshberg, L., Barton, M. L., & Regan, C. (1994). Mothers'
representations of their infants are concordant with infant attachment.
Developmental Issues in Psychiatry and Psychology, 1, 1-14.

Zeanah, C. H., Danis, B., Hirshberg, L., Benoit, D., Miller, D., & Heller, S. S. (1999).
Disorganized attachment associated with partner violence: A research note. Infant
Mental Health Journal, 20(1), 77-86.

Zhou, Q., Eisenberg, N., Losoya, S. H., F abes, R. A., Reiser, M., Guthrie, I. K., et al.
(2002). The relations of parental warmth and positive expressiveness children's
empathy-related responding and social functioning. Child Development, 73(3),

893-915.

84