 

 

 

.64... L . 59. _>

.w{%

J. '9‘

. t

v...

nut-i- I."
I.

 

.7“ ‘ ”GAL.

79‘ (V

 

LIBRARY
Michigan State
University

~—

 

This is to certify that the
dissertation entitled

COMPLEX POSTTRAUMATIC STRESS SYMPTOMS
AMONG A COMMUNITY SAMPLE OF BATTERED WOMEN

presented by

KERRY LYNN LEAHY

has been accepted towards fulfillment
of the requirements for the

PhD. degree in Psychology

 

 

V; “U

,qum

Date

MSU is an affirmative-action, equal-opportunity employer

«-—.------.-o—-—-— — A

,-c-----n-o-.-o--c-----

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.

 

DAIEDUE

DAIEDUE

DAIEDUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

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

 

C OMPL
(

 

COMPLEX POSTTRAUMATIC STRESS SYMPTOMS AMONG A
COMMUNITY SAMPLE OF BATTERED WOMEN

By

Kerry Lynn Leahy

A DISSERTATION

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

DOCTOR OF PHILOSOPHY
Department of Psychology

2007

COMPI

Objective: The air:
violence (Dl'l \ieti

complex posttraum;

 

 

 

potential mediating
representations of e
detelopment of (‘P
Method: Cluster at
€Xperiences of DV

(threats of agareSsi

ABSTRACT

COMPLEX POSTTRAUMATIC STRESS SYMPTOMS AMONG A
COMMUNITY SAMPLE OF BATTERED WOMEN

By

Kerry Lynn Leahy
Objective: The aim of this study was to identify longitudinal patterns of domestic
violence (DV) victimization and to examine the relationship between these patterns and
complex posttraumatic stress disorder (CP) symptoms. An additional aim was to test the
potential mediating role of social support and the potential moderating roles of adult
representations of childhood attachment and child maltreatment history in the
development of CP symptoms among a community sample of female victims of DV.
Method: Cluster analysis techniques were used to summarize 164 women’s individual
experiences of DV victimization over six years. Frequency, severity, and type of abuse
(threats of aggression, physical DV, and sexual DV) were considered separately in the
cluster analyses. Regression analyses were used to examine the relationships between
DV trajectories and CP symptoms, and DV duration and CP symptoms. Baron and
Kenny’s (1986) procedures for testing mediation and moderation were applied to
examine social support as an intervening variable and adult representations of childhood
attachment and child maltreatment history as vulnerability factors in the proposed DV-CP
Symptom link. Results: Cluster analyses based on DV frequency, DV severity, threats of
DV, and physical DV produced nearly equivalent solutions; each produced Minimal,
Moderate, and High DV subgroups. The cluster analysis based on sexual DV produced a
2-group solution: Minimal and High sexual DV. The relationships between all cluster

solutions and CP symptoms, as well as DV duration and CP symptoms were signiﬁcant,

 

 

 

 

such that greater I)

 

of social support: 'ti
DV and CP sympti.

link. but adult repr

  

suggest that female
atiectixe and helm
accounted for by C
exacerbate DV-rel;
support suggests tl
community-d“ ell i
understood. Resu

such that greater DV was related to more CP symptoms. DV was related to poor quality
of social support; however, social isolation did not account for the relationship between
DV and CP symptoms. Child maltreatment history moderated the DV-CP symptoms
link, but adult representations of childhood attachment did not. Conclusions: Findings
suggest that female victims of DV (even those with moderate levels of DV) are at risk for
affective and behavioral dysregulation that transcends simple PTSD and may be better
accounted for by CP. In addition, the experience of abuse in childhood appears to
exacerbate DV-related CP symptoms. The lack of signiﬁcant ﬁndings regarding social
support suggests that mechanisms for the development of CP symptoms among
community-dwelling women who experience mild to moderate levels of DV are not well
understood. Results have signiﬁcant clinical implications for the psychological treatment

of women who experience DV.

 

First and iorerr
her thoughtful
Her mentoring
been inx'aluab
my commute
Alex to“ E}.
to Support m
L'I‘liV'ersity c
aim 10 the g
foi- this sum
[0 Share the
Finally. I a

ACKNOWLEDGMENTS

First and foremost, I would like to thank my dissertation chair, Alytia Levendosky, for
her thoughtful guidance, ﬂexibility, and patience throughout the production of this work.
Her mentoring and dedication to my success far exceed that which is expected and have
been invaluable in my professional and personal development. I would also like to thank
my committee members: Anne Bogat for her theoretical knowledge and editing skills,
Alex von Eye for his statistical expertise, and Tom Luster for his generosity in agreeing
to support my work. I am indebted to Natalia Katenka, a graduate student at the
University of Michigan, who taught me about cluster analysis techniques. Thank you
also to the graduate and undergraduate research assistants who assisted in data collection
for this study, and to the Mother-Infant Study participants for their time and willingness
to share their life experiences — without them, this research would not have been possible.
Finally, I am very grateful to my husband and family who provided continued support

and encouragement throughout this process.

iv

IISI OF TABLE}

IISI OF FIGL'RI.

   
  

Introduction ...........
Problem of'Dl
Mental Health ('
.4 Theoretical .ll.

Direct Effect (
Social lsoltitit
Personal l'ulr
Eridencefbr a (
Social Support.
Adult Represent
Child Maltreatt
Ht'pot/reses (“It
Htpmlrcs‘w‘ r
lltpotlrexis j
Htpotltesis j
Ht‘potlres‘is _
.l'pOtltt'sis .
.lpotlzt's‘is

 

 

Methods .........
Participants
rocedures
ecruitmcr
. ampling ..
ata C(lllL
raining...

I ‘ aniple RC

TABLE OF CONTENTS

LIST OF TABLES ........................................................................................................... viii
LIST OF FIGURES ........................................................................................................... xi
Introduction ......................................................................................................................... 1
Problem of DV ................................................................................................................ 6
Mental Health Consequences of DV ............................................................................... 7

A Theoretical Model of Complex Posttraumatic Stress Disorder for DV Victims ....... 13
Direct Effect of DV on CP Symptoms ....................................................................... 14
Social Isolation as an Explanatory F actor for the Development of CP ................... 21
Personal Vulnerability Factors for the Development of C P ..................................... 22
Evidence for a Complex PT SD Syndrome .................................................................... 27
Social Support ............................................................................................................... 36
Adult Representations of Childhood Attachment .......................................................... 40
Child Maltreatment History .......................................................................................... 44
Hypotheses and Rationale ............................................................................................. 45
Hypotheses 1a-1 d ...................................................................................................... 46
Hypothesis 2 .............................................................................................................. 49
Hypothesis 3a ............................................................................................................ 49
Hypothesis 3b ............................................................................................................ 50
Hypothesis 4 .............................................................................................................. 51
Hypothesis 5 .............................................................................................................. 52
Methods ............................................................................................................................. 53
Participants ................................................................................................................... 53
Procedures .................................................................................................................... 53
Recruitment ............................................................................................................... 53
Sampling ................................................................................................................... 54
Data Collection ......................................................................................................... 55
Training ..................................................................................................................... 56
Sample Retention ...................................................................................................... 57
Measures ....................................................................................................................... 58
Demographics ........................................................................................................... 58
Trauma Exposure ...................................................................................................... 58
Complex PT SD .......................................................................................................... 61
Social Support ........................................................................................................... 63
Adult Representations of Childhood Attachment ...................................................... 64
Results ............................................................................................................................... 67
Missing Data ................................................................................................................. 67
Hypothesis Testing ........................................................................................................ 68
Hypotheses Ia-Id ...................................................................................................... 68

Analyses for Hypothesis 1 ......................................................................................... 68

Hypothesis 2 .............................................................................................................. 72
Analysis for Hypothesis 2 .......................................................................................... 73
Hypothesis 3a ............................................................................................................ 81
Analyses for Hypothesis 3a ....................................................................................... 82
Hypothesis 3b ............................................................................................................ 86
Analyses for Hypothesis 3b ....................................................................................... 86
Hypothesis 4 .............................................................................................................. 92
Analyses for Hypothesis 4 ......................................................................................... 92
Hypothesis 5 .............................................................................................................. 94
Analyses for Hypothesis 5 ......................................................................................... 95
Discussion ......................................................................................................................... 98
Longitudinal Trajectories of DV Victimization ............................................................. 98
Relationship between DV Victimization and CP Symptoms ....................................... 104
Effect of Partnership Status on C P Symptoms ............................................................ 108
Social Support ............................................................................................................. 1 10
Adult Representations of Childhood Attachment and Child Maltreatment History 1 14
Summary ..................................................................................................................... 119
Study Strengths ........................................................................................................ 120
Study Limitations and Future Research Implications ............................................. 121
Clinical Implications ............................................................................................... 123
APPENDIX A: Demographics ....................................................................................... 172
APPENDIX B: Severity of Violence Against Women Scales ........................................ 174
APPENDIX C: Childhood Trauma Questionnaire — Short Form ................................... 176
APPENDIX D: Self-Report Inventory for Disorders of Extreme Stress ........................ 179
APPENDIX E: Norbeck Social Support Scale ............................................................... 185
APPENDIX F : Perceptions of Adult Attachment Questionnaire ................................... 188
APPENDIX G: Imputed Variables with Corresponding Matching Variables ................ 193
APPENDIX H: Clustering Results for Severity of DV, Threats of DV, and Physical DV ..
......................................................................................................................................... 196
APPENDIX 1: Cluster Results for Severity of DV, Threats of DV, and Physical DV

among Women in Long-term DV Relationships ................................... 202

APPENDIX J: Cluster Results for DV Severity and Threats of DV among Women in
Short-term DV Relationships ................................................................. 208

vi

REFERENCES ............................................................................................................... 212

vii

 

Table 1: Nu

ant

Table 2: De
Table 3: Ch
Table 4: C“
Table 5: Crt
Table 6: C“
Table 7: Cr:
Table 8: Cr.
Table 9: Cr
Table 10; CI
Table 1 1: CI
Rt

Table 11C]
Ctr

Table 13. C
81

a

Table 15: C

Table 1:

Table 2:

Table 3:

Table 4:

Table 5:

Table 6:

Table 7:

Table 8:

Table 9:

Table 10:

Table 1 1:

Table 12:

Table 13:

Table 14:

Table 15:

LIST OF TABLES

Number of Participants Missing Variable-level Data by Study Variable
and Time Period ...................................................................................... 128

Descriptive Data for Continuous Variables in Conceptual Model ......... 129
Cluster Mean Proﬁles for DV Frequency ............................................... 133

Cross-tabulation of DV Frequency and DV Severity Cluster Solutions ......
................................................................................................................. 134

Cross-tabulation of DV Frequency and Threats of DV Cluster Solutions
................................................................................................................. 135

Cross-tabulation of DV Frequency and Physical DV Cluster Solutions ......
................................................................................................................. 136

Cross-tabulation of DV Severity and Threats of DV Cluster Solutions .......
................................................................................................................. 137

Cross-tabulation of DV Severity and Physical DV Cluster Solutions... 138

Cross-tabulation of Threats of DV and Physical DV Cluster Solutions .......
................................................................................................................. 139

Cluster Mean Proﬁles for Sexual DV ..................................................... 140

Cluster Mean Proﬁles for DV Frequency among Women in Long-term DV
Relationships ........................................................................................... 141

Cross-tabulation of DV Frequency and DV Severity Cluster Solutions
among Women in Long-term DV Relationships .................................... 142

Cross-tabulation of DV Frequency and Threats of DV Cluster Solutions
among Women in Long-term DV Relationships .................................... 143

Cross-tabulation of DV Frequency and Physical DV Cluster Solutions
among Women in Long-term DV Relationships .................................... 144

Cross-tabulation of DV Severity and Threats of DV Cluster Solutions
among Women in Long-term DV Relationships .................................... 145

viii

Table 16:

Table 17:

Table 18:

Table 19:

Table 20:

Table 21:

Table 22:

Table 23:

Table 24:

Table 25:

Table 26:

Table 27:

Table 28:

Table 29:

Table 30:

Table 31:

Cross-tabulation of DV Severity and Physical DV Cluster Solutions
among Women in Long-term DV Relationships .................................... 146

Cross-tabulation of Threats of DV and Physical DV Cluster Solutions

among Women in Long-term DV Relationships .................................... 147
Cluster Mean Proﬁles for Sexual DV among Women in Long-term DV
Relationships ........................................................................................... 148

Cluster Mean Proﬁles for DV Frequency among Women in Short-term
DV Relationships .................................................................................... 149

Cluster Mean Proﬁles for Physical DV among Women in Short-term DV
Relationships ........................................................................................... 1 50

Cross-tabulation of DV Severity and Threats of DV Cluster Solutions
among Women in Short-term DV Relationships .................................... 151

Cross-tabulation of DV Severity and Physical DV Cluster Solutions
among Women in Short-term DV Relationships .................................... 152

Cross-tabulation of Threats of DV and Physical DV Cluster Solutions
among Women in Short-term DV Relationships .................................... 153

Cluster Mean Proﬁles for Sexual DV among Women in Short-term DV
Relationships ........................................................................................... 1 54

Means and Standard Deviations for Number of Network Supporters by
DV Frequency Cluster ............................................................................ 155

Means and Standard Deviations for Number of Network Supporters by
Sexual DV Cluster .................................................................................. 156

Means and Standard Deviations for Number of Network Supporters by
DV Duration Group ................................................................................ 157

Means and Standard Deviations for Emotional Support by DV Frequency
Cluster ..................................................................................................... 158

Means and Standard Deviations for Emotional Support by Sexual DV
Cluster ..................................................................................................... 159

Means and Standard Deviations for Emotional Support by UV Duration
Group ...................................................................................................... 160

Complex PTSD Symptom Scores as a function of DV Frequency Cluster
and Attachment Status ............................................................................ 161

ix

 

 

Table 33:

Table Ill:

Table H2:

Table H3:

Table ll:

Table II:

Table 13:

Table ll :

Table l2:

CUT”;
Aria.

Cllni

 

(Tu
Rel.

Table 32:

Table H1:

Table H2:

Table H3:

Table 11:

Table 12:

Table 13:

Table J 1:

Table J2:

Complex PTSD Symptom Scores as a function of Sexual DV Cluster and

Attachment Status ................................................................................... 162
Cluster Mean Proﬁles for DV Severity ................................................... 196
Cluster Mean Profiles for Threats of DV ................................................ 197
Cluster Mean Proﬁles for Physical DV .................................................. 198

Cluster Mean Proﬁles for DV Severity among Women in Long-term DV
Relationships ........................................................................................... 202

Cluster Mean Proﬁles for Threats of DV among Women in Long-term DV
Relationships ........................................................................................... 203

Cluster Mean Proﬁles for Physical DV among Women in Long-term DV
Relationships ........................................................................................... 204

Cluster Mean Proﬁles for DV Severity among Women in Short-term DV
Relationships ........................................................................................... 208

Cluster Mean Proﬁles for Threats of DV among Women in Short-term DV
Relationships ........................................................................................... 209

 

 

 

 

Tigure l:

figure I:
Figure 3:

Figure 4:
Figure 5:
F1Sure 6:
TiElite 7:
Time 8:

Figure 9;

Figme H1:
Figure H2:

Figure H3;

FIEUre 11 '.

Figure 12

FlgUre 13 ‘

“glue J1

 

Theoretit
Victims

Cluster
Cluster

Cluster
TCILIIIU‘

CIUSK
TCI'JIIt

CIUSI
relari

(‘IU\
PCT at

Clu

Tel:

.‘xrc

Cl

C

Figure 1:

Figure 2:
Figure 3:

Figure 4:

Figure 5:

Figure 6:

Figure 7:

Figure 8:

Figure 9:

Figure H1:
Figure H2:
Figure H3:

Figure 11:

Figure 12:

Figure 13:

Figure .1 1:

LIST OF FIGURES

Theoretical model for the development of CP Symptoms in female

Victims of DV ......................................................................................... 163
Cluster mean proﬁles for DV frequency ................................................ 164
Cluster mean proﬁles for sexual DV ...................................................... 165

Cluster mean proﬁles for DV frequency among women in long-term DV
relationships ............................................................................................ 166

Cluster mean proﬁles for sexual DV among women in long-term DV
relationships ............................................................................................ 167

Cluster mean proﬁles for DV frequency among women in short-term DV

relatibnships ............................................................................................ 168
Cluster mean proﬁles for physical DV among women in short-term DV
relationships ............................................................................................ 1 69
Cluster mean proﬁles for sexual DV among women in short-term DV
relationships ............................................................................................ l 70
Mean number of network supporters by DV frequency cluster and time ”.7"
................................................................................................................. l 1
Cluster mean proﬁles for DV Severity ................................................... 199
Cluster mean proﬁles for Threats of DV ................................................ 200
Cluster mean proﬁles for Physical DV ................................................... 201

Cluster mean proﬁles for DV Severity among women in long-term DV
relationships ............................................................................................ 205

Cluster mean proﬁles for Threats of DV among women in long-term DV
relationships ............................................................................................ 206

Cluster mean proﬁles for Physical DV among women in long-term DV
relationships ............................................................................................ 207

Cluster mean proﬁles for DV severity among women in short-term DV
relationships ............................................................................................ 21 0

xi

 

 

 

Figure 13:

 

 

 

CIUS

relat;

 

 

Figure J2: Cluster mean profiles for threats-of DV among women in short-term DV
relationships ............................................................................................ 21 1

xii

 

Domestic Violcn.

mittens of aggressiur

 

public health problem \\
risk for developing a \ :1
disorder tPTSDl. a ssm
exhibited by indi\ idua“.
actual or threatened \ it
Horanitz. 8; Raxxlings
abned \\ omen are his
31-.199S:Kemp et al.
nor appear to fully ca

ta'ticularlv given tlt
c % ‘

PTSD mm, r
the Criteria Were (1
CV

Cir .
tumscnbed tram

SV
'ndrom com I
P c
Calmll‘e The p
S\

Introduction

Domestic violence (DV; deﬁned here as male physical violence, sexual violence,
and threats of aggression perpetrated against female romantic partners) is a signiﬁcant
public health problem with serious mental health consequences. Abused women are at
risk for developing a variety of mental health problems, including posttraumatic stress
disorder (PTSD), a syndrome developed to account for the mental health symptoms
exhibited by individuals who experience a traumatic event (i.e., an event that involves
actual or threatened violence or death, Cascardi & O'Leary, 1992; Kemp, Green,
Hovanitz, & Rawlings, 1995; Kemp, Rawlings, & Green, 1991). Rates of PTSD in
abused women are high, ranging from 45% to 84% (Houskamp & Foy, 1991; Kemp et
al., 1995; Kemp et al., 1991; Vitanza, Vogel, & Marshall, 1995). However, PTSD does
not appear to fully capture the psychological sequelae of DV for all battered women,
particularly given the high rates of co-morbid psychiatric disorders experienced by many
abused women (e. g., Nixon, Resick, & Nishith, 2004).

PTSD may not fully account for the mental health consequences of DV because
the criteria were deve10ped to reﬂect the psychological response to an acute or
circumscribed traumatic event in which the primary stressor is no longer present (e. g.,
automobile accident, natural disaster), whereas DV is a repeated, prolonged interpersonal
trauma (M. A. Dutton & Goodman, 1994; Herman, 1992a; 1992b; van der Kolk, 2002a).
Repeated, prolonged interpersonal trauma is deﬁned as interpersonal physical, sexual
and/or psychological abuse without a discrete beginning or ending. An alternative
syndrome, complex posttraumatic stress disorder (CP) was developed speciﬁcally to

capture the psychological consequences of repeated, prolonged interpersonal trauma and

therefore, may better account for the mental health effects of DV experienced by some
women. CP is characterized by psychological disturbances in three major areas: 1) a
poly-symptomatic psychiatric presentation, including somatization, dissociation, and
depression, 2) poor self-perceptions and interpersonal functioning, and 3) vulnerability to
experiences of future harm, both self-inﬂicted and perpetrated by others (Herman, 19923;
1992b).

The present research proposes and examines a theoretical model for the
development of CP symptoms in female victims of DV. This model includes a direct link
from DV to CP symptoms with DV frequency, severity, duration, and abuse type
differentially affecting the development of CP symptoms. Speciﬁcally, in this model, CP
symptoms are proposed to be the direct result of DV exposure with women who
experience high levels of DV frequency, severity, and duration developing more
symptoms as compared to women who experience less frequent, severe, and chronic
abuse. In addition, sexual DV is more likely to cause CP than physical DV or threats of
DV. The model also assumes that women in long-term DV relationships are more likely
to develop CP symptoms than women in one or more short-term DV relationship(s).
Furthermore, the model includes an indirect link from DV to CP symptoms through
social support, suggesting that the development of CP symptoms is partially due to poor
social support related to DV. Finally, the model includes two moderator variables—
insecure adult representations of childhood attachment and child maltreatment history—
which are expected to exacerbate DV-related CP symptoms (see Figure 1).

This model for the development of CP symptoms in battered women is inﬂuenced

by Herman’s (1992a; 1992b) conceptualization of CP. Herman (1992a; 1992b) asserts

that repeated, prolonged interpersonal trauma occurs when a relationship of coercive
control develops between the perpetrator and victim. The perpetrator establishes a
coercive relationship with the victim over time through his abusive actions, such as
threats of violence, actual violence, and psychological abuse techniques. In a coercive
relationship, the perpetrator has the power to exercise continued abuse and to shape the
attitudes, beliefs, thoughts, feelings, and behaviors of the victim, eventually undermining
the victim’s psychological resistance and mental health. Thus, subordination to coercive
control results in CP, a complex constellation of observable, mental health symptoms that
transcend the arousal, intrusive, and avoidant symptoms of PTSD (Herman, 1992a;
1992b)

Although the model for this study draws primarily on Herman’s (1992a; 1992b)
formulation of CP for theoretical support, Herman’s (1992a; 1992b) theory has
signiﬁcant limitations. Speciﬁcally, Herman (1992a; 1992b) does not adequately
describe how and why victims of interpersonal trauma are vulnerable to future harm
perpetrated by others. In addition, CP theory does not explicitly address how the speciﬁc
conﬁguration of abuse experienced by victims may contribute to differential levels or
patterns of CP symptoms. Furthermore, individual differences in risk for CP are not
discussed. Thus, the theory does not fully explain the mechanisms underlying risk for
repeated harm and does not specify how abuse characteristics and factors apart from the
abuse may directly affect or interact to affect CP symptoms. As a result, the current
study draws on constructivist, dynamic-maturational, and attachment theories to provide

further support for the proposed model that suggests that the speciﬁc conﬁguration of

abuse differentially affects the development of CP symptoms and includes two factors
that affect vulnerability to DV-related CP symptoms.

Empirical research has provided evidence for a CP syndrome among repeated,
prolonged interpersonal abuse victims in general, particularly among victims of sexual
abuse (McLean & Gallop, 2003; Pelcovitz et al., 1997; Roth, Newman, Pelcovitz, van der
Kolk, & Mandel, 1997). Studies suggest that CP is speciﬁc to trauma victims (Pelcovitz
etal., 1997), with exposure to an interpersonal stressor, severity and duration of abuse,
sexual abuse, and early-onset of abuse presenting particular risks for the development of
CP (McLean & Gallop, 2003; Pelcovitz et al., 1997; Roth etal., 1997). The syndrome
has subsequently been referred to under a variety of names, including “Disorders of
Extreme Stress” (DES: Davidson, 1993), “Disorders of Extreme Stress Not Otherwise
Speciﬁed” (DESNOS: Pelcovitz et al., 1997; van der Kolk et al., 1996), and “lasting
personality changes following catastrophic stress” (World Health Organization [WHO],
1992, p.91). In the current DSM-I V nomenclature (American Psychological Association
[APA], 1994), the CP symptom constellation is incorporated under “Associated Features
of PTSD.”

Although ﬁndings provide evidence for the existence of a CP syndrome, there are
some noteworthy limitations in the existing empirical literature. For example, prior
studies of CP have relied primarily on treatment-seeking Caucasian samples, and have
rarely identiﬁed the relationship between victim and perpetrator. This methodology has
perhaps limited the application of ﬁndings to the general population of interpersonal
abuse victims, the majority of whom do not seek treatment, and has precluded the

identiﬁcation of the speciﬁc nature of the violence. Therefore, the validity of the CP

construct has not been explored among battered women speciﬁcally, even though DV
qualiﬁes as an interpersonal trauma, which is often re-occurring and chronic in nature.
Previous studies have also failed to test the assumptions and mechanisms outlined in CP
theory, instead relying on ﬁndings supporting the presence of CP symptoms among
interpersonal abuse victims as providing conﬁrmation for Herman’s (1992a; 1992b)
formulation of CP theory.

The present research begins to address these limitations, and thus was designed to
contribute to the further development of an empirical and theoretical literature on CP by
proposing a theoretical model for the development of CP symptoms in victims of DV
based on trauma and attachment theories, and testing this model in a community sample
of abused women with lifetime histories of DV ranging in severity from mild to severe.
Women were recruited during pregnancy as part of a larger, on-going longitudinal
investigation of the impact of DV on women and children (The Mother-Infant Study:
Levendosky, Bogat, Davidson, & von Eye, 2000). The current study examines CP
symptoms as opposed to diagnostic status due to the non-clinical nature of the sample.
Clinically signiﬁcant CP is expected to be rare in this population, but the core features of
CP are expected to occur with sufﬁcient frequency to determine their relationship to DV.

Speciﬁcally, this study utilizes a person-oriented, developmental approach by
examining CP symptoms in relation to groups of women who have experienced similar
longitudinal patterns of DV victimization based on a variety of abuse characteristics (e.g.,
DV severity) across six waves of data collection. In addition, the study examines the
importance of DV partnership and the process of social isolation in the development of

CP symptoms. Finally, the study tests social support as a mediator and adult

representations of childhood attachment and child maltreatment history as moderators of
the hypothesized DV-CP symptom link. Study ﬁndings are expected to add speciﬁcity to
the existing body of knowledge on CP, speciﬁcally in relation to DV, and have important
clinical implications for understanding the experiences of battered women, such that
psychological symptoms can be linked, where appropriate, to histories of abuse.
Problem of DV

DV is a signiﬁcant public health problem, accounting for 20-64% of all violence
against women according to estimates from two national crime surveys (Rennison, 2003;
Tjaden & Thoennes, 2000a). Lifetime prevalence estimates for DV are high and similar
across studies with rates of spousal or partner rape ranging from 8-10% (Finkelhor &
Yllo, 1987; Russell, 1990; Tjaden & Thoennes, 20003) and rates of physical assault
steady at about 22% (Bureau of Justice Statistics [BJ S], 1998; Tjaden & Thoennes,
2000a). Twelve-month prevalence estimates of DV are more variable. Annual
prevalence rates based on national surveys range from less than 2% to almost 14%
(Rennison, 2003; Schafer, Caetano, & Clark, 1998; Tjaden & Thoennes, 1998, 2000b,
2000c). Community-based studies employing a broader deﬁnition of DV, including
psychological abuse, produce higher estimates ranging from 18-57% (Meisel, Chandler,
& Rienzi, 2003; Smith, Thornton, DeVellis, Earp, & Coker, 2002; Waltermaurer, Ortega,
& McNutt, 2003). Based on US. Census estimates for the number of women age 18 or
older, these ﬁndings suggest that at minimum 1.5 million women experience DV annually
in the US with approximately 4.8 million incidents of DV perpetrated against women
each year in the US. (Tjaden & Thoennes, 1998, 2000b). Furthermore, battered women

are at risk for injury, (Rennison, 2001; Tjaden & Thoennes, 1998, 2000a, 2000b), mental

health problems (e.g., Kemp et al., 1995), and even death at the hands of an intimate
partner (Rennison, 2003). Thus, ﬁndings from both national and community-based
surveys indicate that DV affects a signiﬁcant percentage of women with great personal
costs to physical and mental health.
Mental Health Consequences of DV

Battered women are at risk for developing PTSD, a syndrome developed to
capture the psychological consequences of traumatic events, including DV (Ceballo,
Ramirez, Castillo, Caballero, & Lozoff, 2004; Coker et al., 2002; Frasier et al., 2004;
Houskarnp & Foy, 1991; Kemp et al., 1995; Kemp et al., 1991; Mertin & Mohr, 2000;
Nixon et al., 2004; N011, Horowitz, Bonanno, Trickett, & Putnam, 2003; Pico-Alfonso,
2005; Sharhabani-Arzy, Amir, Kotler, & Liran, 2003; Stein & Kennedy, 2001; Stein,
Kennedy, & Twamley, 2002). Rates of PTSD among battered women are high, ranging
from 45% - 81% depending on sample characteristics, type of DV, and history of prior
interpersonal trauma (Houskarnp & F oy, 1991; Kemp et al., 1995; Sharhabani-Arzy et al.,
2003; Stein & Kennedy, 2001). A recent meta-analysis found that the weighted mean
prevalence of PTSD among abused women was 64%, providing a combined prevalence
estimate across 11 individual studies of PTSD (Golding, 1999). DV appears to be the
greatest predictor of current PTSD symptomatology even after controlling for the effects
of other interpersonal abuse experiences, including childhood abuse and non-partner
related adulthood victimization (Pico-Alfonso, 2005). In addition, a limited number of
studies suggest that frequency and severity of DV are related to risk and severity of
PTSD (Cascardi & O'Leary, 1992; Houskarnp & Foy, 1991; Mertin & Mohr, 2000;

Mitchell & Hodson, 1983; Pico-Alfonso, 2005). However, it is difﬁcult to obtain reliable

prevalence estimates of PTSD by DV severity because the majority of the DV literature
does not report on severity of violence. Thus, rates of PTSD among women experiencing
mild to moderate levels of DV are unclear. Finally, Bogat, Levendosky, Theran, von
Eye, and Davidson (2003) found that chronic experiences of DV were related to higher
PTSD symptoms scores than less chronic DV experiences, suggesting an effect of
duration of abuse on PTSD symptomatology.

Although PTSD accounts for some of the psychiatric consequences of DV,
battered women experience co-morbid psychopathology and a range of psychiatric
symptoms not captured in the PTSD construct, including depression, generalized anxiety,
somatization, and dissociation. For example, a large number of studies consistently
indicate that battered women report symptoms of depression (Cascardi, Langhinrichsen,
& Vivian, 1992; Cascardi & O'Leary, 1992; Christopoulos et al., 1987; Coker et al.,
2002; Doyle, Frank, Saltzman, McMahon, & Fielding, 1999; Frank & Dingle, 1999;
Kemp et al., 1991; Mitchell & Hodson, 1983; Moore, Pepler, Mae, & Michele, 1989;
Nixon et al., 2004; Olson et al., 2003; Stein & Kennedy, 2001; Wolfe, Zak, Wilson, &
Jaffe, 1986). Clinical depression and PTSD are often co-morbid. For instance, Stein and
Kennedy (2001) found that PTSD and major depressive disorder (MDD) co-occurred
more often than would be expected by chance among women with recent experiences of
DV recruited from DV agencies and medical clinics. Speciﬁcally, 43% of women with
current DV-related PTSD also suffered from MDD. Similarly, Nixon et a1. (2004) found
that PTSD and MDD co-occurred in 49% of a help-seeking sample of battered women
with recent abuse experiences. Co-morbid PTSD and MDD diagnoses were more

common than PTSD alone (27%). In addition, individuals diagnosed with both disorders

reported more severe PTSD and MDD symptoms than individuals with PTSD alone
(Nixon et al., 2004), indicating that severity of distress increases with multiple diagnoses.

The literature on DV also reliably indicates that battered women experience more
general symptoms of anxiety in addition to those subsumed under the PTSD diagnosis
(Coker et al., 2002; Kemp et al., 1991; Moore et al., 1989; Olson et al., 2003; Wolfe et
al., 1986), as well as somatic symptoms (Frasier et al., 2004; Lown & Vega, 2001). For
example, Lown and Vega (2001) found that a community sample of battered women was
more likely than a non-abused sample of women to report one or more signiﬁcant
somatic symptom(s) for ﬁve symptom groups: gastrointestinal, cardiopulmonary,
neurological, sexual, and reproductive (Lown & Vega, 2001). To illustrate, women with
histories of DV were twelve times more likely to report urinary retention, nine times
more likely to report general pain, amnesia, and paralysis, and four to six times more
likely to report sexual or reproductive symptoms, such as painful intercourse or
menstruation, as compared to non-abused women (Lown & Vega, 2001). Other
frequently cited somatic complaints among battered women include headaches; dizziness;
gastrointestinal disturbances, such as abdominal pain, nausea, and indigestion; joint pain;
numbness, tingling, or pain in the extremities; racing or pounding heartbeat; shortness of
breath; and chest pain (Frasier et al., 2004; Lown & Vega, 2001).

Though not well researched, one study suggests that DV may be related to
dissociation. Dorahy, Lewis, and Wolfe (2007) compared 33 women who resided in a
shelter for abused women to a sample of 33 non-abused women recruited from the
general population in Northern Ireland. The researchers found that women in the DV

group scored signiﬁcantly higher on scales assessing dissociation as compared to women

in the non-abused group. Furthermore, peritraumatic dissociative episodes (i.e., those
occurring during domestic assaults) were related to higher on-going dissociative
experiences. However, these results must be interpreted with caution as the DV group
also reported more childhood maltreatment than the non-abused group of women. Thus,
it is difﬁcult to parse apart the effects of childhood maltreatment and DV on dissociation.

In addition to placing battered women at risk for depression, anxiety, and
dissociation, DV may lead to disruptions in personality and behavioral functioning, two
areas of disturbance that are not well accounted for by PTSD. Speciﬁcally, DV may lead
to alterations in personal identity, including the development of poor self-perceptions,
and may place women at risk for future harm both self-inﬂicted and perpetrated by
others. Battered women interviewed for psychiatric evaluation reported low self-esteem,
as well as feeling incompetent, unworthy, and unlovable (Hilberman & Munson, 1978).
Research studies support clinical work, indicating that battered women suffer from low
self-esteem (D. G. Dutton & Painter, 1993; Kemp et al., 1991; Mitchell & Hodson,
1983). In addition, qualitative data and empirical research indicate that DV is related to
feelings of shame and guilt (Hilberman & Munson, 1978; Street & Arias, 2001). Shame,
which refers to global, negative evaluations of the self, may alter personal identity.
Furthermore, clinical reports on battered women (Walker, 1984) and ﬁndings from a
national sample of women (Umberson, Anderson, Glick, & Shapiro, 1998) suggest that
victims of DV have a diminished sense of personal control, which results in feelings of
powerlessness and helplessness.

Preliminary ﬁndings also suggest a possible link between DV and personality

disorders. Stanley and Penhale (1999) conducted a small pilot study of 13 mothers with

10

severe mental health problems whose children were involved in the child protection
system. A detailed analysis of the nature of the mental health problems revealed that 6 of
the women had been formally diagnosed with a personality disorder, and all 13 women
had experienced a history of DV. Furthermore, Sansone, Reddington, Sky and
Wiederman (2007) reported a relationship between borderline personality disorder and
DV among a sample of female primary care patients. Speciﬁcally, 64.0% of abused
women scored in the clinical range on a measure of borderline personality disorder, as
compared to only 11.1% of non-abused women. These preliminary ﬁndings implicate
DV as an important contextual factor for women’s mental health problems, particularly
personality disorders.

Poor self-esteem coupled with depression and anxiety may lead to self-harm in an
attempt to manage intense painful affect or to self-validate negative perceptions of the
self. Compared to nonabused women, battered women are more likely to engage in risky,
damaging health behaviors, such as cigarette smoking (Frasier et al., 2004; Weaver &
Etzel, 2003), and alcohol (Roberts, Lawrence, O'Toole, & Raphael, 1997) and substance
abuse (Coker et al., 2002; Doyle et al., 1999; Plichta, 1996), and those who are more
severely abused are more vulnerable to the negative effects of the unhealthy behaviors.
For example, Weaver and Etzel (2003) found that almost 60% of a sample of severely
battered women reported that they were current smokers. Among those current smokers,
women who experienced more severe sexual violence, and greater domination and
isolation were more likely to exhibit greater symptoms of nicotine-related physical
dependence (Weaver & Etzel, 2003). Risky sexual behaviors, such as having multiple

sexual partners and engaging in high-risk types of sexual activity have also been

11

observed in victims of DV (Champion, Shain, Piper, & Perdue, 2001 ). In addition to
behaviors, sexually permissive attitudes and sexual preoccupation (e. g., positive attitudes
toward pornography and thinking about sex frequently) were positively related to
interpersonal victimization, including DV, in a study of young women (N 011 et al., 2003).
These risky behaviors and attitudes are self-destructive, and may also indirectly increase
the likelihood of encountering a perpetrator of assault through impaired judgment (in the
case of substance use) and exposure to highly sexual situations (in the case of sexual
behaviors and attitudes).

In its extreme form, self-harm may manifest as suicidal ideation and behavior, a
common sequel to DV. Research suggests that battered women are more likely to report
suicidal ideation (Coker et al., 2002), as well as suicide attempts (Doyle et al., 1999;
Frank & Dingle, 1999; Kernic, Wolf, & Holt, 2000; Olson et al., 2003; Roberts et al.,
1997) than non-abused women. Speciﬁcally, battered women are ﬁve times more likely
to attempt suicide than non-abused women (Stark & Flitcraft, 1996).

In addition to self-harm, survivors of DV are at risk for interpersonal
revictimization inﬂicted by others. Women involved in abusive romantic relationships
are frequently exposed to other experiences of adult victimization (Pico-Alfonso, 2005).
Pico-Alfonso (2005) found that female victims of DV had higher rates of non-partner
related, adulthood physical, psychological, and sexual victimization than non-battered
women.

