a!>vtl

{in I

uvv Uh.
l.

. ,. :
1...: . 456.11”de
3:1.Hfi ...: . , {a}; Juntndhr.
Suns...) : V 1.53 ‘ t. .

V) :-
213.3%.uhmm».

. .
n l .0... .33
nu»... a.
a v1;
«fur : twat:
fl...-
..

.
4 v’: l2‘l‘
fitf....z..d..1..£...u.ufl
xlnh..tfln¥.nf.’!' n: 2

n .25., I. y
. .3. ~92... ... Er
.5 5

01‘:

‘3‘
v A. -
xvi“

I-

.5, . . . , . . . I.
.5: f. . . V . if, :1...

.‘aZEZl. .9 up. . , Jnv’l

L...

a .-

. A i:

51...! .
.. . P

I
K
t :
$.31

2.51%...

an:

r..«.r.!3t .1
, ,ilftu. .
1.3.1. .27.“.
.. “in...
..\

vn-QIOHOun ‘
d4

r. 5:
:3... .I.

5%.
y

.4
£3, 5.?

s

I.
u 1

.w
.

It
L:

xi? L..‘.‘&
.15 3.5.5.:
. .. .

Hun. ......n .p

Mmifimfififiifi ‘ , . ‘ ., . , ‘ , , .‘

 

 

Are umvensm usmne
ll!

Illnlllitl:Jill .llllillllllllll

3 1293 01410 1855

 

 

This is to certify that the

dissertation entitled

THE IMPACT OF ECONOMIC RESOURCES ON FEMALE
SURVIVORS OF INTIMATE MALE VIOLENCE

presented by

MAUREEN HILARY RUMPTZ

has been accepted towards fulfillment
of the requirements for

Ph.D. Psychology

degree in

 

 

We? M1442
Major professor

William 5. Davidson, II

Date July 13, 1995

 

MSU is an Affirmative Action/Equal Opportunity Institution 0- 12771

 

LIBRARY
Michigan State
University

 

 

 

PLACE IN RETURN BOX to romovo this chockout from your rocord.
TO AVOID FINES rotum on or Moro duo duo.

DATE DUE DATE DUE DATE DUE

     

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

THE IMPACT OF ECONOMIC RESOURCES
ON FEMALE SURVIVORS OF INTIMATE MALE VIOLENCE

By

Maureen Hilary Rumptz

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Psychology

1995

ABSTRACT

THE IMPACT OF ECONOMIC RESOURCES
ON FEMALE SURVIVORS OF INTIMATE MALE VIOLENCE

By

Maureen Hilary Rumptz

Many women turn to domestic violence shelters for safe housing and support after
being physically abused by an intimate partner because they do not have any other
economic or social resources, yet there is a lack of research examining the combined
impact of abuse and economic resources on women’s lives after their shelter stay. The
present longitudinal study examined the impact of abuse and economic resources on 225
women who stayed at a battered women’s shelter. Participants were interviewed at three,
nine, and fifteen months post-shelter. Descriptive, correlational, and hierarchical and
logistic regression analyses were employed to develop a model for understanding two
interrelated sets of relationships over time: 1) the relationship between abuse, economic
resources, and women’s overall psychological well-being, and 2) the relationship
between women’s level of economic resources, dependence on others for their economic
resources, and assailant involvement.

The results showed that women who were economically independent had lower
economic resources than women who were economically dependent. However, women
who were economically independent were also less likely to be involved with their

assailants. The results indicated that women who were involved with their assailants

experienced higher levels of continued abuse than women who were no longer involved
with their assailants. Further, results showed that women who experienced further abuse
reported greater dissatisfaction with their overall psychological well-being.

The concurrent and over time hypotheses that predicted women with fewer
economic resources would report greater dissatisfaction with their overall psychological
well-being were not supported. Further, women’s level of economic resources did not
mediate the relationship between abuse and psychological well-being. Economic
dependence on others did, however, moderate the relationship between women’s level of
economic resources and involvement with their assailants at nine and fifteen months
post-shelter, and across these two time periods suggesting that economic dependence was
an important factor that kept women trapped in abusive relationships. Feminist theory is
presented as perspective that provides a rationale for exploring the interconnectedness of
gender and economic oppression in woman battering research. Implications of the

findings for future research, intervention, and policy are discussed.

ACKNOWLEDGEMENTS

I consider myself extremely fortunate to have been involved in the mother study,
the Community Advocacy Project, from its beginning in 1989 through the completion of
my dissertation in 1995. I was moved, in large part, to conduct this study based on my
experiences with the Community Advocacy Project and all of the incredible people I
worked with. First and foremost, I am truly grateful to and inspired by the strength,
honesty, and generosity of the many women who shared their experiences with our
interviewers. Thank you to Cris Sullivan for ”knowing I was the right person for the job”
and for teaching me so many things, both academic and non-academic, throughout our
years as coworkers and fi'iends. I was also inspired, encouraged, and supported in this
work by my other colleagues on the Community Advocacy Project. My deepest
appreciation to the many interviewers and advocates whose dedication, intelligence,
loyalty, and openness to change made the project what it was and my life so much easier
and happier. Without them, this work would not have been possible. These friends and
work partners continually influenced the path to this study.

In addition, many other individuals significantly contributed to the development
of this research study. I cannot express enough my appreciation and thanks to William
Davidson, the chair of my committee, for teaching me so much including how to believe

iv

in myself. Cris Sullivan, Deb Bybee, Ellen Strommen, and Bob Caldwell were also
invaluable members of my committee. Thanks, also, to Tom Reischl who earlier was
extremely influential in teaching me about community psychology, research methods,
and how to write a research paper.

As is commonly true with all dissertations, this process was often very
emotionally and intellectually difficult, accompanied by moments of true inspiration and
success. Many people were close in my heart and in physical proximity throughout the
years that I worked on my dissertation. Words will fail to express how I feel about each
of these people but I believe that each of them know what is in my heart already. The
love and support I received from Tim throughout my entire graduate career, but most
especially these last couple of years, holds a large part of my heart and mind and will
affect me forever. I have never known such an unselfish, giving person in my life and
feel very lucky that we have shared our lives together over the last eight years. Members
of my dissertation support group, Barb Schillo, Madeline Wordes, and Kurt Ribisl, were
all extremely important in my life and in the development of this study. Barb has been
the truest definition of a friend over the last seven years. Without her love and support I
can honestly say I would not have finished my dissertation at this time. For all her “all I
can say”’s and “trust me”’s I will always be truly grateful. Kim - visionary, commited,
feminist, coworker, and friend - perhaps the most dedicated woman I know and the best
listener and problem solver I have ever met. Without our many talks about woman
battering and our support of one another in our commitment to keep on working I would
not be at the point I am at today. And who would I rather walk, talk, and have coffee and

V

sweets with than my sister and friend, Susan Jackson-Walker. And to Madeline, who is
a part of almost every single memory I have of graduate school and East Lansing,
Michigan, I can only say I am thankful every day that our paths crossed and we were able
to spend those years together.

Finally, thank you to my family for giving me, early on, an appreciation for
education, a desire to learn, and for providing me with an environment and opportunities

for developing myself.

vi

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................... x
LIST OF FIGURES ............................................................................................................ xi
INTRODUCTION ................................................................................................................ 1
Review of the Literature ........................................................................................... 2
Overview .................................................................................................................. 3
Physical Consequences of Woman Abuse ............................................................... 5
Psychological Consequences of Woman Abuse ...................................................... 7

The Impact of Economic Resources and Economic Dependence on Womens’
Decisions About Leaving An Abusive Relationship and Overall Well-Being ...... 11
Contemporary Explanations of the Etiology of Woman Abuse ............................. 16
Individual/psychological perspective on woman abuse ............................. l8
Sociological/family systems perspective on woman abuse ........................ 20
Feminist perspective on woman abuse ....................................................... 22
Rationale for the Present Study .............................................................................. 25
The Present Study ................................................................................................... 30
Research Objectives and Hypotheses ..................................................................... 31
METHOD ........................................................................................................................... 41
Design ..................................................................................................................... 41
Research Participants ............................................................................................. 42
Demographic and economic characteristics of participants ....................... 44
Procedure ................................................................................................................ 47
Interviewer training .................................................................................... 47
Advocacy Intervention ........................................................................................... 48
Measures ................................................................................................................. 49
Economic resources .................................................................................... 51
Abuse .......................................................................................................... 51
Psychological well-being ........................................................................... 61
Economic dependence ................................................................................ 62
Involvement with assailant ......................................................................... 63
Experimental condition .............................................................................. 63
Data Analysis Strategy ........................................................................................... 64

vii

RESULTS .......................................................................................................................... 68

Descriptive Information ......................................................................................... 68
Women’s psychological well-being over time ........................................... 69
Women’s experience of abuse over time .................................................. .70
Women’s level of economic resources over time ...................................... 74
Women’s level of economic dependence on others over time ................... 74
Involvement with the assailant over time ................................................... 76
Summary of descriptive information .......................................................... 76

Intercorrelations Among Psychological Well-Being, Involvement With

Assailant, Abuse, Economic Resources, and Experimental Condition .................. 79

Summary of the intercorrelations among psychological well-being,
involvement with assailant, abuse, economic resources, economic

dependence, and experimental condition ................................................... 83
Regressions With Psychological Well-Being as the Dependent Variable ............. 85
Abuse as the predictor ................................................................................ 86
Economic resources as the predictor .......................................................... 91
Economic resources as the mediator .......................................................... 96
Regressions With Assailant Involvement as the Dependent Variable ................. 111
DISCUSSION .................................................................................................................. 123
The Experience of Abuse and Women’s Psychological Well-Being ................... 124
Level of Economic Resources and Women’s Psychological Well-Being ........... 125
Economic Resources as a Mediator of the Relationship Between Experience
of Abuse and Women’s Psychological Well-Being ............................................. 127
The Impact of Economic Resources and Economic Dependence on
Women’s Decisions About Remaining Involved With Their Assailant .............. 129
Limitations of the Current Study .......................................................................... 133
Implications for Future Work ............................................................................... 135
Intervention ............................................................................................. -136
Research .................................................................................................. .138
Theory ..................................................................................................... .141
APPENDIX A
Items That Compose the Economic Resources Scale ......................................... 143
APPENDIX B
Revised Conflict-Tactics Scale ............................................................................ 144
APPENDIX C
Injuries Measure .................................................................................................. 145

viii

APPENDIX D

Threats Measure ................................................................................................... 146
APPENDIX E

Index of Psychological Abuse .............................................................................. 147
APPENDIX F

Revised Quality of Life Measure .......................................................................... 149
APPENDIX G

Revised CES-D (Depression Measure) ................................................................ 150
APPENDIX H

Economic Dependence Item ................................................................................. 151
APPENDIX I

Involvement With Assailant Item ......................................................................... 152
LIST OF REFERENCES ................................................................................................. 153

ix

PE”

wsav

10.

11.

12.

13.

14.

15.