The poly-symptomatic psychiatric presentation, poor interpersonal functioning,
and negative self-perceptions commonly observed among battered women, as well as

research indicating that battered women are at risk for self-harm and other experiences of

12

interpersonal victimization suggests that PTSD does not sufﬁciently capture the
psychological sequelae of DV for all abused women. Many battered women experience a
range of mental health problems that exceed the re-experiencing, avoidance, and arousal
symptoms of PTSD, and may co-occur. In order to fully account for the mental health
outcomes documented among victims of DV and other repeated, interpersonal traumas
that transcend the symptoms of PTSD, Herman (1992a; 1992b) proposed a psychological
syndrome of repeated, prolonged, interpersonal trauma—complex posttraumatic stress
disorder (CP).

A Theoretical MOdel of Complex Posttraumatic Stress Disorder for DV Victims

The current study proposes a theoretical model for the development of CP

symptoms in women exposed to DV, and tests this model by examining longitudinal
trajectories of DV victimization in relation to CP symptoms. This model includes a
direct link between DV victimization and CP symptoms, and assumes that particular
abuse characteristics and DV relationship length differentially affect the development of
CP symptoms in abused women. Speciﬁcally, the model suggests that women who
experience greater frequency, severity, and duration of abuse, as well as sexual abuse are
at greater risk for developing CP symptoms than women who experience less frequent,
chronic, and severe forms of abuse. The model further assumes that women involved in
long-term DV relationships are more vulnerable to developing CP symptoms as
compared to women in one or more short-term DV relationship(s). In addition, the model
includes an indirect link from DV to CP through social support, suggesting that the
development of CP symptoms is due, in part, to poor social support related to DV

victimization. Finally, the model includes two moderator variables that increase

13

vulnerability to CP symptoms in response to DV: insecure adult representations of
childhood attachment and experience of child maltreatment (see Figure 1).
Direct Effect of DV on CP Symptoms

The proposed model is integrative and draws on trauma and attachment theories
to explain the development of CP symptoms in DV victims. First, in this model, DV has
a direct effect on CP symptoms. According to Herman’s (1992a; 1992b)
conceptualization of CP, the perpetrator uses systematic and repeated abuse to create a
coercive relationship, in which the perpetrator becomes powerful and the victim has little
power. Over time, the perpetrator’s use of coercive methods, which may include a
combination of physical violence, sexual violence, threats of aggression, and/or
psychological abuse, depletes the victim’s external resources (e. g., social support) and
internal resources (e. g., initiative, feelings of control), ultimately resulting in a complex
group of mental health symptoms characterized by behavioral and affect dysregulation
(Herman, 1992a; 1992b).

Impact of DV on interpersonal functioning and personal identity. The
development of interpersonal and identity disturbances involves a complex process. The
perpetrator uses violence and threats of aggression to create an environment of fear, in
which the victim is afraid to assert power by not complying with the perpetrator’s wishes
or following his rules. The perpetrator capitalizes on this environment and uses a variety
of psychological abuse tactics to decrease the victim’s independence and social support,
which further limits the victim’s power. For example, the perpetrator may limit the
victim’s economic independence by making important ﬁnancial decisions without

consulting the victim or by restricting her access to money. The perpetrator may

14

decrease the victim’s social support by intercepting telephone calls and letters,
monitoring his victim’s whereabouts, and ruining sentimental items, such as pictures and
gifts that are symbolic of the victim’s connection to others. Thus, the perpetrator uses
DV to undermine the victim’s sense of autonomy and to limit her social connections,
forcing the victim to become dependent on the perpetrator for economic security,
information, emotional sustenance, and in severe cases, bodily needs (Herman, 1992a).

Over time, forced dependency on the perpetrator results in constriction of
initiative, referred to as “passivity” and “learned helplessness” among battered women, as
well as a bond of identiﬁcation with the perpetrator called “traumatic bonding” (Herman,
1992a, 1992b; Walker, 1979). The victim possesses ambivalent feelings toward her
perpetrator; she may fear him, and paradoxically, feel confused, worthless, and helpless
without him (Walker, 1979). Eventually, the victim may view all relationships “through
a lens of extremity” as she experiences her relationship with the perpetrator (Herman,
1992b, p. 385). This new understanding of relationships may lead to poor interpersonal
functioning, in which the victim behaves ambivalently with others, alternating between
extreme dependency and withdrawal.

Impact of DV on mental health. In addition to interpersonal and identity
disturbances, Herman’s (1992a; 1992b) conceptualization of CP suggests that victims of
repeated, interpersonal trauma, including DV, suffer from a range of psychiatric
symptoms, including somatization, dissociation, and depression. According to Herman
(1992a; 1992b), the experience of prolonged, repeated, and unpredictable violence, such
as DV, results in the chronic expectation of danger, which in turn, causes hyperarousal—

chronic arousal of the nervous system. Hyperarousal leads to observable somatic

15

complaints (physical or bodily problems that defy medical explanation) including
insomnia, headaches, and gastrointestinal problems, as well as irritability and agitation
(Grinker & Spiegel, 1945; Herman, 1992a, 1992b; Kardiner & Spiegel, 1947).

The need to cope with overwhelming affect associated with abuse activates
dissociation, a basic psychological defense mechanism that impairs capacity for cognitive
integration (Terr, 1991). Dissociation is initially an adaptable, conscious form of self-
hypnosis employed to mentally avoid misery and pain, as information about traumatic
experiences is disconnected, disjointed, and relegated to separate aspects of
consciousness preventing interference with everyday level of consciousness (Luxenberg,
Spinazzola, & van der Kolk, 2001). However, with repeated abuse, the dissociative
process becomes automatic, involuntary, and relatively ﬁxed (Terr, 1991). Prolonged use
of dissociation may result in amnesia, depersonalization, concentration and attention
difﬁculties, and a disrupted sense of time. In its extreme form, dissociation may lead
deformations of identity such as those characteristic of dissociative identity disorder
(Herman, 1992a, 1992b).

Abuse and betrayal at the hands of an intimate partner (one who has theoretically
committed to loving and protecting the victim) may disrupt the victim’s spiritual faith
and belief systems. This is particularly true when others are aware of the abuse and do
not intervene appropriately, allowing the violence to continue, as well as when others
engage in victim-blaming or denial of the abuse. Information that is incompatible with
previously formed belief systems results in cognitive and affective changes, most
prominently a state of depression (Herman, 1992a, 1992b; Rorty & Yager, 1996). In

addition, the internalization of unexpressed anger toward the perpetrator (which the

16

victim is unable to express appropriately for fear of retribution) may manifest as
symptoms of depression, low self-worth, and in extreme cases, deliberate self-harm, such
as self-mutilation and suicidal preoccupation (Herman, 1992b).

Risk for repeated harm among DV victims. Finally, Herman’s (1992a; 1992b) CP
theory suggests that the self-defeating behaviors associated with internalized anger (i.e.,
self-injurious behavior and suicidality), as well as other forms of compulsive self-harm,
including alcohol and substance use and risky sexual behavior may represent attempts to
manage painful affects by numbing or distracting the self from emotional pain. It is also
possible that self-harming behaviors serve to communicate pain to others, and/or serve a
self-validating function by conﬁrming negative beliefs about the self (e. g., “I am bad or
evil,” “I deserve to be punished”). Unfortunately, these coping strategies are maladaptive
and may lead to impaired judgment (in the case of alcohol and substance use) and
involvement in dangerous situations (in the case of risky sexual behavior) which increase
vulnerability to harm perpetrated by others (e. g., stranger perpetrated harm and
involvement in repeated abusive relationships). Vulnerability to self-inﬂicted harm and
harm inﬂicted by others may also represent unconscious behavioral reenactments of
traumatic experiences, in which the victim unsuccessfully attempts to achieve cognitive
and affective mastery over the trauma (Freud, 1995; Herman, 1992a, 1992b).

Although Herman’s (1992a; 1992b) formulation of CP describes how the
perpetrator comes to control his victim and break down her psychological resistance,
resulting in pathological relationship functioning, identity disturbances, and multiple
psychiatric symptoms, CP theory does not fully explain why victims of repeated,

prolonged abuse are vulnerable to repeated harm perpetrated by others. Two additional

17

theoretical perspectives on trauma provide support for the link between DV and the
proclivity to reenact abusive relationships, a core symptom of CP. Speciﬁcally, a
constructivist theory of trauma suggests that individuals develop constructs or templates
to understand and predict their own behavior as well as the behavior of others (Harter &
Neimeyer, 1995). Women living in DV relationships develop idiosyncratic constructions
of relationships based on their experiences with an abusive partner. Therefore, non-
abusive relationships may lie outside of the abused women’s range of constructions
useful for anticipating behavior. As a result, abused women may unwittingly choose
similar abusive partners because the consequences of such relationships, although
negative, are at least predictable (Harter & Neimeyer, 1995). Alternatively, in her
formulation of a dynamic-maturational approach to trauma, Crittenden (1997) suggests
that the victim of DV learns after repeated failures to protect herself from abuse that her
actions are futile and that she is incapable of keeping her physical or psychological self
safe. At this point, the woman becomes completely vulnerable to repeated harm because
she has reached a state of learned helplessness in which all efforts at self-protection have
ceased.

Overall, trauma theories (Crittenden, 1997; Freud, 1995; Harter & Neimeyer,
1995; Herman, 1992a, 1992b; Terr, 1991; Walker, 1979, 1984) suggest a direct link
between DV and CP symptoms. However, not all women share the same experiences of
DV, and different experiences of abuse may differentially inﬂuence CP symptoms. In
her formulation of CP, Herman (1992a; 1992b) does not articulate how the speciﬁc
conﬁguration of DV and other coercive techniques may vary across batterers and

contribute to differential levels or patterns of CP symptoms among abused women. In

18

contrast, the proposed model suggests that certain abuse characteristics (i.e., DV severity,
frequency, duration, type, and partnership) differentially affect the development of CP
symptoms in response to DV, and draws on other trauma theories to support such
hypotheses.

Differential effects of abuse characteristics on CP. Speciﬁcally, severity of DV is
one abuse characteristic in the proposed model that inﬂuences the development of CP
symptoms, such that women with patterns of DV characterized by severe levels of abuse
develop more CP symptoms as compared to women with less severe trajectories of abuse.
The development of a coercive relationship, which ultimately undermines mental health,
is most likely to occur when abuse is severe. Women who experience severe DV have
few options to escape perpetrator control due to the risk of serious danger.
Insubordination, attempts to escape, or any exertion of power on behalf of the victim may
result in serious bodily injury or death when the abuse is severe. Therefore, severe abuse
may be an effective method of achieving and maintaining control over the victim.

Likewise, frequency of abuse differentially inﬂuences the development of CP
symptoms in the proposed model. Women who experience frequent DV may learn
through punishment that any behavior that is not sanctioned by the perpetrator is likely to
result in abuse. Women in this situation are at the mercy of the perpetrator because
protection of physical safety requires submission. In addition, when abuse is frequent,
the victim lives in a constant state of fear and physiological arousal resulting in high
levels of somatization, and may employ dissociation on a regular basis in an attempt to
escape psychological and physical pain. Women who experience frequent DV may feel

completely helpless because any initiative of self will almost indeﬁnitely result in abuse.

19

Feelings of ineffectiveness may result in'a state of learned helplessness, leaving women
vulnerable to future harm because all efforts at self-protection cease (Crittenden, 1997).
In contrast, if violence is relatively infrequent, women may be able to assert more power
and take more “risks” because assertiveness rarely results in DV. Thus, in cases of
infrequent abuse, women may be able to resist perpetrator control.

Type of DV is another abuse characteristic relevant to the development of CP in
the proposed model. Sexual abuse may place women at relatively greater risk for the
development of CP than other types of abuse because sexual abuse represents the most
intimate form of betrayal and an assault on reproductive control. Thus, in addition to
representing an intimate invasion of the self, sexual abuse undermines a primary function
of human evolution (Crittenden, 1997), leaving sexually abused women powerless to
control even the most basic human function.

Furthermore, the proposed model suggests that duration of DV is a characteristic
of abuse, which inﬂuences the development of CP symptoms. The CP construct was
developed to reﬂect the impact of traumatic experiences, which are “prolonged” in nature
without a well-deﬁned beginning or ending, in contrast to psychological disorder
resulting from a single traumatic event. According to trauma theory, a perpetrator
establishes a relationship of coercive control, which ultimately leads to CP symptoms,
over time with repeated use of coercive techniques (Herman, 1992a, 1992b). For
example, the process of social isolation occurs over time through the gradual
implementation of restrictive rules preventing social contact with others.

Finally, the proposed model suggests that relationship length with the perpetrator

is relevant to the development of CP symptoms. Herman’s (1992a; 1992b) CP theory

20

implies that a relationship of coercive control and the resulting CP symptoms develop
over time with a single perpetrator. Although the perpetrator may be able to create fear
in his victim after one abusive incident, autonomy, social support, and spiritual faith, for
example, do not deteriorate considerably after one episode of abuse. The creation of a
relationship of coercive control and the eroding of the victim’s psychological resistance is
a gradual process. Involvement in multiple short DV relationships where subordination
to coercive control never begins is less likely to result in CF symptoms. Thus, the current
study examines longitudinal trajectories of DV based on DV severity, frequency, type,
and duration in relation to CP symptoms separately for women in long-term DV
relationships (deﬁned here as three or more consecutive years) and for women in one or
more short-term DV relationship(s).
Social Isolation as an Explanatory F actor for the Development of CP

Although limited data and sample size preclude a test of all of the mechanisms
underlying the theoretical relationship between DV exposure and CP symptoms, the
current research examines the process of social isolation, and its role in the DV-CP
symptom link. Speciﬁcally, the proposed model suggests that social support partially
mediates the relationship between DV victimization and CP symptoms. According to
trauma theory (Herman, 1992a; 1992b), the DV perpetrator decreases his victim’s social
contacts and support through the use of psychological abuse, such as discouraging visits
with family and ﬁiends. Because the victim is isolated from others, the perpetrator
becomes the victim’s primary source of support. As dependency on the perpetrator
increases, the victim may view the perpetrator as both a source of fear and a source of

solace. Eventually, the victim may generalize this ambivalence to other relationships,

21

resulting in unhealthy interpersonal functioning, in which the victim alternates between
dependency and withdrawal in relationships with others, a primary disturbance
characteristic of CP (Herman, 1992a, 1992b). In addition, as the victim’s social support
deteriorates, she is less likely to have the external resources to leave a violent
relationship, making her vulnerable to continued abuse and its direct mental health
consequences, such as somatization and dissociation. Finally, the deterioration of social
support reinforces the isolation and withdrawal symptoms of depression. Thus, DV leads
to social isolation, which in turn, affects the woman’s mental health. In this study, social
support is examined over time to test whether the process of social isolation occurs in
victims of DV, and thus, whether poor social support can be considered a reasonable
indicator of social isolation and examined as a mediator in DV-CP symptom link.
Personal Vulnerability F actors for the Development of CP

In addition to testing for mediation, the model addresses individual differences in
risk for developing CP symptoms in response to DV. Herman’s (1992a; 1992b)
conceptualization of CP does not address individual differences in vulnerability to CP
symptoms. However, characteristics of traumatic events, as well as characteristics of the
individual likely interact to inﬂuence adaptation to trauma. Perhaps the failure to address
personal vulnerability to CP symptoms was a deliberate decision to emphasize that
psychological problems experienced by trauma victims are effects, not causes of abuse.
Nonetheless, preexisting individual differences certainly contribute to the particular
psychological response to abuse. As a result, the proposed model includes two factors

suggested by attachment theory that may inﬂuence personal vulnerability to trauma

22

exposure. such a

 

maltreatment hi .\

Adult ref
that an individua
develops in relati

risk for psychopa

 

.\'lohr.2001). Spc

 

vulnerability to tr:

According
by an innate dispt‘
system is aCIIVaIL‘l
illness. and by ext
the primary attach
and inner security

Theoretica

available and SCUK

exposure, such as DV: adult representations of childhood attachment and child
maltreatment history.

Adult representations of childhood attachment. First, attachment theory suggests
that an individual’s strategy for responding to conditions of danger, which originally
develops in relation to the primary caregiver, may be related to individual differences in
risk for psychopathology in combination with other risk factors, such as DV (Cassidy &
Mohr, 2001). Speciﬁcally, an insecure attachment organization may create psychological
vulnerability to trauma exposure, including DV.

According to Bowlby (1969), the founder of attachment theory, infants are driven
by an innate disposition to form an attachment system with primary caregivers. This
system is activated under conditions of danger and distress by internal cues, such as
illness, and by external threats to safety, which drive the infant to seek protection from
the primary attachment figure. Caregiver responses determine the infant’s sense of safety
and inner security.

Theoretically, infants with caregivers who are physically and emotionally
available and sensitive develop primary, secure attachment strategies. In situations of
threat, securely attached infants seek caregiver protection and then return to exploration
when the distress or threat has passed (Main, 1990). However, infants with unresponsive,
unavailable, or insensitive caregivers adopt secondary defensive behavioral strategies for
gaining caregiver access and preserving the attachment relationship. Insecure infants
either deactivate the attachment system, shifting attention away from a rejecting caregiver

and minimizing displays of distress to avoid pain and conﬂict within the attachment

23

 

 

svstem. 0f hyper;

   
   
  

insistent demand>
Early atta
mental representa

the attacllmsmI ﬁ

"”3

behavior in subse
interpersonal relati

attachment relatior
as a relatively stab

In addition
interpersonal beha
regulation ('Cassid;
in the contexr of a

Intemalize self-re

t‘
:

about internal caln

giving bond derail
attachm

system, or hyperactivate the attachment system with exaggerated displays of distress and
insistent demands for care and attention to maintain caregiver contact (Main, 1990).

Early attachment experiences with the primary caregiver crystallize into internal
mental representations or working models representing the relationship among the self,
the attachment figure, and the external world. Internal working models guide social
behavior in subsequent relationships, and affect expectations and perceptions about future
interpersonal relationships throughout the lifespan (Bowlby, 1977, 1980). Thus, the early
attachment relationship plays a vital role in the development of a sense of self and serves
as a relatively stable template for future relationships.

In addition to inﬂuencing an understanding of the interpersonal realm and shaping
interpersonal behavior, early attachment experiences affect the development of emotion
regulation (Cassidy & Mohr, 2001; van-der-Kolk & F isler, 1994). Children who develop
in the context of a supportive relationship with the primary caregiver gradually
internalize self-regulating functions initially provided by the caregiver designed to bring
about internal calm under conditions of stress. However, disruption of the early care-
giving bond derails this critical developmental process. Individuals with a damaged
attachment system may not successfully learn how to internally regulate affective
impulses or engage in adaptive self-soothing behaviors. In response to situations of
overwhelming affect, such as trauma exposure, these individuals may be forced to rely on
basic defenses, including dissociation, or engage in dangerous external behaviors, such as
self-mutilation or sexual promiscuity in attempt to modulate affect (Cassidy & Mohr,
2001; van-der-Kolk & Fisler, 1994). Therefore, a damaged attachment system disrupts

essential developmental processes, including the development of healthy emotion and

24

 

behavior regular
1

theoretically plat
especially in con
attachment is a r
affective. relatio:
attachment stratc
emergence of p53
insecure attachm
a predisposing in
|

Child ma;

(defined here as l
neglect experieng
mOderator of the

1993a 1993b) su

 

child maltreatmer
such as DV; how:

experience of earl

behavior regulation, adaptive coping strategies, and mature interpersonal functioning,
theoretically placing a child at risk for short- and long-term psychological problems,
especially in combination with other risk factors (Bowlby, 1977). As a result, disrupted
attachment is a pathway to diverse forms of psychopathology, including a range of
affective, relational, and behavioral disturbances captured by CP. Therefore, an insecure
attachment strategy may be a risk factor for sensitivity to later risk conditions and the
emergence of psychopathology, including CP symptoms. The present research examines
insecure attachment based on adult recollections of childhood care-giving experiences as
a predisposing individual-level factor for CP symptoms in relation to DV.

Child maltreatment history. Finally, in this model, child maltreatment history
(deﬁned here as physical, sexual, and emotional abuse, and physical and emotional
neglect experienced during childhood and adolescence) is prOposed as a possible
moderator of the hypothesized DV-CP symptom relationship. Trauma theory (Herman,
1992a, 1992b) suggests that the experience of prolonged, interpersonal trauma, including
child maltreatment, renders victims vulnerable to repeated experiences of later abuse,
such as DV; however, trauma theory does not conceptually address how and why the
experience of early abuse may affect psychological response to later abuse. Attachment
theory provides some insight into this shortcoming of trauma theory.

Speciﬁcally, attachment theory suggests that disrupted attachment to the primary
caregiver, as often occurs in situations of child maltreatment (D. Barnett, Ganiban, &
Cicchetti, 1999; E. A. Carlson, 1998), derails the critical developmental process of
learning to internally modulate physiological arousal and affective states. As noted

above, failure to internalize self-regulating functions forces the individual to rely on basic

25

 

 

defenses. such a
substance abuse
early reliance or
development of
experiences of c

negatively affect

 

many individual:
in which the indi
coping strategies
of the attachmen
psychologically

poor coping mec
more adaptive st
context in which
Effects of later e:

the proposed Stu

DV-related CP 5

defenses, such as dissociation, and external, maladaptive coping mechanisms, such as
substance abuse, to calm internal discomfort (van-der-Kolk & F isler, 1994). In addition,
early reliance on external coping strategies and immature defenses may prevent the
development of more psychologically sophisticated defensive strategies. Thus,
experiences of child maltreatment may disrupt the attachment system, which in turn,
negatively affects capacity to regulate emotions. Furthermore, Gold (2000) asserts that
many individuals who experience child maltreatment also grow up in ineffective families
in which the individual does not learn necessary skills for living, including adaptive
coping strategies for modulating emotions. Poor emotional regulation, whether the result
of the attachment process or a deﬁcient learning environment, may leave individuals
psychologically vulnerable to both current and future traumatic experiences, as the use of
poor coping mechanisms and basic defensive strategies becomes automatic over time and
more adaptive strategies are not learned. Thus, early experiences of abuse and the
context in which it occurs may make an individual more sensitive to the psychological
effects of later experiences of DV, increasing the risk of DV-related CP symptoms. In
the proposed study child maltreatment is examined as a potential vulnerability factor for
DV-related CP symptoms.

By specifying the relationships among abuse characteristics and CP, and
including a potential mediator and two potential moderators of the hypothesized DV-CP
symptom link, the current research agenda was designed to add to a theoretical literature
on CP, particularly in relation to DV. This model is supported by trauma and attachment
theories, as well as a growing body of empirical literature linking previously isolated

mental health symptoms, which can be understood in the context of a traumatic reaction

26

to repeated. prob

attributed to mull

 

A handful
providing evident
trauma confirmin
reexperiencing. a
psychological im
Several studies e:
of the DSM-Il ' fr
among a combin.
adolescents and z
1997; van der K(

For exam

dissociation. son‘

to repeated, prolonged interpersonal abuse, including DV, rather than erroneously
attributed to multiple psychological conditions and enduring personality factors.
Evidence for a Complex PT SD Syndrome

A handful of empirical studies have examined CP among trauma victims,
providing evidence for a CP syndrome in victims of repeated, prolonged interpersonal
trauma conﬁrming Herman’s (1992a; 1992b) theoretical proposition that the
reexperiencing, arousal, and avoidance symptoms of PTSD do not adequately capture the
psychological impact of interpersonal trauma exhibited by a subset of individuals.
Several studies examining complex human adaptation to trauma were conducted as a part
of the DSM-I V ﬁeld trial for PTSD, a multisite investigation of post-trauma pathology
among a combination of treatment-seeking (n=395) and community recruited (n=125)
adolescents and adults (Newman, Riggs, & Roth, 1997; Pelcovitz et al., 1997; Roth et al.,
1997; van der Kolk et al., 1996).

For example, van der Kolk and colleagues (1996) investigated PTSD in relation to
dissociation, somatization, and affect dysregulation—the primary psychiatric symptoms
of CP. Correlations revealed that PTSD and its associated features are highly related, and
to a greater extent than may be suggested in the DSM-I V (APA, 1994). Participants with
a diagnosis of lifetime PTSD scored signiﬁcantly lower on scales assessing dissociation,
somatization, and affect dysregulation than those with current PTSD, but signiﬁcantly
higher than those who never had PTSD. Subjects who developed PTSD in response to
interpersonal trauma (physical and/or sexual abuse) in adolescence or adulthood had
signiﬁcantly fewer associated features than those who endured childhood abuse, but

signiﬁcantly more than those who experienced disaster. Thus, PTSD, dissociation,

27

somatization. 3
characteriSIiCS (
trauma (van de’
Based o
ﬁfty trauma EXT
structured inten
SIDES) to operz
Kolk et al. (199!
several ﬁeld tria
studies generally
victims of early-
or before age 13
example, Pelcov
victims and 38° c
compared to onl;
diagnosed among
stressors that did
FIEld trial

example’ Netvma

somatization, and affect dysregulation often occur together, in part as a function of abuse
characteristics (i.e., age of onset, nature of trauma), representing a complex response to
trauma (van der Kolk et al., 1996).

Based on a systematic review of the literature and contributions from a survey of
ﬁfty trauma experts, Pelcovitz and colleagues (1997) developed a criterion set and
structured interview (i.e., the Structured Interview for Disorders of Extreme Stress:
SIDES) to operationally deﬁne the complex symptom constellation described by van der
Kolk et al. (1996) and Herman (1992a; 1992b). The SIDES was subsequently used in
several ﬁeld trial studies to examine the utility of the CP construct. Findings from these
studies generally indicate that CP is a syndrome speciﬁc to trauma that primarily affects
victims of early-onset interpersonal abuse (i.e., physical and/or sexual abuse beginning at
or before age 13: Newman et al., 1997; Pelcovitz et al., 1997; Roth et al., 1997). For
example, Pelcovitz et al. (1997) reported that 63% of early-onset interpersonal abuse
victims and 38% of late-onset interpersonal abuse victims met criteria for CP, as
compared to only 10% of those who witnessed a disaster. Furthermore, CP was rarely
diagnosed among non-trauma exposed individuals (3% among individuals exposed to
stressors that did not qualify for DSM-I V PTSD criteria).

Field trial studies also suggest that CP is highly co-morbid with PTSD. For
example, Newman (1997) reported that 53% of trauma-exposed individuals who met
criteria for current PTSD also met criteria for CP. The rate of co-morbid PTSD and CP
increased to 72% when victims of interpersonal abuse were examined separetely from

individuals who experienced other types of trauma (Roth et al., 1997). Results also

28

 

indicate that C I’

the Nevvman et L

Addition,

ofthe current P’l

lie. the inability
current lives) as t
(that individuals \v
thematic disruptir
trauma; interpersc
Cerrience of a m
e.‘tperience and the

subset (22:50) oftl

 

symptoms were or
{1 .
ndrng suggests th

CP construct may 1

Category (Newman

Findings frr

abuse differentially

of sexually abused
of PTSD also met c

indicate that CP typically occurs in conjuction with PTSD; less than 3% of participants in
the Newman et al. (1997) study were diagnosed with CP alone.

Additional ﬁndings from Newman et al.’s (1997) study speak to the heterogeneity
of the current PTSD construct. Individuals with CP exhibited greater thematic disruption
(i.e., the inability to resolve trauma-related issues and create adaptive meaning in their
current lives) as compared to those with pure PTSD and those without PTSD, suggesting
that individuals with CP are the most pervasively affected by trauma. Differences in
thematic disruption among the three groups appeared to be a function of the nature of the
trauma; interpersonal abuse was associated with greater thematic disruption than was the
experience of a non-interpersonal stressor, controlling for age of ﬁrst traumatic
experience and the number of traumatic experiences. Further analyses performed on a
subset (n=50) of the same sample indicated that severity of intrusive and avoidant
symptoms were greatest in the group of individuals with co-morbid PTSD and CP. This
ﬁnding suggests that those with CP represent the most severe cases of PTSD. Thus, the
CP construct may represent a marker of severity, rather than a distinct diagnostic
category (Newman et al., 1997).

Findings from a ﬁnal DSM-I V ﬁeld trial study suggest that type of interpersonal
abuse differentially affects risk for CP. Speciﬁcally, Roth et a1. (1997) reported that 76%
of sexually abused women (regardless of physical abuse status) with a lifetime diagnosis
of PTSD also met criteria for lifetime CP, as compared to 53% of those women who
reported physical abuse only. Multiple regression analyses revealed that women who
reported both sexual and physical abuse were 14.5 times more likely to meet a diagnosis

of CP than those who had not experienced both types of abuse, suggesting a cumulative

29

 

impact of abuse .

   
   

compared to no
women. especial
higher risk of de

CP occurred at a

abuse demonstra“

Given the
CP S-‘mPIOmS e)
dEGmF‘ Dickin:
tit-omen in prim;
childhood Sexu;
the research t ea
suffered from s
and averaged n
CSA and dUrat:

and family hist

impact of abuse. In addition, sexual abuse resulted in a risk of CP 4.4 times as great as
compared to women who experienced physical abuse only. Thus, sexually abused
women, especially those who had also experienced physical abuse, had a signiﬁcantly
higher risk of developing CP than those who experienced physical abuse alone; though,
CP occurred at a high base rate among physically abused women (27%). Taken together,
these results suggest that physical and sexual abuse are risk factors for CP, with sexual
abuse demonstrating greater speciﬁcity.

Given the speciﬁc risk for CP associated with sexual abuse, two studies examined
CP symptoms exclusively among sexual abuse survivors. For example, Dickinson,
deGruy, Dickinson, and Candib (1998) provided evidence of a CP syndrome among
women in primary care outpatient settings with a history of sexual abuse, including
childhood sexual abuse (CSA) and/or adult sexual assault (n=99). Using cluster analysis,
the research team identiﬁed a homogenous subgroup of sexual abuse victims who
suffered ﬁ'om symptoms of depression, PTSD, dissociation, and somatoform disorder,
and averaged more than six psychiatric diagnoses. Furthermore, increased severity of
CSA and duration of abuse, within family perpetration, greater number of perpetrators,
and family history of antisocial behavior were associated with greater overall
symptomatology. This symptom proﬁle may be better characterized by CP, as opposed
to a plethora of distinct, discrete, apparently unrelated disorders.

Building on these ﬁndings, McLean and Gallop (2003) examined the full CP
construct among a small sample (n=65) of adult female CSA survivors obtaining services
from urban mental health clinics in Canada. The researchers found a signiﬁcant degree

of overlap between CP and borderline personality disorder with virtually all women

30

 

    
    
  

meeting criteria
significantly hill
lie. abuse occu.
years of age). su-
borderline persoi
onset of abuse al.
higher lifetime re
onset group. mak
the development
revealed that part
Incidents) were 5

In a more
SimPIOms amon
experiencing CU}
Sample; hOWQVC
adolescence exh
features of (11330

absence of PTS I

meeting criteria for both disorders. Furthermore, diagnoses of both disorders were
signiﬁcantly higher among participants reporting early- versus late-onset sexual abuse
(i.e., abuse occurring at or prior to 12 years of age versus abuse occurring at or beyond 13
years of age), suggesting a relationship between early abuse and co-morbid CP and
borderline personality disorder. However, the group of women who experienced an early
onset of abuse also reported higher paternal rates of abuse, no single incidents of abuse,
higher lifetime revictimization, and higher biparental neglect as compared to the late-
onset group, making it difﬁcult to disentangle the relative contribution of these factors to
the development of CP and borderline personality disorder. A separate set of analyses
revealed that paternal sexual abuse and frequency of abuse (i.e., experiencing 10 or more
incidents) were signiﬁcant predictors of diagnostic status.

In a more recent study, Ford, Stockton, Kaltman, and Green (2006) assessed CP
symptoms among a community sample of female college students who were not
experiencing current trauma. Findings suggest that full CP is rare among a community
sample; however, women who experienced interpersonal trauma during childhood or
adolescence exhibited elevated levels of CP symptoms independent of PTSD (i.e., the CP
features of dissociation, altered relationships, and systems of meaning occurred in the
absence of PTSD: 32-43% discordance). Furthermore, women who experienced repeated
interpersonal trauma reported more CP symptoms than women who experienced a single
incident of non-interpersonal or interpersonal trauma, and women who experienced a
single incident interpersonal trauma reported more CP symptoms than women who
experienced non-interpersonal trauma or no trauma. Generally, few CP symptoms were

reported by women with no trauma history.

31

 

These fr

 

syndrome speci
interpersonal in
a severe form o=

constitute a dist:

  
 

response to traut
by acute stress d
between.
Finally. '.
confirmation fo
“psychometric 1
(MCMI) among
(p.289). C onsi
included a com
Chronic depresj
Profound isolat
directed aggreg
increased Sym]
Other t)‘lths of r

levels (J_ G A'

These ﬁndings are consistent with ﬁeld trial studies in suggesting that CP is a
syndrome speciﬁc to trauma, and that risk for CP is greater when abuse is frequent and
interpersonal in nature. However, unlike ﬁeld trial studies that suggest that CP represents
a severe form of PTSD (e.g., Newman et al., 1997), these ﬁndings suggest that CP may
constitute a distinct posttraumatic reaction to interpersonal abuse. Thus, the human
response to trauma may be more accurately conceptualized along a continuum anchored
by acute stress disorder at one end, CP at the other end, and simple PTSD somewhere in
between.

Finally, literature on personality assessment of trauma victims provides additional
conﬁrmation for a CP syndrome. Allen, Coyne, and Huntoon (1998) delineated a
“psychometric prototype of CP” based on the Million Clinical Multiaxial Inventory-III
(MCMI) among a sample of psychiatric inpatient females with childhood abuse histories
(p.289). Consistent with Herman’s (1992a; 1992b) formulation of CP, this prototype
included a complex pattern of symptomatic and personality disturbances: severe and
chronic depression and anxiety, high prevalence of somatization, low self-esteem, and
profound isolation coupled with insecure attachment, masochism, dependency, and self-
directed aggression. In addition, CSA demonstrated the strongest relationship to
increased symptom scores on the clinical syndrome scales of the MCMI as compared to
other types of childhood abuse, and duration of abuse predicted higher clinical symptom
levels (J. G. Allen et al., 1998).

Another study on the same sample of women revealed ﬁve distinct clusters based
on MCMI personality disorder scales: four clusters exhibited personality proﬁles

consistent with CP and one showed an absence of personality disturbance (J. G. Allen,

32

 

Huntoon. 8; EV
relative promin
symptomatic at
most severel," C
and dependenc
lovv self-esteen
substance use.
Allen et al.. 19
exhibit signifii
resilience.
Overal
childhood trat
PTSD, and fit
support the e)
imerpersonal
develODment
severity of p—
also Suggem j
that Cp can (
How,
Primarily Ca
addition, a

ll

general Were

Huntoon, & Evans, 1999). Although the four disturbed clusters were distinguished by the
relative prominence of particular personality problems, all the groups demonstrated
symptomatic and personality disturbances captured by the descriptive label of CP. The
most severely disturbed cluster of women showed a pattern of depression, masochism,
and dependency (as did women in the remaining disturbed clusters), profound isolation,
low self-esteem, paranoia, dissociation, aggression, PTSD, anxiety, somatization,
substance use, psychoticism, poor coping skills, and poor interpersonal functioning (J. G.
Allen etal., 1999). The identiﬁcation of a subgroup of abused women who did not
exhibit signiﬁcant psychopathology in response to trauma may be indicative of personal
resilience.

Overall, results from studies examining the long-term impact of sexual abuse or
childhood trauma, those from studies conducted as a part of the DSM—I V ﬁeld trial for
PTSD, and ﬁndings from the personality assessment of victims of childhood abuse
support the existence of a CP syndrome among victims of repeated, prolonged
interpersonal abuse. Several abuse characteristics appear to be related to the
development of CP, such as age of onset, severity, duration, and type of abuse, as well as
severity of PTSD symptoms and experiencing more than one type of abuse. Findings
also suggest high rates of co-occurring PTSD and CP, though there is some indication
that CP can occur in the absence of PTSD.

However, these studies have several limitations, including the reliance on
primarily Caucasian, treatment-seeking samples and retrospective reports of abuse. In
addition, all but one of the studies examining CP among interpersonal abuse victims in

general were based on the same sample of women—the ﬁeld trial’s sample—or a subset

33

of that larger sample. These limitations may restrict the external validity of ﬁndings and
introduce the possibility of recall bias. Furthermore, relationships between perpetrators
and victims were not speciﬁed in any of the studies, with the exception of the McLean
and Gallop (2003) study, in which paternal sexual abuse was identiﬁed as a risk variable.
Thus, the extent to which battered women were included in the Dickinson et al. (1998)
study of sexual victimization or the ﬁeld trial sample is unclear, making it difﬁcult to
determine the applicability of these ﬁndings to victims of DV.

Although a consensus is emerging in the literature suggesting the integrity of CP
as a diagnostic construct for understanding the psychological responses of victims of
repeated, prolonged interpersonal trauma, particularly CSA victims, Herman’s (1992a;
1992b) formulation of CP does not focus exclusively on sexual abuse. Rather, CP can
theoretically result from a wide-range of repeated, prolonged interpersonal abuse
experiences, including DV (Herman, 1992a). The present study is the ﬁrst to examine
CP symptoms among a community sample of battered women from relatively diverse
backgrounds with a range of lifetime DV experiences. Speciﬁcally, this study examines
CP symptoms in relation to various longitudinal patterns of DV victimization based on
several theoretically- and empirically-derived abuse characteristics (e. g., DV severity).
Furthermore, symptoms are examined separately for women in long and short-term DV
relationships to address the relevance of DV partnership in the development of CP
symptoms. The current study examines the full range of CP symptoms using a set of
criteria speciﬁcally designed to assess the CP syndrome (CP symptoms are reported
rather than diagnostic status given that the sample is community-based and fewer women

are likely to meet clinical thresholds than would be expected among treatment-seeking

34

 

(n

In

samples). This study is designed to provide information about how individual differences
in DV victimization, including whether the violence occurs in the context of one or
multiple violent relationships, relate to CP symptoms, further specifying a theoretical
model for the development of CP in battered women.