LIST OF TABLES

Methodological Limitations of Research Investigating the Impact of
Economic Resources and Economic Dependence on Women Who Have

Been Battered ......................................................................................................... 27
Demographic and Economic Characteristics of Participants at Time 1,

Time 2, and Time 3 (N=225) .................................................................................. 4 5
Measurement Model ............................................................................................... 50
Corrected Item-Total Correlations for the Original Assailant and New
Relationship Physical Abuse Scales ....................................................................... 53
Corrected Item-Total Correlations for the Overall Abuse Scale ............................ 58
Descriptive Information on Psychological Well-Being Over Time ....................... 70
Descriptive Information on Abuse Over Time ....................................................... 72
Zero-Order Correlations of the Predictor and Outcome Variables in the
Regression Analyses at Time 1, Time 2, and Time 3 ............................................. 80

Summary of the Hierarchical Regression Analyses to Test Hypothesis 1:

Extent of Abuse Predicting Psychological Well-Being, Controlling for
Experimental Condition and Prior Level of Psychological Well-Being ................ 87
Summary of the Hierarchical Regression Analyses to Test Hypothesis 2:
Economic Resources Predicting Psychological Well-Being, Controlling

for Experimental Condition and Prior Level of Psychological Well-Being .......... 93
Summary of the Hierarchical Regression Analyses to Test Hypothesis 3:
Economic Resources at Time 1 Mediating the Effect Extent of Abuse at

Time 1 has on Psychological Well-Being at Time 1 ............................................ 100
Summary of the Hierarchical Regression Analyses to Test Hypothesis 3:
Economic Resources at Time 2 Mediating the Effect Extent of Abuse at

Time 2 has on Psychological Well-Being at Time 2 ............................................ 102
Summary of the Hierarchical Regression Analyses to Test Hypothesis 3:
Economic Resources at Time 3 Mediating the Effect Extent of Abuse at

Time 3 has on Psychological Well-Being at Time 3 ............................................ 104
Summary of the Hierarchical Regression Analyses to Test Hypothesis 3:
Economic Resources at Time 1 Mediating the Effect Extent of Abuse at

Time 1 has on Psychological Well-Being at Time 2 ............................................ 106
Summary of the Hierarchical Regression Analyses to Test Hypothesis 3:
Economic Resources at Time 2 Mediating the Effect Extent of Abuse at

Time 2 has on Psychological Well-Being at Time 3 ............................................ 108

X

16.

17.

Summary of the Hierarchical Regression Analyses to Test Hypothesis 3:
Economic Resources at Time 2 Mediating the Effect Extent of Abuse at

Time 1 has on Psychological Well-Being at Time 3 ............................................ 110
Summary of Logistic Regression Analyses to Test Hypothesis 4: Economic
Dependence as a Moderator of Economic Resources on Psychological

Well-Being ............................................................................................................ 113

xi

LIST OF FIGURES

Heuristic Model of the Hypothesized Relationships Between Abuse and

Psychological Well-Being Within and Across Time ............................................. 34
Heuristic Model of the Hypothesized Relationships Between Economic
Resources and Psychological Well-Being Within and Across Time ..................... 35

Heuristic Model of the Hypothesized Relationships of Economic

Resources Mediating the Effect Extent of Abuse Has on Psychological
Well-Being Within and Across Time ..................................................................... 37
Heuristic Model of the Hypothesized Relationships of Economic

Dependence Moderating the Effect of Economic Resources on Assailant
Involvement ............................................................................................................ 39
Economic Dependence at Time 2 as a Moderator of the Relationship

Between Level of Economic Resources at Time 2 and the Log Odds of
Involvement With the Assailant at Time 2 ........................................................... 116
Economic Dependence at Time 3 as a Moderator of the Relationship

Between Level of Economic Reosurces at Time 3 and the Log Odds of
Involvement With the Assailant at Time 3 ........................................................... 118
Economic Dependence at Time 2 as a Moderator of the Relationship

Between Level of Economic Resources at Time 2 and the Log Odds of
Involvement With the Assailant at Time 3 ........................................................... 121

xii

INTRODUCTION

Many women who turn to domestic violence shelters for safe housing and support
after being physically abused by an intimate partner do so because they do not have any
other economic or social resources to utilize (Okun, 1986; Sullivan & Rumptz, 1994).
Many studies on the consequences of woman battering have been conducted with
convenient samples found at battered women's shelters or other community agencies (i.e.,
women with few resources). None of these, however, have empirically examined the
impact that level of economic resources, in addition to abuse, had on women's overall
psychological well-being over time. Additionally, there is a dearth of empirical,
longitudinal research on the relationship between battered women's level of economic
resources, economic dependence, and consequential involvement with the assailant.
Feminist theory provides a rationale for research that examines the complexities of
women with abusive partners’ lives which, for the majority of women who have stayed at
a battered women’s shelter, includes not only abuse but economic hardships. The purpose
of this study is twofold: to longitudinally investigate how both abuse and economic
resources affect women's overall psychological well-being and to longitudinally examine
how women’s level of economic resources and level of economic dependence affects

their ability to leave an abusive relationship.

Review of the Literature

The purpose of this review is to examine the literature that documents the
frequency and seriousness of male violence against female partners in the United States,
establishes the physical and psychological consequences of battering on women's lives,
and documents the strong relationship between level of economic resources and battered
women's psychological well-being and ability to leave an abusive relationship. This
literature review is organized into five major sections. The first three sections provide
the background for understanding the prevalence and effects of woman battering.
Specifically, the first section of this review provides an overview of the scope of the
problem of woman battering and the critical nature of intimate male violence against
women. The second section of this review demonstrates the serious physical
consequences of intimate male violence. The third section of this review details the
equally serious psychological consequences of intimate male violence.

The fourth and fifth sections of this review provide the empirical and theoretical
rationale for examining the impact level of economic resources and economic
dependence have on women’s lives. Specifically, the fourth section of this review
focuses on the role that economic resources play in understanding and predicting
women's ability to leave an abusive relationship as well as the role that economic
resources play in predicting women's overall psychological well-being over time. The
fifth section of this review focuses on the application of theory to this area of research

and examines the major perspectives on the etiology of woman battering, suggesting

3

feminist theory as a perspective that provides a rationale for exploring the
interconnectedness of gender and economic oppression in woman battering research.
Overview

Woman battering is pandemic. Many women are battered by men they are, or
once were, involved in a relationship with. Estimates suggest that at least 2 to 4 million
women are assaulted by their male partners or ex-partners each year in the United States
alone (Browne, 1993; Straus & Gelles, 1990; Straus, Gelles, & Steinmetz, 1980), and
between 21% and 34% of all women in the United States will be physically abused by an
intimate adult partner in their lifetime (Browne, 1993; Frieze, Knoble, Washbum, &
Zomnir, 1980; Koss, 1990; Stark & Flitcraft, 1988). In a nationally representative survey
of couples in 1985, Straus and Gelles (1990) reported that nearly one-eighth of the
husbands had carried out one or more acts of physical violence against their wives in the
past year. Other research estimates that up to 50% of all wives are physically abused by
their husbands (Stark & Flitcrafi, 1988). As a society we have come to understand that
intimate male violence against women is common; it is by no means an unusual or
exceptional "family" problem, but a form of danger that is familiar, frequent, typical, and
ordinary for many women.

These statistics, widely recognized to be conservative estimates (Browne, 1993;
Straus & Gelles, 1988), make it clear that many women are battered by an intimate
partner in their lifetime. In fact, women are more likely to be assaulted, killed, or raped

by a male partner or ex-partner than by all other categories of assailants combined

4
(Browne & Williams, 1989; Finkelhor & Yllo, 1985; Goodman, Koss, & Russo, 1993;
Koss, 1985).

An additional alarming aspect of intimate male violence against women is the
substantial percentage of women who are abused and forced to remain in, or return to,
violent relationships. Many women attempt to leave their assailants numerous times
before being successful in escaping them permanently (Dobash & Dobash, 1979; Gelles,
1979; Gondolf, 1988; Hofeller, 1982; Strube & Barbour, 1983). A critical reason why so
many women remain with or return to their assailants is lack of economic and community
resources (Aguirre, 1985; Gondolf, 1988; Hofeller, 1982; Hilbert & Hilbert, 1984;
Horton, Simonidis, & Simonidis, 1987; Mitchell & Hodson, 1983; Sidel, 1986; Strube &
Barbour, 1986). In addition to these factors, women who are battered report many other
reasons for staying with abusive men such as fear of further or worse violence, concern
for their children's well-being, and assailants' promise of love and change (Dobash &
Dobash, 1979; Hofeller, 1982; Strube & Barbour, 1984). Women also stay because the
very agencies designed to help people in need (e.g., criminal justice, social services,
health care) are largely unresponsive to requests for protection and assistance that would
enable women to leave and remain free from their assailants (Dobash, Dobash, &
Cavanagh, 1985; Dutton, 1987; Gelles, 1979; Stark & Flitcraft, 1988). As a result of
male violence and inappropriate helping response, women's options are cut off and they
become entrapped in violent homes (Stark & Flitcrafi, 1988).

Remaining with an assailant or returning to an assailant can be very dangerous for

women. Research indicates that violence in intimate relationships increases in severity

5

and frequency over time (Hilbert & Hilbert, 1984; Okun, 1986; Walker, 1985). Although
women who return to their abusers are considered to be at higher risk for further violence
than women who do not return, violence does not necessarily end when women are able
to leave abusive relationships. Data indicate that separated and divorced women are also
vulnerable to battering by their ex-partners (Stark & Flitcraft, 1988).
Physical Consequences of Woman Abuse

Evidence is convincing that many women experience abuse in intimate
relationships and are often trapped in these relationships because of a range of individual
and societal barriers to leaving. The definition of abuse will be important to the
subsequent discussion of how intimate male violence negatively impacts women's lives
as well as how other factors, such as economic resources and economic dependence, also
affect women's lives. Over the last three decades there has been a near exclusive focus
on physical violence in defining woman abuse (Campbell & Fischer, under review).
Campbell & Fischer (under review) argue that for any definition to be inclusive of
women's experiences it must include not only physical behaviors of the assailant, but also
a discussion of resulting injuries, psychological violence, and sexual violence. For this
study, abuse is defined as any act that is harmful to its victim and includes physical,
sexual, and psychological battering. However, most of the literature reviewed herein is
primarily based on the consequences of physical abuse because there has not been much
research conducted on the consequences of psychological abuse.

The literature strongly supports the hypothesis that physical abuse negatively

affects women's overall physical health (Browne, 1993; Dobash & Dobash, 1979,

6

Randall, 1990; Sullivan, Basta, Tan, & Davidson, 1992). Physical battering takes many
forms ranging from pushing, grabbing, slapping, hitting with a fist or object, kicking,
choking, burning, physically restraining, to stabbing and shooting (Binney, Harkell, &
Nixon, 1981; Bowker, 1983; Browne, 1987, 1993; Dobash & Dobash, 1979, 1984;
Hilberman & Munson, 1977-78; Sullivan et al., 1992). Research suggests that sexual
abuse also occurs in at least 40% of all cases of battering (Campbell, 1989a; Finkelhor &
Yllo, 1985; Hanneke, Shields, & McCall, 1986). Injuries resulting from physical and or
sexual violence may include (but are not limited to) cuts, scrapes, or bruises, black eyes,
soreness without bruises, torn hair, burns, loose or broken teeth, broken bones or
fractures, concussions, hearing or vision impairments, internal injuries, strains, sprains, or
torn ligaments, dislocated joints, pregnancy complications or miscarriage, permanent
scarring, and knife or gunshot wounds (Browne, 1993; Dobash & Dobash, 1979, 1984;
Sullivan et al., 1992). Women who are abused when they are pregnant face the risk of
especially severe outcomes because they are less able to maneuver to escape attacks and
are at risk for injuries to the fetus as well as injuries to themselves (Goodman, Koss,
Fitzgerald, Russo, & Keita, 1993; Goodman, Koss, & Russo, 1993; Saltzrnan, 1990).

Being beaten by a male partner or ex-partner is the number one cause of serious
physical injury to women in the United States (Stark & Flitcrafi, 1987, 1988). A review
of the medical records of 3,676 randomly selected patients seen at an urban area
emergency room during a one-year period disclosed the seriousness of intimate male
violence against women; Twenty-one percent of all women using emergency surgical

services were there because they were abused by a male partner and over one half of all

7
rapes to women over the age of 30 were partner rapes (Stark, Flitcrafi, Zuckerrnan, Grey,
Robison, & Frazier, 1981). Other researchers concur that as high as 25% of emergency
room visits by women are made because they have been battered by a male partner
(Pagelow, 1992). Pagelow (1992) suggests that these statistics are almost certainly
underestimates due to inaccurate detection or incorrect reporting by health care
professionals.

Tragically, many women are injured to the point of being killed by the men who
sought to control them. Browne (1993) concluded that women are more likely to be
killed by their partners than by all other categories of persons combined. Browne &
Williams (1993) found that more than half of all women murdered in the United States
during the first half of the 1980's were killed by their male partners. Browne & Williams
(1993) analyzed all criminally negligent homicides from 1976 through 1987 and found
that the deaths of approximately 38,648 individuals aged 16 and above resulted when one
partner killed another - 61 % were females killed by male partners and 39 % were males
killed by female partners. It is important to point out, however, that most women who
kill their male partners do so in self-defense (Browne, 1993).

Pachological Consequences of Woman Abuse

The literature also maintains that there is a negative relationship between physical
abuse and women's overall psychological well-being (Gelles & Harrop, 1989; Goodman,
Koss, Russo, 1993; Koss, 1990). Symptoms include memory loss, cognitive
dissociations, sleep and appetite disturbances, chronic fatigue and tension, listlessness,

somatic problems, re-experiencing of the traumatic event when exposed to associated

8
stimuli, thoughts of suicide, and suicide attempts (Browne, 1987; Dobash & Dobash,
1979; Dutton, 1992; Hilberman, 1980; Hilberman & Munson, 1977-78; Hoff, 1990; Jaffe,
Wolfe, Wilson, & Zak, 1986; Stark & Flitcrafi, 1987, 1988; Walker, 1979, 1983, 1984).
Researchers further have been able to document a positive relationship between the
frequency and severity of the abuse and the frequency and severity of the psychological
symptoms women experience as a result (F ollingstaad, Brennan, Hause, Polek, &
Rutledge, 1991; Mitchell & Hodson, 1983).

Gelles & Harrop (1989) conducted a random survey of 6,002 households which
yielded a nationally representative sample of battered and non-battered women. They
found that women who had been physically abused reported higher levels of moderate
and severe psychological distress than women who had not been physically abused. In
reviewing several surveys, Koss (1990) also concluded that women who were physically
abused showed higher degrees of mental health problems when compared with women
who had not been physically battered. Specifically, she noted that battered women were
more likely to be diagnosed with depression, alcohol and drug dependence or abuse,
generalized anxiety, obsessive-compulsive disorder, eating disorders, posttraumatic stress
disorder, and other psychological diagnoses than non-battered women (Koss, 1990).

Other researchers have reached similar conclusions. Sato & Heiby (1992) found
that almost half of their sample (47%) reported clinically significant levels of symptoms
associated with depression, leading them to conclude that women who have been
physically abused may be at higher risk for depression and other symptoms of

psychological distress. Gleason (1993) reported a higher prevalence of depression,

9
posttraumatic stress disorder, generalized anxiety disorder, psychosexual dysfunction,
and obsessive compulsive disorder among 62 battered women receiving assistance from a
community agency than in a national study of 10,953 randomly sampled women.

Of all of the psychological symptoms associated with battering, depression is one
of the most prominent (Hilberman & Munson, 1977-78; Walker, 1984). Consistently,
studies suggest that women who have been physically battered report higher rates of
depression than either the general population or samples of women who have not been
physically abused (American Psychiatric Association, 1987; Campbell, 1989b; Gleason,
1993; Walker, 1984). Sato & Heiby (1992) found that approximately half of their sample
(47%) of women involved in battered women's groups, staying at shelters, or presenting
for help at a community agency reported clinically significant rates of depression.

A longitudinal study examining the relationship between physical abuse and
depression on a sample of women who had stayed at a shelter for women with abusive
partners found further evidence that physical and emotional abuse are correlated with
depression (Campbell, Sullivan, & Davidson, 1995). Campbell et a1. (1995) found that
83% of their sample reported feelings of depression immediately after exiting the shelter
and only 58% reported feelings of depression 10 weeks later. At a 6-month follow-up,
59% reported feelings of depression.

Recently, the literature on the psychological consequences of woman abuse has
reported posttraumatic stress disorder (PTSD) as a diagnosis for the psychological
symptoms women report after being abused. The most common trauma suggested as a

necessary precursor to PTSD is very applicable to a battered woman's situation; that is, a

10
serious threat or harm to a person's life or physical integrity including that of their
children (American Psychiatric Association, 1987). Some researchers suggest that PTSD
more adequately describes the range of psychological symptoms many women who have
been battered experience than other diagnoses such as depression (Browne, 1993;
Goodman, Koss, Fitzgerald, Russo, & Keita, 1993; Koss, 1990; Walker, 1984).
Commonly reported symptoms by women such as fear and anxiety, agitation, memory
loss, reexperiencing the trauma, nightmares, and sleep disturbances can be explained
within a PTSD diagnosis (Browne, 1993). The prevalence of PTSD in women who have
been physically abused range from 33% (using a conservative estimate of PTSD) (Astin,
Lawrence, & Foy, 1993) to 84% (Kemp, Rawlings, & Green, 1991). In both of these
studies, the level of PTSD significantly correlated with the extent and severity of
exposure to violence.

Another serious psychological consequence for many women who have been
abused by an intimate male partner are suicidal thoughts and attempts (Gleason, 1993;
Hilberman & Munson, 1977-78; King, 1981; Koss, 1990; Stark & Flitcraft, 1982, 1987;
Straus & Gelles, 1987). Estimates suggest that 26% of the women who attempt suicide
are victims of domestic violence (Randall, 1990). Gayford (1975) reported that 50% of a
sample of women who had been battered had tried to commit suicide at least once.

Although the bulk of evidence has led researchers to conclude that a history of
victimization is a strong risk factor for the development of psychological problems
(Kilpatrick, Saunders, Veronen, Best, & Von, 1987), the longitudinal study of depression

in women who have been battered by Campbell et al. (1995) suggests the importance of

11
examining the relationships over time in order to adequately understand the phenomena.
For example, it is unclear whether women who have been battered suffer from actual
clinical depression or whether their level of depression represents a healthy, temporary,
and/or normal response to a stressful and negative life event such as physical abuse by an
intimate male partner. The findings of Campbell et al. (1995) suggest that the latter may
be true. Further longitudinal research is necessary to document the exact nature of the
relationship (Campbell et al., 1995; Sato & Heiby, 1992).
The Impact of Economic Resources and Economic Demndence on Womens’ Decisions
About Leaving an Abusive Relationship and Overall Well-Being.

It is clear from a review of the literature that experiences of woman battering
result in both primary and secondary psychological and physical problems and have
serious detrimental effects on women's overall well-being. Of particular interest in this
study is the additional impact that women’s level of economic resources and economic
dependence has on their psychological adjustment and decisions about staying with the
assailant. Prior research on woman battering suggests that lack of economic and
community resources and economic dependence impacts womens’ decision process
about whether to remain with an assailant thereby often serving to trap them in
relationships where they are further abused (Aguirre, 1985; Carlson, 1977; Dobash &
Dobash, 1979; Gelles, 1976; Greaves, Heapy, & Wylie, 1988; Hofeller, 1982; McDonald,
1989; Pagelow, 1981; Roy, 1977; Strube & Barbour, 1983, 1984). There is also a strong
literature which documents that lack of economic resources has a negative impact on

women’s overall well-being. To further understand the relevance of economic resources

12
on battered women’s lives, following is a review of the literature documenting the
relationship between economic resources and womens' decisions about continued
involvement with their assailants and a review of the literature on the significant impact
level of economic resources can have on womens' overall psychological well-being.

In further trying to understand the full range of consequences of woman abuse
and the process involved in ending the abuse, researchers have established that many
women continue to endure physical attacks fi'om their male partners because the
alternative of separating often means living in poverty or living with considerably fewer
economic resources than before the separation (Aguirre, 1985; Carlson, 1977; Dobash &
Dobash, 1979; Gelles, 1976; Greaves, Heapy, & Wylie, 1988; Hofeller, 1982; McDonald,
1989; Pagelow, 1981; Roy, 1977; Strube & Barbour, 1983, 1984). The sexist economic
and occupational structure in the United States limits women's alternatives when trying to
escape an abusive partner (Straus, 1977). In the United States, women are more likely
than men to be poor, more likely to remain poor once they enter poverty, and are
increasingly likely to be left with the responsibility for children without access to the
resources of the other parent (Wilson, 1987). Wilson (1987) also documented that
women who were the sole provider in their homes were nearly five times as likely to be
poor as men who were the sole provider in their households. Without access to good,
decent paying jobs with adequate child care and health care, women are often forced to
be dependent on their male partners.

Many researchers have pointed out that economic dependence on male partners is

a major obstacle that prevents women from leaving abusive relationships (Aguirre, 1985;

13
Carlson, 1977; Dobash & Dobash, 1979; Gelles, 1976; Greaves, Heapy, & Wylie, 1988;
Hofeller, 1982; McDonald, 1989; Pagelow, 1981; Roy, 1977; Strube & Barbour, 1983,
1984). Lack of money and/or employment was one of three main factors which Gelles
(1976) found associated with abused women remaining with their partners. McDonald
(1989) suggested that lack of employment and inadequate finances limited the
alternatives available to women and increased their entrapment in abusive relationships.
Hofeller (1982) found that 58% of a sample of women who had been abused remained
with their assailants because of economic reasons; that is, they felt that they could not
support themselves (and their children, if applicable) on their own. Strube & Barbour
(1983) also found that employment and economic hardship contributed to the decision of
whether to stay or leave an abusive partner. Aguirre (1985) examined four antecedent
variables (previous experiences with violence, number of injuries, issues experienced
during abuse, and economic dependence on husband) and determined that the only
variable that influenced womens' decision whether to return to an abuser was economic
dependence on him.

It is important to also consider that batterers often purposefully increase this
dependence by successfully keeping many women unemployed (Shepard & Pence, 1988)
and isolated from their family and friends (Mitchell & Hodson, 1983). Many batterers
also threaten economic and physical reprisal against women and their children if they
attempt to leave (Dobash, Dobash, & Cavanagh, 1985).

Research also suggests that the more economic resources women have (such as

employment, income, material goods and services that money can buy, and level of

14
education), the more likely they are to seek outside intervention (Carlson, 1977), and the
better their chances of leaving their assailant (Sullivan, 1991). Women with more
resources can more easily afford to move or stay in a hotel, are more likely to have a
number of friends or family who also have more financial resources and can assist in
some way, and are more likely to have cars to leave the area if necessary.

For women without many economic resources a domestic violence shelter may be
the only safe, available, and accessible place to go to when escaping an abusive man.
Although battering cuts across all socioeconomic, racial, ethnic, and religious lines,
women who utilize domestic violence shelters tend to be more economically
impoverished than the general population (Okun, 1986; Sullivan & Rumptz, 1994). In
addition to needing economic or financial resources, women who use domestic violence
shelters have been found to need a variety of community resources such as housing, legal
services, health care, material goods and services, and childcare (Sullivan et al., 1992).

For women who escape violence to seek refuge in a domestic violence shelter,
limited economic resources are often a significant factor that not only influences whether
they return to their assailants (Aguirre, 1985; Strube & Barbour, 1983), but may
generally influence their overall psychological well-being as well. Research has
established the strong relationship between poverty and women's psychosocial well-
being. Generally, individuals with economic resources fare much better than those
without, as economic resources increase the coping options in many stressful situations,
including intimate male violence against women. As Dohrenwend (1978) pointed out,

"other things being equal, an individual whose financial or other material resources are

15
strained by the demands of a stressful life event is likely to have a worse outcome than a
person with adequate material resources" (p. 5). Individuals who live under high stress
and have few coping resources - including women living in poverty (Belle, 1990) - are at
especially high risk for negative psychological sequelae (Goodman, Koss, Russo, 1993).

Reviewing the epidemiological and ecological literature on the relationship of
economic factors to psychosocial dysfunction, Seidman & Rapkin (1983) found that at
both the macro- and the microlevel, indices of economic conditions (e.g., unemployment
rates, social class status, common stock prices, business indicators, individual job loss,
etc.) are correlated with rates of numerous psychosocial maladaptations (suicide rates,
hospital admissions, outpatient treatment, etc.). As they pointed out, "economic stressors
function as precursors of preconditions contributing to the prevalence and incidence of
any number of psychosocial maladaptations" (Seidman & Rapkin, 1983).