Although the CP construct as a whole has not yet been examined among battered
women speciﬁcally, clinical and empirical literature regarding the impact of DV on
women suggest the presence of symptoms consistent with the constellation of
symptomatic, relational, and behavioral disturbances subsumed under the CP
nomenclature with frequency, severity, and duration of abuse predicting severity of
mental health symptoms (see above for review of Mental Health Consequences of DV).
In addition to noting symptoms consistent with the CP syndrome among abused women,
studies of DV indicate that battered women report more structural and subjective
dependency on the perpetrator, view themselves as powerlessness, and experience greater
social isolation as compared to their non-battered counterparts (Farris & F enaughty,
2002; Forte, F ranks, Forte, & Rigsby, 1996; Rokach, 2006). For example, Farris and
F enaughty (2002) found that physically abused women reported that their partners
discouraged friendships and leaving the home, which resulted in social isolation. These
ﬁndings provide some support for potential explanatory processes underlying the
relationship between prolonged interpersonal abuse and CP proposed by Herman (1992a;
1992b) in her theoretical conceptualization of CP. Although limited data and sample size
prevent the examination of each of these potential mechanisms underlying the DV-CP
link, the current study examines the process of social isolation, and the potential

mediating role of social support in the hypothesized DV-CP symptom relationship.

35

 

relationsl
D\-’ on sr
levendo
consistcr
tempura
Carlson.
severity
member
support;
then di
&th
For 6x3;
“Omen
dining 1
qUality

When [I

Writro]
elamir
infOIm

SUPDOr

Social Support

Previous empirical studies have not examined the role of social support in the
relationship between DV and CP; however, several studies ﬁnd a direct negative effect of
DV on social support (0. W. Barnett, Martinez, & Keyson, 1996; Curry & Harvey, 1998;
Levendosky et al., 2004; Mitchell & Hodson, 1983, 1986; Thompson et al., 2000)
consistent with the proposed theoretical model of CP, with only one study ﬁnding
comparable levels of social support between battered and non-battered women (B. E.
Carlson, McNutt, Choi, & Rose, 2002). Generally, studies indicate that frequency and
severity of DV is related to poor social support, such as few social support network
members; low perceived social support, including emotional, informational, and practical
support; and decreased likelihood of receiving social support from friends and relatives
when disclosing abuse (Matud-Aznar, Aguilera-Avila, Marrero-Quevedo, Moraza-Pulla,
& Carballeira-Abella, 2003; Mitchell & Hodson, 1983, 1986; Thompson et al., 2000).
For example, El-Bassel, Gilbert, Rajah, F oleno, and Frye (2001) found that abused
women reported few individuals to turn to for support, experienced insufﬁcient support
during their relationships with abusive partners, and expressed dissatisfaction with the
quality of support they had received. These women also considered the threat of DV
when making choices about seeking support.

Although study ﬁndings suggest that DV perpetrators may use dominance and
control to isolate women and prevent them from accessing social support, these studies
examined social support at one point in time. Cross-sectional studies do not provide
information on the direction of causality; thus, it is possible that women with poor social

support are more likely to experience DV, rather than DV precipitating loss in support.

36

 

Furu
test ‘
over
viol:
SUP;
med

in tt

rela
sup
maj
abu
One
hea
rel;
Ion
In:

hea

Furthermore, examining social support at one point in time does not permit researchers to
test whether social support deteriorates in response to DV. Examination of social support
over time is necessary to determine whether the process of social isolation occurs in
violent relationships. Thus, the current study examines longitudinal trajectories of social
support in relation to DV victimization. Furthermore, social support is examined as a
mediator in the DV-CP symptom link, such that DV predicts poor social support, which
in turn, is related to CP symptoms.

Although prior studies have not examined the role of social support in the
relationship between DV and CP, several prior studies have examined the role of social
support in the relationship between DV and other symptoms characteristic of CP. The
majority of these studies have treated social support as a moderating variable, buffering
abused women from the negative effects of DV (e.g., Levendosky et al., 2004); however,
one study has examined social support as an intervening variable between DV and mental
health outcomes. Thompson et al. (2000) found that social support mediated the
relationship between DV and psychological distress and PTSD symptoms in a sample of
low-income, African American women seeking routine medical care at an urban hospital
(n=l 38). Findings from this study suggest that the negative effects of DV on mental
health outcomes may be explained by low levels of social support.

Other studies of DV, social support, and mental health outcomes also support the
possibility of a mediation model. Several studies ﬁnd a direct effect of DV on social
support (e.g., Levendosky et al., 2004: see above for review), and an association between
social support and mental health outcomes in the context of DV (e. g., Coker, Watkins,

Smith, & Brandt, 2003). Studies examining the relationship between social support and

37

mental health symptoms in battered women generally indicate a positive association
between support and psychological well-being, and a negative association between
support and mental health problems captured by CP (Coker et al., 2002; Coker et al.,
2003; Kemp et al., 1995; Tan, Basta, Sullivan, & Davidson, 1995; Thompson et al.,
2000). Speciﬁcally, studies have linked structural and functional social support to
general levels of psychological functioning (Thompson et al., 2000), depression
(Anderson, Saunders, Yoshihama, Bybee, & Sullivan, 2003; B. E. Carlson et al., 2002;
Fowler & Hill, 2004; Mitchell & Hodson, 1983; Tan et al., 1995), anxiety (B. E. Carlson
et al., 2002), somatization (Matud-Aznar et al., 2003), suicide attempts (Meadows,
Kaslow, Thompson, & Jurkovic, 2005), self-esteem (Mitchell & Hodson, 1983), and
PTSD (Astin, Lawrence, & Foy, 1993; Thompson et al., 2000) in community and shelter
samples of abused women.

For example, Levendosky et al. (2004) found that structural and functional
support predicted mental health outcomes, controlling for abuse severity. Three
structural support variables—total number of supporters, abuse disclosure, and
nonhomophily (i.e., fewer network supporters experiencing DV)—predicted better
mental health among pregnant women in violent romantic relationships. Functional
support was also a signiﬁcant predictor for the mental health of battered women.
Speciﬁcally, more criticism predicted lower self-esteem, and greater practical aid
predicted higher self-esteem and lower anxiety. In addition, Coker and colleagues (2002)
reported that abused women with greater emotional support were less likely to report
current poor mental health, anxiety, depression, PTSD symptoms, and suicidal ideation

and actions than were abused women reporting lower emotional support and controlling

38

 

for

35

00

1h.

for frequency and severity of abuse, race, age, and Medicaid insurance status. A study on
a sub-sample of the same participants (n=19l) revealed similar ﬁndings. Higher scores
on emotional support were associated with better self-perceived mental health among
those women reporting current physical, sexual, or psychological abuse (Coker et al.,
2003). F urtherrnore, a woman’s experience of battering was indirectly associated with
mental health through emotional support; however, mediation was not adequately tested
(i.e., direct effects and mediation models were not statistically compared to determine the
difference in model ﬁt). Finally, Matud et al. (2003) reported that social support
correlated positively with self-esteem, and negatively with depression, anxiety, insomnia,
and somatic symptoms in battered women.

Overall, studies suggest that social support plays an important role in the
relationship between DV and mental health symptoms, including those symptoms
characteristic of CP. Speciﬁcally, studies support a link between DV and social support,
and a link between social support and mental health outcomes in battered women.
Moreover, one study ﬁnds that poor social support accounts for increased psychological
distress and PTSD symptoms in response to DV (Thompson et al., 2000). Taken
together, these ﬁndings are consistent with a mediation model. However, only one prior
study has tested for mediation and no studies have examined social support in relation to
CF speciﬁcally. The current study examines whether both structural and ﬁrnctional
aspects of social support mediate the relationship between longitudinal patterns of DV
victimization and CP symptoms.

In addition to investigating the process of social isolation and the role of social

support in the proposed relationship between DV and CP symptoms, the current study

39

examines two potential factors that may moderate or affect the strength of the
hypothesized DV-CP symptom link: adult representations of childhood attachment and
child maltreatment history. These factors are proposed to increase personal risk for CP
symptoms in response to DV victimization.
Adult Representations of Childhood Attachment

One factor believed to qualify the impact of partner abuse on CP symptoms is
insecure attachment based on adult recollections of childhood care-giving experiences.
The study of adult representations of childhood attachment involves the assessment of
internal working models, which can be classiﬁed as secure or insecure, parallel to infant
attachment organization (Main & Goldwyn, 1994). Adult attachment classiﬁcations are
typically identiﬁed based on the quality and quantity of discourse while describing and
evaluating childhood attachment-related experiences (Main & Goldwyn, 1994). Adult
attachment classiﬁcations represent an index of self-representations and strategies for
processing emotions and thoughts related to childhood attachment relationships.
Individuals with secure adult representations of childhood attachment emphasize the
importance of early relationships and describe relationships with caregivers in a
thoughtful, coherent, consistent, and succinct manner. In contrast, individuals with
insecure adult representations of childhood attachment provide descriptions of
relationships with caregivers that are incoherent, inconsistent, conﬁised, or conﬂicted.
Such individuals minimize the inﬂuence of childhood attachment experiences by
idealizing, normalizing, or devaluing childhood attachment relations, or provide
confusing, lengthy descriptions of care-giving experiences, which indicate an angry,

passive preoccupation with childhood attachment ﬁgures (Main & Goldwyn, 1994).

40

Although researchers are beginning to consider adult representations of childhood
attachment in studies of childhood experiences of trauma and psychopathology (e.g., J.
G. Allen et al., 1998; see Cassidy & Mohr, 2001 for review), researchers have not
examined the role of adult representations of childhood attachment in the relationship
between adult experiences of trauma, such as DV, and psychopathology. However, a
growing body of theoretically-based empirical literature links insecure adult
representations of childhood attachment in normal and patient samples of adolescents and
adults to a wide range of psychological symptoms reﬂected in the CP construct,
suggesting that insecure adult attachment in relation to early care-giving experiences is a
risk factor for poor mental health.

For example, Pianta, Egeland, and Adam (1996) found a relation between
attachment insecurity and overall symptoms of psychopathology among a high-risk,
multi-problem sample of pregnant women, the large majority of whom had histories of
DV and child maltreatment. Speciﬁcally, in comparison to women classiﬁed as secure,
women classiﬁed as insecure scored signiﬁcantly higher on the Psychopathic Deviation,
Paranoia, and Schizophrenia scales of the Minnesota Multiphasic Personality Inventory —
2 (Butcher, Dahlstrom, Graham, Tellegen, & Kraemmer, 1989) endorsing symptoms of
impulsivity, insensitivity to others and social norms, suspiciousness, hostility,
persecution, isolation, self-preoccupation, and inferiority. Furthermore, self-reported
symptom levels exceeded the clinical cutoff, consistent with mental illness (Pianta et al.,
1996). Insecure attachment was also linked to clinical status in a study of adolescent

psychiatric patients; Adam et al. (1996) found a low prevalence of secure attachment

41

representations in the psychiatric sample as a whole (16%) in comparison to normative
data on adolescents (56%: van Ijzendoom & Bakermans-Kranenburg, 1996).

Studies assessing speciﬁc psychopathology outcomes indicate that adult
representations of childhood attachment are related to affective disorders, self-harm,
impulsive behaviors, and alterations in relational functioning—symptoms consistent with
CP (Adam et al., 1996; J. P. Allen, Hauser, & Borman Spurrell, 1996; Cole-Detke &
Kobak, 1996; Fonagy et al., 1996; Rosenstein & Horowitz, 1996; van Ijzendoom &
Bakermans-Kranenburg, 1996). Two studies provide evidence for a relationship between
insecure adult representations of attachment and affective disorders, including major
depressive, dysthymic, schizoaffective, and eating disorders (Cole-Detke & Kobak, 1996;
Rosenstein & Horowitz, 1996). For example, Cole-Detke and Kobak (1996) found that
76% of college women reporting elevated symptom levels of eating pathology and
depression were classiﬁed as insecure, whereas 62% of women reporting no disorder
were secure.

Evidence also suggests that insecure adult representations of childhood
attachment are related to self-harm and impulsive, risky behavior. For example, Adam,
Sheldon-Keller, and West (1996) conducted a case-comparison study of 133 adolescents
in psychiatric treatment, about half of whom had a history of severe suicidal ideation or
behavior, and found that insecure attachment organization increased the probability of
membership in the case group differentiating adolescents with a history of suicidal
behaviors. Two additional studies of psychiatric patients linked insecure adult
representations of childhood attachment to substance abuse disorder (J. P. Allen et al.,

1996; Rosenstein & Horowitz, 1996).

42

In addition to Axis I psychopathology, evidence suggests a relationship between
insecure adult representations of childhood attachment and relational disturbances
reﬂected in Axis II psychopathology. Two studies found signiﬁcant associations between
adult representations of childhood attachment and borderline personality disorder
(Fonagy et al., 1996; Rosenstein & Horowitz, 1996). For example, Fonagy et al. (1996)
reported that adult psychiatric inpatients diagnosed with borderline personality disorder
were signiﬁcantly more likely to have insecure adult representations of childhood
attachment (e.g., 75% of participants with borderline personality disorder were classiﬁed
as insecure) as compared to matched controls. Furthermore, Allen et al. (1998) suggested
a relationship between insecure adult representations of childhood attachment and other
Axis II features, including paranoid and schizotypal behavior, based on the personality
assessment of psychiatric inpatient females with histories of child abuse.

Finally, van IJzendoom and Bakermans-Kranenburg (1996) provided a summary
of the literature on the relation between attachment processes and psychiatric disorders in
adolescents and adults by conducting a meta-analysis comparing the distributions of
current representations of childhood attachment between participants in 14 clinical
samples and 13 normative samples. The researchers found a deviant distribution of
attachment classiﬁcations among the mentally ill, such that samples of adolescents and
adults with oppositional, depressive, and borderline psychopathology were more likely to
have insecure representations than the normative group; the large majority of the
mentally ill were classiﬁed as insecure.

Although no studies have investigated the relation between adult representations

of childhood attachment and post-trauma psychopathology in the context of DV or other

43

interpersonal traumas, theoretical arguments and empirical ﬁndings suggest that disrupted
attachment is a pathway to diverse forms of psychopathology, including a range of
affective, relational, and behavioral disturbances captured by CP. Therefore, an insecure
attachment strategy may be a risk factor for sensitivity to later risk conditions and the
emergence of psychopathology, including CP symptoms. The current study tests the
potential of attachment insecurity as a predisposing individual-level factor for CP
symptoms in relation to DV.
Child Maltreatment History

Although previous studies have not examined child maltreatment as a potential
vulnerability factor for DV-related CP symptoms, several studies have linked child
maltreatment to CP. Speciﬁcally, two groups of researchers found a high prevalence rate
of CP diagnoses among child maltreatment victims utilizing the full CP structured criteria
set (McLean & Gallop, 2003; Pelcovitz et al., 1997), one study found a relationship
between child maltreatment and elevated CP symptom levels (Ford et al., 2006), and
three studies noted complex patterns of symptomatic and personality problems consistent
with the CP syndrome among victims of childhood maltreatment (J. G. Allen et al., 1998;
Dickinson et al., 1998; van der Kolk et al., 1996: see section on Evidence for a Complex
PTSD Syndrome for details of these studies). Although ﬁndings from these studies
suggest a connection between early experiences of abuse and CP, the role of child
maltreatment in the relationship between DV and CP has not previously been explored.
The current study examines child maltreatment as a moderator of the hypothesized

relationship between DV and CP symptoms.

44

Hypotheses and Rationale

Several mental health professionals have addressed post-trauma psychopathology
that exceeds the avoidance, reexperiencing, and arousal symptoms constituting the
diagnostic construct of PTSD. Most notable are Herman (1992a; 1992b), who was the
ﬁrst to deﬁne a CP syndrome, and Pelcovitz and colleagues (1997), who developed a
diagnostic criterion set to assess CP symptoms. These authors argue that the current
diagnostic formulation of PTSD fails to capture the complex psychological consequences
of experiencing prolonged, repeated, interpersonal trauma, which occurs when the victim
is captive, under control of the perpetrator. Instead, Herman (1992a; 1992b) proposed the
construct CP to capture the profound impact of repeated, prolonged interpersonal
traumatic experiences on emotion- and behavior-regulation, self-concept, and cognitive
and interpersonal functioning. Research has validated the CP construct among victims of
interpersonal abuse; however, the syndrome has yet to be explored among battered
women speciﬁcally. The current study proposes a theoretical model for the development
of CP symptoms in women exposed to DV based on trauma and attachment theories. The
model includes a direct link from DV to CP symptoms, as well as an indirect link through
social support. In addition, this model suggests that abuse characteristics and DV
partnership may differentially inﬂuence CP symptom development. Finally, the model
includes two moderator variables, attachment insecurity and experience of child
maltreatment, which are expected to increase personal vulnerability to CP symptoms in
response to DV. This model is tested by identifying longitudinal patterns of DV

victimization based on several important abuse characteristics, and examining these

45

patterns in relation to CF symptoms among a community sample of female DV victims.
This study has several theoretically-driven hypotheses:
Hypotheses I a-I d

Women with longitudinal patterns of DV victimization based on a variety of
abuse characteristics will suffer from CP symptoms. Speciﬁcally, women classiﬁed into
groups with high levels of DV frequency will report more CP symptoms than women
classiﬁed into groups with lower levels of DV frequency (1a). Women classiﬁed into
groups with high levels of DV severity will report more CP symptoms than women
classiﬁed into groups with lower levels of DV severity (1b). Women classiﬁed into
groups with high levels of sexual DV will report more CP symptoms than women
classiﬁed into groups based on threats of DV or physical DV (1c). Finally, women with
longer duration of DV will report more symptoms of CP than women with shorter
duration of DV (1d). Herman’s (1992a; 1992b) formulation of CP suggests that
interpersonal abuse victims, including battered women are at risk for the development of
CP, characterized by co-occurring affective, somatic, identity, relational, and behavioral
disturbances (see Theoretical Model section for details about the development of CP
symptoms in response to DV).

Although CP has not been examined in battered women speciﬁcally, two lines of
clinical and empirical inquiry support the general theoretical link between DV and CP.
First, research examining the relation between interpersonal victimization and CP
provides evidence for a CP syndrome among victims of interpersonal abuse in general
(e.g., Roth et al., 1997). Second, the DV literature indicates that abused women are at

risk for co-morbid psychiatric disorders (e. g., Nixon etal., 2004), personality problems

46

(e.g., Walker, 1984), and experiences of repeated harm (Pico-Alfonso, 2005)—primary
areas of disturbance in the CP syndrome.

Although trauma theories and the empirical literature suggest that battered women
are at risk for CP, battered women likely have varying experiences of abuse. Particular
patterns of DV victimization due to theoretically important abuse characteristics may
differentially inﬂuence the development of CP symptoms. (1a). For example, severity of
abuse may be relevant in the development of CP symptoms. According to trauma theory,
a relationship of coercive control, which ultimately breaks down the psychological
resistance of the victim resulting in CP symptoms, is more likely to occur in cases of
severe abuse where the victim has few options to escape perpetrator control due to the
risk of serious danger than in cases of less severe abuse (Herman, 1992a, 1992b).
Consistent with theoretical predictions, Dickinson et al. (1998) found that severity of
sexual abuse was related to greater overall symptomatology reﬂected in the CP
syndrome, including depression, somatization, PTSD, and dissociation. In addition, the
DV literature suggests that DV severity is related to post-trauma psychopathology,
speciﬁcally PTSD prevalence, in abused women (Houskamp & F oy, 1991).

(lb). Similarly, frequency of DV may be an inﬂuential DV characteristic on the
development of CP symptoms. Subordination to coercive control leading to CP
symptoms is more likely to occur in cases of frequent abuse than in cases of infrequent
abuse because deﬁance against the perpetrator usually results in abuse. Thus, self-
protection requires submission. In addition, frequent abuse increases the use of
dissociation to escape emotional and physical pain, as well as arousal levels resulting in

somatic symptoms. Women who experience frequent DV may feel helpless because any

47

initiative of self will almost indeﬁnitely result in abuse. Feelings of futility may result in
a state of learned helplessness, leaving the women vulnerable to future harm (Crittenden,
1997). Empirical evidence suggests that frequency of DV is an important contributor to
CP. For example, one study found that frequency of sexual abuse was a signiﬁcant
predictor of CP (McLean & Gallop, 2003). In addition, studies suggest that frequency of
DV increases the risk of depression and low self-esteem (Cascardi & O'Leary, 1992)—
both symptoms of CP.

(1c). Type of DV is another abuse characteristic relevant to the development of
CP. According to Herman (1992a), CP is a syndrome experienced by survivors of
prolonged, repeated interpersonal trauma, regardless of the speciﬁc nature of the
interpersonal trauma. However, sexual abuse may place women at relatively greater risk
for the development of CP because sexual abuse represents the most intimate form of
betrayal and an assault on reproductive control, thereby undermining a primary function
of human evolution (Crittenden, 1997). Consistent with this idea, Roth et al. (1997)
found that sexual abuse posed a greater risk for CP than physical abuse, suggesting that
CP may demonstrate greater speciﬁcity for sexual abuse.

(1d). Finally, duration of DV may be another characteristic of abuse, which
inﬂuences the development of CP symptoms. Herman’s (1992a; 1992b) CP construct
was developed to capture the consequences of “prolonged” traumatic experiences. A
perpetrator establishes a relationship of coercive control, which ultimately leads to CP
symptoms, over time with repeated use of coercive techniques. Consistent with this
theoretical argument, Dickenson et al. (1998) found that duration of sexual abuse was

related to multiple symptoms included in the CP syndrome. F urtherrnore, Bogat et al.

48

(2003) reported that chronic experiences of DV were related to worse psychosocial
outcomes captured by CP than less chronic abuse experiences.
Hypothesis 2

Women in long-term DV relationships (deﬁned here as three or more consecutive
years in the same relationship) will be more vulnerable to CF symptoms related to
longitudinal patterns of DV victimization as compared to women in one or more short-
term DV relationship(s). Herman’s (1992a; 1992b) formulation of CP implies that the
development of a relationship of coercive control, which ultimately leads to symptoms of
CP, occurs over time with a single perpetrator. Thus, involvement in one or multiple
short-term DV relationships where subordination to coercive control never begins is less
likely to result in CF symptoms. Although empirical studies suggest that duration of
interpersonal abuse is related to CF symptoms (J. G. Allen et al., 1998; Dickinson et al.,
1998), previous studies have not addressed whether or not the violence occurs in the
context of one abusive relationship or multiple violent relationships. The current study is
the ﬁrst to investigate whether duration of DV with a single perpetrator is an important
factor in the development of CP symptoms.
Hypothesis 3a

Women who experience DV will lose social support over time. In addition,
women classiﬁed into groups with high levels of DV frequency, DV severity, threats of
DV, physical DV or sexual DV will report lower levels of emotional support and fewer
network supporters across all time periods than women classiﬁed into groups with lower
levels of DV frequency, DV severity, threats of DV, physical DV, or sexual DV. Women

with longer duration of DV will also report lower levels of emotional support and fewer

49

network supporters over time than women with shorter duration of abuse. According to
Herman’s (1992a; 1992b) CP theory, the perpetrator uses DV to socially isolate the
victim. Previous studies have noted a relationship between frequency and severity of DV
and impaired social support (e. g., Levendosky et al., 2004; Mitchell & Hodson, 1983,
1986); however, longitudinal trajectories of social support have not been studied in
relation to longitudinal patterns of DV victimization. Examination of social support over
time will be an indicator of whether or not it is possible that DV perpetrators use social
isolation as a method of achieving control over battered women as suggested by the
proposed theoretical model.
Hypothesis 3b

Social support will mediate the relationship between longitudinal patterns of DV
victimization and CP symptoms, such that small network sizes and low levels of
emotional support will partially account for CP symptoms in victims of DV.
Theoretically, poor interpersonal functioning (a primary characteristic of CP) is the
outcome of traumatic bonding between perpetrator and victim, a process which is enabled
when the perpetrator isolates the victim, such that the victim must depend on the
perpetrator for support and other needs (Herman, 1992a, 1992b). In addition, limited
external support resources make it difﬁcult to leave an abusive relationship, thereby
subjecting the victim to continued abuse and its mental health consequences. Thus, the
perpetrator of DV socially isolates his victim, which in turn, leads to the development of
CP symptoms. Although this is the ﬁrst study to examine the role of poor social support
in the development of CP symptoms among battered women, one study found that social

support mediated the relationship between DV and psychological distress and PTSD

50

symptoms (Thompson et al., 2000). Furthermore, several studies have linked DV to poor
social support (e. g., Levendosky et al., 2004), and poor social support in abused women
to mental health problems captured by the CP construct (e. g., Meadows et al., 2005).
Hypothesis 4

Adult representations of childhood attachment will moderate the relationship
between longitudinal patterns of DV victimization and CP. Speciﬁcally, insecure adult
representations of childhood attachment will be a risk factor for the relationship between
DV and CP symptoms. Attachment theory suggests the importance of early experiences
with the primary caregiver in critical developmental processes, such as the development
of a sense of self, internalized mental representations of relationships, and capacity to
regulate affective states (Bowlby, 1977). A damaged attachment system disrupts these
important processes leading to unhealthy interpersonal functioning and the utilization of
basic defenses and maladaptive coping strategies to regulate affective states (van-der-
Kolk & F isler, 1994). Thus, disrupted attachment to the primary caregiver has long-term
negative consequences for social, emotional, and behavioral functioning, which may
affect an individual’s response to traumatic events, such as DV. Although this is the ﬁrst
study to examine the role of attachment in the development of CP among battered
women, prior literature suggests strong connections between a damaged attachment
system and several psychiatric problems constituting the CP construct (e.g., Cole-Detke
& Kobak, 1996).
Hypothesis 5

Child maltreatment history will moderate the relationship between longitudinal

patterns of DV victimization and CP symptoms, such that women with histories of

51

frequent and severe child maltreatment will be more vulnerable to DV—related CP
symptoms as compared to women with less frequent and severe experiences of childhood
maltreatment. Attachment theory suggests that disrupted attachment to the primary
caregiver, as often occurs in situations of child maltreatment (D. Barnett et al., 1999; E.
A. Carlson, 1998), derails the crucial developmental process of learning to internally
modulate physiological arousal and affective states (Bowlby, 1977). Thus, experiences
of child maltreatment may disrupt the attachment system, which in turn, negatively
affects capacity to regulate emotions. Poor emotional regulation may leave individuals
psychologically vulnerable to both current and future traumatic experiences, as the use of
basic defensive and maladaptive coping strategies become automatic over time and more
adaptive strategies are not learned. Thus, early experiences of abuse may make an
individual more sensitive to the psychological effects of later experiences of DV,
increasing the risk of DV-related CP symptoms. Empirical research suggests a link
between child maltreatment and CP (e.g., J. G. Allen et al., 1998). However, this is the
ﬁrst study to examine how child maltreatment may inﬂuence psychological response to

adult trauma, speciﬁcally DV, in terms of CP symptoms.

52

 

 

Pa
inesuga
olD\'an
3000'). l
througho
Soenn
mxnm
particip.
death (7
tenant
63.800

Amen
Lhtav
Single
nido,
0r eq]

dcgre

Methods
Participants

Participants included a subsample of the 206 women recruited for a larger
investigation, the Mother-Infant Study—a prospective, longitudinal study of the impact
of DV and other risk factors on women and children’s functioning (Levendosky et al.,
2000). The subsample consisted of women who reported experiencing DV at any time
throughout the course of the larger study as deﬁned by endorsement on any item on the
Severity of Violence Against Women Scales (Marshall, 1992: see Measures section for
description of scale). Analyses indicated that the total DV group consisted of 181
participants. However, 17 of these women were excluded from the current study due to
death (n=2) or failure to participate in the ﬁnal three waves of data collection (n=15),
resulting in a ﬁnal sample of 164 women. The 164 women identiﬁed themselves as
62.8% Caucasian, 25.0% African American, 5.5% Latina, 4.9% multi-racial, 1.2% Native
American, and .6% Asian American. Based on demographics collected at study entry,
the average age of women was 25.0 years (SD=4.9); 55.5% of the women were
single/never married, 34.1% were married, and 10.4% were separated, divorced, or
widowed; and 17.1% did not complete high school, 29.3% earned a high school diploma
or equivalent, 42.7% had some college or trade school, and 11.0% earned a bachelor’s

degree or higher. Finally, the median monthly income was $1,300 (range: $0-$7,000).

Procedures
Recruitment

Participants for the larger study were recruited from three counties in mid-

Michigan (representing urban, rural, and suburban areas) through contact with ﬁfty-one

53

sites, including obstetric/ gynecological clinics or other women’s health clinics (39%),
social service programs (26%), childbirth classes (5%), legal agencies (2%), and DV
shelters (1%), as well as by posting ﬂyers in public areas, such as libraries, laundromats,
and grocery stores (27%). Two ﬂyers were distributed: one invited pregnant women to
participate in a study about mother-infant relationships, and the other targeted women
who had DV experiences (i.e. being pushed, slapped, kicked, punched, or worse) during
pregnancy.

Sampling

To participate in the study, women were required to be between eighteen and
forty years of age, in their third trimester of pregnancy, and involved in a romantic
relationship that lasted for at least six weeks sometime during pregnancy. Women with a
limited knowledge of the English language were excluded from the study if it was
believed that they would have difﬁculty understanding assessment protocol.

The sampling procedure for the larger study involved a ﬁve-minute telephone
interview to screen for eligibility based on demographic information and the presence of
a romantic relationship during pregnancy. After approximately half of the sample had
been enrolled in the study (n=96), additional screening procedures were instituted to
ensure that a reasonable number of subjects experienced DV during pregnancy. During
the telephone screen, the Conﬂict Tactics Scale (Straus, 1979) was administered in order
to track the number of abused and non-abused women. For the purpose of recruitment,
women were categorized as abused if they experienced physical violence during
pregnancy according to the CTS. This screen was used to exclude women who had not

experienced DV during pregnancy once it was determined that enough non-abused

54

women were already enrolled. After 137 women had been recruited and interviewed, it
was discovered that many of the “non-abused” 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 DV during their current
pregnancy. Overall, 161 women were excluded because they did not meet age, language,
relationship status, or battering experience criteria, or due to interview scheduling
difﬁculties. There were no signiﬁcant demographic differences between excluded
women and participants on age, race, education, marital status, or abuse status.

Data Collection

Data for the Mother-Infant study was collected at six waves, beginning in the
women’s third trimester of pregnancy (T1) and continuing every year thereafter through
the child’s ﬁfth birthday (T2-T6). For the current study, data for each variable from all
available waves was utilized: DV and social support data from T1-T6, attachment data
from T1, and childhood maltreatment and CP data from T6.

Procedures for data collection were similar at each time period to ensure
standardization. Every twelve months, women were contacted by phone, letter, or
electronic mail to schedule data collection appointments. Assessment protocol was
administered in an interview format by graduate and undergraduate research assistants.
Interviews were conducted in a private space at participant homes or at university project
ofﬁces. At each appointment, research assistants explained interview procedures and
informed participants about anonymity and conﬁdentiality policies. Participants
completed an informed consent form that speciﬁed voluntary participation with the

option of withdrawing from the study at any time without negative consequences.

55

Interviewers administered questionnaires aloud and recorded participant responses in
order to control for varying levels of literacy among participants. Research assistants
were blind to the abuse status of the women, which was guaranteed by administering
questionnaires referring to DV near the end of the protocol. Interviews lasted
approximately one to four hours to complete depending on the wave of data collection.
Women were paid cash to thank them for their participation and to reimburse them for
their time. Payment amounts varied depending upon the length of the interview. All
women were provided with a list of community resources at the end of each interview.

Conﬁdentiality was maintained by assigning all participants an identiﬁcation
number, which was placed on all data rather than participant names. Participant names
and corresponding identiﬁcation numbers were kept in a separate tracking ﬁle apart from
all data. The tracking ﬁle and data were kept in locked project ofﬁces, in locked ﬁling
cabinets, and on password-protected computers in password-protected ﬁles.
Training

Undergraduate and graduate research assistants were trained to strictly adhere to
assessment protocol, conﬁdentiality, and tracking procedures. Duration of the training
was three to seven months depending on the nature of each interview. Training consisted
of reviewing questionnaires and interview procedures, role-playing, reading and
discussing relevant research articles, conducting mock interviews, and observing real
interviews completed by experienced interviewers. In addition, interviewers learned
skills to be empathic, non-judgmental, and to troubleshoot in difﬁcult situations (e. g., the
intrusion of a partner who is unhappy about a woman’s participation in the study).

Finally, interviewers conducted two to ﬁve interviews supervised by senior graduate

56

student research assistants to ensure consistent and standard administration of measures.
Supervisors reviewed completed interviews to ensure accurate data collection. Research
assistants attended weekly training meetings throughout the data collection process to
discuss concerns or problems, and to receive qualitative feedback from experienced
interviewers. This meeting also served as a debrieﬁng period for research assistants who
experienced difﬁcult interviews.
Sample Retention

In order to limit attrition and preserve a high retention rate, participants were
contacted every ninety days. Participants were mailed a letter and a blank contact
information sheet to be completed and returned in a postage-paid envelope. Letters were
mailed in envelopes with “address service requested” such that the post ofﬁce would send
an updated address for those participants who had relocated and registered a new address.
If participants did not return the information sheet or if the post ofﬁce did not send
updated address information within three weeks, research assistants attempted to contact
the participant via alternative methods. First, participants were contacted directly by
telephone or electronic mail. If women could not be reached directly (due to a
disconnected phone, for example), efforts were made to contact them through three
recontact people given to us by each participant at the T1 interview. If all other tracking
methods failed, research assistants would go to participants’ and recontact people’s

homes. Women were paid ten dollars for their participation in each tracking contact.

57

Measures

Demographics

A brief questionnaire was administered to obtain basic demographic information
from participants, including age, ethnicity, romantic relationship history,
marital/relationship status, education level, and family income (see Appendix A).
Trauma Exposure

Severity of Violence Against Women Scales (S VA WS.‘ Marshall, 1992). The
SVAWS is a 46-item instrument designed to assess DV, deﬁned as adult male violence
and threats of violence against female romantic partners (see Appendix B). The
questionnaire is composed of nine dimensions: symbolic violence; mild, moderate, and
serious threats of physical violence; minor, mild, moderate, and serious actual physical
violence; and sexual violence. Example items include, “Hit or kicked a wall, door, or
furniture,” “Threatened to hurt you,” “Pushed or shoved you,” “Beat you up,” and
“Physically forced you to have sex.” Women reported on their experiences of DV 1)
with their most recent previous partner at the time of the pregnancy interview, 2) with
their current partner during the year prior to pregnancy, 3) with their current partner
during pregnancy, and 4) with all romantic partners during the ﬁve years following the
birth of their child. Ratings are on a 4-point scale with response choices ranging from
“Never” to “Many times.”

To capture various abuse characteristics, each participant was assigned DV
frequency, DV severity, threats of DV, physical DV, and sexual DV scores corresponding
to each wave of data collection (Tl-T6). DV scores were summed across partners at each

wave when a participant endorsed DV with more than one romantic partner during the

58

wave. DV frequency was calculated by summing frequency scores for individual items
(possible scores range from 0-138 for any one partner). DV severity was calculated by
multiplying frequency scores for individual items by physical impact weights provided
by the author, and summing the weighted frequency scores for individual items (possible
scores range from 0-96). Threats of aggression was calculated by summing frequency
scores for individual items pertaining to threats (possible scores range from 0-5 7).
Physical DV was calculated by summing frequency scores for individual items pertaining
to physical abuse (possible scores range from 0-63). Sexual DV was calculated by
summing frequency scores for individual items pertaining to sexual abuse (possible
scores range from 0-18). In addition, each participant was assigned an overall DV
duration score ranging from 1 to 8 corresponding to the number of data collection waves
at which DV was endorsed (with the caveat that T1 could represent 3 observation points:
current partner during pregnancy, current partner the year prior to pregnancy, and most
recent previous partner).

The SVAWS has good psychometric properties. Marshall (1992) reported
coefﬁcient alphas for the nine subscales ranging from .89 for symbolic violence to .96 for
both mild and serious physical violence. In the current study, coefﬁcient alphas for the
entire scale ranged from a low of .92 for T7 partner 1 to a high of .99 for T3 partner 2.
Dutton, Landolt, Starzomski, and Bodnarchuk (2001) found signiﬁcant correlations
between women’s reports of abuse on the SVAWS and their partners’ reports of abuse
according to the Propensity for Abusiveness Scale, providing evidence for the validity of

the SVAWS.

59

Childhood Trauma Questionnaire —— Short Form (CTQ—SF: Bernstein et al.,
2003). The CTQ-SF is a 28-item, self-report measure for retrospectively assessing
multiple types of abuse and neglect experienced during childhood and adolescence (see
Appendix C). The questionnaire has ﬁve clinical scales each consisting of ﬁve items:
physical, sexual, and emotional abuse, and physical and emotional neglect. The
instnunent also includes a 3-item minimization/denial scale to detect under-reporting of
maltreatment. Only the 25 items on the clinical scales were used for purposes of the
current study. Example questions include, “People in my family hit me so hard it left me
with bruises or marks,” “Someone tried to touch me in a sexual way,” and “People in my
family called me things like ‘stupid,’ ‘lazy,’ or ‘ugly.”’ Items were rated on a 5-point
Likert scale with response options ranging from “Never” to “Very Often True.” A
composite child maltreatment score was calculated for each woman by summing
frequency scores for individual items on all clinical scales. Possible total scores range
from 25 to 125 with higher scores reﬂecting greater frequency and severity of abuse.