Research on women and poverty suggest that lack of economic resources has both
long and short term consequences for women's lives including placing them at risk for
psychological and physical health problems. Belle, et al. (1982) found that low-income
women have higher rates of depression and other psychological problems compared to
population base rates. Studies reviewed by Belle (1990) found high levels of depressive
symptoms among unemployed and socially isolated women, among poor women with
responsibility for young children, and among women experiencing economic difficulties.
Belle (1990) pointed out that the association between low income with psychological

problems is not surprising given that poverty itself is a stressor, bringing with it persistent

16
undesirable chronic conditions such as inadequate housing, inadequate nutrition,
dangerous neighborhoods, poor medical care, and financial uncertainties.

At the same time that poverty is a source of stress, it also undermines the social
support system of the individual (Belle, 1990). Relatives and fi'iends of poor women are
themselves likely to have economic problems and to be stressed. Furthermore, poor
women have to depend largely on overburdened and unresponsive societal institutions for
financial, legal, health, and social services that may contribute to their diminished
psychological well-being (Belle, 1990).

Because the bulk of research on woman battering is conducted with women who
utilize domestic violence shelters, it seems especially relevant to decipher the
relationship that level of economic resources has on women’s lives as they cope with the
abuse and make decisions about their future. To understand the complicated process and
outcomes for women surviving abuse from their partner or ex-partner, it is necessary to
examine not only the impact of abuse on women’s lives, but the impact of other relevant
factors such as level of economic resources and economic dependence. The feminist
perspective on the etiology of woman battering provides a conceptual basis for such an
analysis.

Contemmm Explanations of the Etiology of Woman Abuse

The preceding sections of this literature review documented the frequency and
seriousness of male violence against female partners in the United States, established the
physical and psychological consequences of battering on women's lives, and documented

the strong relationship between economic resources, economic dependence,

l7
psychological well-being, and assailant involvement. To best understand the relationship
between abuse, economic resources, involvement with assailant, and psychological well-
being, it is important to examine them in some theoretical context.

As woman battering has gained more attention over the past three decades there
has been an evolution of sorts in terms of the theories social scientists use to understand
this social problem. The most popular theories of woman battering have emerged from
psychological/individual, sociological/family systems, and/or sociocultural/feminist
perspectives. These three theoretical perspectives have not necessarily evolved
chronologically, but have evolved out of different perspectives on the root causes of
social problems. The basic tenets of each of the three primary perspectives on woman
battering focuses on a different level of conceptualization regarding the etiology of
woman battering and society's response to it. To better understand the relevance of
studying the impact of economic resources and economic dependence on battered
women's lives over time, all three theories are reviewed below. Each theoretical
perspective is reviewed for the evidence it has to support that women’s level of economic
resources and economic dependence influences women’s psychological well-being and
decisions about leaving their assailant after exiting a battered women’s shelter.

It is important to note that some researchers have chosen to work within a fourth
perspective, or ecological perspective, which integrates what has been learned within
each of the abovementioned three frameworks and suggests that the only way to more
fully understand how individuals, families, institutions, and culture are involved in

woman battering is to explore the interactions of each of these systems (Carlson, 1984;

18
Dutton, 1985; Sullivan, 1989). As this perspective is an integration of the primary three
theoretical perspectives, it will not be reviewed here.

Individual/psychological pgrspective on woman abuse. Much of the early
research literature on woman abuse focused on individual level variables, especially
individual characteristics of women that were hypothesized to make them particularly
vulnerable to and responsible for abuse. There are three major analyses within this
perspective. The first analyses, and one of the earliest theories to arise in the area of
woman battering, was that men batter their female partners when their female partners
provoke them to do so. Gelles (1972) differentiated between victim-precipitated violence
and non-victim-precipitated violence and argued that victim-precipitated violence occurs
when women contribute to their assault through actions that provoke their assailant.
Gayford (1978) elaborated on this theory and suggested that women are at least partially
responsible for the abuse through nagging, being too quiet, being careless with money, or
through doing nothing at all. This theory has been widely criticized for focusing on the
victim's behavior instead of the assailants (Martin, 1979), and for ignoring the fact that
often there aren’t any precipitating factors to woman abuse (Dobash & Dobash, 1979;
Martin, 1976; Okun, 1986).

A second analysis within the psychological or individual level theory of woman
battering is that men who batter and women who are battered hold more traditional views
of gender roles than men who do not batter and women who are not battered. The
hypothesis is that men who batter women and women who are abused come from violent

families of origin and therefore consider abuse normal, common, and unavoidable (Roy,

19
1977). To date, the most widely agreed upon cause of family violence cited in the
woman battering literature is witnessing parental violence in one's family of origin
(Carlson, 1984; Straus, 1977). Aside from being female, Hotaling and Sugarman (1986)
found witnessing violence in women's family of origin to be the only consistent risk
marker that predicted abuse for women. For men, three primary risk markers were
identified as witnessing violence as a child, violence toward one's children, and sexually
abusing one's wife. Caution should be taken, however, in interpreting these results.
Some researchers suggest that witnessing abuse as a child should be viewed cautiously as
a vulnerability, not a causal factor. For example, Stark and F litcraft (1979) suggested
that for every one boy who witnesses abuse and grows up to be abusive, two boys who
witness abuse will not grow up to be abusive.

Hotaling and Sugannan (1986) found additional risk markers for men included
occupation, education, income, lack of assertiveness, and alcoholism. Other research
also concluded that alcoholism was an individual level cause of family violence and also
mentioned status incompatibility (women having more personal resources than men and
thereby threatening their dominant position in the family) and stress as contributors
(Carlson, 1984; Dutton, 1985).

A third psychological or individual level analysis for why men batter women is
that women learn to be helpless in their assaultive relationship thereby allowing the abuse
to continue. Walker's (1977-78, 1979) theory of learned helplessness purports that
battered women develop poor self-concepts from having been abused and consequently

believe they are powerless to end or escape from their partners violence. Carlson (1984)

20
also suggested that women with abusive partners exhibit low self-esteem. However,
other researchers have pointed out that low self-esteem was not necessarily the cause of
battering but the result of battering (Campbell, 1992). Walker (1977-78, 1979) suggested
that women with abusive partners need to change their cognitive set to believe their
competent actions can change their life situation.

As with the other individual or psychological level theories, the learned
helplessness analysis puts the blame on the victim. The barriers that women face when
attempting to leave their assailants are ignored. All three analyses within this perspective
fail to address how economic resources, economic dependence, lack of housing, fear for
their lives and the lives of their children, lack of employment and/or education, and social
pressure all contribute to a woman's entrapment in an abusive relationship. Given the
societal barriers facing women such as employment discrimination, lack of adequate and
affordable child care, and inadequate police protection from the abuser, it is very
reasonable for a battered woman to believe that her competent actions are not enough to
end the abuse.

Sociological/family systems persmctive on woman abuse. The psychological or
individual focus shifted in the early 1980's as the research was influenced by a large
national random sample survey conducted by Straus, Gelles, and Steinmetz (1980). Their
research suggested that woman battering by male partners and ex-partners was only a
part of a larger problem of societal norms and mores that work to perpetuate violence in
the family and should be viewed in an environmental context rather than as an individual

problem. Within this perspective, generalized violence in society, using parental

21
violence to discipline children, violence among siblings, women's subordination, and the
increasing social isolation of families are all acknowledged as contributing factors in
family violence and responsible for transmitting violence from one generation to another
(Carlson, 1984; Dutton, 1985; Kurz, 1993; Straus, 1977; Straus, et al., 1980).

While there is some validity to the argument that social factors contribute to
violence within the family, they can only be a partial explanation for woman battering.
For instance, although it is true that many cultural norms support violence within
families, there also exist cultural norms that condemn such behavior (e.g., there are
certainly societal expectations that violence against one's wife is unacceptable and that
families should be safe, respectful, and free from violence). The sociological or family
systems perspective does not adequately explain the interrelationship between these
contradictory norms or how one may come to dominate over the other.

Another shortcoming with this perspective is that it focuses on violence as a result
of family conflict but does not explain why so many more men than women are violent,
since both are subjected to a great deal of stress at home and at work. Nor does it take
into account that a great deal of battering is not precipitated by the level of conflict
within the family (Dobash & Dobash, 1979; Hofeller, 1982; Okun, 1986).

Another sociological argument about the causes of woman battering is the
personal resources analysis of battering. This analysis contends that men use violence to
maintain power in relationships (Allen & Straus, 1980). Allen & Straus (1980) found a
correlation between resources and power for working class men; that is, working class

men who scored very low in resources also scored higher on the use of violence against

22
their partners. This suggests that in families where men have a lower level of power and
resources, they will be more prone to violence in order to gain additional power. A
problem with this theory is the definition of power. In this study, Allen and Straus
(1980) asked couples questions pertaining to who decides what they will eat for dinner or
what they will watch on television rather than about which partner had more access to
actual societal resources such as a job, an income, status in the community, and various
other advantages. Viewing power in this manner minimizes the very real disadvantages
that the vast majority of women face in contemporary marriages compared to their
husbands.

The sociological/family systems perspective on woman abuse comes closer to
addressing the importance of looking at the problem of woman battering from a broader
perspective, thereby including multiple forces (e.g., women’s subordination, economic
resources) as potentially relevant factors in causing men to batter women. But, as with
the individual level perspective, the sociological/family systems theory does not
adequately address the societal barriers battered women face (e.g., economic resources,
economic dependence, lack of housing, fear for their lives and the lives of their children,
lack of employment and/or education, and social pressure) and how these barriers impact
their lives.

Feminist pgrsp_ective on woman abuse. Feminist theorists and researchers who
have studied woman abuse for three decades agree that violence against women belongs
in a larger social context than the individual or the family (Davis, 1985; King & Bohom,

1993; Schechter, 1982). The feminist perspective on woman abuse argues that the norm

23
of male dominance within our broader culture and incorporated into our formal
institutions are the roots of woman battering. Men batter women because they are
socialized within a patriarchal social structure in which institutional sexism contributes to
the denigration and abuse of women in society (Schechter, 1982). Proponents of this
perspective cite such factors as women's subordinate role in society, institutionalized
sexism, traditional sex role stereotyping, and a history of male domination as reasons for
why men use violence to control women and assert power in their relationships. Further,
feminist theorists suggest that intimate male violence is legitimized and perpetuated by
major societal institutions such as the health care, criminal justice, and social service
systems through their failure to identify, help, and protect female survivors (Bowker,
1987; Kurz, 1993; Okun, 1986).

The feminist perspective is most relevant to this study in that it seeks to
understand and explain the relationship between abuse and economic resources. The
feminist perspective supposes that women have less power in society than men do and
are economically disadvantaged and, therefore, often forced to be economically
dependent on men. Feminist theory on woman battering argues that this inequitable
distribution of economic resources between men and women contributes to women's
economic dependence on men, limiting women's options and often forcing them to
remain in abusive relationships (Corcoran, Duncan, & Hill, 1984; Dobash & Dobash,
1992; Edwards, 1985; Hofeller, 1982; Okun, 1986).

Cross cultural research supports the argument that battering is related to the

economic oppression of women. Levinson (1989) found that wife-beating is more likely

2 4
in societies in which men control the family wealth. In their cross cultural research on
woman abuse, Counts, Brown, & Campbell (1992) found that in most cultures, women
are oppressed by men and intimate male violence against women is common. However,
in cultures where women had more power, they found intimate male violence against
women to be less common.

The feminist perspective further suggests that woman abuse should be categorized
with other types of violence against women such as rape, incest, and pornography.
Feminist theory further proposes that intimate male violence against women in the United
States should be categorized with other forms of violence under capitalism such as racism
and classism (Williams, 1981). Therefore, it is essential to see woman battering within
the framework of misogyny, the feminization of poverty, forced homelessness, and racial
discrimination. Feminist theory contends that because multiple oppressions influence
womens's lives and all are rooted in the patriarchal social structure (e.g., women are
oppressed because they are women but also may be oppressed because they are poor or
women of color), any analyses of social problems affecting womens lives would benefit
by analyzing multiple factors.

Feminist theory is more applicable to the problem of intimate male violence
against women than other theoretical perspectives that address inequalities, such as
conflict theory or social structuralism, because it places gender as the basis for such
inequalities. Domestic violence is overwhelmingly a crime commited by men against
their intimate female partners and, as such, must be understood within the context of

gender.