The CTQ is a reliable and valid measure of childhood trauma across a variety of
individuals who differ in terms of age, sex, ethnicity, socioeconomic status,
psychopathology, and life experiences. Test-retest reliabilities range from .79 for the
physical neglect scale to .86 for the entire measure in a sample of adult substance abusers
(Bernstein & Fink, 1998). Bernstein and Fink (1998) reported internal consistency
reliability coefﬁcients ranging from a median of .66 for physical neglect to a median of
.92 for sexual abuse across a variety of clinical and non-clinical samples. Coefﬁcient
alpha for the entire CTQ was reported to be .91 in a large community sample (Scher,

Stein, Asmundson, McCreary, & Forde, 2001) and was .95 in the current study.

60

Conﬁrmatory factor analyses support the invariance of the 5-factor solution across six
diverse groups of people with differing maltreatment histories, including adult substance
abusers from a northeast metropolitan city and southwest rural area, adolescent
psychiatric inpatients, female HMO members from the northwest, and two normative
community samples (Bernstein & Fink, 1998; Bernstein etal., 2003; Scher et al., 2001).
The CTQ represents an improvement in content validity due to its broader content
coverage reﬂecting multiple domains of abuse described in the literature (Bernstein &
Fink, 1998). Finally, the CTQ demonstrated convergent validity with a clinician-rated,
semi-structured interview of child abuse (Bernstein & Fink, 1998) and therapists’
independent ratings of abuse (Bernstein & Fink, 1998; Bernstein et al., 2003).
Complex PT SD

Self-Report Inventory for Disorders of Extreme Stress (SIDES - SR: van der Kolk,
2002b). The SIDES-SR is a 45-item, self-report inventory designed to assess CP or
DESNOS (see Appendix D). The measure consists of 6 major scales: (1) alteration in
regulation of affect and impulses, (2) alterations in attention or consciousness, (3)
alterations in self-perception, (4) alterations in relations with others, (5) somatization, and
(6) alterations in systems of meaning. Example items include, “When I feel upset, I have
trouble ﬁnding ways to calm myself down,” “I am active sexually in ways that I know
put me in danger,” “I sometimes feel so unreal that it is as if I am living in a dream, or
not really there, or behind a glass wall,” and “I suffer from chronic pain in my back, yet
doctors could not ﬁnd a clear cause for it.” Women rated lifetime presence or absence of
DV-related CP symptoms. Thus, in order to isolate the relationship between DV and CP

symptoms, the effects of other interpersonal traumas (e. g., child abuse) on CP were

61

methodologically controlled by instructing women to report CP symptoms speciﬁcally
related to experiences of DV. For the current study, participants received a continuous
score reﬂecting the number of CP symptoms endorsed.

The SIDES is a useful tool for the investigation of psychological response to
extreme stress. The measure has good behavioral anchors and internal consistency (van
der Kolk, 2003). van der Kolk (2003) reported coefﬁcient alphas ranging from .68 to .82
for the subscales with a full-scale alpha of .93 among study participants with a diverse
history of trauma. In the current study, coefﬁcient alpha for the entire scale was .95. van
der Kolk (2003) also indicated that the SIDES self-report and interview versions
demonstrated convergent validity: Pearson r correlations were signiﬁcant and ranged
from .60 -.78 for the subscales with a full-scale correlation coefﬁcient of .86. These
correlations represent 74% shared variability between the two instruments (van der Kolk,
2003). Findings from a second study further support the convergent and discriminant
validity of the SIDES instruments: the interview and self-report versions of the SIDES
correlated more strongly with each other (r=.85) than with self-report and clinician-rated
measures of PTSD symptoms (correlation range: 51-62). In addition, the construct
validity of the SIDES has been supported in several studies (Blaustein, Spinazzola,
Simpson, & van der Kolk, 2000; Spinazzola et al., 1994, November; van der Kolk et al.,
1996). For example, in one study individuals showing impairment on the SIDES as
compared to those with PTSD alone exhibited more comorbid Axis I and Axis II
psychopathology, as well as a speciﬁc pattern of DSM-IV diagnoses reﬂective of shared

underlying deﬁcits with the DESNOS construct (Blaustein et al., 2000).

62

Social Support

Norbeck Social Support Scale WSSQ: Norbeck, Lindsey, & Carrieri, 1981). The
NSSQ is designed to measure multiple dimensions of social support, including emotional
support, practical aid, and network structure (i.e., size of support network, duration of the
relationships, frequency of contact, and recent loss of important relationships). Example
items include, “How much does this person make you feel liked or loved,” and “If you
needed to borrow $10, a ride to the doctor, or some other immediate help, how much
could this person usually help?” Participants listed each signiﬁcant person in their lives,
and rated the amount of social support each person provided on a 5-point scale ranging
from “Not at all” to “A great deal.” For purposes of the current study, women received
two social support scores for each available wave of data collection: 1) number of
network supporters, and 2) mean level of emotional support (determined by summing
participant ratings across network members for the corresponding questions and dividing
by the total number of supporters). The possible number of network supporters ranges
from 0-24. Possible scores for emotional support range from 0-4 with higher scores
indicating greater emotional support. See Appendix E for questions assessing emotional
support.

The NSSQ provides a reliable assessment of social support. Norbeck and
colleagues (1981) reported l-week test-retest reliability correlations ranging from .85 to
.92 across measured dimensions of social support. Stability over a 7-month interval is
high, with correlations ranging from .58 to.78 (Norbeck, Lindsey, & Carrieri, 1983). The
instrument also appears to be internally consistent; correlations between pairs of related

items range from .89 to .97 (Norbeck et al., 1981). Validity of the NSSQ has also been

63

established (Norbeck et al., 1981, 1983). The measure was reportedly relatively free
from social desirability response bias among a group of nursing graduate students
(N orbeck et al., 1981). In addition, moderate correlations were found between the NSSQ
and two other instruments designed to measure social support (Norbeck et al., 1981,
1983). Findings of positive correlations between interpersonal needs for inclusion and
affection and amount of available social support provided evidence for the construct
validity of the NSSQ (Norbeck et al., 1983). Finally, Norbeck et al. (1983) found
conﬁrmation for the predictive validity of the NSSQ, reﬂecting the stress-buffering role
of social support.
Adult Representations of Childhood Attachment

Perceptions of Adult Attachment Questionnaire (PAA Q, previously referred to as
the Inventory of Adult Attachment: Lichtenstein & Cassidy, 1991). The PAAQ is a 60-
item questionnaire for measuring representations of early attachment experiences (see
Appendix F). The instrument assesses 3 broad areas: childhood relationship with the
primary caregiver (e. g., “My mother was not very affectionate”), current relationship
with the primary caregiver (e.g., “No one gets under my skin like my mother”), and
accessibility of childhood memories (e. g., It’s hard for me to remember my early
relationship with my mother in any detail). Items form 8 subscales, several of which
were designed to be equivalent to scales on the Adult Attachment Interview (AAI: Main
& Goldwyn, 1994): 1) Rejection by mother, 2) Loved by mother, 3) Role Reversal with
mother, 4) Anger towards mother, 5) Derogation of attachment experiences, 6) Forgiving
of childhood problems, 7) feelings of Vulnerability about the relationship with mother,

and 8) No Memory of childhood attachment experiences. Items are rated on a 5-point

64

 

all.

sul

{It

scale with response options ranging from “Strongly Disagree” to “Strongly Agree.”
Higher scores on each subscale indicate higher levels of the measured construct (e. g.,
more rejection, more love, etc.).

Based on an earlier conﬁrmatory factor analysis of the PAAQ performed as a part
of the larger Mother-Infant Study, six subscales were used here to measure adult
representations of childhood attachment: Rejection, Love, Anger, Derogation,
Vulnerability, and No Memory. The previous factor analysis revealed that these six
subscales loaded onto a general attachment factor reﬂecting attachment security (Huth-
Bocks, Levendosky, Bogat, & von Eye, 2004). For the current study, participant
attachment representations were determined by summing item totals on these six
subscales (after reverse-scoring necessary subscales, such that higher scores reﬂected
greater security), and then dichotomizing composite scores, such that the lower third of
cases represented “insecure” and the upper two-thirds of cases represented “secure” adult
representations of childhood attachment.

Lichtenstein and Cassidy (1991) established the concurrent validity of the PAAQ
by comparing PAAQ responses to AAI responses in a sample of mothers. Analogous
scales were highly correlated with each other, with the exception of the Role Reversal
and Derogation subscales (Lichtenstein & Cassidy, 1991). The PAAQ is also a reliable
instrument. Lichtenstein and Cassidy (1991) reported test-retest reliabilities ranging from
.64 (Forgiving) to .86 (Loved) for a sample of college students. Alpha coefﬁcients
ranged from a .62 (Derogation) to .90 (No Memory) in the same sample, and from .51

(Derogation) to .94 (N 0 Memory) in a sample of mothers. In the current study,

65

coefﬁcient alpha for the composite scale was .94. These ﬁndings suggest that the PAAQ

has good psychometric properties.

66

Results
Missing Data

There were two types of missing data due to attrition and data collection errors:
data missing at the item level and data missing at the variable level. Data missing for
single item variables (e.g., income) were considered missing at the variable level. See
Table 1 for missing data at the variable level. Data missing at the item level included the
following: one participant was missing an individual item on the SVAWS at T1, two
participants at T1 and one participant at T4 were missing an individual item on the social
support scale, and one participant was missing four items on the social support scale at
T5. Scale scores with missing data at the item level were prorated. Missing data at the
variable level were estimated using the hot deck method available in Prelis 8.72
(J oreskog & Sorbom, 2005). The hot deck method substitutes real values for missing
values by obtaining real values from another participant whose scores on speciﬁed
variables most closely match those of the participant with missing data. In this study,
variables from prior waves of data collection corresponding to missing variables were
speciﬁed as matching variables (e. g., T2 DV Frequency was imputed based on T1 DV
Frequency). For those missing variables in which corresponding data was not collected
at a prior wave or imputation was not successful based only on corresponding variables
from previous observations, matching variables were selected based on conceptual
closeness and correlational data. See Appendix G for a list of imputed variables and
corresponding matching variables. Descriptive data for measures before and after
estimation were similar, suggesting that imputation procedures were reasonable (see

Table 2).

67

Hypothesis Testing

Hypotheses Ia-I d

Women with longitudinal patterns of DV victimization based on a variety of
abuse characteristics will suffer from CP symptoms. Speciﬁcally, women classiﬁed into
groups with high levels of DV frequency will report more CP symptoms than women
classiﬁed into groups with lower levels of DV frequency (1a). Women classiﬁed into
groups with high levels of DV severity will report more CP symptoms than women
classiﬁed into groups with lower levels of DV severity (1b). Women classiﬁed into
groups with high levels of sexual DV will report more CP symptoms than women
classiﬁed into groups based on threats of DV or physical DV (1c). Finally, women with
longer duration of DV will report more symptoms of CP than women with shorter
duration of DV (1d).
Analyses for Hypothesis 1

To classify participants into homogeneous groups based on longitudinal patterns
of DV victimization, ﬁve sets of cluster analyses were conducted using R 2.3.1
(Maechler, Rousseeuw, Struyf, & Hubert, 2005; R Development Core Team, 2006).
Cluster analysis is a person-oriented approach to data analysis that classiﬁes individuals
into groups or clusters based on pre-speciﬁed clustering variables, such that individuals
within a cluster are more similar to each other than they are to individuals in other
clusters. Cluster analysis is useful for longitudinal studies because it reduces the large
number of possible developmental trajectories into a manageable number of groups while
minimizing the loss of important information about individual differences (von Eye,

Mun, & Indurkhya, 2004).

68

For the present study, cluster solutions were produced separately for DV
frequency, DV severity, threats of DV, physical DV, and sexual DV. Clustering methods
were the same for all analyses and decisions about the choice of clustering methods were
guided by von Eye et al.’s (2004) discussion of characteristics of clustering methods.
Hierarchical agglomerative clustering procedures were employed using Ward’s method
for the clustering algorithm and Euclidean distance as the proximity measure. Ward’s
method was selected because it maximizes within-group homogeneity and is less
sensitive to outliers than other agglomerative methods. Euclidean distance was a natural
choice for a base measure because it indicates the geometric distance between individuals
in the multidimensional space of the clustering variables. Selection of the number of
interpretable clusters was based on examination of the dendrogram and silhouette plots,
as well as whether the clusters made theoretical sense.

In the ﬁrst cluster analysis, DV frequency scores for each of the 6 waves of data
collection were entered as clustering variables. This cluster analysis suggested a 3-group
solution. Cluster mean proﬁles for DV frequency are presented in Table 3. Cluster 1
(n=89) was named Minimal DV because these women reported low levels of DV
frequency across all time periods. Cluster 2 (n=62) was named Moderate DV; women in
this cluster experienced a moderate level of DV frequency at T1 , decreasing to lower
levels of DV from T2-T6. Cluster 3 (n=13) was named High DVbecause DV frequency
was high at T1, although frequency of DV decreased to lower levels over time. (See
Table 3, Figure 2).

The next 3 sets of cluster analyses based on severity of DV, threats of DV, and

physical DV also suggested 3-group solutions with similar longitudinal proﬁles of DV

69

victimization as those produced by DV frequency. Thus, to determine whether these
groups were equivalent across the 3 solutions and to those produced by the DV frequency
cluster analysis, cross-tabulations were performed between all possible pairs of cluster
solutions. Results indicated that the women who comprised the Minimal, Moderate, and
High DV groups were relatively similar across solutions with the percentage of overlap
ranging from 77% for the threats of DV and physical DV solutions to 92% for the DV
severity and physical DV solutions (see Tables 4, 5, 6, 7, 8, and 9). Therefore, the cluster
solution for DV frequency is the only solution from the ﬁrst 4 sets of cluster analyses
discussed further (see Appendix H for results obtained for severity of DV, threats of DV,
and physical DV cluster analyses), though later ﬁndings pertaining to DV frequency can
be generalized to DV severity, threats of DV, and physical DV since the solutions were
nearly equivalent regardless of the abuse characteristic used to create the groups.

For the ﬁnal cluster analysis, sexual abuse scores for each of the 6 waves were
entered as clustering variables. In contrast to previous analyses, this cluster analysis
suggested a 2-group solution: one large cluster (n=159) and one small cluster (n=5). The
large cluster was called Minimal Sexual DVwith women in this group reporting almost
no sexual abuse at any of the observation periods. The small cluster was called High
Sexual DV because women in this cluster reported sexual abuse at all time periods, with
the exception of T5 and T6, in which no abuse was reported. Cluster mean proﬁles for
sexual abuse are presented in Table 10 and Figure 3.

To examine the relationship between DV victimization and CP symptoms, two
analyses of variance and one regression were conducted. First, a one-way, between-

subjects analysis of variance (ANOVA) was used to test whether number of CP

70

symptoms differed as a function of DV frequency cluster membership (Minimal DV
versus Moderate DV versus High DV). The ANOVA indicated a statistically signiﬁcant
difference, F (2,161) = 14.92, p<.05, in number of CP symptoms across the 3 groups. The
Minimal DV group had a mean CP symptom score of 3.78 (SD=7.08), the Moderate DV
group had a mean score of 7.13 (SD=7.63), and the High DV group had a mean score of
16.00 (SD=12.42). Post-hoe Tukey tests indicated that the Moderate and High DV
groups differed signiﬁcantly from the Minimal DV group (p<.05 in both cases) and that
the High DV group differed signiﬁcantly from the Moderate DV group (p<.05) in the
expected directions. Thus, women who experienced greater frequency of DV reported a
greater number of CP symptoms relative to women who experienced less frequent DV.
Furthermore, this result can be generalized to DV severity, threats of DV, and physical
DV, such that women who reported greater DV severity, more threats of abuse, and more
physical DV reported more CP symptoms relative to women who experienced less severe
DV, fewer threats of abuse, and less physical DV, given that the cluster analyses
produced nearly equivalent solutions for DV frequency, DV severity, threats of abuse,
and physical abuse.

A second one-way, between-subjects ANOVA was performed to test whether
number of CP symptoms differed as a function of sexual abuse groupings (Minimal
Sexual DV versus High Sexual DV). The ANOVA indicated a statistically signiﬁcant
difference, F (1 ,162)=9.l6, p<.05, in number of CP symptoms across the 2 groups. The
Minimal Sexual DV group had a mean CP symptom score of 5.67 (SD=8.10) and the

High Sexual DV group had a mean score of l 7 (SD=12.79); thus, women who reported

71

more sexual abuse reported more CP symptoms than women who experienced less sexual
DV.

To examine whether sexual DV demonstrated better speciﬁcity for CP symptoms
than threats of DV and physical DV, the R2 resulting from the sexual DV ANOVA and
the R2 resulting from the DV frequency ANOVA (which would be nearly equivalent to
that produced by an ANOVA with threats of DV or physical DV groupings as the
between-subj ects factor) were compared. The ANOVA with DV frequency cluster
membership as the between-subjects factor produced an R2 of .16, whereas the ANOVA
with sexual DV cluster membership as the between-subjects factor resulted in an R2 of
.05. Thus, results suggest that sexual DV does not demonstrate better speciﬁcity for CP
than threats of DV or physical DV.

Finally, a linear regression was conducted to determine whether duration of DV
predicted number of CP symptoms. In this model, CP symptoms was regressed on DV
duration. The model resulted in an R2 =.16, F(l,162)=30.06, p<.05 with an
unstandardized estimate b=1.66, t(162)=5.48, p<.05. Thus, for each additional time
period that DV was reported, number of CP symptoms increased by 1.66.

Hypothesis 2

Women in long-term DV relationships (deﬁned here as three or more consecutive
years) will be more vulnerable to CP symptoms related to longitudinal patterns of DV
victimization and DV duration as compared to women in one or more short-term DV

relationship(s).

72

Analysis for Hypothesis 2

First, participants were divided into two groups: women involved in at least one
DV relationship for three or more consecutive years (n=132) and women involved in one
or more short-term DV relationships lasting less than three consecutive years (n=32).
Next, the analyses described above for Hypothesis 1 were repeated separately for the two
groups of women.

Group 1 .' Women in long-term DVrelationships. For the ﬁrst cluster analysis,
DV frequency scores for each of the 6 waves were entered as clustering variables. This
cluster analysis suggested a 3-group solution. Cluster mean proﬁles for DV frequency
are presented in Table 11. Cluster 1 (n=76) was named Minimal DV because these
women reported low levels of DV frequency across all time periods. Cluster 2 (n=34)
was named Moderate DV; women in this cluster experienced a moderate level of DV
frequency at T1, decreasing to lower levels of DV from T2-T6. Cluster 3 (n=22) was
named High DV because DV frequency was high at T1, although frequency of DV
decreased to lower levels over time. (See Table 11, Figure 4).

The next 3 sets of cluster analyses based on severity of DV, threats of DV, and
physical DV also suggested 3-group solutions with similar longitudinal proﬁles of DV
victimization as those produced by DV frequency. Thus, to determine whether these
groups were equivalent across the 3 solutions and to those produced by the DV frequency
cluster analysis, cross-tabulations were performed between all possible pairs of cluster
solutions. Results indicated that the women who comprised the Minimal, Moderate, and
High DV groups were relatively similar across solutions with the percentage of overlap

ranging from 75% for the DV frequency and severity solutions to 89% for the DV

73

severity and physical DV solutions (see Tables 12, 13, 14, 15, 16, and 17). Therefore, the
cluster solution for DV frequency is the only solution from the ﬁrst 4 sets of cluster
analyses discussed further (see Appendix I for results obtained for severity of DV, threats
of DV, and physical DV cluster analyses).

For the ﬁnal cluster analysis, sexual DV scores for each of the 6 waves were
entered as clustering variables. In contrast to previous analyses, this cluster analysis
suggested a 2-group solution: one large cluster (n=127) and one small cluster (n=5). The
large cluster was called Minimal Sexual DV with women in this group reporting almost
no sexual abuse at any of the observation periods. The small cluster was called High
Sexual DV because women in this cluster reported sexual abuse at all time periods, with
the exception of T5 and T6, in which no abuse was reported. Cluster mean proﬁles for
sexual DV are presented in Table 18 and Figure 5.

To examine the relationship between DV victimization and CP symptoms, two
analyses of variance and one regression were conducted. First, a one-way, between-
subjects AN OVA was used to test whether number of CP symptoms differed as a

function of DV frequency cluster membership (Minimal DV versus Moderate DV versus
High DV). The ANOVA indicated a statistically signiﬁcant difference, F (2,129) = 9.24,
p<.05, in number of CP symptoms across the 3 groups. The Minimal DV group had a
mean CP symptom score of 4.21 (SD=7.58), the Moderate DV group had a mean score of
7.38 (SD=8.79), and the High DV group had a mean score of 12.68 (SD=9.71). Post-hoe
TUkey tests indicated that the High DV group differed signiﬁcantly from the Minimal DV
group (p<.05) in the expected direction. Furthermore, the difference between the High

and Moderate DV groups approached signiﬁcance (p=.05). There was no difference in

74

mean CP symptom scores between the Moderate and Minimal DV groups. Thus, the
pattern in means suggests that women in long-term relationships who experience frequent
DV suffer from more CP symptoms than women in long-term relationships who
experience a moderate or infrequent level of DV; however, there appears to be no
difference in the number of CP symptoms between women in long-term relationships
who experience moderate or minimal levels of DV. Furthermore, this result can be
generalized to DV severity, threats of DV, and physical DV, such that women in long-
term DV relationships who reported high levels of DV severity, threats of abuse, and
physical DV suffer more CP symptoms as compared to women who experienced
moderate or low levels of DV severity, threats of abuse, or physical DV, given that the
cluster analyses produced nearly equivalent solutions for DV frequency, DV severity,
threats of abuse, and physical abuse.

A second one-way, between-subjects ANOVA was performed to test whether
number of CP symptoms differed as a function of sexual abuse groupings (Minimal
Sexual DV versus High Sexual DV). The ANOVA indicated a statistically signiﬁcant
difference, F (1,130) = 7.91, p<.05, in number of CP symptoms across the 2 groups. The
Minimal Sexual DV group had a mean CP symptom score of 6.02 (SD=8.39) and the
High Sexual DV group had a mean score of 17.0 (SD=12.79); thus, women in long-term
DV relationships who reported more sexual abuse reported more CP symptoms than
women in long-term DV relationships who reported less sexual DV.

Finally, a linear regression was conducted to determine whether duration of DV
predicted number of CP symptoms. In this model, CP symptoms was regressed on DV

duration. The model resulted in an R2 =.17, F(l,130) = 26.18, p<.05 with an

75

unstandardized estimate b=1.80, t(130)=5. 12, p<.05. Thus, for each additional time
period that DV was reported, number of CP symptoms increased by 1.80.

Group 2: Women in short-term DVrelationships. For the ﬁrst cluster analysis,
DV frequency scores for each of the 6 waves were entered as clustering variables. This
cluster analysis suggested a 3-group solution. Cluster mean proﬁles for DV frequency
are presented in Table 19. Cluster 1 (n=15) was named Minimal DV because these
women reported low levels of DV frequency across all time periods. Cluster 2 (n=l3)
was named Moderate DV; women in this cluster experienced a moderate level of DV
frequency at T1, decreasing to lower levels of DV from T2-T6. Cluster 3 (n=4) was
named High DV because DV frequency was high at T1 and T2, although frequency of
DV decreased to lower levels over time. (See Table 19, Figure 6).

For the second cluster analysis, physical DV scores for each of the 6 waves were
entered as clustering variables. This cluster analysis also suggested a 3-group solution,
though the longitudinal proﬁles of DV victimization differed from those produced in the
DV frequency analysis. Cluster mean proﬁles for physical abuse are presented in Table
20. Cluster 1 (n=19) was named Minimal Physical DVbecause these women reported
low levels of physical abuse across all time periods. Cluster 2 (n=5) was named Early
Moderate-High Physical DV because women in this cluster experienced a moderate level
of physical abuse at T], a high level of physical abuse at T2, and then almost no physical
abuse for the remaining time periods. Cluster 3 (n=8) was named Early High Physical
DV because physical abuse was high at T1 dropping to minimal levels of abuse for T2-

T6. (See Table 20, Figure 7).

76

The next 2 sets of cluster analyses based on severity of DV and threats of DV
suggested 3-group solutions with similar longitudinal proﬁles of DV victimization as
those produced by physical abuse. Thus, to determine whether these groups were
equivalent across the 2 solutions and to those produced by the physical DV cluster
analysis, cross-tabulations were performed between all possible pairs of cluster solutions.
Results indicated that the women who comprised the Minimal, Early Moderate-High, and
Early High DV groups were relatively similar across solutions with the percentage of
overlap ranging from 78% for the DV severity and threats of abuse solutions to 88% for
the DV severity and physical DV solutions (see Tables 21, 22, and 23). Therefore, a
decision was made to not report any further on the DV severity and threats of DV cluster
solutions (see Appendix J for results obtained for severity of DV and threats of DV
cluster analyses).

For the ﬁnal cluster analysis, sexual abuse scores for each of the 6 waves were
entered as clustering variables. In contrast to previous analyses, this cluster analysis
suggested a 2-group solution, one large cluster (n=29) and one small cluster (n=3). The
large cluster was called Minimal Sexual DVbecause women in this group reported little
to no sexual abuse at any of the observation periods. The small cluster was called Early
High Sexual DV because women in this cluster reported sexual abuse at T1 and T2 with
no abuse at any of the remaining time periods. Cluster mean proﬁles for sexual abuse are
presented in Table 24 and Figure 8.

To examine the relationship between DV victimization and CP symptoms, three
analyses of variance and one regression were conducted. First, a one-way, between-

subjects ANOVA was used to test whether number of CP symptoms differed as a

77

function of DV frequency cluster membership (Minimal DVversus Moderate DV versus
High DV). The ANOVA indicated a statistically signiﬁcant difference, F (2,29) = 13.66,
p<.05, in number of CP symptoms across the 3 groups. The Minimal DV group had a
mean CP symptom score of 2.47 (SD=3.42), the Moderate DV group had a mean score of
2.54 (SD=4.03), and the High DV group had a mean score of 16.50 (SD=11.91). Post-hoe
Tukey tests indicated that the High DV group differed signiﬁcantly from the Moderate
and Minimal DV groups (p<.05 for both instances) in the expected directions. However,
there was no difference in mean CP symptom scores between the Moderate and Minimal
DV groups. Thus, the pattern in means suggests that women in short-term relationships
who experience frequent DV suffer from more CP symptoms than women in short-term
relationships who experience a moderate or infrequent level of DV; however, there
appears to be no difference in the number of CP symptoms between women in short-term
relationships who experience moderate or minimal levels of DV.

A second one-way, between-subjects ANOVA was performed to test whether
number of CP symptoms differed as a ﬁmction of physical abuse groupings (Minimal
versus Early Moderate-High versus Early High Physical DV). The ANOVA indicated a
statistically signiﬁcant difference, F (2,29) = 23.98, p<.05, in number of CP symptoms
across the 3 groups. The Minimal Physical DV group had a mean CP symptom score of
2.32 (SD=3.35), the Early Moderate—High Physical DV group had a mean score of 16.4
(SD=7.80), and the Early High Physical DV group had a mean score of 1.25 (SD=3.54).
Post-hoe Tukey tests indicated that the Early Moderate-High group differed signiﬁcantly
from the Minimal group (p<.05) in the expected direction. Furthermore, the Early High

group differed signiﬁcantly from the Early Moderate-High group in the expected

78

direction. There was no difference in mean CP symptom scores between the Early High
and Minimal groups.

Thus, the pattern in means suggests that women in short-term relationships who
experience moderate to high levels of physical abuse over time suffer from more CP
symptoms than women in short-term relationships who experience minimal physical
abuse. In addition, women in short-term DV relationships who experience moderate
levels of DV at one time point and high levels at another time report more CP symptoms
than women who report high levels of DV at just one point in time. This ﬁnding is
expected given that women in the Early High group experienced less DV overall as
compared to those in the Early Moderate-High group. Finally, there appears to be no
difference in the number of CP symptoms between women in short—term relationships
who experience high levels of physical abuse at one point in time but none at others and
those who experience minimal physical abuse at all time periods.

These ﬁndings can also be generalized to DV severity and threats of DV given
that the cluster analyses produced nearly equivalent solutions for physical DV, DV
severity, and threats of abuse. Therefore, women in short-term relationships who
experience moderate to high levels of DV severity or threats of abuse over time suffer
from more CP symptoms than women in short-term relationships who experience
minimal levels of DV severity or threats of abuse. In addition, women in short-term DV
relationships who experience moderate levels of DV severity or threats of DV at one time
point and high levels at another time suffer more CP symptoms than women who report
high levels of DV severity or threats of abuse at just one point in time. Finally, there is

no difference in the number of CP symptoms between women in short-term relationships

79

who experience high levels of DV severity or threats of abuse at one point in time but
none at others and those who experience minimal DV severity or threats of abuse across
all time periods.

A third one-way, between-subjects ANOVA was performed to test whether
number of CP symptoms differed as a function of sexual abuse groupings (Minimal
Sexual DV versus Early High Sexual DV). The ANOVA indicated a statistically
signiﬁcant difference, F (1,30) = 84.33, p<.05, in number of CP symptoms across the 2
groups. The Minimal Sexual DV group had a mean CP symptom score of 2.41 (SD=3.61)
and the High Sexual DV group had a mean score of 22.00 (SD=1.73); thus, women who
reported more sexual abuse reported more CP symptoms than women who report less
sexual DV.

Finally, a linear regression was conducted to determine whether duration of DV
predicted number of CP symptoms. In this model, CP symptoms was regressed on DV
duration. The model resulted in an R2=.05, F (l ,30)=1 .52, n.s. with an unstandardized
regression estimate b=.88, t(30)=1.23, n.s. Thus, duration of DV was not related to CP
symptoms among women in one or more short-term DV relationships.

Findings regarding the relationship between DV and CP symptoms for women in
long- and short-term relationships were compared by examining differences in the
magnitude of R2 for each analysis. For women in long-term DV relationships, DV
frequency cluster membership (which was deemed equivalent to DV severity, threats of
DV, and physical DV cluster membership) accounted for 13% in the variation in number
of CP symptoms, sexual DV cluster membership accounted for 6% in the variation in

number of CP symptoms, and duration of DV accounted for 18% in the variation of

80

number of CP symptoms. For women in short-term DV relationships, DV frequency
cluster membership accounted for 49% in the variation in number of CP symptoms,
physical DV cluster membership (which was deemed equivalent to DV severity and
threats of DV cluster membership) accounted for 62% in the variation in number of CP
symptoms, sexual DV cluster membership accounted for 74% in the variation in number
of CP symptoms, and duration of DV accounted for 5% in the variation of number of CP
symptoms. Thus, it appears that DV was generally a better predictor of CP symptoms for
women in short-term DV relationships compared to those in long-term DV relationships.
The only exception to this was with duration of DV, which accounted for more variation
in CF among women in long-term DV relationships than in women in short-term DV
relationships.

Although it appears that DV may be a better predictor of CP among women in
short-term than in long-term DV relationships, the cluster solutions produced in the
analyses for Hypothesis 1 were used in all remaining analyses because those cluster
solutions were based on the entire sample of women and the larger sample size was
needed to increase power in remaining analyses.

Hypothesis 3a

Women who experience DV will lose social support over time. In addition,
women classiﬁed into groups with high levels of DV frequency, DV severity, threats of
DV, physical DV or sexual DV will report lower levels of emotional support and fewer
network supporters across all time periods than women classiﬁed into groups with lower

levels of DV frequency, DV severity, threats of DV, physical DV, or sexual DV. Women

81

with longer duration of DV will also report lower levels of emotional support and fewer
network supporters over time than women with shorter duration of abuse.
Analyses for Hypothesis 3a

Repeated measures analyses were conducted to examine whether abused women
lose social support over time. Analyses were conducted separately for DV frequency
cluster membership, sexual DV cluster membership, and duration of DV, as well as for
number of network supporters and emotional support. The cluster solutions produced for
the overall sample of women (i.e., those produced in the Analysis for Hypothesis 1) were
used for these analyses. Overall, results did not support study hypotheses regarding main
effects and the effect of time on social support. However, details of these results are
reported below.

Number of Network Supporters. A two-way repeated measures ANOVA was
conducted with DV frequency cluster membership as the between-subjects factor and
time as the within-subjects factor. The Geisser-Greenhouse correction was applied for
within-subjects tests to adjust for asphericity. The ANOVA indicated a signiﬁcant main
effect for time [F (4.02, 646.35)=28.07, p<.05], such that number of supporters decreased
over time from T1 to T6 for all groups of abused women. The ANOVA also indicated a
marginally signiﬁcant main effect for DV frequency cluster membership [F (2, 161)=2.97,
p=.05]. The interaction of time and cluster membership was not statistically signiﬁcant
[F (8.03, 646.35)=8.03, n.s.]. Means and standard deviations for number of network
supporters as a function of cluster membership and time are presented in Table 25. Mean
number of network supporters by cluster membership and time is also illustrated

graphically in Figure 9.

82

Games-Howell comparisons were used to follow-up the main effect of cluster
membership. The comparisons between the Minimal and High DV groups and between
the Moderate and High groups were not signiﬁcant. However, the comparison between
the Minimal and Moderate DV clusters was signiﬁcant (p<.05). The pattern in means
suggests that women who experience moderate levels of DV have fewer supporters
averaged across time as compared to women with low levels of DV. However, contrary
to expectation, women with high levels of DV do not have signiﬁcantly fewer supporters
than women with low or moderate levels of DV frequency.

A two-way repeated measures AN OVA was conducted with sexual abuse cluster
membership as the between-subjects factor and time as the within-subjects factor. The
Geisser-Greenhouse correction was applied for within-subjects tests to adjust for
asphericity. The ANOVA indicated a signiﬁcant main effect for time [F (4.05,
655.89)=4.25, p<.05]. The main effect of sexual abuse cluster membership was not
statistically signiﬁcant [F (1 , l62)=.06, n.s.]. The interaction of time and cluster
membership was also not statistically signiﬁcant [F (4.05, 655.89)=.59, n.s.]. Means and
standard deviations for number of network supporters as a function of cluster
membership and time are presented in Table 26. Thus, results indicate that number of
network supporters decreased over time from T1 to T6 for all groups of abused women;
however, there was no mean difference in number of supporters averaged across time
between women with high versus low sexual DV.

In order to perform a repeated measures ANOVA with duration of DV, the
continuous DV duration variable was converted to a categorical variable by dividing

participants into 4 groups. Group 1 consisted of women who experienced DV for a

83

minimum of 1 time period and a maximum of 2 time periods (n=3 8), Group 2 consisted
of women who experienced DV for a minimum of 3 time periods and a maximum of 4
time periods (n=50), Group 3 consisted of women who experienced DV for a minimum
of 5 time periods and a maximum of 6 time periods (n=53), and Group 4 consisted of
women who experienced DV for a minimum of 7 time periods and a maximum of 8 time
periods (n=23). Next, a two-way repeated measures ANOVA was conducted with
duration group membership as the between-subj ects factor and time as the within-
subjects factor. The Geisser—Greenhouse correction was applied for within-subjects tests
to adjust for asphericity. The ANOVA indicated a signiﬁcant main effect for time
[F (4.02, 643.13)=57.23, p<.05]. The main effect for group membership was not
statistically signiﬁcant [F (3, 160)=1.27, n.s.]. The interaction of time and group
membership was also not statistically signiﬁcant [F (12.06, 643.13)=.82, n.s.]. Means and
standard deviations for number of network supporters as a function of cluster
membership and time are presented in Table 27. Thus, number of network supporters
decreased over time from T1 to T6 for all groups of abused women; however, there was
no mean difference in number of supporters averaged across time among women with
chronic versus less persistent experiences of DV.

Emotional Support. A two-way repeated measures ANOVA was conducted with
DV frequency cluster membership as the between-subj ects factor and time as the within-
subjects factor. The Geisser-Greenhouse correction was applied for within-subjects tests
to adjust for asphericity. The ANOVA indicated a signiﬁcant main effect for time
[F(4.42, 711.33)=5.54, p<.05] and a marginally signiﬁcant main effect for DV frequency

Cluster membership [F(2, 161)=2.66, p=.07]. The interaction of time and cluster

84

membership was not statistically signiﬁcant [F(8.84, 711.33)=1.15, n.s.]. Means and
standard deviations for emotional support as a function of cluster membership and time
are presented in Table 28. In contrast to expectation, ﬁndings indicate that emotional
support increased over time from T1 to T6 for abused women.

A two-way repeated measures ANOVA was conducted with sexual abuse cluster
membership as the between-subjects factor and time as the within-subjects factor. The
Geisser-Greenhouse correction was applied for within-subjects tests to adjust for
asphericity. The ANOVA indicated a signiﬁcant main effect for cluster membership
[F (1 , 162)=4.66, p<.05], such that women with high sexual abuse report less emotional
support than women with low sexual abuse. However, the main effect of time was not
statistically signiﬁcant [F(4.39, 710.56)=1.64, n.s.], suggesting no change in mean level
of emotional support over time. The interaction of time and cluster membership was also
not statistically signiﬁcant [F (4.39, 710.56)=.77, n.s.]. Means and standard deviations for
emotional support as a function of cluster membership and time are presented in Table
29.