2 5
Rationale for the Present Study

There is a considerable body of knowledge within the woman battering literature
that documents the pervasiveness and seriousness of intimate male violence against
women and the various negative physical and psychological outcomes that abuse
(whether physical, sexual, or psychological) has on women's lives. The literature also
contends that women's level of economic resources and whether they are dependent on
another person for those resources plays a central role in influencing their: 1) decision
about continuing involvement with their assailants (Aguirre, 1985; Carlson, 1977;
Dobash & Dobash, 1979; Gelles, 1979; Greaves, Heapy, & Wylie, 1988; Hofeller, 1982;
McDonald, 1989; Pagelow, 1981; Roy, 1977; Strube & Barbour, 1983, 1984), and 2)
psychological well-being over time (Belle, 1990; Belle, Dill, Feld, Greywolf, Reese, &
Steele, 1982; Dohrenwend, 1978; Goodman, Koss, Russo, 1993; Seidman & Rapkin,
1983)

As the literature review suggested, women who stay at domestic violence shelters
tend to have fewer resources to begin with than women who do not turn to domestic
violence shelters after being battered (Okun, 1986; Sullivan & Rumptz, 1994). Much of
the research in the area of woman battering has been conducted with women at domestic
violence shelters but has not incorporated an analysis of the impact of economic
resources and economic dependence on women’s lives. Therefore, the relationships
researchers currently suggest between abuse and other constructs like involvement with
assailant or psychological well-being may be confounded by level of economic resources

or economic dependence. Russo (1990) suggests the importance of future research that

26
examines what variables mitigate the impact of abuse on women's lives in order to
understand the factors that influence the differential outcomes of woman abuse. For
example, to understand what factors affect women’s psychological well-being and
decisions about staying involved with the assailant. The present study contributes to the
literature by empirically exploring these relationships over time in order to begin to
conceptualize a model for how level of economic resources and economic dependence
impacts women who have been battered by their partners or ex-partners. Research on
these factors could have important implications for future theory building, intervention
strategies, research, and policy decisions.

In addition to beginning to develop a model to explain the role of economic
resources and economic dependence in predicting women’s psychological well-being and
involvement with their assailants, the present study contributes to the literature on
woman battering by remedying several common methodological weaknesses that limit
the causal inferences and definitive conclusions of past research. Table 1 provides a
summary of the methodological limitations of the research reviewed herein that has
investigated the impact of economic resources and economic dependence on women who
have been battered. To date, much of the existing research has been descriptive or
anecdotal in nature. That is, results are often limited to simple tabulation of frequencies
and percentages. The present study employed correlational and regression analyses to

further explore the relationships among key variables in this area of research.

 

27

 

95328:. 533-5 0: o: 3; 3» $3: 53:3,.
5036.5
2.23 on cocoaé o: o: .3 8» A33: 3.5m
$3235
0:23 a. $33.5 o: o: «a 8» Anna: 3.5m
.333 costs 8» 8% 2.2- _ 3 porn.» no» 333 2325
83 32:03 no.» no» of mo» :3 3 .81
San Ecouutcmoo
”32235 .883
-5 R033 55.? we» 8» p23.» 8» Ema 3 sac—$5
>955 52.5, me.» o: 8 no» G: S =ono=2
Burr—85 83$ cow—oi mo» 8» vmm 8» Anna: 8320
303.52: gangs: mo.» 8» on 33:3 GS: 8:00
3033:: 5.4395 no» we» 3. 8» Am; C .8280
zeta 59.5 8» o: N. m 8% Anna C oE=w<
amasé 8:83... an; E52 :3
on»... zeta—m .mcouoom $.20 0331080 325 35 295m 29:5 252—350 35.2. DEE

 

 

_ use.

28

Most of the research to date has also been cross-sectional, and therefore limits the
conclusions about the stability of the relationship between major outcome variables.
Interviewing women at only one point in time while they are in shelters or immediately
upon shelter exit provides only limited information about the effects of battering on
women's overall well-being (Campbell et al., 1995; Sullivan, Campbell, Angelique, Eby,
& Davidson, 1994). For example, the information cross-sectional studies provides can be
very misleading regarding the psychological well-being of women with abusive partners.
Campbell et al. (1995) and Sullivan et al. (1994) found that in a sample of shelter users,
women reported much higher levels of depression immediately post-shelter than they did
six months later. Although interviewing women during a crisis period can be important
to our understanding of critical life events, care must be taken not to overgeneralize these
findings to battered women's long-term psychological well-being. Currently, there is an
insufficient amount of information known about the process women undergo over time as
they attempt to escape abusive men with whom they are or have been involved with. The
present study included multiple data collection points (3 months post-shelter, 9 months
post-shelter, and 15 months post-shelter) to help sort out what factors influence women's
decisions and psychological well-being over time.

Another weakness throughout the woman abuse literature is a lack of agreement
on a single definition of abuse and differential emphasis on the dimensions of abuse
(Campbell & Fischer, under review). Most often, physical battering is examined to the
near exclusion of psychological and sexual battering as part of woman abuse (F inkelhor

& Yllo, 1983; Follingstad et al., 1990; Russell, 1990). As there is ample evidence that

29
psychological and/or sexual abuse oflen accompany physical violence and are at least as
damaging as physical violence, the definition of abuse must be expanded in our research
(Browne, 1987; Campbell, 1989; Campbell & Fischer, under review; Follingstad,
Brennan, Hause, Polek & Rutledge, 1991; F ollingstad, Rutledge, Berg, Hause, & Polek,
1990; Frieze, 1983; Hanneke, Shields, & McCall, 1986; Hilberman & Munson, 1977-78;
Pagelow, 1984; Walker, 1984). The present study incorporated physical (including one
item assessing sexual abuse) and psychological abuse in the measure of abuse and
documented the conceptual and empirical rationales behind the development of the scale.

In addition to the failure to longitudinally and empirically examine the impact of
abuse and economic resources on women's psychological well-being and return to the
assailant, there are also wide gaps in our knowledge because theory is rarely integrated
with research. Becerra and Zambrana (1985) concluded that data and a research agenda
are meaningless without a context in which to interpret them. To date, there is a lack of
theory-informed research on the actual relationship between shelter users' economic
resources, economic dependence, involvement with the assailant, further abuse, and
psychological well-being over time.

Women in shelters are often both poor and battered. Their situation symbolizes
the complexity of the broader patterns of institutionalized economic and sexual
exploitation, political discrimination, and male domination against women which serves
as the basis for woman battering. Although all women who have been abused by a male

partner would share the experience of intimate male battering in common, differing

30
levels of economic resources or economic dependence might be factors that influence
differential outcomes.

The present study contributes to the literature by presenting feminist theory as
support for the importance of looking at economic resources and economic dependence
within the context of woman abuse. To date, researchers have maintained a narrow
definition of violence against women. In the struggle to end intimate male violence
against women, many researchers focus only on male violence instead of the broader
system of capitalism and its institutions that perpetuate both sexism and classism in their
very structure (Williams, 1981). It is important, however, to understand abuse of
impoverished women in relation to both their economic resources and gender, for each
interpenetrate one another. The present study investigated woman abuse within the
context of economic oppression facing women who have resided at a domestic violence
shelter.

The Present Study

The present study was part of a larger ongoing study. Feminist theory and
previous research findings suggest that when women leave a domestic violence shelter,
both abuse and economic resources/dependence may play important roles in predicting
their overall psychological well-being over time (Belle, 1990; Belle, Dill, Feld,
Greywolf, Reese, & Steele, 1982; Dohrenwend, 1978; Gelles & Harrop, 1989; Goodman,
Koss, Russo, 1993; Koss, 1990; Seidman & Rapkin, 1983). Research also suggests that
economic resources or economic dependence on the assailant play a prominent role in

predicting return to the assailant (Aguirre, 1985; Carlson, 1977; Dobash & Dobash, 1979;

31

Gelles, 1976; Greaves, Heapy, & Wylie, 1988; Hofeller, 1982; Pagelow, 1981; Roy,
1977; Strube & Barbour, 1983, 1984) . Feminist theory supports that to more clearly
understand the dynamic process involved for women post—shelter and to gain further
insight into the multiple factors impacting women's psychological well-being, we must
begin to examine the complexities of women's lives and examine the multiple
oppressions that impact them over time (Corcoran, Duncan, & Hill, 1984; Davis, 1985;
Dill, 1983; Dobash & Dobash, 1992; Edwards, 1985; Hofeller, 1982; Okun, 1986). This
research was designed to explore the complex relationships between abuse, economic
resources, economic dependence, involvement with the assailant, and overall
psychological well-being as described by the following three objectives and specific
hypotheses. Information was collected through face-to-face interviews with women at
three time points: 3, 9, and 15 months after leaving a domestic violence shelter.
Research Objectives and Hypotheses

Given the lack of agreement on definitions of abuse and psychological well-being
in the woman abuse literature, and the lack of measures of economic resources, the first
objective was to conceptually and empirically create meaningful scales to measure these
constructs. With regard to measuring abuse, the literature suggests that both physical
abuse by the original assailant or any new relationship and psychological abuse by a
current partner (either the original assailant or a new relationship) may contribute to an
overall abuse score. Therefore, items measuring physical abuse (including physical
abuse, sexual abuse, and injuries) by the original assailant or new relationship, threats,

and psychological abuse by whoever women were currently in a relationship with were

32
combined in the present study to create an overall abuse measure. Overall quality of life
and depression have been suggested as indices of overall psychological well-being and
items assessing both of these constructs were combined in the present study to create an
overall psychological well-being measure. Items assessing level of economic resources
were combined to create an economic resources measure.

Given the lack of empirical, longitudinal studies on the relationship between
abuse, level of economic resources and economic dependence, involvement with the
assailant, and psychological well-being of women who were battered and resided at a
domestic violence shelter, the second objective was to provide a descriptive account of
abuse, women's economic resources, economic dependence, involvement with assailant,
and psychological well-being. The descriptive information provides the contextual
background for this exploratory model development study.

The third, and primary, objective was to examine the relationships between abuse,
economic resources, economic dependence, involvement with the assailant, and
psychological well-being. As the first step in examining these relationships, this research
determined the correlations between the five primary variables within each time point
(Time 1 = 3 months post-shelter, Time 2 = 9 months post-shelter, and Time 3 = 15
months post-shelter) and across all three time points.

As the second step in examining these relationships, the present study predicted
relationships between abuse, economic resources, and psychological well-being within
each of the three time periods and across time. Of particular interest were the relative

strength of abuse and economic resources in predicting women's psychological well-

33

being and whether economic resources mediated the relationship between abuse and
psychological well-being. The present study also explored the relationships between
economic resources, economic dependence, and involvement with the assailant within
Time 1, Time 2, and Time 3 and across time. Of particular interest in these analyses was
whether economic dependence moderated the relationship between economic resources
and involvement with the assailant.

The following specific hypotheses were tested:
Hypothesis 1: The literature and previous research suggest that abuse from intimate male
partners is a major stressor that affects women's overall psychological well-being. The
more severe the abuse, the lower women's psychological well-being. Figure 1 shows a
heuristic model of the hypothesized relationships between abuse and psychological ‘well-

being within and across time.

34

Figure 1
ammmmnmmmmmmmmrmmmnmmm
andAcrosinme

 

 

[ __ .———— _

 

 

 

 

 

 

 

 

  

 

 

 

1 Abuse T1 Psychological Well-Being T1
1 .
5 Abuse T2 Psychological Well-Being T2 .
i !
1 Abuse T3 Psychological Well-Being T3
L______.' l E
1. The experience of abuse at Time 1 has a direct negative effect on the
psychological well-being of women at Time 1.
2. The experience of abuse at Time 2 has a direct negative effect on the
psychological well-being of women at Time 2.
3. The experience of abuse at Time 3 has a direct negative effect on the
psychological well-being of women at Time 3.
4. The experience of abuse at Time 1 has a direct negative effect on the
psychological well-being of women at Time 2.
5. The experience of abuse at Time 2 has a direct negative effect on the

psychological well-being of women at Time 3.
Hypothesis 2: The literature and previous research suggest that level of economic

resources impacts women's overall psychological well-being. The lower the economic

35
resources, the lower women's psychological well-being. Figure 2 displays a heuristic
model of the hypothesized relationships between economic resources and psychological
well-being within and across time.
Figure 2

HeufisficMcxleLofjheprchesizedRelationshipsBeMeenficonomic Resourcesandllsxchological
ll! ll-B . Till I g I'

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

E 1' j
E Economic Resources Tl " Psychological Well-Being T1 1
l i
: EconomicResourcesIZ Psychological Well-Being T2 .
L %___ , _,,_..
lmmxgsgmm.‘ Psychological Wen-Being 1'3 i
l l l I

L____ __ ___.,___-.z _ _._‘_

l. The level of economic resources at Time 1 has a direct positive effect on

the psychological well-being of women at Time 1.