In order to perform a repeated measures analysis with duration of DV as a
predictor variable, the 4-group categorical duration variable as described above was used
as the between-subjects factor in a two-way repeated measures ANOVA with time as the
within-subjects factor. The Geisser-Greenhouse correction was applied for within-
subjects tests to adjust for asphericity. The ANOVA indicated signiﬁcant main effects
for both time [F (4.39, 701 .57)=7.88, p<.05] and group membership [F (3, l60)=4.07,
p<.05]. The interaction of time and group membership was not statistically signiﬁcant

[F (13.15, 701.57)=1.20, n.s.]. Means and standard deviations for emotional support as a

85

function of group and time are presented in Table 30. Contrary to expectation, results
indicate that mean emotional support increased over time from T1 to T6 for abused
women. Games-Howell comparisons were used to better understand the main effect of
group membership. Comparisons between the 1 and 2, 1 and 3, 2 and 3, and 2 and 4
groups were not signiﬁcant. However, the comparison between groups 1 and 4 was
signiﬁcant (p<.05), indicating that emotional support was generally lower among women
who experienced chronic DV occurring for at least 7 observation periods as compared to
women who reported DV for 2 observation periods or less.
Hypothesis 3b

Social support will mediate the relationship between longitudinal patterns of DV
victimization and CP symptoms, as well as the relationship between DV duration and CP
symptoms.
A nalysis for Hypothesis 3b

According to Baron and Kenny (1986), to test for mediation a series of three
regression equations must be estimated. First, the dependent variable is regressed on the
independent variable. Second, the mediator is regressed on the independent variable.
Third, the dependent variable is regressed on both the independent variable and the
mediator. To establish mediation, four conditions must be satisﬁed: 1) the independent
variable must predict the dependent variable in the ﬁrst equation, 2) the independent
variable must predict the mediator in the second equation, 3) the mediator must predict
the dependent variable in the third equation, and 4) the effect of the independent variable
on the dependent variable in the third equation must be zero (for complete mediation) or

less than the effect in the ﬁrst equation (for partial mediation).

86

To test the mediating role of social support, Baron and Kenny’s (1986)
procedures for establishing mediation were followed: ﬁrst, regressing CP on DV; second,
regressing social support on DV; and third, regressing CP on both DV and social support.
This series of regression equations was conducted separately for various facets of DV
(i.e., DV frequency and sexual DV clusters as produced in the Analysis for Hypothesis 1,
and duration of DV) and for two different indicators of social support: number of network
supporters at T6 and emotional support at T6. Overall, analyses did not support
mediation, though details of these results are reported below.

Number of Network Supporters. The ﬁrst series of equations tested whether
number of network supporters mediated the relationship between DV frequency and CP
symptoms. Because DV frequency is a categorical variable (with values representing the
Minimal, Moderate, or High DV frequency clusters), a dummy variable coding scheme
designating the Minimal DVcluster as the comparison group was incorporated into the
series of regression equations. First, CP was regressed on DV frequency. The
unstandardized regression estimates were 3.78 for the Minimal DV group [t(161)=4.56,
p<.05], 3.35 for the Moderate DV group [t(161)=2.60, p<.05], and 12.22 for the High DV
group [t(161)=5.27, p<.05]. The R2 was .16 [F(2,161)=14.92, p<.05]. Second, number
of network supporters was regressed on DV frequency. The unstandardized regression
estimates were 7.45 for the Minimal DV group [t(161)=18.45, p<.05], -1.40 for the
Moderate DV group [t(161)=-2.22, p<.05], and -1.45 for the High DV group [t(161)=-
1.28, n.s.]. The R2 was .03 [F(2,161)=2.79, p<.10]. Finally, CP was regressed on DV
frequency and number of network supporters. The unstandardized regression estimates

were 2.92 for the Minimal DV group [t(l60)=l .99, p<.05], 3.51 for the Moderate DV

87

group [t(160)=2.68,p<.05], 12.39 for the High DVgroup [t(160)=5.31,p<.05], and .12
for number of network supporters [t(160)=.71, n.s.]. The R2 was .16 [F (3,160)=10.08,
p<.05]. Although the overall models speciﬁed in each of the three equations were
signiﬁcant or marginally signiﬁcant, the main effect of number of network supporters
was not signiﬁcant in the third equation. Furthermore, the relationship between DV
frequency and CP symptoms was not zero or reduced in the third equation. Therefore,
the mediation hypothesis was not supported.

The second series of equatiOns tested whether number of network supporters
mediated the relationship between sexual abuse and CP symptoms. Although sexual
abuse is a categorical value, there are only two levels (representing the Minimal and High
Sexual DV clusters). Therefore, dummy coding is unnecessary. However, the variable
was recoded, such that a value of zero represents the Minimal Sexual DV cluster and a
value of one represents the High Sexual DV cluster. First, CP was regressed on sexual
abuse. The unstandardized regression estimates were 5.67 for the Minimal Sexual DV
group [t(162)=8.67, p<.05] and 11.33 for the High Sexual DV group [t(162)=3.03, p<.05].
The R2 was .05 [F (1,162)=9.16, p<.05]. Second, number of network supporters was
regressed on sexual abuse. The regression estimates were 6.82 for the Minimal Sexual
DV group [t(162)=22.27, p<.05] and -.42 for the High Sexual DV group [t(162)=-.24,
n.s.]. The R2 was 0 [F(l,162)=.06, n.s.]. Finally, CP was regressed on sexual abuse and
number of network supporters. The unstandardized regression estimates were 5.65 for
the Minimal Sexual DV group [t(161)=4.27, p<.05], 1 1.33 for the High Sexual DV group
[t(161)=3.02, p<.05], and 0 for number of network supporters [t(161)=.02, n.s.]. The R2

was .05 [F (2,161)=4.55, p<.05]. The overall model specifying sexual abuse as a

88

predictor of number of network supporters was not statistically signiﬁcant. F urtherrnore,
number of network supporters was not a signiﬁcant predictor of CP symptoms, and the
relationship between sexual abuse and CP symptoms was not zero or reduced in the third
equation. Therefore, the mediation hypothesis was not supported.

The third series of equations tested whether number of network supporters
mediated the relationship between DV duration and CP symptoms. First, CP was
regressed on DV duration [b=1.66, t(162)=5.48, p<.05; R2=.l6 [F(l,162)=30.06, p<.05].
Second, number of network supporters was regressed on DV duration [b=-.31, t(162)=-
2.10, p<.05; R2=.03 [F(l,162)=4.39, p<.05]. Third, CP was regressed on DV duration
and number of network supporters. The unstandardized regression estimates were 1.70
for DV duration [t(161)=5.55, p<.05] and .14 for number of network supporters
[t(161)=.86, n.s.]. The R2 was .16 [F(2,161)=15.38,p<.05]. Although the ﬁrst two
conditions were satisﬁed, number of network supporters was not a signiﬁcant predictor of
CP symptoms, and the relationship between DV duration and CP symptoms was not zero
or reduced in the third equation. Therefore, the mediation hypothesis was not supported.

Emotional Support. The fourth series of equations tested whether emotional
support mediated the relationship between DV frequency and CP symptoms. Because
DV frequency is a categorical variable (with values representing the Minimal, Moderate,
or High DV frequency clusters), a dummy variable coding scheme designating the
Minimal DV cluster as the comparison group was incorporated into the series of
regression equations. First, CP was regressed on DV frequency. The unstandardized
regression estimates were 3.78 for the Minimal DV group [t(161)=4.56, p<.05], 3.35 for

the Moderate DV group [t(161)=2.60, p<.05], and 12.22 for the High DV group

89

[t(161)=5.27,p<.05]. The R2 was .16 [F(2,161)=14.92, p<.05]. Second, emotional
support was regressed on DV frequency. The unstandardized regression estimates were
3.47 for the Minimal DV group [t(161)=67.72, p<.05], -.05 for the Moderate DV group
[t(l61)=-.66, n.s.], and -.12 for the High DV group [t(161)=-.85, n.s.]. The R2 was .01

[F (2,161)=.48, n.s.]. Finally, CP was regressed on DV frequency and emotional support.
The unstandardized regression estimates were 11.15 for the Minimal DV group
[t(l60)=2.50, p<.05], 3.24 for the Moderate DV group [t(l60)=2.52, p<.05], 11.97 for the
High DV group [t(160)=5.18, p<.05], and -2.13 for emotional support [t(160)=-1.68,
p<.10.]. The R2 was .17 [F (3,160)=1 1.00, p<.05]. The overall model speciﬁed in
equation 2 was not statistically signiﬁcant. In addition, the main effect of social support
was only marginally signiﬁcant, and the relationship between DV frequency and CP
symptoms was not zero and only slightly reduced (for the Moderate and High DV groups)
in the third equation. Therefore, the mediation hypothesis was not supported.

The ﬁfth series of equations tested whether emotional support mediated the
relationship between sexual abuse and CP symptoms. Although sexual abuse is a
categorical value, there are only two levels (representing the Minimal and High Sexual
DVclusters). Therefore, dummy coding is unnecessary. However, the variable was
recoded, such that a value of zero represents the Minimal Sexual DV cluster and a value
of one represents the High Sexual DV cluster. First, CP was regressed on sexual abuse.
The unstandardized regression estimates were 5.67 for the Minimal Sexual DV group
[t(162)=8.67, p<.05] and 11.33 for the High Sexual DV group [t(162)=3.03, p<.05]. The
R2 was .05 [F (l,162)=9.16, p<.05]. Second, emotional support was regressed on sexual

abuse. The unstandardized regression estimates were 3.45 for the Minimal Sexual DV

90

group [t(162)=90.57, p<.05] and -.31 for the High Sexual DV group [t(162)=—1.43, n.s.].
The R2 was .01 [F (1,162)=2.05, n.s.]. Finally, CP was regressed on sexual abuse and
emotional support. The unstandardized regression estimates were 13.26 for the Minimal
Sexual DV group [t(161)=2.84, p<.05], 10.65 for the High Sexual DV group [t(161)=2.84,
p<.05], and -2.20 for emotional support [t(l61)=-1.64,p=.10, n.s.]. The overall model
specifying sexual abuse as a predictor of emotional support was not statistically
signiﬁcant. Furthermore, emotional support was only a marginally signiﬁcant predictor
of CP symptoms, and the relationship between sexual abuse and CP symptoms was not
zero and only slightly reduced (for the High Sexual DV group) in the third equation.
Therefore, the mediation hypothesis was not supported.

The sixth series of equations tested whether emotional support mediated the
relationship between DV duration and CP symptoms. First, CP was regressed on DV
duration [b=1.66, t(162)=5.48, p<.05; R2=.16, F(l,162)=30.06, p<.05]. Second,
emotional support was regressed on DV duration [b=—.04, t(162)=-2.19, p<.05; R2=.03,

F (l ,162)=4.78, p<.05]. Third, CP was regressed on DV duration and emotional support.
The unstandardized regression estimates were 1.60 for DV duration [t(161)=5.21, p<.05]
and -l.50 for emotional support [t(161)=-l.17, n.s.]. The R2 was .16 [F(2,161)=15.75,
p<.05]. Although the ﬁrst two conditions were satisﬁed and the relationship between DV
duration and CP symptoms was slightly reduced in the third equation, emotional support
was not a signiﬁcant predictor of CP symptoms. Therefore, the mediation hypothesis

was not supported.

91

Hypothesis 4

Adult representations of childhood attachment will moderate the relationship
between longitudinal patterns of DV victimization and CP symptoms, as well as the
relationship between DV duration and CP symptoms, such that insecure attachment will
exacerbate CP symptoms related to DV victimization.
Analysis for Hypothesis 4

The differential effect of DV on CP symptoms as a function of attachment was
examined separately for three facets of DV (frequency, sexual abuse, and duration) using
Baron and Kenny’s (1986) recommended procedures for testing moderation. According
to Baron and Kenny (1986), the speciﬁc analytic method for testing moderation varies
according to the categorical or continuous nature of the independent and moderator
variables under study. In this study, the moderator variable attachment is categorical.
However, the nature of the independent variable DV changes depending on the particular
facet of DV examined; DV frequency and sexual abuse are both categorical variables
representing cluster membership as determined in the Analysis for Hypothesis 1, while
DV duration is a continuous variable. Therefore, two cases are presented here. In case 1,
both the independent and moderator variables are categorical, whereas in case 2, the
independent variable is continuous and the moderator variable is categorical. Thus,
analytic procedures and results are reported separately for each case.

Case I : Categorical Independent Variable and Categorical Moderator. To test
for moderation when both the independent and moderator variables are categorical, Baron
and Kenny (1986) recommend conducting a two-way, between-subjects ANOVA. A

moderator effect represented by the product of the independent and moderator variables

92

is included in the ANOVA with a signiﬁcant interaction indicating moderation. This
procedure was followed to test the potential moderating role of attachment in the relation
between DV and CP symptoms for both DV frequency and sexual abuse. [It should be
noted that the use of a multiplicative term to represent the moderator effect only makes
sense if the levels of the independent variable follow a natural order, which is the case in
this study. For example, DV frequency cluster 1 represents Minimal abuse, cluster 2
represents Moderate abuse, and cluster 3 represents High abuse]

First, a two-way, between-subjects AN OVA was computed to test the differential
effect of DV frequency (Minimal, Moderate, High) on number of CP symptoms as a
function of attachment (insecure versus secure). Cell means are presented in Table 31.
The interaction term was not signiﬁcant [F (2,158)=.09, n.s.]. Thus, a moderator
hypothesis was not supported.

This same procedure was followed to test the differential effect of sexual abuse
(Minimal, High) on number of CP symptoms as a function of attachment (insecure versus
secure). Cell means are presented in Table 32. Likewise, the interaction term was not
signiﬁcant [F (l ,160)=. l 3, n.s.], indicating that the moderator hypothesis was not
supported.

Case 2: Continuous Independent Variable and Categorical Moderator. To test
for moderation when the independent variable is continuous and the moderator is
categorical, Baron and Kenny (1986) suggest using regression to estimate the effect of
the independent variable on the dependent variable separately for each level of the
moderator and then comparing the resulting unstandardized regression coefﬁcients

among the independent samples. This procedure was followed to examine the potential

93

moderating role of attachment in the relationship between DV duration and CP
symptoms.

First, CP was regressed on DV duration for participants with insecure
representations of attachment. The unstandardized regression estimate was 1.67
[t(52)=2.87, p<.05], and the R2 was .14, [F(1,52)=8.21, p<.05]. Next, CP was regressed
on DV duration for participants with secure representations of attachment. The
unstandardized regression estimate was 1.61 [t(108)=4.54, p<.05], and the R2 was .16,
[F (1,108)=20.57, p<.05]. Finally, these estimates were compared to test the null
hypothesis Ho: B5=B,, where Bs is the unstandardized regression coefﬁcient for
participants with secure representations of attachment and Bi corresponds to the
unstandardized regression coefﬁcient for participants with insecure attachment. To
perform this analysis, the attachment variable was coded, such that a value of 1
represented secure attachment and a value of 0 represented insecure attachment. Then,
attachment, DV duration, and the interaction of these two variables were entered as
predictors in a regression equation. The interaction term in this equation tests the null
hypothesis Ho: BS=B,. Results failed to reject the null hypothesis [t(160)=-.10, n.s.;
R2=.16, F (3,160)=10.24, p<.05], indicating that the unstandardized regression coefﬁcient
for DV duration predicting CP symptoms was not signiﬁcantly different for participants
with insecure versus secure representations of attachment. Thus, moderation was not
supported.

Hypothesis 5
Child maltreatment history will moderate the relationship between longitudinal

patterns of DV victimization and CP symptoms, as well as the relationship between DV

94

duration and CP symptoms, such that the experience of child maltreatment will
exacerbate the relationship between DV and CP symptoms.
Analyses for Hypotheses 5

To examine the potential moderating role of child maltreatment history in the
relationship between DV and CP symptoms, Baron and Kenny’s (1986) procedures for
testing moderation were employed. Again, this hypothesis was examined separately for
three independent DV variables: frequency cluster membership, sexual abuse cluster
membership, and duration. The procedures used to test moderation, however, differ
somewhat from those described above to examine the moderating role of attachment
because child maltreatment history is a continuous variable. Thus, we have two different
cases altogether. In the ﬁrst case (case 3), the independent variable is categorical and the
moderator is continuous. In the second case (case 4), both the independent and
moderator variables are continuous. Analyses and results are reported separately for each
case.

Case 3: Categorical Independent Variable and Continuous Moderator. To test
for moderation when the independent variable is categorical and the moderator variable is
continuous, Baron and Kenny (1986) suggest conducting a linear regression, in which the
dependent variable is regressed on the independent variable, the moderator variable, and
the product of the independent and moderator variables. The moderator hypothesis is
supported if the interaction is signiﬁcant after controlling for the main effects of the
independent and moderator variables. This was followed to test the potential moderating
role of child maltreatment history in the relation between DV and CP symptoms for both

DV frequency and sexual abuse.

95

 

 

 

First, regression was used to test whether the effect of DV frequency on CP
symptoms changes linearly with respect to child maltreatment history. Because DV
frequency is a categorical variable (with values representing the Minimal, Moderate, or
High DV frequency clusters produced in the Analysis for Hypothesis 1), a dummy
variable coding scheme designating the Minimal DV cluster as the comparison group was
incorporated into the regression equation. This regression produced two interaction terms
(Moderate DV frequency * child maltreatment history) and (High DV frequency * child
maltreatment history). The unstandardized regression estimates were -.13 for the
interaction of Moderate DV and child maltreatment history [t(158)=-2.04, p<.05] and .27
for the interaction of High DVand child maltreatment history [t(158)=2.28,p<.05], and
the R2 for the model was .33 [F (5,158)=15.24, p<.05]. The moderator effect was
signiﬁcant, as indicated by the signiﬁcant DV by child maltreatment interactions after
controlling for the main effects of DV and child maltreatment. Thus, the moderation
hypothesis was supported, such that child maltreatment exacerbated DV—related CP
symptoms with the greatest exacerbation in the High DV group.

Likewise, regression was used to test whether the effect of DV sexual abuse on
CP symptoms changes linearly with respect to child maltreatment history. Although
sexual abuse is a categorical value, there are only two levels (representing the Minimal
and High Sexual DV clusters produced in the Analysis for Hypothesis 1). Therefore,
dummy coding is unnecessary. However, the variable was recoded, such that a value of 0
represented the Minimal Sexual DV cluster and a value of 1 represented the High Sexual
DVcluster. The unstandardized regression estimate was .43 for the interaction term,

which was marginally signiﬁcant [t(160)=1.81, p<.07], and the R2 for the model was .21

96

[F (3,160)=14.57, p<.05]. Thus, results suggest that child maltreatment history moderates
the relationship between sexual abuse cluster membership and CP symptoms, such that
child maltreatment exacerbates DV-related CP symptoms with women in the High Sexual
DVcluster at the greatest risk for DV-related CP symptoms as child maltreatment
increases.

Case 4: Continuous Independent Variable and Continuous Moderator. Although
this case is unique in that both the independent and moderator variables are continuous,
the test for moderation is the same as the test speciﬁed above in case 3, with one
exception: a dummy coding scheme for the independent variable is unnecessary.
Therefore, to test whether the effect of DV duration on CP symptoms changes in a linear
fashion with respect to child maltreatment history, CP was regressed on DV duration,
child maltreatment history, and the interaction of these two variables. The overall model
was signiﬁcant [R2=.27, F (3,160)=1 9.79, p<.05]; however, the unstandardized regression
estimate for the interaction of DV duration and child maltreatment was not signiﬁcant

[b=.02, t(160)=1.14, n.s.]. Thus, the moderator hypothesis was not supported in this case.

97

Discussion

Overall, ﬁndings offer partial support for the proposed model of the development
of CP symptoms in a community sample of women with lifetime experiences of DV.
Findings suggest that abused women report elevated symptoms of CP, and that women
who experience childhood maltreatment may be particularly vulnerable to CF symptoms
in response to DV. DV was related to poor quality of social support; however, social
isolation did not account for the DV-CP symptom relationship. Thus, risk factors for CP
are partially explained here, though mechanisms for the development of CP symptoms
among abused women living in the community are not yet understood. Speciﬁc ﬁndings
pertaining to each hypothesis are discussed below.

Longitudinal Trajectories of DV Victimization

Results from cluster analyses suggest that women’s individual experiences of DV
over time based on frequency, severity, threats of aggression, or physical abuse scores
can be summarized by three longitudinal proﬁles of victimization. The ﬁrst proﬁle is
characterized by low levels of DV across time. The second proﬁle is characterized by
moderate levels of DV at T1 with lower levels of DV across time. The ﬁnal proﬁle is
characterized by high levels of DV at T1 decreasing to lower levels over time. These
proﬁles suggest that abused women differ initially in terms of level of DV, but violence
generally decreases over time for all abused women, regardless of initial level of DV.
Furthermore, at the end of the study period (T6), there are no mean differences in level of
DV among the three patterns.

There are several possible explanations for the observation that DV is generally

greater at T1 with declining levels of DV over time. First, this ﬁnding may reﬂect the

98

sampling strategy utilized for the Mother-Infant Study. Speciﬁcally, participants were
selected for an overrepresentation of DV at T1 to ensure that an adequate number of
women who experienced DV were recruited for the study. Second, participants were
recruited during pregnancy, and the T1 assessment of violence combined DV experienced
during and prior to pregnancy. Previous research suggests that women may be
particularly vulnerable to violence prior to and during pregnancy with a decreased risk
following delivery (Martin, Mackie, Kupper, Buescher, & Moracco, 2001; Torres et al.,
2000). For example, Martin et al. (2001) found that the prevalence rate of physical abuse
against women was 6.9% in the 12 months prior to pregnancy, 6.1% during pregnancy,
and 3.2% approximately 4 months postpartum. In addition, the absence of abuse before
and during pregnancy was strongly protective against postpartum abuse. These ﬁndings
are consistent with trajectories observed in the current study; women who reported low
levels of violence at T1 reported low levels of violence across time, and those who
reported moderate to high levels of DV at T1 reported decreasing levels of abuse over
time. However, the Martin et al. (2001) study was cross-sectional, relied on retrospective
reports of abuse, and assessed physical violence against women more generally (though
in the large majority of cases, violence was perpetrated by a romantic partner). Thus, the
current study is unique in that ﬁndings are speciﬁc to DV, the assessment of violence
included threats of aggression in addition to actual violence, and the study design was
longitudinal and prospective.

Research on women who are not pregnant also ﬁnds decreasing rates of DV
across time, consistent with results from the current study. For example, Jasinski (2001)

found that prevalence of DV dropped from 11.5% to 8.9% over a 5-year period among

99

romantic couples who were together during the length of the study. Furthermore,
epidemiological research suggests that following early adulthood (age 24), risk for DV
declines with age (Rennison, 2001). In the present research, the mean age of participants
was 25.0 years (SD=4.9) at study entry; thus, based on epidemiological research, it is
consistent that DV decreased as time elapsed and participants aged.

From a theoretical perspective, these trajectories can be understood in a variety of
ways. For example, pregnancy may be a time of increased stress for a couple and as a
result, increase the risk for DV (J asinski, 2001). This may particularly be true of couples
who will be new parents, couples who are expecting an unplanned child, couples with
ﬁnancial hardship, or couples in which the male partner is not the father of the child
(Jasinski, 2001; Stewart & Cecutti, 1993; Torres et al., 2000). It has also been suggested
that violent male partners may experience feelings of jealousy in relation to the unborn
child, increasing risk of pregnancy DV (Campbell, Harris, & Lee, 1995). For some
couples, DV may decrease postpartum because the birth of a child triggers violence
cessation; for example, J asinski (2001) found that men who ceased to perpetrate DV
against female partners over the course of their ﬁve-year study were almost twice as
likely to have experienced the birth of their ﬁrst child during the ﬁrst wave of data
collection.

The present ﬁnding that DV generally decreases over time is also consistent with
Herman’s (1992a; 1992b) theory that acts of violence within a relationship may decrease
across time because the perpetrator no longer needs to use violence in order to control the
victim. The perpetrator may use violence initially to instill fear and establish a

relationship of coercive control, but once he has achieved this objective, actual violence

100

romantic couples who were together during the length of the study. Furthermore,
epidemiological research suggests that following early adulthood (age 24), risk for DV
declines with age (Rennison, 2001). In the present research, the mean age of participants
was 25.0 years (SD=4.9) at study entry; thus, based on epidemiological research, it is
consistent that DV decreased as time elapsed and participants aged.

From a theoretical perspective, these trajectories can be understood in a variety of
ways. For example, pregnancy may be a time of increased stress for a couple and as a
result, increase the risk for DV (Jasinski, 2001). This may particularly be true of couples
who will be new parents, couples who are expecting an unplanned child, couples with
ﬁnancial hardship, or couples in which the male partner is not the father of the child
(J asinski, 2001; Stewart & Cecutti, 1993; Torres et al., 2000). It has also been suggested
that violent male partners may experience feelings of jealousy in relation to the unborn
child, increasing risk of pregnancy DV (Campbell, Harris, & Lee, 1995). For some
couples, DV may decrease postpartum because the birth of a child triggers violence
cessation; for example, J asinski (2001) found that men who ceased to perpetrate DV
against female partners over the course of their ﬁve-year study were almost twice as
likely to have experienced the birth of their ﬁrst child during the ﬁrst wave of data
collection.

The present ﬁnding that DV generally decreases over time is also consistent with
Herman’s (1992a; 1992b) theory that acts of violence within a relationship may decrease
across time because the perpetrator no longer needs to use violence in order to control the
victim. The perpetrator may use violence initially to instill fear and establish a

relationship of coercive control, but once he has achieved this objective, actual violence

100

may no longer be necessary to maintain power over the victim. Other methods of control
not measured in the current study, such as control over economic and legal resources, or
undermining or discrediting the victim may continue throughout the course of a
relationship. Thus, psychological abuse tactics may be effective due to prior violence.
This position is supported by Marshall’s (1996; 1999) investigations on the effects of
psychological abuse, which suggest that a perpetrator can be abusive and controlling
without necessarily being physically or sexually violent, as well as other studies that
suggest that psychological abuse can occur independent of violence (Alexander, 1993; D.
G. Dutton & Painter, 1993). Further examination of alternative methods of coercion over
time may serve to further clarify women’s individual experiences in abusive
relationships.

Findings also indicate that longitudinal proﬁles of DV victimization are nearly
equivalent regardless of whether the proﬁle is based on DV frequency, DV severity,
threats of DV, or physical DV scores. Thus, each of these indicators of abuse provides
similar information about women’s experiences of DV victimization over time,
suggesting that measurement of only one of these indicators may be necessary to deﬁne
developmental trajectories of DV. This ﬁnding is not surprising given that each of these
scores were derived from the same measure and were highly correlated (e.g., Tl DV
frequency highly correlated with T1 DV severity, T1 threats of DV, and T1 physical
DV). DV frequency and severity scores were positively related because the calculation
of these scores was based on endorsement of the same items. However, threats of DV
and physical DV scores were based on summing frequency counts for separate items.

Thus, the positive correlation between these two types of scores indicates that threatened

101

and actual violence do not often occur in the absence of each other; women who
experience threatened violence are also likely to experience physical violence. This
ﬁnding is consistent with two prior studies of large community samples of women in
long-term heterosexual relationships conducted by Marshall (1996; 1999), who found
that threats of aggression and acts of violence as assessed by the SVAWS were highly
correlated (r=.82 and r=.85).

In contrast, the cluster solution produced for sexual DV was not equivalent to the
solutions produced for DV frequency, DV severity, threats of DV, and physical DV.
Instead of a three-group solution representing mild, moderate, and high levels of
violence, the cluster analysis based on sexual DV scores produced a two-group solution.
Women who comprised Group I experienced almost no sexual abuse across time,
whereas women in Group 2 experienced high sexual DV at T1 with sexual abuse
decreasing to lower levels over time. The decline in sexual violence over time in Group
2 is consistent with the decline observed in DV frequency, DV severity, threats of DV,
and physical DV groupings, and is supported by the empirical and theoretical literature
noted above, which suggests a decline in DV over time (Herman, 1992a, 1992b; Jasinski,
2001; Rennison, 2001).

In addition, results indicate that sexual DV is not as highly correlated with the
other indicators of abuse (e. g., the correlation between sexual DV and threats of DV at T6
was .23, the correlation between sexual DV and physical DV at T6 was .29, and the
correlation between threats of DV and physical DV at T6 was .71). This is consistent
with Marshall’s (1996) study of women in long-term heterosexual relationships that

found that the correlation between sexual DV and threats of DV was .49, and the

102

correlation between sexual DV and physical DV was .50 (as compared to the correlation
of .82 observed between threats and physical DV). Thus, sexual DV may be qualitatively
different than threats or actual physical aggression. Speciﬁcally, sexual abuse may
represent the most intimate form of abuse and is an assault on reproductive control, a
basic human function (Crittenden, 1997). The distinctive nature of sexual abuse may
partially explain why the cluster solution for sexual DV was not equivalent to those
produced by other indicators of abuse, and thus, should be considered in addition to other
indicators of abuse when trying to understand women’s experiences of DV and the
differential effects of DV based on the constellation of harmful acts sustained. This
recommendation is supported by Marshall (1996) who identiﬁed differential effects of
DV on women’s physical and mental health, help-seeking behaviors, and perceptions of
relationships depending on the type and conﬁguration of abuse sustained. For example,
sexual aggression was related to participation in psychotherapy, whereas threats and
physical violence were not. Furthermore, sexual DV contributed to women’s help-
seeking behaviors signiﬁcantly more often than physical violence.

It is also possible that the sexual DV cluster analysis did not produce a moderate-
level DV group similar to that produced with other abuse indicators because sexual abuse
occurred infrequently for the majority of women. Although 54.9% of women reported
experiencing sexual abuse at some point throughout the course of the study, the
frequency of these acts was low (the highest mean score on sexual abuse was 2.31 at T1).
Thus, sexual DV may be more common than previously believed among battered women
residing in the community, but acts of sexual aggression are infrequent for most abused

women, consistent with results obtained in previous DV research among community

103

samples (e.g., Marshall, 1996). It is important to keep in mind, however, that although
results here indicate that individual trajectories of sexual DV victimization can be
summarized by two longitudinal patterns, this solution may not generalize to shelter
samples of abused women who may experience greater frequency and severity of sexual
abuse.

Relationship between DV Victimization and CP Symptoms

Consistent with the proposed model, results suggest a signiﬁcant relationship
between DV victimization and CP symptoms. Speciﬁcally, ﬁndings indicate that mean
number of CP symptoms differed as a function of DV cluster membership, such that
women who experienced greater levels of DV (regardless of the speciﬁc indicator of
abuse) suffered from more CP symptoms as compared to women in clusters characterized
by lower levels of DV victimization. Thus, women in groups characterized by greater
DV frequency, DV severity, and threats, physical, and sexual DV reported more
symptoms as compared to women in groups characterized by lower DV frequency, DV
severity, and threats, physical, and sexual DV, respectively. Furthermore, duration of DV
was related to CP symptoms, such that the more observation periods in which women
reported abuse, the greater the number of CP symptoms.

The relationship observed between DV and CP symptoms is consistent with CP,
dynamic-maturational, and contructivist theories of trauma, which together suggest that
victims of prolonged, repeated interpersonal trauma, including DV, suffer from a
complex constellation of problems, including somatization, depression, dissociation,
negative self-perceptions, chaotic interpersonal functioning, re-victimization, and self-

inﬂicted harm—problems that exceed the re-experiencing, avoidance, and hyperarousal

104

symptoms of PTSD (Crittenden, 1997; Harter & Neimeyer, 1995; Herman, 1992a,
1992b). Speciﬁcally, Harter and Neimeyer (1995) suggest that individuals develop
templates for relationships based on experiences with an abusive partner resulting in CP
symptoms, particularly re-victimization. Crittenden (1997) argues that the victim of
repeated abuse reaches a state of learned helplessness after repeated failures to protect
oneself, which leaves the victim vulnerable to coercive control by the perpetrator. It is
this relationship of coercive control, in which the perpetrator undermines the victim’s
autonomy and gains control over the victim’s cognitions, actions, and feelings through
repeated threats of violence, actual violence, and other coercive techniques (e. g., severing
the victim’s interpersonal connections with others) that Herman (1992a; 1992b) argues
leaves victims psychologically vulnerable to CP.

According to theory, the relationship of coercive control, which ultimately breaks
down the psychological resistance of the victim leading to CP is more likely to occur
when abuse is chronic, severe, or frequent, consistent with ﬁndings from the current
study. Speciﬁcally, the development of a coercive relationship occurs over time with
repeated use of coercive techniques and therefore, is more likely to occur when abuse is
chronic. Submission to coercive control is more likely to occur when abuse is severe or
frequent because deﬁance against the perpetrator places the victim in grave danger (in the
case of severe DV) or inevitably leads to physical or psychological injury (in the case of
frequent DV). Furthermore, frequent abuse may lead to chronic arousal of the ﬁght/ﬂight
response, resulting in anxiety and somatic symptoms reﬂected in CP. Though the current

study did not directly measure the development of a coercive relationship, violence is one

105

technique used to achieve control over an individual and thus, may be one indicator of the
existence of a coercive relationship.

Results supporting a relationship between DV and CP are also consistent with
empirical studies which suggest that DV is related to a range of mental health problems
reﬂected in the CP construct (e.g., Champion et al., 2001; Coker et al., 2002; Dorahy et
al., 2007; D. G. Dutton & Painter, 1993; Lown & Vega, 2001; Olson et al., 2003; Pico-
Alfonso, 2005; Sansone et al., 2007; Stein & Kennedy, 2001; Umberson et al., 1998), as
well as to studies which have validated the CP construct among interpersonal abuse
victims in general, and particularly among CSA survivors (e.g., Dickinson et al., 1998;
Ford et al., 2006; McLean & Gallop, 2003; Pelcovitz et al., 1997; Roth et al., 1997).
Studies of CP among interpersonal abuse victims also suggest that abuse characteristics
such as frequency (McLean & Gallop, 2003), severity (Dickinson et al., 1998), and
duration of abuse (J. G. Allen et al., 1999; Dickinson etal., 1998) are related to an
increased risk for CP symptoms, consistent with ﬁndings from this study.

However, the current study adds to the existing empirical literature in several
important ways. For example, this is the ﬁrst known study to examine and ﬁnd evidence
for CP symptoms speciﬁcally among victims of DV. In addition, the current study uses a
person-centered approach summarizing women’s individual experiences of DV over time
and linking these longitudinal trajectories of DV to CF symptoms. Furthermore, in the
current study, CP was explored among a community sample of women with a range of
DV experiences in order to investigate whether women with relatively less severe and
frequent abuse as compared to women residing in battered women’s shelters would report

core features of CP.

106

Although ﬁndings suggest that greater DV was associated with more CP
symptoms as predicted, results also indicate that even women with moderate levels of DV
suffer from elevated levels of CP symptoms. Thus, CP may not occur exclusively among
individuals who experience severe levels of abuse, as might be expected based on an
understanding of current CP theory. This ﬁnding is similar to results reported in a prior
study by Ford et al. (2006), in which infrequent interpersonal trauma was associated with
elevated levels of CP symptoms among a college sample of women. Though full CP may
be rare in community samples of interpersonal abuse victims, subclinical levels of CP
may nonetheless cause signiﬁcant impairment in functioning. For example, previous
research has indicated that subclinical PTSD is related to signiﬁcant psychosocial and
physical health impairment (Schnurr & Green, 2004; Stein et al., 1996). Furthermore,
subthreshold psychiatric symptomatology has been linked to future psychiatric morbidity
(Katzelnick et al., 2001).

Contrary to expectation, the current study found that sexual DV did not
demonstrate greater speciﬁcity for CP symptoms as compared to other types of abuse
(threats and physical DV). This ﬁnding runs contrary to Crittenden’s (1997) position that
sexual abuse may have the most devastating consequences for mental health as compared
to other types of abuse because it represents the most intimate form of betrayal, as well as
to the existing empirical literature on CP, which suggests that sexual abuse is a stronger
risk factor for CP than other types of violence. Speciﬁcally, Allen et al. (1998) found
that adult women who experienced CSA had higher scores on symptoms consistent with

CP than women who did not experience CSA. Furthermore, Roth and colleagues (1997)

107

found that sexual abuse placed individuals at greater risk for developing full CP than
physical abuse alone.

However, the Allen et al. (1998) study only included victims of childhood sexual
abuse, and it is unclear in the Roth et al. (1997) study whether participants experienced
sexual abuse during childhood or adulthood. Findings in the current study may be a
function of when the sexual abuse occurred during the lifespan. It is possible that the
experience of sexual abuse in adulthood, particularly within a romantic relationship in
which sexual intimacy is appropriate, does not have as detrimental of an impact as it does
when the sexual abuse is experienced during childhood. Thus, the impact of sexual abuse
may be equivalent to other forms of aggression, such as physical abuse, when it occurs
later in life. In fact, Herman (1992a; 1992b) does not specify that sexual abuse has a
differential effect on CP as compared to other forms of prolonged interpersonal abuse,
and Marshall (1999) found that psychological abuse made a greater contribution to
women’s mental health problems than sexual DV, though CP was not assessed.
Alternatively, this ﬁnding may be a function of the infrequency with which sexual
violence occurred in the current sample; the high sexual DV group contained only ﬁve
women as the majority of participants reported a low frequency of sexual aggression.
Additional studies including more women who experience frequent sexual DV may be
needed to better understand this ﬁnding.

Effect of Partnership Status on CP Symptoms

Conceptually, DV contributes to CP via the development of a relationship of

coercive control, which the perpetrator establishes over time (Herman, 1992a, 1992b).

Thus, DV was hypothesized to be more predictive of CP symptoms for women in long-

108

term violent relationships as compared to women in short-term violent relationships in
which there is limited opportunity for a relationship of coercive control to develop. As
predicted, duration of DV among women in long-term DV relationships accounted for
more variance in CP symptoms than among women in short-term DV relationships. In
contrast, DV groupings (patterns of DV victimization) among women in short-term DV
relationships accounted for more variance in CF symptoms as compared to DV groupings
among women in long—term DV relationships. Therefore, although DV clusters were
predictive of CP for both women in long- and short-term violent relationships, the present
research seems to suggest that DV is generally a better predictor of CP among women in
short-term violent relationships. According to Herman (1992a; 1992b), the perpetrator
establishes a relationship of control leading to CP over time through a m of coercive
techniques. Thus, it is possible that in long-term relationships, other factors (not
measured here) in addition to DV, such as psychological abuse and economic and legal
subordination, contribute to CP, whereas in short-term violent relationships, it is
primarily DV that drives CP.

Another interesting ﬁnding pertains to group differences in CP symptom levels
for women in short- versus long-term DV relationships. Speciﬁcally, mean number of
CP symptoms differed signiﬁcantly in the expected direction between the High and
Minimal DV groups, but not between the Moderate and High DV groups among women
in long-term violent relationships. The pattern in means was different among women in
short-term violent relationships. For these women, mean number of CP symptoms
differed signiﬁcantly in the expected direction between the High and Moderate and the

High and Minimal DV groups, but not between the Moderate and Minimal DV groups.

109

These ﬁndings appear to suggest that DV only has to be moderate (not high) for women
in long-term violent relationships to experience an elevation in CP symptoms, whereas
the violence has to be high for women in short-term violent relationships to experience an
elevation in CP symptoms. It is possible that women involved in long-term violent
relationships may experience signiﬁcant stress symptoms even if violence is not high
because a relationship of coercive control has been established over time and the
perpetrator relies on other coercive techniques rather than violence to maintain control,
an explanation which would support current CP theory.
Social Support

Overall, results regarding social support did not support the theoretical model that
social isolation partially accounts for the relationship between DV and CP symptoms.
However, there were several interesting ﬁndings pertaining to DV and patterns in social
support, which are discussed in detail below. For example, results generally suggest a
main effect for emotional support, such that women in groups characterized by greater
DV have less emotional support averaged across time as compared to women with lower
DV. However, there was no main effect for number of supporters; thus, women in
groups characterized by greater DV did not have signiﬁcantly fewer supporters than
women in groups with less DV. Taken together, these ﬁndings suggest that the average
quality rather than quantity of social support differs according to level of DV. These
ﬁndings are consistent with theory and prior literature that suggests that DV is related to
poor social support (e. g., Levendosky et al., 2004; Mitchell & Hodson, 1983; Thompson
et al., 2000), however, they run contrary to prevailing theory and literature that suggests

that poor social support among battered women is due to social isolation (O. W. Barnett

110

et al., 1996; Herman, 1992a, 1992b; Hilberman & Munson, 1978; Mitchell & Hodson,
1983; Tan et al., 1995; Walker, 1979). This ﬁnding may be a function of the varying
severity of DV experienced by women across samples. Speciﬁcally, the majority of
studies that ﬁnd that poor social support is a function of social isolation utilize shelter
samples, which may include a large percentage of women who experience severe levels
of DV. In contrast, the current study examined abused women residing in community
settings, most of who reported mild to moderate levels of DV. This explanation is
difﬁcult to test, however, as most DV studies do not include information on the severity
of violence experienced by participants.

Alternatively, poor quality of social support may stem from abused women’s
failure to reach out to network members for support due to shame, embarrassment, fear of
perpetrator retribution, or concern about endangering the safety of family and friends—
reasons frequently cited by female participants in a qualitative study about social support
and DV for not appealing to family and friends for support (El-Bassel et al., 2001). It is
also possible that abused women request support from others in their network, but are
dissatisﬁed with the quality of support received because family and friends respond with
a lack of empathy or are unable to provide adequate support due to their own personal
circumstances. For example, results from a study on the entire sample of Mother-Infant
Study participants during pregnancy found that abused women, as compared to
nonabused women, had more network supporters who experienced DV themselves
(homophily: Levendosky et al., 2004). In addition, homophily was positively related to
disclosure of abuse and negatively associated with emotional support within the DV

group. This suggests that abused women may turn to others for support, but do not

111

receive adequate support, perhaps because supporters who are also abused may have few
psychological resources to provide quality support. Thus, impaired social support among
abused women may be best explained by homophily, rather than social isolation.

Repeated measures results regarding the effect of time were signiﬁcant for both
number of supporters and emotional support. As predicted in the proposed model,
repeated measures analyses indicate that number of supporters decreased over time
averaged across groups of abused women (regardless of the indicator of abuse used in the
analyses). In contrast, emotional support increased over time averaged across DV
frequency clusters and DV duration groups (though the effect of time was not signiﬁcant
when sexual DV clusters were deﬁned as the between-subjects factor, perhaps because of
the small sample size in the High Sexual DV group, n=5). Thus, ﬁndings regarding the
effect of time generally suggest that abused women lose supporters over time, but that
perceived quality or satisfaction with support increases. Findings regarding the effect of
time for emotional support may be best understood in light of current study ﬁndings on
longitudinal trajectories of DV, which suggest that DV declines over time for all groups
of abused women. Thus, as DV decreases over time, quality of support increases, which
is consistent with ﬁndings from previous cross-sectional studies citing an inverse
relationship between quality of social support and DV (e. g., Thompson et al., 2000).

In contrast, number of supporters decreased over time. Although this ﬁnding was
predicted in the proposed model, it is believed that the decrease in network size may be a
measurement problem, rather than a valid indicator of social isolation, particularly given
the lack of a main effect for number of supporters. The social support questionnaire

utilized in the current study required participants to answer a number of questions

112

pertaining to each network member. Thus, over time, participants may have learned that
the interview takes less time to complete if fewer supporters are identiﬁed, which may
explain the observed decrease in network size. Alternatively, it is possible that the
decrease in network members is a function of the normal developmental trajectory of
social support across the lifespan. For example, studies of age differences in social
support generally indicate a decline in quantity of available supporters with age (e. g.,
Fischer, 1982). Therefore, the decrease in network members observed in the current
study may be a function of participants aging, rather than an indicator of social isolation
by the DV perpetrator. Furthermore, even though network size may erode over time,
studies on perceived social support suggest that quality of social support received is
relatively stable with age or increases in relation to particular sources of support (e. g.,
children: Coventry, Gillespie, Heath, & Martin, 2004). Thus, the longitudinal pattern in
social support observed in the present research (i.e., decrease in number of network
members and increase in emotional support) may be explained by normative
developmental patterns in social support, rather than DV-related factors.

Number of supporters and emotional support were expected to mediate the
relationship between DV and CP symptoms; however, given the ﬁndings noted above,
social isolation does not appear to be a primary route to DV-related CP symptoms, at
least among a community sample of abused women. Thus, it is not surprising that
mediation was not supported. Although ﬁndings do not necessarily discount that social
support may be an explanatory factor for CP among women with more severe levels of
DV in which the victim may be held captive by the perpetrator, the absence of mediation

suggests that social isolation or even the perception of isolation, reﬂected in poor quality

113

of support, do not explain the development of behaviors characteristic of CP for abused
women living in the community. It may be that coercive techniques other than
disconnection from ﬁiends and family are central in these violent relationships, such as
psychological abuse (e.g., discrediting the victim) or control over economic and legal
resources. These coercive techniques may serve to undermine the victim’s autonomy,
power, and initiative, creating dependency and traumatic bonding with the perpetrator,
even though the victim is not isolated. Perhaps perceived level of dependency on the
perpetrator, rather than social isolation, may better explain the link between DV and the
interpersonal and identity disturbances captured by CP for abused women residing in the
community.

Adult Representations of Childhood Attachment and Child Maltreatment History

The present research examined two factors that were hypothesized to moderate

the DV-CP symptom link: adult representations of childhood attachment and child
maltreatment. The inclusion of both of these factors was based primarily on attachment
theory, which suggests that disruption of the early care-giving bond (reﬂected in
attachment insecurity and that often occurs in situations of child maltreatment) disrupts
critical developmental processes (e. g., emotion and behavior regulation), which leave
individuals psychologically vulnerable to traumatic experiences in adulthood, including
DV (D. Barnett et al., 1999; Bowlby, 1977; E. A. Carlson, 1998; van-der-Kolk & Fisler,
1994). Interestingly, results indicate that child maltreatment moderated the relationship
between DV clusters (regardless of the abuse characteristic used as the clustering
variable) and CP symptoms, such that child maltreatment exacerbated DV-related CP

symptoms with the greatest exacerbation among women with the highest levels of DV;

114

 

h0‘

€F§

at

at

rel
so
Gr
nta
ha\
eve
For
disc.
givir

disru

rESea
Sittia

however, attachment insecurity based on adult representations of early care-giving
experiences did not.

One possible explanation for the lack of signiﬁcant ﬁndings regarding attachment
is that adult representations of care-giving experiences may not adequately reﬂect early
attachment experiences that disrupt developmental processes due to discontinuity in
attachment over time. In fact, several longitudinal studies have found a low concordance
rate between infant attachment organization and adult attachment status based on
recollections of care-giving experiences among adolescents using the AAI and AAI Q-
sort procedure (Jones, 1996; Weinﬁeld, Sroufe, & Egeland, 2000; Zimmermann,
Grossmann, & Fremmer—Bombik, 1998), suggesting that for many individuals attachment
may not be stable from childhood through adulthood. Furthermore, additional studies
have found that attachment is less stable among individuals who experience traumatic life
events (Waters, Merrick, Treboux, Crowell, & Albersheim, 2000; Weinﬁeld et al., 2000).
For example, Weinﬁeld (2000) found that child maltreatment accounted for some
discontinuity in attachment status over time. Thus, adult representations of early care-
giving experiences (particularly in the presences of traumatic events) may not capture
disruption in infant attachment that negatively affects critical developmental processes
that create psychological vulnerability to conditions of stress, such as DV.

However, literature regarding the stability of attachment is inconsistent. Other
researchers have observed 70% to 77% correspondence rates between childhood Strange
Situation behavior and adolescent and adult attachment status with regard to care-giving
experiences, suggesting that attachment is stable over time (Benoit & Parker, 1994;

Hamilton, 2000; Waters et al., 2000). If attachment is in fact stable over time, an

115

alternative explanation may be plausible. It is possible that adult representations of
childhood attachment accurately reﬂect disruptions in early developmental processes;
however, the PAAQ may not be a good measure of adult attachment with regard to care-
giving experiences. Longitudinal studies that suggest continuity in attachment used the
AAI (e. g., Waters et al., 2000), which is the gold standard of adult attachment measures
(regarding care-giving experiences) with the greatest psychometric support. In addition,
studies cited here to support the moderation argument linking adult attachment with
regard to care-giving experiences to psychopathology (J. P. Allen et al., 1996; Cole-
Detke & Kobak, 1996; F onagy et al., 1996; Rosenstein & Horowitz, 1996) utilized the
AAI. The PAAQ was selected for the current study to increase efﬁciency due to the time
intensive nature of training interviewers, and administering and scoring the AAI.
Although the PAAQ was developed to predict adult attachment status with regard to care-
giving as assessed by the AAI with signiﬁcant correlations found between 4 of the 6
overlapping subscales, and preliminary research suggests that the PAAQ has adequate
validity and reliability (Lichtenstein & Cassidy, 1991), the PAAQ has rarely been used;
therefore, its psychometric properties have not been investigated thoroughly. In addition,
a clinical cutoff for insecure attachment on the PAAQ has not been established in the
prior literature.

Furthermore, studies comparing self-identiﬁed adult attachment based on care-
giving experiences to AAI status have not found signiﬁcant associations (Crowell,
Treboux, & Waters, 1999; Hesse, 1999). For example, Hesse (1999) found that
participants who were classiﬁed as insecure according to the AAI were likely to identify

themselves as secure on questionnaire measures, and individuals who were secure on the

116

AAI were likely to identify themselves as insecure. F urtherrnore, (Hamilton, 1995)
found no relation between self-reported adult attachment based on care-giving
experiences and infant attachment among adolescents using the Inventory of Peer and
Parent Attachment (IPPA: Arrnsden & Greenberg, 1987), but in another study found that
infant strange situation behavior was related to AAI status (Hamilton, 2000). Thus, adult
representations of early care-giving experiences may correspond to infant attachment, but
not be satisfactorily identiﬁed through self-report measures. Unlike the AAI which
assesses attachment based on coherence of discourse rather than content, self-report
measures of adult attachment with respect to care-giving experiences, such as the PAAQ
are subject to deﬁcits in autobiographical memory about early relationship experiences
with the primary caregiver, as well as social desirability bias. Finally, women may report
on attachment experiences more favorably as a way to avoid negative feelings associated
with acknowledgment that early care-giving experiences were poor.

However, it is alternatively possible that the PAAQ adequately measures
disrupted attachment, but the clinical cutoff for insecure attachment utilized in the current
study may not have accurately captured women with severe enough attachment
disturbances that would be indicative of disruption in developmental processes believed
to contribute to psychological vulnerability to traumatic experiences. In contrast, it is
arguable that child maltreatment, which is an extreme circumstance, is a more sensitive
measure of disrupted care-giving experiences that would fail to provide opportunities for
normal development. If this is the case, the ﬁnding that child maltreatment moderated
the DV-CP symptom link is consistent with theoretical literature suggesting that

disruptions in early developmental processes of emotion and behavior regulation

117

stemming from disrupted attachment lead to psychological vulnerability to later
conditions of risk (e.g., Bowlby, 1977).

It is also possible that the assessment and cutoff selected for attachment in the
current study was not problematic, but rather that attachment insecurity alone is simply
not a strong enough risk factor to be related to CP symptoms, which are suggestive of
severe psychological problems. Child maltreatment however, is a strong enough risk
factor to contribute to CF because it is a more severe experience than disrupted
attachment alone. The experience of childhood abuse is often indicative of a very
disturbed childhood, which is often associated with insecure attachment as well as a host
of other risk factors for psychopathology (e. g., poverty, parental substance abuse,
parental psychopathology, domestic violence: see Zielinski & Bradshaw, 2006 for
review), which is not necessarily the case with insecure attachment in and of itself. Thus,
it may be the combination of childhood maltreatment and other risk factors associated
with the environment within which childhood maltreatment occurs that increases
vulnerability to CP in response to DV. This explanation is consistent with Gold’s (2000)
understanding of the impact of child maltreatment. Speciﬁcally, Gold (2000) proposes
that it is not child maltreatment alone, but rather the combination of child maltreatment
and contextual factors (though not measured here) associated with early abuse that
contribute to psychological vulnerability in adulthood.

Furthermore, Gold (2000) suggests that poor emotion regulation, which increases
vulnerability for mental health problems in response to conditions of risk (e. g., DV), may
stem from the deﬁcient learning environment associated with child maltreatment. He

argues that many individuals who experience child maltreatment are raised in ineffective

118

families with caregivers who do not teach effective skills for living. Thus, poor emotion
regulation that increases vulnerability to future trauma may result from the failure to
learn adaptive coping strategies for modulating emotions in one’s family of origin.
Therefore, the ﬁnding that child maltreatment exacerbates DV-related CP symptoms may
be due the absence of opportunities to learn effective strategies for regulating emotions,
rather than due to problematic attachment.

Although the exact explanation as to why child maltreatment moderates the DV-
CP symptom link may not yet be clear (e. g., is moderation the result of disrupted
developmental processes due to insecure attachment, due to the cumulative impact of risk
factors associated with child maltreatment, or related to the inability to learn emotion
regulation skills because of the deﬁcient family environment associated with child
maltreatment), moderation is consistent with several previous studies linking child
maltreatment and CP in adulthood (J. G. Allen etal., 1998; Dickinson et al., 1998;
McLean & Gallop, 2003; Pelcovitz et al., 1997; van der Kolk et al., 1996).

Summary

Overall, ﬁndings from the current study provide partial support for the proposed
model. Speciﬁcally, results support the theoretical link between DV and CP with level
and type of abuse differentially affecting CP, and suggest that child maltreatment history
moderates the relationship between DV and CP. Thus, this study is an important ﬁrst
step toward establishing an empirical link between experiences of DV and CP, and
uncovering a personal vulnerability factor for experiencing DV—related CP symptoms.
However, DV appears only to be a risk factor for women who experience moderate to

high levels of DV. Women who reported mild levels of DV across time (Minimal DV

119

group) reported few CP symptoms, suggesting that this group of women may be similar
to women who do not experience DV in that they do not appear to be at increased risk for
developing signiﬁcant posttraumatic sequelae.

Although an association was generally found between DV and an elevation in CP
symptoms, it is clear that the mechanisms for developing CP in response to DV are not
fully understood as poor social support was not explained by social isolation, and social
support did not mediate the association between DV and CP. Thus, explanatory factors
for the development of CP symptoms in response to DV require further theoretical and
empirical consideration. Nonetheless, ﬁndings do not discount prevailing trauma theory
and literature, but rather suggest that such theories may not apply to all abused women,
particularly those living in community settings with mild to moderate levels of DV.

Study Strengths

This study had several strengths. Speciﬁcally, this study combined person- and
variable-centered analytic strategies to classify women’s individual experiences of DV
victimization over a six-year period and to explore the relationship between these patterns
and CP symptoms. Although prior studies have examined CP among victims of repeated,
interpersonal trauma, some of whom may have experienced DV (e.g., Pelcovitz et al.,
1997), this is the ﬁrst known study to assess CP speciﬁcally in victims of DV. In
addition, prior studies have relied on retrospective reports of abuse, whereas the current
study assessed DV prospectively, eliminating the potential for recall bias. This is also the
ﬁrst known study to test Herman’s (1992a; 1992b) assertion that social isolation is a
primary route to developing CP. Again, although this idea should not be discounted,

current ﬁndings suggest that it may not be applicable to women who experience mild to

120

moderate levels of DV. Finally, the cmrent research was a ﬁrst step toward identifying
personal vulnerability factors that may exacerbate CP symptoms related to prolonged,
interpersonal trauma.
Study Limitations and Future Research Implications

Although this research has several important ﬁndings, results must be interpreted
in light of study limitations with future research addressing such limitations. For
example, the current study did not statistically control for the potential effects of other
traumas on CP. Prior literature has suggested that CP demonstrates greater speciﬁcity for
chronic interpersonal trauma as compared to single-incident, non-interpersonal traumas,
such as a natural disaster (Pelcovitz et al., 1997); however, participants may have
experienced other types of interpersonal trauma that were not measured (e.g., adult sexual
assault by a non-romantic partner) that may contribute to particular symptoms of CP. In
an attempt to partially control for the effects of other traumas, participants were asked to
endorse CP symptoms related speciﬁcally to experiences of DV, but women who
experience multiple interpersonal traumas may not be able to separate the effects of
speciﬁc traumas. For example, a woman with lifetime experiences of DV and adult
sexual assault by a non-romantic partner may be unable to distinguish whether the DV,
sexual assault, or both contributed to depressive affect or chronic physiological arousal
resulting in somatization. In future studies, DV researchers should consider assessing
and parsing out the effects of other interpersonal traumas on CP, or alternatively testing a
cumulative stress model of CP, particularly given that trauma survivors are often subject

to repeated victimization.

121

Another weakness of the current study was the cross-sectional measurement of
CP symptoms. In contrast to DV, which was measured longitudinally, CP was only
measured at the end of the study period (T6). Thus, the immediate effects of DV on CP
symptomatology could not be assessed, and women who may have experienced CP
symptoms earlier in the study period, but not at T6, were not identiﬁed. Prospective,
longitudinal measurement of CP symptoms would help researchers identify patterns in
CF in relation to changes in DV.

Furthermore, it is possible that DV-related CP symptoms ﬂuctuate in relation to
other life stressors or mental health intervention. The inclusion of additional risk factors
or factors restoring mental health in future studies may serve to clarify the relationship
between DV and CP and contribute to the further development of a theoretical literature
on the development of CP among women who experience DV.

Finally, the sample size in the current study was relatively small, which may have
precluded the identiﬁcation of relationships among study variables, particularly with
regard to sexual abuse. Future studies should include larger sample sizes to increase
power and allow a latent variable approach to testing models for the development of DV-
related CP. A latent variable approach would permit the inclusion of multiple indicators
to reduce the bias of measurement error, as well as allow researchers to include all
relevant factors in one model to examine the interaction among variables. A larger
sample size would also allow researchers to include all types of abuse, including
psychological abuse and other methods of coercion in one cluster analysis, rather than

multiple separate cluster analyses. This would permit researchers to more precisely

122

 

def

abu

lim
ﬁnc
suci
exp
and
inve
isolz
Hen
seve
for tl
COntr
help
to ta,

Clim'

Provj

 

inte
dlag

Chara

 

deﬁne conﬁgurations of abuse and identify the differential impact of such patterns of
abuse on CP.

In addition to future research implications noted above in relation to study
limitations, studies with other samples of abused women are needed to replicate the
ﬁnding that DV is related to CP, particularly given that this is the ﬁrst study to support
such a link. As the relationship between DV and CP is further supported empirically,
explanatory factors for the development of CP in response to DV need to be considered
and tested. Although social isolation was not found to be a mediating variable in this
investigation, researchers should examine the potential explanatory role of social
isolation in studies that include women with a broader range of DV experiences as
Herman’s (1992a; 1992b) theory may be more applicable to women who experience
severe DV. Finally, future studies should include additional risk and protective factors
for the development of CP among abused women. This type of research would
contribute to the development of effective prevention and intervention programs, and
help mental health professionals identify individuals who may be at greatest risk for CP
to target for such programs.

Clinical Implications

The current study provides empirical evidence for a link between DV and CP,
providing additional cause for challenging our current view of posttraumatic stress
reactions in order to facilitate an improved understanding of the psychological impact of
interpersonal trauma, including DV. Currently, clinicians must rely on established
diagnostic concepts, such as personality disorders, to describe the complex symptom and

character presentations of women with trauma histories, including battered women.

123

 

 

According to the DSM-I V, the application of personality disorders to trauma victims is
not incorrect; victims may meet symptom criteria as they are currently deﬁned, especially
as constriction and negative symptoms are predominant in the clinical presentation.

However, the use of personality disorders to describe trauma victims may result in
the failure to recognize the traumatic origins of symptoms, the failure to understand the
full impact of victimization on the trauma victim, and the mistake of attributing
symptoms to enduring personality traits (Herman, 1992a). The risk of misunderstanding
may be particularly high among the group of interpersonal trauma survivors who no
longer meet criteria for PTSD, but exhibit other trauma-related sequelae captured by CP
(see Dickinson et al., 1998 for a complete description of this proﬁle of women). In these
cases, the link between symptoms and traumatic origins may be easily lost with the
current classiﬁcation system.

In addition to the risk of misunderstanding the experience and underlying
pathology of trauma victims, the application of personality disorders may serve to further
stigmatize a group of women who may already experience a great deal of embarrassment,
shame, and self-blame. Several personality diagnoses have acquired a pejorative tone
among clinicians; borderline personality disorder, a diagnosis that commonly overlaps
with CP (McLean & Gallop, 2003), is particularly uncomplimentary and laden with
negative connotations.

The diagnostic picture for victims of interpersonal abuse is further complicated by
the fact that clinicians are often forced to diagnose several co-occurring Axis I and Axis
11 conditions in an attempt to capture the multiple and pervasive disturbances frequently

present among trauma victims. Multiple diagnoses may muddle the conceptualization,

124

ﬁrrther misrepresenting the underlying pathology, and add to the ﬂawed sense of self
often observed among interpersonal abuse victims.

Misunderstanding the experience of interpersonal abuse victims and the
consequent assignment of co-morbid conditions has signiﬁcant implications for treatment
planning and success. Some may argue that the issue is one of labeling—what names we
assign to particular mental health disturbances. However, it would be naive to assume
that labels serve only a descriptive and communicative function in a profession that is
becoming increasing dominated by empirically-validated, manualized treatments
designed for particular disorders. The current classiﬁcation system, without the
recognition of CP, contributes to the application of multiple diagnostic constructs, which
may not recognize the underlying link between trauma and symptoms, and thus, may lead
to inadequate treatment planning and transient improvement among trauma victims. The
CP construct provides clinically meaningﬁrl information, which may aid in treatment
planning.

Given the signiﬁcant degree of diagnostic overlap between personality disorders
and CP among victims of interpersonal abuse (e.g., McLean & Gallop, 2003), it may be
clinically useful to separate this group of trauma victims from Axis II psychopathology
and subsume them under CP, a diagnosis which combines state and trait
psychopathology. This would eliminate the problems of stigma, co-morbidity,
misunderstanding (if not misdiagnosis), and possible mistreatment for this particular
group.

Recognition of symptoms common among abuse victims and knowledge

regarding the range of human responses to trauma and exploitation would improve

125

treatment planning. The assessment of CP would help clinicians identify areas of
psychological impairment that are essential for treatment planning and may cut across
current diagnostic boundaries. Once clinicians understand the totality of the impact of
trauma on psychological and interpersonal functioning, they can work with victims to

,9 ‘6

“reconnect fragments, reconstruct history,” and “make meaning of present symptoms in
the light of past events” (Herman, 1992a, p. 3). Thus, given the association observed in
the present research between CP and DV, mental health professionals can begin to
develop a greater understanding of the full impact of DV on women, placing them in a
better position to develop effective treatments with long-lasting success.

Ide and Paez (2000) advocate for a multidimensional and integrative treatment
approach for victims with CP, which now appears to be relevant to battered women. This
approach to treatment involves a variety of tactics, including social skills training,
behavior and affect management, alcohol and substance use rehabilitation, and improving
family functioning, in addition to the exploration of traumatic themes. Inpatient
treatment may be indicated to provide adequate safety for victims with suicidal ideation
or self-harm tendencies prior to outpatient treatment (Ide & Paez, 2000).

Although social support was not found to mediate the relationship between DV
and CP, frequency of DV was related to poor social support. Therefore, it is appropriate
for practitioners to target social support in treatment for abused women by incorporating
strategies to enhance the quality of informal social support. Creating an environment of
support rather than blame will empower women whose sense of control and initiative

have been depleted, and communicate a societal commitment to zero tolerance for partner

abuse. Prevention programs that facilitate the strengthening of social ties for women at

126

risk for DV (e. g., women with a history of childhood maltreatment) may be beneﬁcial.
These efforts could potentially reduce the likelihood of involvement in an abusive
relationship characterized by coercive control.

A comprehensive treatment program for battered woman should directly address
post-trauma psychOpathology as well as factors that play an important role in the
relationship between DV and mental health problems, such as child maltreatment.
Consequently, a thorough treatment plan for an abused woman should include an
assessment of childhood maltreatment history and exploration of traumatic themes related
to childhood abuse if indicated, in addition to those related to DV. Ideally, victims of
child maltreatment should be identiﬁed and targeted for early intervention in order to

reduce vulnerability to CP.

127

Table 1

Number of Participants Missing Variable-Level Data by Study Variable and Time Period

 

 

Variable T1 T2 T3 T4 T5 T6
DV Frequency 0 5 5 9 10 13
DV Severity 0 5 5 9 10 13
Threats of DV 0 5 5 9 10 13
Physical DV 0 5 5 9 1 0 1 3
Sexual DV 0 5 5 9 10 13
Child Maltreatment 17
CP Symptoms 17
No. Supporters l 5 5 9 10 13
Emotional Support 1 5 5 9 10 1 3
Long/Short—term DV 6
Partner

Income 1

 

Note. There was no missing data for study variables not included in the table. Blank
cells indicate that data for study variable was not collected at that particular time period

and therefore, missingness is not applicable.

128

 

 

N_.m Q..— NoTo vmdmé 84. $4 bto>om >Q 8.
who mN.N N26 3.3.: m2. Nm.~ bto>om >Q mu.
mm.m ca; N26 5.36 3.0 gm bigom >Q E.
05.x omd N26 cadmé 8.x Rd bigom >Q E.
5.2 mmé ~26 95.3.9 02: one £55m >0 NH
swam om._m wmmé node—6 swam om._m bto>om >Q E.
wmw vmd onmé $6 Worm cod 55:3.“— >D E.
m: _ m; 33 was :2: :1. 55:3; >o we
who mm.m emmé 36 No.2 mo;o moses—cor; >Q vb
Ed 8% £3 to on: own sconce". >a E
3.2 9; cnmé 2:5 002 S; caucus—coun— >Q NH
3.34 3.2 Evé 37o 3.3 mm.mm 5:262”— >Q Tr
AcocmEzmobav Acoszzmobav owcmm owcmm Acorn—:38 $83 323653-303

033:3

QM. :32 o_£mmom 3334. QM. 5.22

 

921$ 33: 3339880 S $35.8; mzoantnob S\SaQ 9:52.9me

N 2an

129

 

 

mom 5N v2 «.3 m2 :3 >Q Ream :-
Ea :2 £3 85 8a 3. >9 Rumba E
v: R; 3:. 9; m3. 8; >2 3%.: 3
win a. 8:. h To NE a. >9 .8755 E-
»? SN 82. 21. 3.0 gm >0 .833; ﬂ
o: 3m £3 85 z: ”2 >3 .8553 S
8.2 ”we a To :5 8.2 ”v.2 >o. samba :
mum m _ .N E I. 3.0 26 92. >o go 38:: 8
$0 m2 3 To $5 m3 0: >9 he 38:: B
one am 3 To 3; go SN >o ho $2.: E
N3 3m 3 To 25 m: w; >96 emeﬁ ﬂ
we.” N3 3 To 35 2.” m3 >o E 385 E
3.: 3.2 :2 :33 $2 $2 >o mo 33:: E
Acoumﬁtmo-oav 323550-23 owcmm owcmm $55650 -303 Acouaﬁcmo-amoav

03%;

a. :82 ”3:8“. .32 mm :32

 

.9253 N 03$

130

 

32.—235

 

:3” 3.0 vmé mm; 56 mad 00 85:52 2.
automazm
mm.m 86 vmé w T. Nag.” 2.0 .«o “3:52 3‘
£88.55
omé ow.» vmé vm-m 3; 3.x 0o 39:52 C.
ﬂotoaasm
cod 0 _ .0. vmé _m-o wwd an no .3532 NH
muotoqum
w; 2.: «Né «m6 ta 3.: 00 59:52 C.
ow. mo. omé 9o co. no. >0 338m oh
em. 2. cmé m6 3. 2. >0 338m mrr
mm; mm. 3.6 o-o vm._ vm. >0 :3me E‘
20 mm. omé w-o m: cm. >0 Esxom 2‘
mm.~ xv. cmé h To OWN cw. >0 :3me NF
€23§ﬁ923 Acoszcmoéav owcmm owcmz €238:va -303 AcotmEBmoémoav

038:;

QM :82 03580 _~Bo< 0m :82

 

.5253 m 2an

131

 

 

M82 2? M: 3 SN 8: >9 8 .8225
on: :w 3 878 3-3. 8:: 3.; 88888282 2:6
83 88 3-8 2-8 2; 5.8 2.882% mu
3. 3% 8.8 3: M8. 3% 228% 8285 E
No 83. 8-8 :8. mm. mom 8&5 .2285 B
Q. 88 «.8 if Q. 9;. 2285 8:285 E
Q. mom 8.8 if 3. 83 228% .8285 E
mm. m3. 8.8 8.8 mm. 2% 228.8 8:285 S
3. an E 382 a» N3 228% 8:285 ;
ﬂotoamzm
SM 38 2-8 E; 2X 88 8 .2882 E
AcotaEcmoéav 928858-23 owcmm owcmm 0532250 -603 Acowwﬁcmoﬁoav

2828>

QM :82 03680 .mEo< QM :32

 

03:53 N DEN—L

132

Table 3

Cluster Mean Proﬁles for DV Frequency

 

 

DV Frequency Cluster T1 T2 T3 T4 T5 T6

Cluster 1: Minimal DV 7.78 1.91 2.56 1.35 2.35 3.17
Cluster 2: Moderate DV 53.47 12.77 9.63 6.82 6.86 2.66
Cluster 3: High DV 138.20 20.08 5.92 9.31 1.77 4.15

 

133

 

|l

 

Table 4

Cross-tabulation ofDVFrequency and DVSeverity Cluster Solutions

 

 

 

DV Frequency Cluster
Cluster 1 Cluster 2 Cluster 3
DV Severity Cluster
Minimum DV Moderate DV High DV
Cluster 1: Minimum DV 89 24 0
Cluster 2: Moderate DV 0 38 5
Cluster 3: High DV 0 0 8

 

Note. Percentage of overlap between DV Frequency and DV Severity cluster solutions =

82%.

134

Table 5

C ross-tabulation of DV Frequency and Threats of DV Cluster Solutions

 

 

 

DV Frequency Cluster
Cluster 1 Cluster 2 Cluster 3
Threats of DV Cluster
Minimum DV Moderate DV High DV
Cluster 1: Minimum DV 85 11 0
Cluster 2: Moderate DV 4 47 1
Cluster 3: High DV 0 4 12

 

Note. Percentage of overlap between DV Frequency and Threats of DV cluster solutions

= 88%.

135

Table 6

Cross-tabulation of DV Frequency and Physical DV Cluster Solutions

 

 

 

DV Frequency Cluster
Cluster 1 Cluster 2 Cluster 3
Physical DV Cluster
Minimum DV Moderate DV High DV
Cluster 1: Minimum DV 89 17 0
Cluster 2: Moderate DV 0 41 3
Cluster 3: High DV 0 4 10

 

Note. Percentage of overlap between DV Frequency and Physical DV cluster solutions =

85%.

136

Table 7

Cross-tabulation of DV Severity and Threats of DV Cluster Solutions

 

 

 

DV Severity Cluster
Cluster 1 Cluster 2 Cluster 3
Threats of DV Cluster
Minimum DV Moderate DV High DV
Cluster 1: Minimum DV 91 5 0
Cluster 2: Moderate DV 22 29 1
Cluster 3: High DV 0 9 7

 

Note. Percentage of overlap between DV Severity and Threats of DV cluster solutions =

77%.

137

Table 8

Cross-tabulation of DVSeverity and Physical DV Cluster Solutions

 

 

 

DV Severity Cluster
Cluster 1 Cluster 2 Cluster 3
Physical DV Cluster
Minimum DV Moderate DV High DV
Cluster 1: Minimum DV 106 0 0
Cluster 2: Moderate DV 7 37 0
Cluster 3: High DV 0 6 8

 

Note. Percentage of overlap between DV Severity and Physical DV cluster solutions =

92%.

138

Table 9

Cross-tabulation of Threats of DV and Physical DV Cluster Solutions

 

 

 

Threats of DV Cluster
Cluster 1 Cluster 2 Cluster 3
Physical DV Cluster
Minimum DV Moderate DV High DV
Cluster 1: Minimum DV 87 19 0
Cluster 2: Moderate DV 9 29 6
Cluster 3: High DV 0 4 10

 

Note. Percentage of overlap between Threats of DV and Physical DV cluster solutions =

77%.

139

Table 10

Cluster Mean Proﬁles for Sexual DV

 

 

Sexual DV Cluster T1 T2 T3 T4 T5 T6
Cluster 1: Minimal Sexual DV 1.88 .30 .22 .08 .10 .08
Cluster 2: High Sexual DV 16.00 5.60 1.60 5.20 O 0

 

140

Table l l

Cluster Mean Proﬁles for DV Frequency among Women in Long-term DV Relationships

 

DV Frequency Cluster T1 T2 T3 T4 T5 T6

 

Cluster 1: Minimal DV 8.32 2.21 3.34 1.75 2.71 3.69
Cluster 2: Moderate DV 45.59 7.74 3.50 7.24 7.91 1.62

Cluster 3: High DV 112.80 15.45 15.77 9.09 6.41 6.36

 

141

Table 12
Cross-tabulation of DV Frequency and DVSeverity Cluster Solutions among Women in

Long-term DV Relationships

 

 

 

DV Frequency Cluster
Cluster 1 Cluster 2 Cluster 3
DV Severity Cluster
Minimum DV Moderate DV High DV
Cluster 1: Minimum DV 75 18 0
Cluster 2: Moderate DV 1 16 14
Cluster 3: High DV 0 0 8

 

Note. Percentage of overlap between DV Frequency and DV Severity cluster solutions =

75%.

142

Table 13
Cross-tabulation of DVF requency and Threats of DV Cluster Solutions among Women in

Long-term DV Relationships

 

 

 

DV Frequency Cluster
Cluster 1 Cluster 2 Cluster 3
Threats of DV Cluster
Minimum DV Moderate DV High DV
Cluster 1: Minimum DV 71 3 0
Cluster 2: Moderate DV 5 31 8
Cluster 3: High DV 0 0 14

 

Note. Percentage of overlap between DV Frequency and Threats of DV cluster solutions

= 88%.

143

Table 14
Cross-tabulation of DV Frequency and Physical DV Cluster Solutions among Women in

Long-term DV Relationships

 

 

 

DV Frequency Cluster
Cluster 1 Cluster 2 Cluster 3
Physical DV Cluster
Minimum DV Moderate DV High DV
Cluster 1: Minimum DV 76 13 1
Cluster 2: Moderate DV 0 21 6
Cluster 3: High DV 0 0 15

 

Note. Percentage of overlap between DV Frequency and Physical DV cluster solutions =

85%.

144

Table 15
C ross-tabulation of DVSeverity and Threats of D V Cluster Solutions among Women in

Long-term D V Relationships

 

DV Severity Cluster

 

 

Cluster 1 Cluster 2 Cluster 3
Threats of DV Cluster
Minimum DV Moderate DV High DV
Cluster 1: Minimum DV 74 0 0
Cluster 2: Moderate DV 19 24 1
Cluster 3: High DV 0 7 7

 

Note. Percentage of overlap between DV Severity and Threats of DV cluster solutions =

80%.

145

Table 16
Cross-tabulation of DVSeverity and Physical DV Cluster Solutions among Women in

Long-term DV Relationships

 

 

 

DV Severity Cluster
Cluster 1 Cluster 2 Cluster 3
Physical DV Cluster
Minimum DV Moderate DV High DV
Cluster 1: Minimum DV 88 2 O
Cluster 2: Moderate DV 5 22 0
Cluster 3: High DV 0 7 8

 

Note. Percentage of overlap between DV Severity and Physical DV cluster solutions =

89%.

146

Table 17
C ross-tabulation of Threats of DV and Physical DV Cluster Solutions among Women in

Long-term D V Relationships

 

 

 

Threats of DV Cluster
Cluster 1 Cluster 2 Cluster 3
Physical DV Cluster
Minimum DV Moderate DV High DV
Cluster 1: Minimum DV 72 17 1
Cluster 2: Moderate DV 2 21 4
Cluster 3: High DV 0 6 9

 

Note. Percentage of overlap between Threats of DV and Physical DV cluster solutions =

77%.