2. The level of economic resources at Time 2 has a direct positive effect on
the psychological well-being of women at Time 2.

3. The level of economic resources at Time 3 has a direct positive effect on
the psychological well-being of women at Time 3.

4. The level of economic resources at Time 1 has a direct positive effect on

the psychological well-being of women at Time 2.

36
5. The level of economic resources at Time 2 has a direct positive effect on
the psychological well-being of women at Time 3.
Hypothesis 3: It is hypothesized that economic resources mediate the relationship
between abuse and psychological well-being. First, abuse has a direct negative impact on
economic resources such that the higher the level of abuse that women experience, the
fewer the economic resources they will have. Second, abuse has a direct negative impact
on psychological well-being such that the higher the level of abuse that women
experience, the lower their psychological well-being. Third, economic resources has a
direct positive effect on psychological well-being such that the more economic resources
women have, the higher their psychological well-being. Figure 3 is heuristic model of
the hypothesized relationships of economic resources mediating the effect extent of abuse

has on psychological well-being within and across time.

37

Figure3
mmmmmmmmcmmmmmof
!l H I! ll . lll’ll-E' lM'l' l! I'

 

 

 

 

1 Abusell l_.,L Economic Resources Tl j__, Psychological Well-Being rim

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I 1 _ ‘

l Abyss-J2 j__,[ EconomicResourcesIZ j -—-) E Psychological Well-Being T2:
| 1

! AhuseTZi 7 _., L EeonomicResomesfl 4 —-+l Psychological Well-Being 13*:
l T

 

 

 

 

 

 

 

 

 

 

 

1. The level of economic resources at Time 1 mediates the relationship
between abuse at Time 1 and psychological well-being at Time 1.

2. The level of economic resources at Time 2 mediates the relationship
between abuse at Time 2 and psychological well-being at Time 2.

3. The level of economic resources at Time 3 mediates the relationship
between abuse at Time 3 and psychological well-being at Time 3.

4. The level of economic resources at Time 1 mediates the relationship
between abuse at Time 1 and psychological well-being at Time 2.

5. The level of economic resources at Time 2 mediates the relationship

between abuse at Time 2 and psychological well-being at Time 3.

38
6. The level of economic resources at Time 2 mediates the relationship
between abuse at Time 1 and psychological well-being at Time 3.

Hmthesis 4: The literature and previous research suggest that level of economic
resources is related to concurrent and future involvement with assailant. Given the
amount of literature that supports the contention that women report economic
dependence as a primary reason for returning to their abusive partners, the relationship
between economic resources and involvement with assailant is hypothesized to be
moderated by whether women are economically dependent. That is, women's level of
economic resources will affect whether they are involved with their assailant only if they
are economically dependent for those resources. Figure 4 illustrates a heuristic model of
the hypothesized relationships of economic dependence moderating the effect of

economic resources on assailant involvement within and across time.

39

Figure 4
WW . B ! 'l I l W!MW

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

rmgkgmfl : T :Tfl- Assailant Involvement Tl . «
LEconomic Dependence Tl _l
W ' 1. t 3 Assailantlnvolvemenm F
liEcmiomiclkpendenceTZ I
I Ecunorniellependencefi l
(Economiukesourcefll 1, r L Afimgflgmggw
f EconomicDependenceIl J
{ W| T i ; W131}
(EconomicDependencezTZ l
1. Economic dependence at Time 1 moderates the relationship between level

of economic resources at Time 1 and involvement with assailant at Time
1.

2. Economic dependence at Time 2 moderates the relationship between level
of economic resources at Time 2 and involvement with assailant at Time
2.

3. Economic dependence at Time 3 moderates the relationship between level

of economic resources at Time 3 and involvement with assailant at Time

40
Economic dependence at Time 1 moderates the relationship between level
of economic resources at Time 1 and involvement with assailant at Time
2.
Economic dependence at Time 2 moderates the relationship between level

of economic resources at Time 2 and involvement with assailant at Time

METHOD
DLsign

The current study is part of a larger ongoing study that is the first to utilize a
longitudinal, experimental design to examine the effects of an advocacy intervention on
battered women's overall psychological well-being and ability to remain free from abuse
over a two-year period. The larger study employed a longitudinal design with six
interviews over time (pre-intervention/immediately upon shelter exit, post-intervention,
and 6, 12, 18, and 24 month follow-up periods). The current study focuses on women’s
psychological well-being and involvement with their assailant afler having been abused
and staying at a shelter for women with abusive partners and not on the intervention in
the larger study. Three time periods of post-shelter information were used in the present
study corresponding with the post-intervention, 6 month follow-up, and 12 month follow-
up interviews in the larger study.

For the purposes of this study, however, these three interviews reflect women’s
experiences 3 months post-shelter (Time 1), 9 months post-shelter (Time 2), and 15
months post-shelter (Time 3). Because the purpose of the present study was to begin to
formulate a model for the relationship between economic resources, economic
dependence, abuse, psychological well-being, and assailant involvement, it was
important to interview women and measure these dimensions when the women were not

in crisis. Three months post-shelter was long enough after the crisis that precipitated

41

42

their shelter stay to gain stable information on continued abuse, level of economic
resources, economic dependence, and involvement with assailant. Furthermore, having a
reasonably short time between leaving the shelter and this interview helped the project
maintain contact with the participants, thereby improving the retention rate. Because the
present study was exploratory, two follow-up time periods were chosen in order to
examine the stability of the relationships and in order to determine if there were
predictive relationships across time. A period of six months between interview one and
interview two and between interview two and interview three was chosen because it was
an adequate length of time to expect changes in the variables of interest (if there were
changes across time).
Research Participants

Research participants were recruited from the Council Against Domestic Assault,
a shelter for women with abusive partners located in a medium-sized Midwestem city.
Women who stayed at the shelter for at least one night and who did not move out of the
metropolitan area were eligible for participation in the study. Potential participants were
informed that participation involved being interviewed three times after their shelter exit:
3 months post-shelter, 9 months post-shelter, and 15 months post-shelter and that they
would be paid $40.00 for the first interview, $60.00 for the second interview, and $80.00
for the third interview. Potential participants were also informed that half of the women
recruited for interviews would be randomly chosen to receive the free services of a
trained advocate for 4-6 hours per week over ten weeks after they left the shelter to help

them access needed community resources.

43

Of 244 women who were recruited and interviewed by the larger study
immediately upon exiting the domestic violence shelter, four women refused to continue
participation with the project, two were suspected to have been murdered, and three
could not be located (despite intensive tracking efforts), leaving a total sample of 235
interviewed 3 months post-shelter. By the next interview 6 months later (9 months post-
shelter) four more women could not be located (N =23 1). At the 15 month post-shelter
interview, four additional women could not be located, and two were in substance abuse
treatment programs and could not be interviewed (N =225). Therefore, the retention rate
for the first interview (3 months post-shelter) was 96%, the retention rate for the next
interview (9 months post-shelter) was 95%, and the retention rate for the 15 month post-
shelter interview was 92%.

Because of the longitudinal nature of this research design and the regression
analyses used to test the hypotheses, the nineteen participants discussed above who were
not interviewed at every time period were dropped fiom the study (N = 225). On all of
the demographic variables and scale scores, the nineteen participants who were not
included in the final sample did not significantly differ from the 225 participants who had
full data at all three time points any more than would be expected by chance. In addition,
data from one woman who exhibited signs of psychological dysfunction which could
have invalidated her answers to some questions was dropped as an outlier in the
correlational and regression analyses (for a final N=224) but is included in descriptive

reports.

44

Demographic and economic characteristics of participants. Table 2 shows the
demographic and economic characteristics of the sample. The mean age of the sample
was 28 years. Women’s ages ranged fi'om 17 years old to 61 years old. Forty-five
percent of the participants were Non-Hispanic White, 44% were African American, 6%
were Hispanic, and the remainder were Asian American, Native American, Arab
American, or of mixed heritage. At all three time periods, three-quarters of the women
had children who were currently living with them. Women reported having anywhere
from one child living with them to having 7 children living with them. On average,
women had two children who were currently living with them at the time of each

interview.

45

Table 2

Demo hic and Economic Characteristics of Partici ants at Time 1 Time 2 and Time 3 =225

 

Demographic and Economic Characteristics Time 1 Time 2 Time 3
Age in years (mean) 28.3 (SD 7.7)
Race/Ethnicity

Non-Hispanic White 45%

African-American 44%

Hispanic 6%

Asian American 1%

Other 4%

Educational level

Less than high school 36%

High school grad/GED 25%

Some college 28%

College graduate 6%

Trade school 4%

Professional degree 1%
Children living with them 76% 75% 75%
Number of children living with participant (mean) 2 (SD 1.4) 2 (SD 1.4) 2 (SD 1.5)
Annual income (mean) $10,317 $10,162 $10,098

(SD 9460.9) (SD 8082.1) (SD 7083.5)

Living in poverty 72% 72% 70%
Currently employed 26% 33% 32%
Receiving governmental assistance 80% 71% 71%
Currently a student 16% 20% 20%
Type of living situation

Rent an apartment, home, or room 81% 73% 79%

Own a home 5% 7% 5%

Stay with friends or relatives 9% 12% 9%

Other (including homeless or in an institution) 5% 8% 7%

Regular access to a car 40% 46% 51%

 

46

Upon exiting the shelter, women were asked about their educational level. Over
one-third of the sample did not complete high school and did not obtain their GED.
Twenty-five percent did have a high school diploma or obtain their GED. Only 6%
graduated from college, but 28% had attended college for some period of time. One
percent of the women pursued higher education and obtained a professional degree. Four
percent of the women completed trade school.

Table 2 illustrates that participants’ economic characteristics (that were measured
at every time period) were very stable across all three time periods of this study. Most of
the women were living below the federal poverty line with an annual income averaging
just over ten thousand dollars. Less than one-third of the sample were employed, while
over 71% needed to rely on governmental assistance. Few of the participants reported
currently being students at any of the interviews (16% - 20%). The majority of women
(73% - 81%) were renting a place to live (a room, apartment, or home), only 5% - 7%
owned their own home, 9% - 12% were staying with fiiends or relatives, and 5% - 8%
were either homeless or in an institution (e.g., jail, prison, psychiatric hospital) at any of
the time periods. Between 49% and 60% of the women did not have consistent access to
a car if they needed it at any time period.

The participants were demographically similar to the samples of comparable
studies (e.g., Astin et al., 1993; Firm, 1985; Gondolf, 1988; Kemp et al., 1991; Mitchell
& Hodson, 1983; Okun, 1986; Pagelow, 1981; Sato & Heiby, 1992). Thus, they are

generally representative of women who utilize domestic violence shelters.

47
Procedure

Interviewer training. Interviews were conducted by highly trained undergraduate
students at a large Midwestem university (Michigan State University) in exchange for
college credits. Students enrolled for two consecutive terms in small groups of four to
eight persons and received an average of nine weeks (the length of training remained
flexible depending on the performance of the group of interviewers) of intensive training
in interviewing techniques and the tenets of community research from an ecological and
feminist perspective. Training was very interactive and included reading articles on
woman battering and training manuals of interviewing techniques, learning and role-
playing the interviews, films, and discussion of community research skills such as
techniques for tracking participants.