147

Table 18

Cluster Mean Proﬁles for Sexual DV among Women in Long-term DVRelationships

 

Sexual DV Cluster T1 T2 T3 T4 T5 T6

 

Cluster 1: Minimal Sexual DV 1.82 .07 .27 .10 .11 .09

Cluster 2: High Sexual DV 16.00 5.60 1.60 5.20 O 0

 

148

Table 19

Cluster Mean Proﬁles for DV Frequency among Women in Short-term DV Relationships

 

DV Frequency Cluster T1 T2 T3 T4 T5 T6

 

Cluster 1: Minimal DV 8.40 1.20 9.00 .20 .73 .13
Cluster 2: Moderate DV 49.77 11.31 3.23 4.92 2.31 1.85

Cluster 3: High DV 91.75 71.75 .25 4.50 0 0

 

149

Table 20

Cluster Mean Proﬁles/or Physical DV among Women in Short-term DVRelationships

 

 

Physical DV Cluster T1 T2 T3 T4 T5 T6
Cluster 1: Minimal Physical DV 3.05 .47 3.90 0 .21 .47
Cluster 2: Early Moderate-High

Physical DV 21.20 33.40 0 2.40 0 0
Cluster 3: Early High DV 27.88 1.75 2.25 .50 .75 0

 

150

Table 21

C ross-tabulation of DV Severity and Threats of DV Cluster Solutions among Women in

Short-term DV Relationships

 

DV Severity Cluster

 

 

Cluster 1 Cluster 2 Cluster 3
Threats of DV Cluster
Minimum DV Moderate DV High DV
Cluster 1: Minimum DV 16 0 l
Cluster 2: Moderate DV 3 2 0
Cluster 3: High DV 3 0 7

 

Note. Percentage of overlap between DV Severity and Threats of DV cluster solutions =

78%.

151

Table 22
Cross-tabulation of DVSeverity and Physical DV Cluster Solutions among Women in

Short-term D V Relationships

 

 

 

DV Severity Cluster
Cluster 1 Cluster 2 Cluster 3
Physical DV Cluster
Minimum DV Moderate DV High DV
Cluster 1: Minimum DV l9 0 0
Cluster 2: Moderate DV 2 2 1
Cluster 3: High DV 1 0 7

 

Note. Percentage of overlap between DV Severity and Physical DV cluster solutions =

88%.

152

ll

Table 23
Cross-tabulation of Threats of DV and Physical DV Cluster Solutions among Women in

Short-term D V Relationships

 

 

 

Threats of DV Cluster
Cluster 1 Cluster 2 Cluster 3
Physical DV Cluster
Minimum DV Moderate DV High DV
Cluster 1: Minimum DV 16 1 2
Cluster 2: Moderate DV 0 4 1
Cluster 3: High DV 1 0 7

 

Note. Percentage of overlap between Threats of DV and Physical DV cluster solutions =

84%.

153

Table 24

Cluster Mean Proﬁles for Sexual DV among Women in Short-term DV Relationships

 

Sexual DV Cluster T1 T2 T3 T4 T5 T6

 

Cluster 1: Minimal Sexual DV 2.07 .10 .03 0 .06 O

Cluster 2: Early High Sexual DV 2.67 11.67 0 0 0 0

 

154

Table 25

Means and Standard Deviations for Number of Network Supporters by DVFrequency

 

 

 

Cluster
Cluster 1: Minimal DV Cluster 2: Moderate DV Cluster 3: High DV

Time

Period Mean SD Mean SD Mean SD
T1 11.70 5.79 10.60 4.33 10.00 4.00
T2 7.74 3.82 6.34 3.32 7.77 5.93
T3 8.74 4.50 8.00 4.17 8.15 5.43
T4 6.39 3.28 5.74 3.09 5.85 3.41
T5 6.80 3.87 5.07 2.13 7.77 5.69
T6 7.45 4.32 6.05 3.09 6.00 3.06

 

155

 

Table 26

Means and Standard Deviations for Number of Network Supporters by Sexual DV Cluster

 

 

Cluster 1: Minimal Sexual DV Cluster 2: High Sexual DV
Time Period Mean SD Mean SD
T1 11.21 5.21 9.00 3.08
T2 7.22 3.91 7.00 3.16
T3 8.43 4.49 8.00 3.16
T4 6.09 3.24 6.40 2.61
T5 6.19 3.65 7.20 1.79
T6 6.82 3.89 6.40 2.51

 

156

Table 27

Means and Standard Deviations for Number of Network Supporters by DV Duration
Group

 

Group] Group 2 Group 3 Group 4

(1-2 Time Periods) (3-4 Time Periods) (5-6 Time Periods) (7-8 Time Periods)

 

Time Mean SD Mean SD Mean SD Mean SD
Period

 

T1 11.84 5.49 11.72 5.85 10.26 4.37 10.78 4.72
T2 7.58 3.51 7.02 3.81 7.43 4.40 6.52 3.44
T3 8.68 4.17 8.60 4.55 8.47 4.93 7.44 3.50
T4 6.58 3.56 6.06 3.49 6.06 3.00 5.52 2.48
T5 7.26 3.86 6.28 3.84 6.06 3.66 4.74 1.71
T6 7.74 4.44 7.14 4.63 5.96 2.76 6.48 2.71

 

157

Table 28

Means and Standard Deviations for Emotional Support by DV Frequency Cluster

 

Cluster 1: Minimal DV Cluster 2: Moderate DV Cluster 3: High DV

 

 

Time Mean SD Mean SD Mean SD
Period

T] 3.29 .44 3.14 .54 3.16 .54
T2 342 .45 3.29 .60 3.25 .53
T3 3.42 .41 3.35 .42 3.32 .64
T4 3.54 .35 3.45 .49 3.35 .52
T5 3.45 .46 3.38 .51 2.97 .86
T6 3.47 .47 3.42 .49 3.35 .51

 

158

Table 29

Means and Standard Deviations for Emotional Support by Sexual DV Cluster

 

Cluster 1: Minimal Sexual DV Cluster 2: High Sexual DV

 

 

Time Period Mean SD Mean SD
Tl 3.23 .48 3.01 .67
T2 3.36 .52 3.12 .68
T3 3.40 .41 2.83 .72
T4 3.50 .42 3.28 .66
T5 3.40 .53 2.96 .39
T6 3.45 .48 3.14 .56

 

159

Table 30

Means and Standard Deviations for Emotional Support by DVDuration Group

 

Group] Group 2 Group 3 Group 4

(1-2 Time Periods) (3-4 Time Periods) (5-6 Time Periods) (7-8 Time Periods)

 

Time Mean SD Mean SD Mean SD Mean SD
Period

 

T1 3.29 .46 3.21 .51 3.20 .52 3.19 .44
T2 3.50 .44 3.41 .39 3.31 .47 3.08 .83
T3 3.46 .41 3.44 .33 3.35 .50 3.22 .45
T4 3.56 .37 3.59 .36 3.42 .44 3.33 .54
T5 3.52 .45 3.50 .38 3.29 .56 3.14 .74
T6 3.57 .40 3.41 .52 3.47 .38 3.23 .63

 

160

Table 31

Complex PTSD Symptom Scores as a function of DV Frequency Cluster and Attachment

 

 

 

Status

Secure Attachment Insecure Attachment
DV Frequency Cluster Mean SD Mean SD
Cluster 1: Minimal DV 3.54 7.09 4.50 7.18
Cluster 2: Moderate DV 6.94 6.87 7.39 8.72
Cluster 3: High DV 14.86 14.30 17.33 10.98

 

161

Table 32

Complex PTSD Symptom Scores as a function of Sexual DV Cluster and Attachment

 

 

 

Status

Secure Attachment Insecure Attachment
Sexual DV Cluster Mean SD Mean SD
Cluster 1: Minimal Sexual DV 5.04 7.66 6.96 8.86
Cluster 2: High Sexual DV 17.33 15.37 16.50 13.44

 

162

. >0 00 08:03 00080.0 E 08899? 00 00 208020000 05 .80 0on 1003000020. .~ 0.5ME

toaazm _0_0om

 

manExm 95.005000 0%
$0.0m .003. 620050
2002:3259. A r £80100. 0
003800 .bt0>0m >0

 

080$

20.500202
EEO

 

 

   
       

  

E08582

goat—EU mo
8208:8053.

:32

163

 

 

 

 

 

 

 

3
*— ‘“— Group3
"""" Group2
8 _ '''' Group1
O
S .—
5‘
C O _
0) co
3
U
2
U- o
> <0 ‘
O
o _
v
o _
N
O\ 8
0- .ﬁ. ' O o o. --
I I I I I I
1 2 3 4 5 6
TIme Period

Figure 2. Cluster mean proﬁles for DV Frequency.

164

 

 

 

 

 

 

 

° —_ Group2
3' """" Group1
m o_
m ‘-
3
.0
<
75
3
X
a)
(D
o
m_ 0
..~
0 .‘............~........
I l l I l I
1 2 3 4 5 6

Two Period

Figure 3. Cluster mean proﬁles for Sexual DV.

165

 

 

 

 

 

 

 

— Group3
O """" Group2
g — """ Group1
o a
co
>
0
C
3
a 8 -
Ll.
>
o
o _
v
a a
‘ °——°\
..._.,8NQ._';_..
1 2 3 4 5 6

Tune Period

Figure 4. Cluster mean proﬁles for DV Frequency among women in long-term DV

relationships.

166

 

 

 

 

 

 

° — Group2
53 ‘ """ Group1
3
3 2 -
.0
<2
E
3
X
G)
U)
>
D
O
o — \O-—-—.'_-—O-—-— n—n-O
l I l l l l
1 2 3 4 5 6

Time Period

Figure 5. Cluster mean proﬁles for Sexual DV among women in long-term DV

relationships.

167

 

 

 

 

 

 

° — Grom3
"""" Group2
8 _ """ Group1
O
o _
> (O
0
C
8 0
U' \
Q 5
LI. 0 __ ‘
> v
o
o _
N
0.
k a .‘\
/ -.' . ...Q
o _ 0' .%.:>.|;;;=8
I I I m I I
1 2 3 4 5 6
TIme Period

Figure 6. Cluster mean proﬁles for DV Frequency among women in short-term DV

relationships.

168

 

 

 

 

 

 

 

 

“—- Group3
o """" Group2
m ‘ """ Group1
to _
N
Q)
U)
3
.o o _
< N
.3
">’. to _
c T-
o.
>
o o _
ID —-I .
/' ‘
-7“""""'.° \ ‘.'-.
o q .Naﬁlmo
I I I I I I
1 2 3 4 5 6
Time Period

Figure 7. Cluster mean proﬁles for Physical DV among women in short-term DV

relationships.

169

 

 

 

 

 

 

$3 _
° _ Group2
""" Group1
g _
00 J
a)
(D
3
.0
<
E co —
3
X
G)
a)
>
o v —
o
N - 0‘
\
\
\
\
O _ .-—-—.-—.——.-—--0
I I I I I I
1 2 3 4 5 6
TIme Period

Figure 8. Cluster mean proﬁles for Sexual DV among women in short-term DV

relationships.

170

 

 

 

D MInImal DV
I Moderate DV
h DV

 

 

 

 

    

    
  

. 0¢.«0¢.i.ioioiOiOioioioi‘oioiOI.‘
.ow..o..wowowowowowowowowowowo4.4.»...«o.

016‘ 1

      

 

    
   

  

, COM.“ 000060000... 0 O 000 0 0 o
t t t . i t i b

. .
no». . . or. . ouowon. 3%.».
, . . ﬂ ,

  

         

 

          

 

   

             

     

. , 5
, , T
........................ . ..
.............'.‘..'..’..
i, , .M

  

               

   

...4.u.- ..'.u.<.n.4.toi.
Vu.»ou...now.u.u.wowou.u

 

                       

          

   

  

.«oduoc
005......

 
         

Q..‘.“.' ‘ ‘.““.‘

0.9.0... .090...

v .0.......... . o.....o......
, :

 

 

T1

— _ iT
4 2 0 8 6 4. 2 o
1 1 1

29:23.6 {9302 we eonﬁaz cues.

 

171

Time Period

 

Figure 9. Mean number of network supporters by DV Frequency cluster and time.

APPENDIX A

Demographics
From T1:
1. Yourdateofbirth: __/__ /__
(m0) (dY) (yr)
2. Choose the one that best describes your current marital/relationship status (choose
only one):
(a) single, never married (see below)
(b) married For how long? (in months)
(c) separated For how long? (in months)
(d) divorced For how long? (in months)
(e) widowed For how long? (in months)
If (a) is circled: Are you currently in a relationship? YES NO
If YES, go to Question 3.
If NO, were you in a relationship that lasted at least
6 weeks during your current pregnancy? YES NO
3. First name of your current partner or the partner you were with for at least 6

weeks during your pregnancy:

 

4. Prior to your current romantic relationship, speciﬁed in Question #3
(a) were you ever married? 1 = YES 2 = NO
(b) did you ever live with a partner? 1 = YES 2 = NO
(0) were you ever separated? l = YES 2 = NO
(d) were you ever divorced? l = YES 2 = NO
(e) were you ever widowed? 1 = YES 2 = NO
5. What is your racial or ethnic group? (Circle one)

1 = Native American

2 = Asian American/Paciﬁc Islander
3 = Black, African American

4 = Latino, Hispanic, Chicano

5 = Biracial (mixed): Specify
6 = Caucasian, White

7 = Other:

 

 

6. What is your total family income per month (estimate)?

 

172

APPENDIX A (cont’d).

 

 

 

 

 

 

 

 

 

7. What is the highest level of education you have completed? (Circle one)
1 = grades 1, 2, 3, 4, 5, or 6 (circle speciﬁc grade)
2 = grades 7, 8, 9, 10, 11, 12, GED (circle speciﬁc grade)
3 = some college Where?
4 = AA degree Where?
5 = BA/BS Where?
6 = some grad school Where?
7 = graduate degree Where?
MA? Ph.D.? Law? MD?
8 = other; Specify (e.g., Beauty School, nursing school)
From T3-T6:
8. When we interviewed you during your pregnancy, you had been involved with
[name of TI partner: ﬁll in from assignment sheet] for at least SIX weeks
during the pregnancy. What has happened to this relationship since we
interviewed you then? (Read all choices and circle one)
(1) I am still in a relationship with him/her.
(2) We have been together off and on since the interview, and we are
currently together.
(3) We have been together off and on since the interview, and we are
currently not together.
(4) I have not had a relationship with him/her since the interview.
9. Since your child was one year old (or two, three, or four years old depending on

wave of data collection), please list the ﬁrst names of the people with whom you
had a romantic relationship that lasted at least 6 weeks. Please start with the most
recent or current relationship and go back in time. Include the person listed in
Question #8 (T 1 partner) if appropriate.

9a. Was this person the same person you were involved
with one year ago? YES NO
Is this person the father of your child? YES NO
Are you currently involved? YES NO

 

Anyone else?
9b. Was this person the same person you were involved

with one year ago? YES NO
Is this person the father of your child? YES NO
Are you currently involved? YES NO

 

Anyone else?
90. Was this person the same person you were involved

with one year ago? YES NO

Is this person the father of your child? YES NO
Are you currently involved? YES NO

 

173

APPENDIX B

Severity of Violence Against Women Scales

INSTRUCTIONS: You and your partner have probably experienced anger or conﬂict. Below is a list of
behaviors he may have done. Describe how often he has done each behavior to you during the last year and
how many times your child saw or heard what happened by choosing a letter from the following scale.
How often: A= never B= once C= a few times D= many times

Stressfulness: 0= event did not occur l= not at all stressful 2= a little stressful 3= very stressful

During the last year

1 Times your child saw or heard what happened

1 Stressfulness

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

i
1. Hit or kicked a wall, door or furniture
2. Threw, smashed or broke an object
3. Driven dangerously with you in the car
4. Threw an object at you
5. Shook a ﬁnger at you
6. Made threatening gestures or faces at you
7. Shook a ﬁst at you
8. Acted like a bully toward you
9. Destroyed something belonging to you
10. Threatened to harm or damage things you care about
1 l. Threatened to destroy property
12. Threatened someone you care about
13. Threatened to hurt you
14. Threatened to kill himself
15. Threatened you with a club-like object
16. Threatened you with a knife or gun
17. Threatened to kill you
18. Threatened you with a weapon
l9. Acted like he wanted to kill you
20. Held you down, pinning you in place
2 I. Pushed or shoved you
22. Shook or roughly handled you

 

 

 

 

174

 

APPENDIX B (cont’d).

How often: A= never B= once C= a few times D= many times
Stressfulness: 0= event did not occur l= not at all stressful 2= a little stressful 3= very stressful
During the last year

1 Times your child saw or heard what happened

i Stressfulness

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I
23. Grabbed you suddenly or forcefully
24. Scratched you
25. Pulled your hair
26. Twisted your arm
27. Spanked you
28. Bit you
29. Slapped you with the palm of his hand
30. Slapped you with the back of his hand
31. Slapped you around your face and head
32. Kicked you
33. Hit you with an object
34. Stomped on you
35. Choked you
36. Punched you
37. Burned you with something
38. Used a club-like object on you
39. Beat you up
40. Used a knife or gun on you
41. Demanded sex whether you wanted to or not
42. Made you have oral sex against your will
43. Made you have sexual intercourse against your will
44. Physically forced you to have sex
45. Made you have anal sex against your will
46. Used an object on you in a sexual way

 

 

 

175

 

APPENDIX C
Childhood Tramna Questionnaire — Short Form

These questions ask about some of your experiences growing up as a child and a
teenager. Although these questions are of a personal nature, please try to answer as
honestly as you can. For each question, circle the response that best describes how
you feel.

WHEN I WAS GROWING UP...

1. I didn’t have enough to eat.

Never True Rarely True Sometimes True Often True Very Often True

2. I knew that there was someone to take care of me and protect me.

Never True Rarely True Sometimes True Often True Very Often True

3. People in my family called me things like “stupid,” “lazy,” or “ugly.”
Never True Rarely True Sometimes True Often True Very Often True

4. My parents were too drunk or high to take care of the family.
Never True Rarely True Sometimes True Often True Very Often True

5. There was someone in my family who helped me feel that I was important or special.

Never True Rarely True Sometimes True Often True Very Often True

6. I had to wear dirty clothes.
Never True Rarely True Sometimes True Often True Very Often True

7. I felt loved.
Never True Rarely True Sometimes True Often True Very Often True

8. I thought that my parents wished I had never been born.
Never True Rarely True Sometimes True Often True Very Often True

9. I got hit so hard by someone in my family that I had to see a doctor or go to the

hospital.
Never True Rarely True Sometimes True Often True Very Often True

176

APPENDIX C (cont’d).

WHEN I WAS GROWING UP...

10. There was nothing I wanted to change about my family.

Never True Rarely True Sometimes True Often True Very Often True

11. People in my family hit me so hard that it left me with bruises or marks.

Never True Rarely True Sometimes True Often True Very Often True

12. I was punished with a belt, a board, a cord, or some other hard object.

Never True Rarely True Sometimes True Often True Very Often True

13. People in my family looked out for each other.

Never True Rarely True Sometimes True Often True Very Often True

14. People in my family said hurtﬁil or insulting things to me.

Never True Rarely True Sometimes True Often True Very Often True

15. I believe that I was physically abused.
Never True Rarely True Sometimes True Often True Very Often True

16. I had the perfect childhood.
Never True Rarely True Sometimes True Often True Very Often True

17. I got hit or beaten so badly that it was noticed by someone like a teacher, neighbor, or
doctor.

Never True Rarely True Sometimes True Often True Very Often True

18. I felt that someone in my family hated me.

Never True Rarely True Sometimes True Often True Very Often True

19. People in my family felt close to each other.

Never True Rarely True Sometimes True Often True Very Often True

20. Someone tried to touch me in a sexual way, or tried to make me touch them.

Never True Rarely True Sometimes True Often True Very Often True

177

APPENDIX C (cont’d).

WHEN I WAS GROWING UP...

21. Someone threatened to hurt me or tell lies about me unless I did something sexual
with them.

Never True Rarely True Sometimes True Often True Very Often True

22. I had the best family in the world.
Never True Rarely True Sometimes True Often True Very Often True

23. Someone tried to make me do sexual things or watch sexual things.

Never True Rarely True Sometimes True Often True Very Often True

24. Someone molested me.

Never True Rarely True Sometimes True Often True Very Often True

25. I believe that I was emotionally abused.

Never True Rarely True Sometimes True Often True Very Often True

26. There was someone to take me to the doctor if I needed it.

Never True Rarely True Sometimes True Often True Very Often True

27. I believe that l was sexually abused.

Never True Rarely True Sometimes True Often True Very Often True

28. My family was a source of strength and support.
Never True Rarely True Sometimes True Often True Very Often True

178

APPENDIX D

Self-Report Inventory for Disorders of Extreme Stress

What follows are descriptions of typical reactions someone could have after
traumatic experiences such as you have had. Please indicate if you had similar
feelings SOON AFTER THE EXPERIENCE OR AS LONG AS YOU CAN
REMEMBER.

1.

Small problems get me very upset? For example, I get angry at a minor
frustration. I cry easily.

YES NO
I ﬁnd it hard to calm myself down after I become upset and have trouble getting
back on track.

YES NO

When I feel upset, I have trouble ﬁnding ways to calm myself down.
YES NO

I feel angry most of the time.
YES NO

I have thoughts or images of hurting somebody else.
YES NO

I have trouble controlling my anger.
YES NO

I worry about people ﬁnding out how angry I am.
YES NO

I have been in accidents or near accidents.
YES NO

I ﬁnd myself careless about making sure that I am safe.
YES NO

179

ll

APPENDIX D (cont’d).

AFTER THE EXPERIENCE OR AS LONG AS I CAN REMEMBER...

10. I have deliberately tried to hurt myself (like burning or cutting myself)

11.

12.

13.

14.

15.

16.

17.

18.

19.

YES NO

I have thought about killing myself.
YES NO

I make active efforts to keep myself from thinking about sex.
YES NO

It bothers me to be touched in general.
YES NO

It bothers me to be touched in a sexual way.
YES NO

I actively avoid sex.
YES NO

I ﬁnd myself thinking about sex more than I want to.
YES NO

I ﬁnd myself driven to engage in sexual activities without really feeling that I had
a choice.

YES NO

I am active sexually in ways that I know put me in danger.
YES NO

I expose myself to situations that might be dangerous. For example, I get
involved with people who might hurt me. I go to places that are not safe. I drive
too fast.

YES NO

180

APPENDIX D (cont’d).

AFTER THE EXPERIENCE OR AS LONG AS I CAN REMEMBER...

20. There are parts of my life that I cannot remember, or I am confused about what
happened, or I am unsure whether certain important things did or did not happen
to me.

YES NO

21. I have difﬁculty keeping track of time in my daily life.
YES NO

22. I ‘space’ out when I feel frightened or under stress.
YES NO

23. I sometimes feel so unreal that it is as if I am living in a dream, or not really there,
or behind a ‘glass wall.’

YES NO

24. I sometimes feel like there are two or more totally different people living inside
me who control how I behave at different times.

YES NO

25. I have the feeling that I basically have no inﬂuence or control over what happens
to me in my life?

YES NO
26. I feel that I have something wrong with me after what happened to me that can
never be ﬁxed?

YES NO

27. I feel chronically guilty about all sorts of things.
YES NO

28. I am too ashamed of myself to let people get to know me.
YES NO

181

APPENDIX D (cont’d).

AFTER THE EXPERIENCE OR AS LONG AS I CAN REMEMBER...

29.

30.

31.

32.

33.

34.

35.

I feel set apart and very different from other people?
YES NO

People make too big a deal about the dangerousness of situations that I get
involved in.

YES NO

I have trouble trusting people.
YES NO

I avoid having relationships with other people.
YES NO

I have difﬁculty working through conﬂicts in relationships.
YES NO

I ﬁnd that other traumatic experiences keep happening to me.
YES NO

I have hurt other people in ways similar to how I was hurt?
YES NO

36. I have trouble with (circle items that apply), yet doctors have not found a clear

cause for it.

YES NO
a. Vomiting
b. Abdominal pain
c. Nausea
d. Diarrhea
e. Intolerance for food

182

APPENDIX D (cont’d).

AFTER THE EXPERIENCE OR AS LONG AS I CAN REMEMBER...

37 . I suffer from chronic pain (circle items that apply), yet doctors have not found a
clear cause for it.

YES
f.

FVVFW

NO

In my arms and legs
In my back

In my joints

During urination
Headaches
Elsewhere

38. I suffer from (circle items that apply), yet doctors have not found a clear cause for

It.
YES

I.
m.
n.
0.

NO

Shortness of breath
Palpitations

Chest pain
Dizziness

39. I suffer from trouble with (circle items that apply), yet doctors have not found a
clear cause for it.

YES

arsseearep

NO

Remembering things

Swallowing

Losing my voice

Blurred vision

Actual blindness

Fainting and losing consciousness
Seizures and convulsions

. Being able to walk

Paralysis or muscle weakness
Urination

183

APPENDIX D (cont’d).

AFTER THE EXPERIENCE OR AS LONG AS I CAN REMEMBER...

40. I suffer from (circle items that apply), yet doctors have not found a clear cause for

41.

42.

43.

44.

45.

1t.

YES NO
2. Burning sensations in my sexual organs or rectum (not during intercourse)
aa. Impotence
bb. Irregular menstrual periods
cc. Excessive pre-menstrual tension
dd. Excessive menstrual bleeding

I feel hopeless and pessimistic about the future.
YES NO

I do not expect to be able to ﬁnd happiness in love relationships.
YES NO

I am unable to ﬁnd satisfaction in work.
YES NO

I believe that life has lost its meaning.
YES NO

There have been changes in my philosophy or religious beliefs or in the religious
beliefs or philosophical beliefs I grew up with.

YES NO

184

APPENDIX E
Norbeck Social Support Scale

Please list each signiﬁcant person in your life below. Consider all the persons who
provide personal support for you or who are important to you. Use only ﬁrst names or
initials.

Use the following list to help you think of the people important to you,
and list as many people as apply in your case.

«spouse or partner

«family members or relatives (specify: mother, father, grandmother, etc.)

«friends

«work or school associates

«neighbors

«health care providers

«counselor or therapist

«minister/priest/rabbi

«God, Allah, or Supreme Being

«other

You do not have to use all 24 spaces. Use as many spaces as you have
important persons in your life.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

l. 8. 15.
2. 9. 16.
3. 10. 17.
4. 11. 18.
5. 12. 19.
6. 13. 20.
7. 14. 21.

 

 

 

185

APPENDIX E (cont’d).

For each person you listed, please answer the following questions by writing in the
number that applies.

0 = not at all

1 = a little

2 = moderately
3 = quite a bit
4 = a great deal

Question 1: How much does this person make you feel liked or loved?

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1. 8. 15.
2. 9. 16.
3. 10. 17.
4. 11. 18.
5. 12. 19.
6. 13. 20.
7. 14. 21.

 

 

 

Question 2: How much does this person make you feel respected or admired?

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1. 8. 15.
2. 9. l6.
3. 10. 17.
4. 11. 18.
5. 12. 19.
6. 13. 20.
7. 14. 21.

 

 

 

186

APPENDIX E (cont’d).

For each person you listed, please answer the following questions by writing in the
number that applies.

0 = not at all

1 = a little

2 = moderately
3 = quite a bit
4 = a great deal

Question 3: How much can you conﬁde in this person?

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1. 8. 15.
2. 9. 16.
3. 10. 17.
4. 11. 18.
5. 12. 19.
6. 13. 20.
7. 14. 21.

 

 

 

Question 4: How much does this person agree with or support your actions or thoughts?

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1. 8. 15.
2. 9. 16.
3. 10. 17.
4. ll. 18.
5. 12. 19.
6. 13. 20.
7. 14. 21.

 

 

 

187

APPENDIX F
Perceptions of Adult Attachment Questionnaire

The majority of the following statements refer to your early childhood relationship with
your mother (when you were approximately 3 to 8 years old). If someone else was the
principal person responsible for your care in childhood, please respond to the questions
which refer to “mother” with that person in mind.

A few of the questions have two parts. For example “when I caused trouble as a child I
knew my mother would forgive me.” Some people might feel like they never caused
trouble as a child; however, they consider their mothers very forgiving. How then do you
answer? Only answer AGREE or STRONGLY AGREE if you agree with both parts of
the statement. If you agree with only one part of the statement, answer NEUTRAL. If
you disagree with both parts of the statement, answer DISAGREE or STRONGLY
DISAGREE.

l= STRONGLY DISAGREE

2 = DISAGREE
3 = NEUTRAL (NEITHER DISAGREE NOR AGREE)
4 = AGREE

S = STRONGLY AGREE
1. In childhood, I felt like I was really treasured by my mother.

2. In childhood I sometimes felt like my mother was really lonely when I
was not with her.

3. My mother was not very affectionate.

4. When I was a young child and little things went wrong, I did not feel sure
I could count on my mother to take care of me.

5. As a child, I couldn’t stand being separated from my mother.

6. My mother can make me feel really good, but when she is not nice to me
she can really tear me apart.

7. In my family of origin, we don’t make a show of expressing our feelings.
We prefer keeping feelings to ourselves.

8. Neither my mother nor myself are perfect, but somehow we made it
through my childhood.

9. I remember when I was frightened as a child, my mother holding me
close.

188

 

 

 

APPENDIX F (cont’d).

l= STRONGLY DISAGREE

2 = DISAGREE

3 = NEUTRAL (NEITHER DISAGREE NOR AGREE)
4 = AGREE

5 = STRONGLY AGREE

10. When I was a child, my mother sometimes told me that if I was not good
she would stop loving me.

11. My mother is selﬁshly caught up in herself to the exclusion of everybody
else.

12. My family was not particularly intimate, but this has never bothered me.

13. It’s hard for me to remember my early relationship with my mother
in any detail.

14. In childhood, I sometimes felt that my mother and I were so alike that I
didn’t know where she ended and I began.

15. If anything happened to my mother, I wonder if I could survive it.

16. I remember as a child feeling a desire to protect my mother.

17. Even though I went through rough times with my mother during my
childhood, somewhere along the line I managed to let go of the majority of
those angry, hurt feelings.

18. In childhood, I knew I was low on my mother’s priority list.

19. My mother was an all-around excellent mother.

20. No one gets under my skin like my mother.

21. As a child, I never thought separations from my parents were any big deal.

22. I often felt responsible for my mother’s welfare.

23. In childhood, my mother sometimes threatened to leave me or to send me
away if I wasn’t good.

24. To this day, my mother has no clue who I am or what I am all about.

189

APPENDIX F (cont’d).

1= STRONGLY DISAGREE

2 = DISAGREE
3 = NEUTRAL (NEITHER DISAGREE NOR AGREE)
4 = AGREE

5 = STRONGLY AGREE

_—

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

Even with all our past difﬁculties, I realize my mother did the best for me
that she could.

I have forgotten what most of my early childhood was like.

I always knew my mother was there for me; no matter what, I could
depend on her.

There are times when I feel like shaking my mother and saying “wake up
and see me for who I am.”

In childhood, I often had the impression that my mother was not listening
to me. She often tuned me out.

During my childhood, I sometimes felt like I was my mother’s whole life.

My mother and I are more accepting of each other’s differences than we
have been in the past.

When I was young, I often feared something dreadful would happen to my
mother or father.

I remember my mother telling me that I didn’t pay enough attention to her
or love her enough.

I often take my mother’s opinions about me to heart and lose sight of my
own opinions about myself.

My mother is a real nag.

My mother and I were so alike we often could ﬁnish each other’s
sentences.

I think people put too much emphasis on the mother/child relationship.
I remember very little about my early childhood (ages three to seven).

The concept of the loving, supporting mother is pure myth.

190

APPENDIX F (cont’d).

l= STRONGLY DISAGREE

 

 

 

2 = DISAGREE
3 = NEUTRAL (NEITHER DISAGREE NOR AGREE)
4 = AGREE
5 = STRONGLY AGREE
40. My relationship with my mother has gone through major changes over the
course of my childhood and adolescence.
41. Even as an adult, I sometimes feel like I will never dig myself out from
under my mother’s inﬂuence.
42. As a child, I sometimes got the feeling that without me my mother would
have fallen apart.
43. I couldn’t have asked for a better mother.
44. If my mother was not fair to me as a child, I realize now it was because
she was dealing with her own problems.
45. If something really bad happened to me in childhood, I did not feel I could
count on my mother to support me.
46. When I was a child, I sometimes got the feeling that my mother wished I
was never born.
47. I remember when I was a child feeling scared that one or both of my
parents would die unexpectedly.
48. My mother can devastate me with her criticisms.
49. In childhood, my mother often told me she was sacriﬁcing herself for me.
50. I don’t think my early childhood relationship with my mother has any
signiﬁcant inﬂuence on who I am today or my present relationships.
51. My mother was always there for me when I needed her.
52. When I acted bad as a child, my mother would, at times, threaten
to send me away.
53. I never felt like my mother gave me enough attention.

191

APPENDIX F (cont’d).

l= STRONGLY DISAGREE

2 = DISAGREE
3 = NEUTRAL (NEITHER DISAGREE NOR AGREE)
4 = AGREE
5 = STRONGLY AGREE
54. For all our past problems, my mother and I can still enjoy a good laugh
together.
55. During my childhood, my mother would often turn to me and tell me lots
of things that upset and bothered me.
56. In childhood, I often worried about my mother’s state of health.
57. I ﬁnd it difﬁcult to remember my early childhood.
58. My mother was a perfect mother.
59. My mother’s issues are still interfering with my life.
60. When I think back to my early childhood experiences, I discover things

about myself and my parents that I’ve never considered before.

192

APPENDIX G

Imputed Variables with Corresponding Matching Variables

 

Imputed Variable

Matching Variables

 

Income

T1 N0. Supporters

T2 N0. Supporters

T3 N0. Supporters

T4 N0. Supporters

T5 N0. Supporters

T6 N0. Supporters

T1 Emotional Support

T2 Emotional Support

T3 Emotional Support

T4 Emotional Support

T5 Emotional Support

T6 Emotional Support

Age, Race, Marital Status, Education

Age, Race, Marital Status, Education, Income, Attachment
Security

T1 No. Supporters, Education, Income

T1 & T2 No. Supporters, Education, Income

T1, T2, & T3 N0. Supporters, Education, Income

T1-T4 No. Supporters, Education, Income

Tl-TS No. Supporters, Education, Income

Age, Race, Marital Status, Education, Income, Attachment
Security

T1 Emotional Support

T1 & T2 Emotional Support

T1-T3 Emotional Support

T1-T4 Emotional Support

Tl-TS Emotional Support

 

193

APPENDIX G (cont’d).

 

Imputed Variable

Matching Variables

 

T2 DV Frequency

T3 DV Frequency

T4 DV Frequency

T5 DV Frequency

T6 DV Frequency

T2 DV Severity

T3 DV Severity

T4 DV Severity

T5 DV Severity

T6 DV Severity

T2 Threats of DV

T3 Threats of DV

T4 Threats of DV

T5 Threats of DV

T1 DV Frequency

T1 & T2 DV Frequency

T1-T3 DV Frequency

T1-T4 DV Frequency

Tl-TS DV Frequency

T1 DV Severity

T1 & T2 Dv Severity

T1-T3 DV Severity

T1-T4 DV Severity

T1-T5 DV Severity

Tl Threats of DV & Tl DV Frequency

T1 & T2 Threats of DV & DV Frequency

T1-T3 Threats of DV & DV Frequency

T1-T4 Threats of DV & DV Frequency

 

194

APPENDIX G (cont’d).

 

Imputed Variable

Matching Variables

 

T6 Threats of DV

T2 Physical DV

T3 Physical DV

T4 Physical DV

T5 Physical DV

T6 Physical DV

T2 Sexual DV

T3 Sexual DV

T4 Sexual DV

T5 Sexual DV

T6 Sexual DV

Child Maltreatment

CP Symptoms

Tl-TS Threats of DV & DV Frequency

Tl Physical DV & DV Frequency

T1 & T2 Physical DV & DV Frequency

T1-T3 Physical DV & DV Frequency

Tl-T4 Physical DV & DV Frequency

T1 -T5 Physical DV & DV Frequency

T1 Sexual DV & DV Frequency

T1 & T2 Sexual DV & DV Frequency

Tl-T3 Sexual DV & DV Frequency

T1-T4 Sexual DV & DV Frequency

T1-T5 Sexual DV & DV Frequency

Attachment Security, T1-T6 DV Frequency

Child Maltreatment, T1-T6 DV Frequency

 

195

APPENDIX H

Clustering Results for Severity of DV, Threats of DV, and Physical DV

Table H1

Cluster Mean Proﬁles for DVSeverity

 

 

DV Severity Cluster T1 T2 T3 T4 T5 T6

Cluster 1: Minimal DV 7.49 1.88 1.60 .75 1.18 1.46
Cluster 2: Moderate DV 43.16 9.22 7.34 5.49 6.27 1.90
Cluster 3: High DV 98.93 11.26 5.03 6.52 1.33 3.33

 

196

APPENDIX H (cont’d).

Table H2

Cluster Mean Proﬁles for Threats of DV

 

 

Threats of DV Cluster T1 T2 T3 T4 T5 T6

Cluster 1: Minimal DV 5.24 1.38 .99 1.24 1.67 1.58
Cluster 2: Moderate DV 25.04 8.79 6.50 4.27 5.21 2.44
Cluster 3: High DV 57.16 10.88 5.56 3.44 1.38 3.00

 

197

 

APPENDIX H (cont’d).

Table H3

Cluster Mean Proﬁles for Physical DV

Physical DV Cluster T1 T2

 

T3 T4 T5 T6
Cluster 1: Minimal DV 2.83 1.05 .54 .26 .50 1.03
Cluster 2: Moderate DV 21.89 5.14 5.64 2.16 3.21 .91
Cluster 3: High DV 56.00 3.86 2.00 1.64 .86 .93

 

198

APPENDIX H (cont’d).

 

 

 

 

 

 

8 _
*- — Group3
"""" Grorp2
""" Group1
o _
(D
e 8 a
a)
>
m
a)
>
o g _.
O _
N
.-\~ 8\:.. -......‘
o \e-—-—:-—-_::>.n-i= '
I I I I
1 2 3 4 5 6
Tune Period

Figure H1. Cluster mean proﬁles for DV Severity.

199

 

APPENDIX H (cont’d).

 

 

 

 

 

 

 

— Group3
""""" Group2
8 _ """ Group1
O _
v
(D
(D
3
.0
<
“a P.—
22
(U
2!
.c
I— o_
N
o _
‘- °.\\
.~\§ 8&8 ...... .....Q
I I I I I I
1 2 3 4 5 6

'I'Ime Period

Figure H2. Cluster mean proﬁles for Threats of DV.

200

APPENDIX H (cont’d).

 

 

 

 

 

 

 

*— Group3
o """" Group2
ID “ """ Group1
o _
v
(D
(I)
3
.0
< o _
_ (‘0
.3
(D
>
.C
o. o _.
N
g _.
...... 0..
.-~-~:\.—......-....°..Qo
I I I I I
1 2 3 4 5 6
Time Period

Figure H3. Cluster mean proﬁles for Physical DV.

201

APPENDIX I
Cluster Results for Severity of DV, Threats of DV, and Physical DV among Women in

Long-term DV Relationships

Table II

Cluster Mean Proﬁles for DV Severity among Women in Long-term DVRelationships

DV Severity Cluster T1 T2 T3 T4 T5 T6

 

Cluster 1: Minimal DV 7.21 1.33 1.74 .90 1.31 1.64

Cluster 2: Moderate DV 43.50 6.41 7.10 6.85 8.35 2.60

Cluster 3: High DV 98.93 11.26 5.03 6.52 1.33 3.33

202

APPENDIX I (cont’d).
Table 12

Cluster Mean Proﬁles for Threats of DV among Women in Long-term DVRelationships

Threats of DV Cluster T1 T2 T3 T4 T5 T6

 

Cluster 1: Minimal DV 4.08 .72 1.30 .73 1.31 1.60

Cluster 2: Moderate DV 24.91 8.09 5.32 5.30 7.02 3.23

Cluster 3: High DV 58.06 9.29 7.64 6.36 1.29 3.43

203

APPENDIX I (cont’d).
Table I3

Cluster Mean Proﬁles for Physical DV among Women in Long-term DV Relationships

 

Physical DV Cluster T1 T2 T3 T4 T5 T6

 

Cluster 1: Minimal DV 3.06 .72 .99 .64 .57 1.17

Cluster 2: Moderate DV 21.11 1.52 1. .33 1.82 3.22 .22

Cluster 3: High DV 54.33 6.33 7.73 1.53 3.87 2.80

 

204

APPENDIX I (cont’d).

 

100
1

—“- Group 3
""""" Group 2
""" Group 1

 

 

80
I

 

 

 

 

O .—
g co
0.)
>
a)
(D
>
o g .-
Q d
N
\0
.~\ 0"\"~GS'——w'- O”"’".‘~...
I T I I I I
1 2 3 4 5 6

TIme Period

Figure 11. Cluster mean proﬁles for DV Severity among women in long-term DV

relationships.

205

APPENDIX I (cont’d).

 

 

 

O
(D
— Group 3
"""" Group 2
o _ """ Group 1
to

DV Threats of Abuse
30
I

 

 

 

 

amt:\8 ...o....
.~ ...... . .s.
o __ \‘\.-_-—e-_-_.:_\e-‘::O
I I I I I I
1 2 3 4 5 6
Time Period

Figure 12. Cluster mean proﬁles for Threats of DV among women in long-term DV

relationships.

206

APPENDIX 1 (cont’d).

 

 

 

 

 

 

 

"— Group3
8 _ """" Group2
""" Group1
o _
v
(D
In
3
.0
<
_ O __
8 m
'5
>
.0
CL
.3 a —
o _
‘— ‘ o
0‘ Q/
o — \~‘8. :__.&H.Ivh= g?:<j.-_a._.;
I
1 2 3 4 5 6

TIme Period

Figure 13. Cluster mean proﬁles for Physical DV among women in long-term DV

relationships.

207

APPENDIX J
Cluster Results for DV Severity and Threats of DV among Women in Short-term DV
Relationships
Table J1

Cluster Mean Proﬁles for DVSeverity among Women in Short-term DVRelationships

 

 

DV Severity Cluster T1 T2 T3 T4 T5 T6
Cluster 1: Minimal DV 9.15 4.08 4.88 .04 .50 .59
Cluster 2: Moderate-High DV 45.18 70.63 0 0 O 0

Cluster 3: Early High DV 49.04 6.96 .88 2.95 1.36 .13

 

208

 

 

 

APPENDIX .1 (cont’d).
Table .12

Cluster Mean Proﬁlesfor Threats of DV among Women in Short-term DVRelationships

 

 

Threats of DV Cluster T1 T2 T3 T4 T5 T6
Cluster 1: Minimal DV 5.61 1.24 3.53 .41 1.06 .88
Cluster 2: Moderate-High DV 17.46 31 0 0 0 0

Cluster 3: Early High DV 32.69 4.80 2.50 1.30 1.10 .30

 

209

APPENDIX J (cont’d).

 

 

 

 

 

 

E E ,1 —— Group3
/ a """" Group2
o : . ''''' Group1
to ‘ ’ s
I
0 0
’ I
o ' ‘
to ‘ o I ‘
0
O s
.2; s
a) O
(D s
a a — x
.0
O |
N '— s
C
I
9 ‘I e\ “
\ ﬁ
0 —- \:-\.:...>:---".m.
I I I I I
1 2 3 4 5 6
TIme Period

Figure J1. Cluster mean proﬁles for DV Severity among women in short-term DV

relationships.

210

APPENDIX .1 (cont’d).

 

 

 

 

 

 

 

— Group3
8 _ """"" Group2
""" Group1
to _
N
a)
(I)
3
.0 O _
< N
“5
{3
9.’ 53*
.C
I'-
>
o g-
m —1
.~
o°\
9-—-—-. e _
O _ ‘9 ...... 83:.--oﬁ'
I I I I I
1 2 3 4 5 6
TIme Period

Figure J2. Cluster mean proﬁles for Threats of DV among women in short-term DV

relationships.

211

 

REFERENCES

Adam, K. S., Sheldon Keller, A. E., & West, M. (1996). Attachment organization and
history of suicidal behavior in clinical adolescents. Journal of Consulting and
Clinical Psychology, 64(2), 264-272.

Alexander, R. (1993). Wife battering: An Australian perspective. Journal of Family
Violence, 8, 229-251.

Allen, J. G., Coyne, L., & Huntoon, J. (1998). Complex posttraumatic stress disorder in

women from a psychometric perspective. Journal of Personality Assessment,
70(2), 277-298.

Allen, J. G., Huntoon, J ., & Evans, R. B. (1999). Complexities in complex posttraumatic
stress disorder in inpatient women: Evidence from cluster analysis of MCMI-III
personality disorder scales. Journal of Personality Assessment, 73(3), 449-471.

Allen, J. P., Hauser, S. T., & Borman Spurrell, E. (1996). Attachment theory as a
framework for understanding sequelae of severe adolescent psychopathology: An

11-year follow-up study. Journal of Consulting and Clinical Psychology, 64(2),
254-263.

American Psychiatric Association. (1994). Diagnostic and statistical manual of mental
disorders (4th ed. ). Washington, DC: American Psychiatric Association.

Anderson, D. K., Saunders, D. G., Yoshihama, M., Bybee, D., & Sullivan, C. M. (2003).
Long-term trends in depression among women separated from abusive partners.
Violence Against Women, 9(7), 807-838.

Araujo, A. B., Mohr, B. A., & McKinlay, J. B. (2004). Changes in sexual function in
middle-aged and older men: Longitudinal data from the Massachusetts male aging
study. Journal of the American Geriatrics Society, 52(9), 1502-1509.

Armsden, G. C., & Greenberg, M. T. (1987). The inventory of parent and peer
attachment: Individual differences and their relationship to psychological well-
being in adolescence. Journal of Youth and Adolescence, 16(5), 427-454.

212

Astin, M. C., Lawrence, K. J ., & Foy, D. W. (1993). Posttraumatic stress disorder among
battered women: Risk and resiliency factors. Violence and Victims, 8(1), 17-28.

Barnett, D., Ganiban, J ., & Cicchetti, D. (1999). Maltreatment, negative expressivity, and
the development of Type D attachments from 12 to 24 months of age.
Monographs of the Society for Research in Child Development, 64, 97-118.

Barnett, O. W., Martinez, T. E., & Keyson, M. (1996). The relationship between
violence, social support, and self-blame in battered women. Journal of
Interpersonal Violence, 11(2), 221-233.

Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in

social psychological research: Conceptual, strategic, and statistical considerations.
Journal of Personality and Social Psychology, 51(6), 1 173-1 182.

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

Bernstein, D. P., & Fink, L. (1998). Childhood Trauma Questionnaire manual. San
Antonio, TX: The Psychological Corporation.

Bernstein, D. P., Stein, J. A., Newcomb, M. D., Walker, E., Pogge, D., Ahluvalia, T., et
al. (2003). Development and validation of a brief screening version of the
Childhood Trauma Questionnaire. Child Abuse and Neglect, 27(2), 169-190.

Blaustein, M. E., Spinazzola, J ., Simpson, W., & van der Kolk, B. (2000). Psychological
sequelae of early trauma: Comorbid diagnoses or diagnostic entity?, 16th annual

meeting of the International Society for Traumatic Stress Studies. San Antonio,
TX.

Bogat, G., Levendosky, A. A., Theran, 8., 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(11), 1271-1291.

Bowlby, J. (1969). Attachment and loss. Vol 1. Attachment. New York: Basic Books.

Bowlby, J. (1977). The making and breaking of affectional bonds: 11. Some principles of
psychotherapy. British Journal of Psychiatry, 130, 421-431.

213

 

Bowlby, J. (1980). Attachment and loss: V01. 3. Loss, sadness, and depression. New
York: Basic Books, Inc.

Bureau of Justice Statistics. (1998). Bureau of Justice Statistics sourcebook of the
criminal justice statistics--1997. US. Department of Justice. NCJ171147 (pp.
198).

Butcher, J ., Dahlstrom, W. G., Graham, J ., Tellegen, A., & Kraemmer, B. (1989). MMPI-
2: Manual for administration and scoring. Minneapolis: University of Minnesota
Press.

Campbell, J. C., Harris, M. J ., & Lee, R. K. (1995). Violence research: An overview.
Scholarly Inquiryfor Nursing Practice, 9(2), 105-126.

Carlson, B. B, McNutt, L. A., Choi, D. Y., & Rose, 1. M. (2002). Intimate partner abuse
and mental health: The role of social support and other protective factors.
Violence Against Women, 8(6), 720-745.

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

Cascardi, M., Langhinrichsen, J ., & Vivian, D. (1992). Marital aggression: Impact,
injury, and health correlates for husbands and wives. Archives of Internal
Medicine, 152, 1178-1184.

Cascardi, M., & O’Leary, K. (1992). Depressive symptomatology, self-esteem, and self-
blame in battered women. Journal of Family Violence, 7(4), 249-259.

Cassidy, J ., & Mohr, J. J. (2001). Unsolvable fear, trauma, and psychopathology: Theory,
research, and clinical considerations related to disorganized attachment across the
life span. Clinical Psychology: Science and Practice, 8(3), 275-298.

Ceballo, R., Ramirez, C., Castillo, M., Caballero, G. A., & Lozoff, B. (2004). Domestic
violence and women's mental health in Chile. Psychology of Women Quarterly,
28(4), 298-308.

Champion, J. D., Shain, R. N., Piper, J ., & Perdue, S. T. (2001). Sexual abuse and sexual
risk behaviors of minority women with sexually transmitted diseases. Western
Journal of Nursing Research, 23(3), 241-254.

214

Christopoulos, C., Cohn, D. A., Shaw, D. 8., Joyce, S., Sullivan-Hanson, J ., Kraft, S., et
al. (1987). Children of abused women: I. Adjustment at time of shelter residence.
Journal of Marriage and the Family, 49(3), 611-619.

Coker, A. L., Smith, P. H., Thompson, M. P., McKeown, R. E., Bethea, L., & Davis, K.
E. (2002). Social support protects against the negative effects of partner violence
on mental health. Journal of Women 's Health and Gender Based Medicine, 11(5),
465-476.

Coker, A. L., Watkins, K. W., Smith, P. H., & Brandt, H. M. (2003). Social support
reduces the impact of partner violence on health: Application of structural

equation models. Preventive Medicine: An International Journal Devoted to
Practice and Theory, 3 7(3), 259-267.

Cole-Detke, H., & Kobak, R. (1996). Attachment processes in eating disorder and
depression. Journal of Consulting and Clinical Psychology, 64(2), 282-290.

Coventry, W. L., Gillespie, N. A., Heath, A. C., & Martin, N. G. (2004). Perceived social
support in a large community sample: Age and sex differences. Social Psychiatry
and Psychiatric Epidemiology, 39, 625-636.

Crittenden, P.-M. (1997). Toward an integrative theory of trauma: A dynamic-maturation
approach. In S. Toth & D. Cicchetti (Eds), Developmental perspectives on
trauma: Theory, research, and intervention (pp. 33-84). Rochester, NY:
University of Rochester Press.

Crowell, J ., Treboux, D., & Waters, E. (1999). The Adult Attachment Interview and the
Relationship Questionnaire: Relations to reports of mothers and partners.

Personal Relationships, 6(1), 1-1 8.

Curry, M. A., & Harvey, S. (1998). Stress related to domestic violence during pregnancy
and infant birth weight. In J. C. Campbell (Ed.), Empowering survivors of abuse:
Health care for battered women and their children (pp. 98-108). Thousand Oaks,
CA: Sage Publications, Inc.

Davidson, J. A. (1993). Issues in the diagnosis of posttraumatic stress disorder. In J.
Oldham, M. Riba & A. Tasman (Eds), Review of psychiatry (V 01. 12, pp. 141-
155). Washington, DC: American Psychiatric Press.

215

 

Dickinson, L., deGruy, F. V., 111, Dickinson, W., & Candib, L. M. (1998). Complex
posttraumatic stress disorder: Evidence from the primary care setting. General
Hospital Psychiatry, 20(4), 214-224.

Dorahy, M. J ., Lewis, C. A., & Wolfe, F. A. M. (2007). Psychological distress associated
with domestic violence in Northern Ireland. Current Psychology: Developmental,
Learning, Personality, Social, 25(4), 295-305.

Doyle, J. R, Frank, E., Saltzman, L. E., McMahon, P. M., & Fielding, B. D. (1999).
Domestic violence and sexual abuse in women physicians: Associated medical,

psychiatric, and professional difﬁculties. Journal of Women 's Health and Gender
Based Medicine, 8(7), 955-965.

Dutton, D. G., Landolt, M. A., Starzomski, A., & Bodnarchuk, M. (2001). Validation of
the propensity for abusiveness scale in diverse male populations. Journal of
Family Violence, 16(1), 59-73.

Dutton, D. G., & Painter, S. (1993). Emotional attachments in abusive relationships: A
test of traumatic bonding theory. Violence and Victims, 8(2), 105-120.

Dutton, M. A., & Goodman, L. A. (1994). Posttraumatic stress disorder among battered

women: Analysis of legal implications. Behavioral Sciences and the Law, 12,
215-234.

El-Bassel, N., Gilbert, L., Rajah, V., Foleno, A., & Frye, V. (2001). Social support
among women in methadone treatment who experience partner violence. Violence
Against Women, 7(3), 246-274.

Farris, C. A., & Fenaughty, A. M. (2002). Social isolation and domestic violence among
female drug users. American Journal of Drug and Alcohol Abuse, 28(2), 339-352.

Finkelhor, D., & Yllo, K. (1987). License to rape: Sexual abuse of wives. New York:
Free Press.

Fischer, C. S. (1982). To Dwell Among Friends. Chicago: University of Chicago Press.

Fonagy, P., Leigh, T., Steele, M., Steele, H., Kennedy, R., Mattoon, G., et al. (1996). The
relation of attachment status, psychiatric classiﬁcation, and response to
psychotherapy. Journal of Consulting and Clinical Psychology. 64(1), 22-31.

216

Ford, J. D., Stockton, P., Kaltman, S., & Green, B. L. (2006). Disorders of extreme stress
(DESNOS) symptoms are associated with type and severity of interpersonal
trauma exposure in a sample of healthy young women. Journal of Interpersonal

Violence, 21(11), 1399-1416.

Forte, J. A., Franks, D. D., Forte, J. A., & Rigsby, D. (1996). Asymmetrical role-taking:
Comparing battered and nonbattered women. Social Work, 41(1), 59-73.

Fowler, D. N., & Hill, H. M. (2004). Social support and spirituality as culturally relevant
factors in coping among African American survivors of partner abuse. Violence
Against Women, 10(11), 1267-1282.

Frank, E., & Dingle, A. D. (1999). Self-reported depression and suicide attempts among
US. women physicians. American Journal of Psychiatry, 156(12), 1887-1894.

Frasier, P. Y., Belton, L., Hooten, E., Campbell, M. K., DeVellis, B., Benedict, S., et al.
(2004). Disaster down east: Using participatory action research to explore
intimate partner violence in eastern North Carolina. Health Education and

Behavior, 31(4), 69s-84s.

Freud, S. (1995). Beyond the pleasure principle (1. Strachey, Trans). New York: W. W.
Norton and Co, Inc. (Original work published 1920).

Gold, S. (2000). Not trauma alone: Therapy for child abuse survivors in family and
social context. Philadelphia: Brunner/Routledge.

Golding, J. M. (1999). Intimate partner violence as a risk factor for mental disorders: A
meta-analysis. Journal of Family Violence, 14(2), 99-132.

Grinker, R. R., & Spiegel, J. P. (1945). Men under stress. Oxford, England: Blakiston.

Hallstrom, T., & Sarnuelsson, S. (1990). Changes in women's sexual desire in middle life:
The longitudinal study of women in Gothenburg. Archives of Sexual Behavior,
19(3), 259-268.

Hamilton, C., E. (1995). Continuity and discontinuity in attachment from infancy through
adolescence, Biennial Meeting of the Society for Research in Child Development.
Indianapolis.

217

 

Hamilton, C., E. (2000). Continuity and discontinuity of attachment from infancy through
adolescence. Child Development, 71(3), 690-694.

Harter, S. L., & Neimeyer, R. A. (1995). Long-term effects of child sexual abuse; Toward

a constructivist theory of trauma and its treatment. Advances in personal construct
psychology, 3, 229-269.

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

Herman, J. L. (1992b). Complex PTSD: A syndrome in survivors of prolonged and
repeated trauma. Journal of Traumatic Stress, 5(3), 377-391.

Hesse, E. (1999). The Adult Attachment Interview: Historical and current perspectives.
In J. Crowell (Ed.), Handbook of attachment: Theory, research, and clinical
applications (pp. 434-465). New York: Guilford Press.

Hilberman, E., & Munson, K. (1978). Sixty battered women. Victimology, 2(3-sup-4),
460-470.

Houskarnp, B. M., & Foy, D. W. (1991). The assessment of posttraumatic stress disorder
in battered women. Journal of Interpersonal Violence, 6(3), 367-375.

Huth-Bocks, A. C., Levendosky, A. A., Bogat, G., & von Eye, A. (2004). The Impact of
Maternal Characteristics and Contextual Variables on Infant-Mother Attachment.
Child Development, 75(2), 480-496.

Ide, N., & Paez, A. (2000). Complex PTSD: A review of current issues. International
Journal of Emergency Mental Health, 2(1), 43 -49.

Jasinski, J. L. (2001). Pregnancy and violence against women: An analysis of
longitudinal data. Journal of Interpersonal Violence, 16(7), 712-733.

Jones, E. E. (1996). Introduction to the special section on attachment and
psychopathology: Part1. Journal of Consulting and Clinical Psychology, 64, 5-7.

Kardiner, A., & Spiegel, H. (1947). War stress and neurotic illness (2nd ed.). Oxford,
England: P. B. Hoeber.

218

Katzelnick, D., Kobak, K. A., DeLeire, T., Henk. H. J ., Greist, J. H., Davidson, J. R., et
al. (2001). Impact of generalized social anxiety disorder in managed care.
American Journal of Psychiatry, 158(12), 1999-2007.

Kemp, A., Green, B. L., Hovanitz, C., & Rawlings, E. I. (1995). Incidence and correlates
of posttraumatic stress disorder in battered women: Shelter and community
samples. Journal of Interpersonal Violence, 10(1), 43-55.

Kemp, A., Rawlings, E. 1., & Green, B. L. (1991). Post-traumatic stress disorder (PTSD)
in battered women: A shelter sample. Journal of Traumatic Stress, 4(1), 137-148.

Kernic, M. A., Wolf, M. E., & Holt, V. L. (2000). Rates and relative risk of hospital
admission among women in violent intimate partner relationships. American
Journal of Public Health, 90(9), 1416-1420.

Levendosky, A. A., Bogat, G., Theran, S. A., Trotter, J. 8., von Eye, A., & Davidson, W.
S., 11. (2004). The social networks of women experiencing domestic violence.
American Journal of Community Psychology, 34(1-2), 95-109.

Levendosky, A. A., Bogat, G. A., Davidson, W. S., & von Eye, A. (2000). Risk and
protective factors for domestic violence. ROI Grant from Centers for Disease
Control, Injury Prevention Center: Total Budget: $875,811 (10/1/00-9/31/03).

Lichtenstein, J ., & Cassidy, J. (1991). The Inventory of Adult Attachment (INVAA):
Validation of a new measure of adult attachment, Society for Research in Child
Development. Seattle, Washington.

Lown, E., & Vega, W. A. (2001). Intimate partner violence and health: Self-assessed
health, chronic health, and somatic symptoms among Mexican American women.
Psychosomatic Medicine, 63(3), 352-360.

Luxenberg, T., Spinazzola, J ., & van der Kolk, B. (2001). Complex trauma and disorders
of extreme stress (DESNOS) diagnosis, part 1: Assessment. Directions in
Psychiatry, 21 , 373-392.

Main, M. (1990). Cross-cultural studies of attaclunent organization: Recent studies,

changing methodologies, and the concept of conditional strategies. Human
Development, 33(1), 48-61.

219

 

Main, M., & Goldwyn, R. (1994). Adult attachment scoring and classiﬁcation
system.Unpublished manuscript, Department of Psychology, University of
California, Berkeley.

Marshall, L. L. (1992). Development of the Severity of Violence Against Women Scales.
Journal of Family Violence, 7(2), 103-121.

Marshall, L. L. (1996). Psychological abuse of women: Six distinct clusters. Journal of
Family Violence, 11(4), 379-409.

Marshall, L. L. (1999). Effects of men's subtle and overt psychological abuse on low-
income women. Violence and Victims, 14(1), 69-88.

Martin, S. L., Mackie, L., Kupper, L. L., Buescher, P. A., & Moracco, K. E. (2001).
Physical abuse of women before, during, and after pregnancy. Journal of the
American Medical Association, 285(12), 1581-1584.

Matud-Aznar, M. P., Aguilera-Avila, L., Marrero-Quevedo, R. J., Moraza-Pulla, O., &
Carballeira-Abella, M. (2003). Social support in battered women. International
Journal of Clinical and Health Psychology, 3(3), 439-459.

McLean, L. M., & Gallop, R. (2003). Implications of childhood sexual abuse for adult
borderline personality disorder and complex posttraumatic stress disorder.
American Journal of Psychiatry, 160(2), 369-371.

Meadows, L. A., Kaslow, N. J ., Thompson, M. P., & Jurkovic, G. J. (2005). Protective
factors against suicide attempt risk among African American women experiencing

intimate partner violence. American Journal of Community Psychology, 3 6(1/2),
109-121.

Meisel, J ., Chandler, D., & Rienzi, B. M. (2003). Domestic violence prevalence and
effects on employment in two California TANF populations. Violence Against
Women, 9(10), 1191-1212.

Mertin, P., & Mohr, P. B. (2000). Incidence and correlates of posttraumatic stress

disorder in Australian victims of domestic violence. Journal of Family Violence,
15(4), 411-422.

220

Mitchell, R. E., & Hodson, C. A. (1983). Coping with domestic violence: Social support
and psychological health among battered women. American Journal of
Community Psychology, 11(6), 629-654.

Mitchell, R. E., & Hodson, C. A. (1986). Coping and social support among battered
women: An ecological perspective. In S. E. Hobfoll (Ed), Stress, social support,
and women: The series in clinical and community psychology (pp. 153-169). New
York: Hemisphere.

Moore, T. E., Pepler, D., Mae, R., & Michele, K. (1989). Effects of family violence on
children: New directions for research and intervention. In G. Cameron & B.
Pressman (Eds), Intervening with assaulted women: Current theory, research,
and practice (pp. 75-91). Hillsdale, NJ: Lawrence Erlbaum Associates.

Newman, E., Riggs, D. S., & Roth, S. (1997). Thematic resolution, PTSD, and complex
PTSD: The relationship between meaning and trauma-related diagnoses. Journal
of Traumatic Stress, 10(2), 197-213.

Nixon, R. D. V., Resick, P. A., & Nishith, P. (2004). An exploration of comorbid
depression among female victims of intimate partner violence with posttraumatic
stress disorder. Journal of Affective Disorders, 82(2), 315-320.

Noll, J. G., Horowitz, L. A., Bonanno, G. A., Trickett, P. K., & Putnam, F. W. (2003).
Revictimization and self-harm in females who experienced childhood sexual

abuse: Results from a prospective study. Journal of Interpersonal Violence,
18(12), 1452-1471.

Norbeck, J. S., Lindsey, A. M., & Carrieri, V. L. (1981). The development of an
instrument to measure social support. Nursing Research, 30(5), 264-269.

Norbeck, J. S., Lindsey, A. M., & Carrieri, V. L. (1983). Further development of the
Norbeck Social Support Questionnaire: Normative data and validity testing.
Nursing Research, 32(1), 4-9.

Olson, B. D., Curtis, C. E., Jason, L. A., Ferrari, J. R., Horin, E. V., Davis, M. 1., et al.
(2003). Physical and sexual trauma, psychiatric symptoms, and sense of
community among women in recovery: Toward a new model of shelter aﬁercare.
Journal of Prevention and Intervention in the Community, 26(1), 67-80.

221

Pelcovitz, D., van der Kolk, B. A., Roth, 8., Mandel, F ., Kaplan, S., & Resick, P. (1997).
Development of a criteria set and a structured interview for disorders of extreme
stress (SIDES). Journal of Traumatic Stress, 10(1), 3-16.

Pianta, R. C., Egeland, B., & Adam, E. K. (1996). Adult attachment classiﬁcation and
self-reported psychiatric symptomatology as assessed by the Minnesota

Multiphasic Personality Inventory«2. Journal of Consulting and Clinical
Psychology, 64(2), 273-281.

Pico-Alfonso, M. A. (2005). Psychological intimate partner violence: The major predictor
of posttraumatic stress disorder in abused women. Neuroscience and
Biobehavioral Reviews, 29(1), 181-193.

Plichta, S. B. (1996). Violence and abuse: Implications for women's health. In K. S.
Collings & M. M. F alik (Eds), Women 's health: The Commonwealth Fund survey
(pp. 23 7-270). Baltimore, MD: Johns Hopkins University Press.

Rennison, C. M. (2001). Intimate partner violence and age of victim, 1993-99. US.
Department of Justice. NCJ 187635 (pp. 1-12): Bureau of Justice Statistics.

Rennison, C. M. (2003). Intimate partner violence, 1993-2001. US. Department of
Justice. NCJ 197838 (pp. 1-2): Bureau of Justice Statistics.

Roberts, G. L., Lawrence, J. M., O'Toole, B. I., & Raphael, B. (1997). Domestic violence
in the emergency department: 1. Two case-control studies of victims. General
Hospital Psychiatry, 19(1), 5-11.

Rokach, A. (2006). Lonliness in domestically abused women. Psychological Reports,
98(2), 367-373.

Rorty, M., & Yager, J. (1996). Histories of childhood trauma and complex post-traumatic
sequelae in women with eating disorders. Psychiatric Clinics of North America,
19(4), 773-791.

Rosenstein, D. S., & Horowitz, H. A. (1996). Adolescent attachment and
psychopathology. Journal of Consulting and Clinical Psychology, 64(2), 244-253.

222

 

Roth, S., Newman, E., Pelcovitz, D., van der Kolk, B., & Mandel, F. S. (1997). Complex
PTSD in victims exposed to sexual and physical abuse: Results from the DSM-IV
ﬁeld trial for posttraumatic stress disorder. Journal of Traumatic Stress, 10(4),
539-555.

Russell, D. E. H. (1990). Rape in marriage (exp. and rev. ed. ): Bloomington, IN, US:
Indiana University Press.

Sackett, L. A., & Saunders, D. G. (1999). The impact of different forms of psychological
abuse on battered women. Violence and Victims, 14(1), 105-117.

Sansone, R. A., Reddington, A., Sky, K., & Wiederman, M. W. (2007). Borderline
personality symptomatology and history of domestic violence among women in
an internal medicine setting. Violence and Victims, 22(1), 120-126.

Schafer, J ., Caetano, R., & Clark, C. L. (1998). Rates of intimate partner violence in the
United States. American Journal of Public Health, 88(11), 1702-1704.

Scher, C. D., Stein, M. B., Asmundson, G. J. G., McCreary, D. R., & Forde, D. R. (2001).
The Childhood Trauma Questionnaire in a community sample: Psychometric
properties and normative data. Journal of Traumatic Stress, 14(4), 843-857.

Schiavi, R. C., Schreiner-Engle, P., Mandeli, J ., Schanzer, H., & Cohen, E. (1990).
Healthy aging and male sexual function. American Journal of Psychiatry, 147(6),
766-771.

Schnurr, P. P., & Green, B. G. (Eds). (2004). Trauma and health: Physical health
consequences of exposure to extreme stress. Washington, DC: American
Psychological Association.

Segraves, R. T., & Segraves, K. B. (1995). Human sexuality and aging. Journal of Sex
Education and Therapy, 21(2), 88-102.

Sharhabani-Arzy, R., Amir, M., Kotler, M., & Liran, R. (2003). The toll of domestic
violence: PTSD among battered women in an Israeli sample. Journal of
Interpersonal Violence, 18(11), 1335-1346.

223

 

Smith, P. H., Thornton, 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.

Spinazzola, J ., Roth, S., Derosa, R., Efrom, L., Lifton, N., & Davidson, J. (1994,
November). The specificity of the Disorders of Extreme Stress construct Poster
session presented at the 10th annual meeting of the International Society for
Traumatic Stress Studies. Chicago, IL.

Stanley, N., & Penhale, B. (1999). The mental health problems of mothers experiencing
the child protection system: Identifying needs and appropriate responses. Child
Abuse Review, 8(1), 34-45.

Stark, E., & F litcraft, A. (1996). Women at risk: Domestic violence and women 's health.
Thousand Oaks, CA: Sage Publications, Inc.

Stein, M. B., & Kennedy, C. (2001). Major depressive and post-traumatic stress disorder
comorbidity in female victims of intimate partner violence. Journal of Aﬂective
Disorders, 66(2-3), 133-138.

Stein, M. B., Kennedy, C. M., & Twamley, E. W. (2002).Neuropsychologica1 function in
female victims of intimate partner violence with and without posttraumatic stress
disorder. Biological Psychiatry, 52(11), 1079-1088.

Stein, M. B., Walker, J ., Anderson, G., Hazen, A., Ross, C., Eldridge, G., et al. (1996).
Childhood physical and sexual abuse in patients with anxiety disorders and in a
community sample. American Journal of Psychiatry, 153(2), 275-277.

Stewart, D. E., & Cecutti, A. (1993). Physical abuse in pregnancy. Canadian Medical
Association Journal, 149(9), 1257-1263.

Straus, M. A. (1979). Measuring intrafamily conﬂict and violence: The Conﬂict Tactics
(CT) Scales. Journal of Marriage and the Family, 41 (1), 75-88.

Street, A. E., & Arias, I. (2001). Psychological abuse and posttraumatic stress disorder in
battered women: Examining the roles of shame and guilt. Violence and Victims,
16(1), 65-78.

224

 

 

 

 

Tan, C., Basta, J ., Sullivan, C. M., & Davidson, W. S. (1995). The role of social support
in the lives of women exiting domestic violence shelters: An experimental study.
Journal of Interpersonal Violence, 10(4), 43 7-45 1.

Terr, L.-C. (1991). Childhood traumas: An outline and overview. American Journal of
Psychiatry, 148, 10-20.

Thompson, M. P., Kaslow, N. J ., Kingree, J. B., Rashid, A., Puett, R., Jacobs, D., et al.
(2000). Partner violence, social support, and distress among inner-city African
American women. American Journal of Community Psychology, 28(1), 127-143.

Tjaden, P., & Thoennes, N. (1998). Prevalence, incidence, and consequences of violence
against women: Findings from the National Violence Against Women Survey.
US. Department of Justice, National Institute of Justice, Centers for Disease
Control and Prevention. NCJ 172837 (pp. 1-16).

 

Tjaden, P., & Thoennes, N. (2000a). Full report of the prevalence, incidence, and
consequences of violence against women: Findings from the National Violence
Against Women Survey. US. Department of Justice, National Institute of Justice.
NCJ 183781. (pp. iii-61).

Tj aden, P., & Thoennes, N. (2000b). Extent, nature, and consequences of intimate partner
violence: Findings from the National Violence Against Women Survey. US.
Department of Justice, National Institue of Justice. NCJ 181867 (pp. iii-57).

Tjaden, P., & Thoennes, N. (2000c). Prevalence and consequences of male-to-female and
female-to-male intimate partner violence as measured by the National Violence
Against Women Survey. Violence Against Women, 6(2), 142-161.

Torres, 8., Campbell, J. C., Campbell, D. W., Ryan, J., King, C., Price, R, et al. (2000).
Abuse during and before pregnancy: Prevalence and cultural correlates. Violence
and Victims, 15(3), 303-321.

Umberson, D., Anderson, K., Glick, J ., & Shapiro, A. (1998). Domestic violence,
personal control, and gender. Journal of Marriage and the Family, 60(2), 442-
452.

van-der-Kolk, B.-A., & Fisler, R.-E. (1994). Childhood abuse and neglect and loss of
self-regulation. Bulletin of the Menninger Clinic, 58(2), 145-168.

225

 

 

van der Kolk, B. A. (2002a). The assessment and treatment of Complex PTSD. In R.
Yehuda (Ed), Treating trauma survivors with PTSD. Washington, DC: American
Psychiatric Press, Inc.

van der Kolk, B. A. (2002b). Self-report Inventory for Disorders of Extreme Stress

(SIDES-SR). In Psychological Trauma Assessment Package. Brookline, MA: The
Trauma Center.

van der Kolk, B. A. (2003). Psychological trauma assessment package. Brookline, MA:
The Trauma Center.

van der Kolk, B. A., Pelcovitz, D., Roth, S., Mandel, F. S., McFarlane, A., & Herman, J.
L. (1996). Dissociation, somatization, and affect dysregulation: The complexity of
adaption to trauma. American Journal of Psychiatry, 153(Suppl), 83-93.

van Ijzendoom, M. H., & Bakermans-Kranenburg, M. J. (1996). Attachment
representations in mothers, fathers, adolescents, and clinical groups: A meta-
analytic search for normative data. Journal of Consulting and Clinical
Psychology, 64(1), 8-21.

Vitanza, S., Vogel, L. C. M., & Marshall, L. L. (1995). Distress and symptoms of
posttraumatic stress disorder in abused women. Violence and Victims, 10(1), 23-
34.

von Eye, A., Mun, E. Y., & Indurkhya, A. (2004). Typifying developmental trajectories -
a decision making process. Psychology Science, 46(1), 65-98.

Walker, L. E. (1979). The battered woman. New York: Harper & Row.

Walker, L. E. (1984). Battered women, psychology, and public policy. American
Psychologist, 39(10), 1 178-1 182.

Waltermaurer, E. M., Ortega, C. A., & McNutt, L. A. (2003). Issues in estimating the
prevalence of intimate partner violence. Assessing the impact of abuse status on
participation bias. Journal of Interpersonal Violence, 18(9), 959-974.

Waters, E., Merrick, S., Treboux, D., Crowell, J ., & Albersheim, L. (2000). Attachment
security in infancy and early adulthood: A twenty-year longitudinal study. Child
Development, 71 (3), 684-689.

226

 

IIIIIII

 

Weaver, T. L., & Etzel, J. C. (2003). Smoking patterns, symptoms of PTSD and
depression: Preliminary ﬁndings from a sample of severely battered women.
Addictive Behaviors, 28(9), 1665-1679.

Weinﬁeld, N. S., Sroufe, L. A., & 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.

Wolfe, D. A., Zak, L., Wilson, S. K., & Jaffe, P. (1986). Child witnesses to violence
between parents: Critical issues in behavioral and social adjustment. Journal of
Abnormal Child Psychology, 14(1), 95-104.

World Health Organization. (1992). The [CD-10 classification of mental and behavioral
disorders: Clinical descriptions and diagnostic guidelines. Geneva: World Health
Organization.

Zielinski, D. S., & Bradshaw, C. P. (2006). Ecological inﬂuences on the sequelae of child
maltreatment: A review of the literature Child Maltreatment, 11(1), 49-62.

Zimmermann, P., Grossmann, K. E., & Fremmer-Bombik, E. (1998). Attachment in
infancy and adolescence: Continuity 0fattachment.Unpublished manuscript,
Regensburg, Germany.

227