At the conclusion of training and before interviewers were permitted to conduct
actual interviews with research participants, inter-rater reliability was calculated. This
involved the four to eight interviewers and supervisor(s) watching, listening to, and
coding the same mock-interview (performed by two group members who simultaneously
acted as interviewer/interviewee and coded). Inter-rater reliability was calculated using
percent agreement. The total number of possible correct responses was totaled as the
denominator (as determined by one supervisors' coding checked against an oral tape of
the interview). The numerator represented the total number of responses each
interviewer coded correctly. The two were divided and multiplied by 100 to determine a

percent agreement score for each individual student. Due to the rigorous interviewer

48
training and repeated practice interviews, percent agreement scores averaged 97% across
all of the terms of training and was never below 94% in any given term.

After training was complete, interviewers were assigned one or two interviews
per week and they were responsible for locating the participant, setting up the interview,
conducting the interview (always in the community, usually at the participant's home),
coding the interview, and listening to the tape of the interview to make sure it was coded
correctly. Interviewers continued to meet weekly in their small groups with one or two
supervisors to discuss the interviews they completed, ask any questions, hand those
interviews in, and receive new interviews to complete the next week. Weekly meetings
were designed to be very interactive and supportive and were an effective way to make
sure all of the interviews were done well and on time. To further ensure accuracy,
interview supervisors (trained graduate students in Ecological-Community Psychology)
listened to the tape recording of every interview, corrected any coding mistakes, and gave
both positive and corrective feedback to the interviewers on a weekly basis.

Advocacy Intervention

Although not a focus of the present study, the larger investigation employed a
randomized experimental design to evaluate the impact of an advocacy program for
women with abusive partners. Because this intervention may have implications for the
present study, a brief description follows. At the recruitment phase of the study, project
staff administered a baseline interview and then randomly assigned women to either the
experimental intervention, the Community Advocacy Project, or to the control group.

Women assigned to the experimental condition worked with a trained paraprofessional

49

advocate 6-8 hours per week for a period of 10 weeks. Each intervention was based on
the needs and circumstances of the women and all advocates were instructed to help
assess the women’s unmet needs, and tailor their interventions to help women access
community resources that would fulfill those needs (e.g., housing, employment, legal
assistance, transportation, childcare, health care, etc.).
Measures

In-person interviews were conducted with the participants in the community 3
months after women left a domestic violence shelter, and then again 6 months later (9
months post-shelter), and then again 6 months later (15 months post-shelter). In order to
gain the most accurate information possible regarding the effects of abuse and economic
resources on womens' overall psychological well-being and involvement with the
assailant research participants were administered extensive verbal and written interviews.
Table 3 describes the measurement model used in the present study. Participants were
asked questions regarding: 1) level of economic resources, 2) experience of abuse, 3)
psychological well-being, 4) involvement with assailant, and 5) economic dependence.

All questions were asked in an oral interview except where noted.

Table 3
Measurement Model
Construct

Economic
Resources

Abuse

Psychological
Well-Being

Involvement
With Assailant

Economic
Dependence

50

Description of Construct

level of economic resources
including annual income,
poverty status, access to a car,
& employment status

frequency of threats by original
assailant and/or new relationship

extent of injuries by original
assailant and/or new relationship

frequency of being pushed, grabbed,
slapped, punched, kicked, hit with
an object, choked, tied up, stabbed,
shot by original assailant and/or
new relationship

frequency of psychological abuse
including ridicule, criticism, &
harassment by whoever woman is
currently in a relationship with

perceived quality of life

depression

involvement with assailant

economic dependence
on others

Measure

Study Interview

Study Interview

Study Interview

Straus' (I 979)
Conflict-Tactics
Scale

Sullivan, Parisian, &
Davidson's Index of
Psychological Abuse
(1991)

revised version of
Andrew & Withey's
(1976) Quality of
Life Measure

Radlofl‘s (1977)
CES-D

Study Interview

Study Interview

5 1

Economic resources. Economic resources were assessed using a 4-item scale
developed for this study. Items assessed total annual income, whether the participant was
classified as living in poverty (U .8. Bureau of the Census, 1990), employment status, and
access to an automobile (see Appendix A). All four items were originally measured with
different response categories. Three of the four variables (poverty status, employment
status, and access to a car) were dichotomous and reverse coded so that a higher value
indicated more resources. Annual income was a continuous variable that was already
consistent with this coding direction. All of the items were standardized (z-scored),
summed, and averaged to determine an economic resources scale score. Internal
consistency of this scale was .63 at Time 1 (3 months post-shelter), .65 at Time 2 (9
months post-shelter), and .70 at Time 3 (15 months post-shelter) with item-total
correlations ranging from .29 to .60. A concern with using economic resource
information from a shelter sample was that there would not be enough variance within
the sample because many women use shelters as a result of not having many other
resources; i.e., most of the women are poor. However, preliminary analyses indicated
that this sample was normally distributed.

Abuse. Originally, several dimensions of abuse were assessed as separate

 

constructs. Three unique measures assessing various aspects of abuse fi'om the original
assailant or a new relationship and one unique measure assessing abuse from a current
partner (either the original assailant or a new partner) were used in this study and their
psychometric properties are presented below. Measures of physical abuse by the original

assailant and a new relationship included a modified version of the Conflict-Tactics Scale

52
(CTS) (Straus, 1979) (used to assess fiequency of physical abuse) and an injuries
measure used to assess types of injuries women experienced from the abuse. Frequency
of psychological abuse was assessed by asking how often women were threatened by
their original assailant or a new relationship and by the Index of Psychological Abuse
(IPA) (Sullivan, Parisian, & Davidson, 1991) which assessed the frequency of ridicule,
harassment, and criticism from a current relationship (either the original assailant or a
new relationship). At the first interview, 3 months post-shelter, the time reference used
for all of the abuse variables was within the last ten weeks. For the second two
interviews, 9 and 15 months post-shelter, the time reference was within the previous six
months.

Because this study was not concerned with the effects of different types of abuse
or on the differential impact of abuse by the original assailant or a new relationship, but
on how any kind of abuse perpetrated by anyone women were intimately involved with
affected their psychological well-being, all seven scales were eventually combined into
one measure of overall abuse. The psychometric properties of the overall Abuse scale are
presented at the end of this section.

Physical abuse by the original assailant and physical abuse by a new relationship
were each measured independently using a modified version of Straus' (1979) M
Tactics Scale Violence subscale (CTS) measuring frequency and severity of violence
experienced (see Appendix B). Two items were dropped fi'om this scale ("burned" and
"drove recklessly so that you felt endangered") due to lack of variance. When the CTS

was used to ask about the original assailant, the reliability of the scale was a = .91 at

53
Time 1, a = .89 at Time 2, and a = .92 at Time 3. When the CTS was used to ask about a
new relationship, the reliability of the measure was a = .91 at Time 1, a = .89 at Time 2,
and a = .89 at Time 3. The corrected item-total correlations for both the original

assailant and new relationship physical abuse (CTS) scales are listed in Table 4.

Table 4

Corrected Item-Total Correlations for the Original Assailant and New Relationship
Physical Abuse Scales

Corrected Item-Total Correlations

 

Item Original Assailant New Relationship
Time 1 Time 2 Time 3 Time 1 Time 2 Time 3

Tore clothing .55 .56 .71 .81 .53 .45
Pushed .81 .81 .78 .83 .83 .70
Grabbed .85 .81 .81 .80 .77 .73
Slapped .82 .70 .79 .32 .54 .55
Hit with fist .76 .70 .78 .80 .79 .70
Kicked .69 .61 .62 .80 .62 .58
Threw object .58 .54 .77 .70 .55 .88
Hit with object .63 .49 .52 .63 .84 .70
Tried to hit .74 .57 .53 .64 .78 .78
Choked .55 .67 .66 .55 .57 .51
Restrained .41 .52 .41 .80 .52 .58
Raped .26 .50 .66 .84 .35 .60
Threat. weapon .40 .47 .56 .74 .66 . l 9
Used weapon .41 .18 .53 .80 * *

a .91 .89 .92 .91 .89 .89

 

* indicates item had zero variance

54

Several of the corrected item-total correlations for the physical abuse measures as
well as for the overall abuse measure (discussed at the end of this section) did not clearly
meet the usual requirements for item inclusion in a scale yet were kept in the scale. That
is, certain items had low item-total correlations outside of the acceptable range. Because
this study was longitudinal and the reliability of the measures was assessed at each time
period, it was possible to determine that items with low item-total correlations at one
time period did not necessarily have low item-total correlations at another time period.
In order to preserve items in the scale if they had good item-total correlations within at
least one time period, and therefore keep the scale consistent over time, items with low
item-total correlations were not deleted.

Also important to this decision was the argument that internal consistency may
not be the best way to evaluate scales that tap “life event” sorts of constructs (Cleary,
1981). The problem with finding a satisfactory, comprehensive list of items is that
scoring high on one does not necessarily mean that an individual would score high on
another even though the items seem to measure the same construct (i.e., a woman who is
slapped is not necessarily likely to have also been kicked or raped). Therefore, the item-
total correlation for several items may be low. Clearly, more work needs to be done to
determine how to best evaluate the reliability of woman abuse measures but that is
beyond the scope of this study. Again, for this study, items that had low corrected item-
total correlations at only one or two time periods and that rationally seemed to measure

the construct were kept in the scale.

55
The third and fourth measures of abuse were the extent of injuries caused by the
original assailant and the extent of injuries caused by a new relationship. The extent of
injuries was measured using the extent of injuries score derived from a checklist of 10
dichotomously coded injuries sustained by the participants from the physical abuse (see
Appendix C). Ifa woman had been harmed at all, as measured by the presence of at least
one violent act against her, she was asked about the injuries she sustained including

,9 66

“cuts, scrapes, or bruises, soreness without bruises,” “burns, including rug burns,”

99 6‘

“loose or broken teeth,” “broken bones or fractures,” “internal injuries, strains or

,9 6‘

sprains,” “dislocated joints, pregnancy complications or miscarriage,” and “knife or
gunshot wound.” The scores ranged from 0 for no injuries to as high as 10 types of
injuries sustained.

The fifth and sixth indicators of abuse were the frequency of threats from the
original assailant and the frequency of threats from a new relationship. Threats were
measured for this study with a single item, frequency of threats. This item asked women
how many times the assailant "threatened you in any way, including saying or doing
things that made you feel scared or in danger, whether in person, over the phone, through
the mail, or through other people" with possible responses from 1 = never to 6 = more
than four times a week (see Appendix D).

The seventh measure of abuse was psychological abuse from whoever the woman
was currently involved with (either the original assailant or a new relationship).

Psychological abuse was measured using the Index of Pachological Abuse (EA)

(Sullivan, Parisian, & Davidson, 1991) which is a 33-item measure of ridicule,

56
harassment, and criticism (see Appendix E). For example, women were asked to rate
how often in the referent time period whoever they were currently (at the time of the
interview) involved with "refused to talk to you", "tried to control your money", and
"broken or destroyed something important to you" on a 4-point Likert-type scale ranging
from "never" to "often." Internal consistency of this scale was .97 at Time 1 (with item-
total correlations ranging from .41 to .88), .98 at Time 2 (with item-total correlations
ranging fi-om .36 to .89), and .97 at Time 3 (with item-total correlations ranging from .28
to .88).

The literature supports the fact that physical abuse, injuries, threats, and
psychological abuse are all harmful to a survivor of intimate male violence and that it
may be meaningful to combine these constructs to assess overall level of abuse. The
present study was concerned with any abuse that may influence women's psychological
well-being or economic resource level after leaving a shelter for women with abusive
partners. Therefore, it was conceptually and empirically relevant to combine the
following items that were originally measured on separate scales: frequency of physical
abuse (including one sexual violence item) by the original assailant, frequency of
physical abuse (including one sexual violence item) by a new relationship, extent of
injuries from the original assailant, extent of injuries from a new relationship, frequency
of threats from the original assailant, frequency of threats from a new relationship, and
frequency of psychological abuse from a current relationship (either the original assailant
or a new relationship). Sixty-five items from each of the original measures were

standardized (z-scored), summed, and averaged to form the final overall abuse scale.

57
Coefficient alpha for the overall abuse scale at Time 1 was .96, at Time 2 was .95, and at
Time 3 was .95. Corrected item-total correlations for the overall abuse scale are

presented in Table 5.

58
Table 5

Corrected Item-Total Correlations for th_e Overall Abuse Scale

 

Corrected Item-Total Correlations

 

 

Original Assailant New Relation_s_hip
Item Time 1 Time 2 Time 3 Time 1 Time 2 Time 3
Physical Abuse
Tore clothing .23 .25 .46 .34 .22 .08
Pushed .46 .46 .55 .33 .29 .16
Grabbed .51 .40 .53 .37 .26 .17
Slapped .36 .34 .48 .07 .20 .16
Hit with fist .30 .39 .48 .42 .34 .18
Kicked .25 .43 .44 .42 .20 .1 8
Threw object .43 .35 .53 .37 .16 .21
Hit with object .40 .31 .38 .34 .32 .19
Tried to hit .36 .38 .45 .31 .28 .24
Choked .33 .31 .41 .33 .17 .14
Restrained .19 .26 .3 8 .42 .19 .18
Raped .23 .38 .43 .36 .20 .18
Threat. weapon .22 .30 .44 .43 .28 .07
Used weapon .22 .17 .38 .42 * *
Injuries .45 .46 .55 .40 .23 .21

Threats .32 .55 .49 .39 .35 .18

 

Table 5 continued

Corrected Item-Total Correlations for the Overall Abuse Scale

59

 

Corrected Item-Total Correlations

Current Relationship (either original
assailant or new relationship)

 

 

Item
Time 1 Time 2 Time 3

Psychological
Abuse
Refused talk .49 .56 .51
Accused of affair .61 .67 .76
Had affair .58 .51 .50
Refused do things .60 .68 .54
Forbid leaving .60 .64 .50
Control money .61 .77 .65
Control activities .70 .78 .75
Withheld approval .74 .74 .59
Lied .68 .74 .64
Contradict. demands .76 .76 .66
Called names .73 .75 .73
Humiliated you .82 .78 .80
Ignored anger .71 .73 .68
Ignored other .73 .76 .74
feelings
Ridiculed in public .65 .68 .59
Insulted beliefs .67 .72 .76

 

60

Table 5 continued

Corrected Item-Total Correlation for the Overall Abuse Scale

 

Corrected Item-Total Correlations

Current Relationship (either original
assailant or new relationship)

 

 

Item
Time 1 Time 2 Time 3

Insulted race .54 .62 .61
Insulted women .64 .71 .68
Criticized strengths .72 .77 .71
Criticized .73 .77 .72
intelligence
Criticized looks .69 .64 .60
Criticized family .66 .73 .75
Harassed family .50 .53 .57
Discouraged .71 .70 .65
family contact
Threat. family .57 .41 .52
Broke object .62 .66 .64
Abused pets .38 .32 .33
Punished kids .53 .53 .54
Threatened take kids .52 .48 .55
Left stranded .49 .52 .35
Threatened end .64 .67 .58
relationship
Forced you to leave .70 .67 .57
home
Threatened suicide .36 .44 .37

a .96 .95 .95

 

* indicates item had zero variance

61

Psychological well-being. Two measures of pSychological well-being, quality of
life and depression, were used in this study and their psychometric properties are each
presented below followed by a rationale for combining them into one measure of overall
psychological well-being. The psychometric properties of the overall psychological
well-being scale are presented at the end of this section.

Quality of life was measured using a modification of Andrews & Withey's (1976)
Quality of Life measure. Nine items measuring respondents’ feelings about life overall,
self, personal safety, level of fim and enjoyment, responsibilities for family members, life
accomplishments, independence or freedom, emotional and psychological well-being,
and leisure activities were selected to predict overall quality of life (coefficient alpha =
.87 at Time 1, .87 at Time 2, and .86 at Time 3 with item-total correlations ranging from
.38 to .72). Response categories ranged from 1 = extremely pleased to 7 = terrible (see
Appendix F).

Depression was measured using a modified version of the CM (Radloff, 1977),
a 16-item self-report checklist of psychological distress within the general population
(coefficient alpha = .92 at Time 1, Time 2, and Time 3 with item-total correlations
ranging from .36 to .80). The original scale included 20 items, four of which composed a
“Happy” subscale. The item-total correlations of the four items in the “Happy” subscale
did not correlate with the items from the Quality of Life scale when the CBS-D and
Quality of Life scales were combined to form Psychological Well-Being. Therefore,
these items were dropped. Respondents were asked to describe how often (1 = rarely or

never to 4 = most or all of the time) they felt “I was bothered by things that usually don't

62
bother me,” “I did not feel like eating; my appetite was poor,” “I felt depressed,” and “I
thought my life had been a failure” (see Appendix G). This measure was the only one
filled out by the respondent herself in order to offer more privacy and hopefully gain
more accurate information.

Conceptually, this study was interested in women's overall psychological well-
being after leaving a shelter for women with abusive partners. The literature suggested
both overall quality of life and depression as indices of psychological well-being and
preliminary analyses illustrated that as separate scales each were highly correlated with
one another (.60 at 3 months post-shelter, .54 at 9 months post-shelter, and .51 at 15
months post-shelter). Therefore, the original measures of quality of life and depression
were combined in the present study to form a Psychological Well-Being scale. Both
scales were reverse coded first so that a higher score reflected greater satisfaction with
one’s psychological well-being (a higher score on Quality of Life reflects greater
satisfaction with one’s quality of life and a higher score on Depression reflects little or no
depression). Responses were standardized (z-scored) and combined to form a 25-item
scale with very high internal consistency and good item-total correlations (Time 1: a =
.93, item-total correlations ranging from .35 to . 74; Time 2: a = .93, item-total
correlations ranging from .33 to .73; and Time 3: a = .92, item-total correlations ranging
from .30 to .77).

Economic dependence. Economic dependence was measured using a single item
that asks "How much of your income do you directly bring into the home, rather than

another adult? This includes ADC or any other government aid that you may get".

63

Originally, the response categories were “none of the income,” “‘/2 of the income or less,”
“over '/2 but less than all of the income,” and “all of the income” (see Appendix H). Due
to lack of variance, the first three categories were collapsed to reflect “economic
dependence (0).” Economic dependence refers, then, to being dependent on someone
(usually the assailant) for all or part of the money that women need to live on. Women
who bring in “all of the income” were coded as “economically independent (1),” meaning
they were not dependent on anyone else for the money they lived on.

Involvement with assailant. Involvement with assailant was measured using a

single item that asks "Are you currently involved with (assailant's

 

name)?". Women who were involved with their assailants at the time of the interview
received a code of “1 " and women who were not involved with their assailants received a
code of “2” (see Appendix 1). Because it was a categorical variable, the “2" was dummy
coded to “0.”

Experimental condition. Prior research with a sub-group of this sample (N=141)
suggested that the experimental intervention that was part of the larger study did have a
significant effect on women’s social support, quality of life, and effectiveness in
obtaining resources 3 months post-shelter and a significant effect on women’s quality of
life 9 months post-shelter (Sullivan, Tan, Basta, Rumptz, & Davidson, 1992; Sullivan,
Campbell, Angelique, Eby, & Davidson, 1994). That is, women who received the
services of an advocate reported being more effective in accessing resources and had
higher levels of social support and overall quality of life post-intervention (3 months

post-shelter) and reported continued higher quality of life six months later. Therefore,

64

although not a focus of the current study, experimental condition was controlled for in all
of the analyses.
Data Analysis Strategy

, A review of the data analysis plan is presented here. Analyses were conducted to
explore the objectives and the specific hypotheses about two sets of relationships: those
between economic resources, abuse, and psychological well-being and those between
economic resources, economic dependence, and involvement with assailant. Descriptive
statistics were conducted first to begin to examine the percentages, means, and standard
deviations of the variables of interest (psychological well-being, abuse, economic
resources, economic dependence, and assailant involvement) across time. Next,
correlation coefficients were computed as a summary index to describe the observed
strength of the associations between psychological well-being, abuse, economic
resources, economic dependence, assailant involvement, and experimental condition both
within and across Time 1, Time 2, and Time 3.

Hierarchical multiple regression analyses were used to examine the first set of
hypotheses that stated that experience of abuse has a direct negative effect on the
psychological well-being of women both within Time 1, Time 2, and Time 3 and across
time (i.e., Time 1 to Time 2 and Time 2 to Time 3). In each of the five equations,
experimental condition was entered first in order to control for the effects of the
intervention. In the four equations where psychological well-being at Time 2 or Time 3
was the dependent variable, prior level of psychological well-being was entered next

(after experimental condition). It was necessary to control for prior level of

65
psychological well-being because it was so highly correlated with psychological well-
being at the next time period. Therefore, the analyses actually investigated the effect of
abuse on change in psychological well-being. The primary independent variable of
interest, abuse, was entered last in every equation.

Hierarchical multiple regression analyses were also used to examine the second
set of hypotheses that stated that level of economic resources has a direct positive effect
on women's psychological well-being both within Time 1, Time 2, and Time 3 and across
time (i.e., Time 1 to Time 2 and Time 2 to Time 3). The order of variable entry was
identical in this set of analyses. That is, in each of the five equations, experimental
condition was entered first in order to control for the effects of the intervention. In the
four equations where psychological well-being at Time 2 or Time 3 was the dependent
variable, prior level of psychological well-being was entered next (after experimental
condition). The primary independent variable of interest, economic resources, was
entered last in every equation.

A series of regression models were estimated to examine the third set of
hypotheses that state that women's level of economic resources mediates the relationship
between extent of abuse and women's psychological well-being. Six sets of relationships
were examined to understand this hypothesized relationship both within and across time.
That is, within Time 1, Time 2, and Time 3 and across Time 1 and Time 2 (i.e., abuse and
economic resources at Time 1 predicting psychological well-being at Time 2), across
Time 2 and Time 3 (i.e., abuse and economic resources at Time 2 predicting

psychological well-being at Time 3), and across all three time periods (i.e., abuse at Time

66
1, mediated by economic resources at Time 2, predicting psychological well-being at
Time 3).

To test for mediation, three regression equations were estimated: first, regressing
the mediator (economic resources) on the independent variable (abuse); second,
regressing the dependent variable (psychological well-being) on the independent variable
(abuse); and third, regressing the dependent variable (psychological well-being) on both
the independent variable (abuse) and on the mediator (economic resources) (Baron &
Kenny, 1986). Baron & Kenny propose that to establish mediation, the following
conditions must hold: First, the independent variable must affect the mediator in the first
equation; second, the independent variable must be shown to affect the dependent
variable in the second equation; and third, the mediator must affect the dependent
variable in the third equation. If these conditions all hold in the predicted direction, then
the effect of the independent variable on the dependent variable must be less in the third
equation than in the second. Perfect mediation holds if the independent variable has no
effect when the mediator is controlled.

Lastly, logistic regression was used to examine the fourth set of hypotheses and
test whether the causal relation between economic resources and involvement with
assailant changed as a function of economic dependence (i.e., was moderated by
economic dependence). Logistic regression was conducted rather than hierarchical
regression because the dependent variable, involvement with the assailant, was
dichotomous (involved/not involved). The moderational hypotheses were tested within

Time 1, Time 2, and Time 3 and across Time 1 and Time 2, and across Time 2 and Time

67
3. To test these five relationships, experimental condition was entered along with the
covariates for each equation. When involvement with assailant was predicted at Time 2
or Time 3, prior involvement with assailant was not controlled for because it did not
seem appropriate to measure change in involvement with assailant which was a
dichotomous variable.

Each of the five statistical analyses must measure and test the differential effect of
women's level of economic resources on involvement with the assailant as a function of
being economically dependent on others (usually the assailant) (Baron & Kenny, 1986).
Economic dependence moderates the relationship between economic resources and
involvement with the assailant if the interaction term makes a significant contribution to

the equation.

RESULTS

The present study was designed to examine the impact of abuse and economic
resources on 225 women who had stayed at a battered women’s shelter. For purposes of
organization, this chapter is divided into sections to coincide with the objectives and
specific hypotheses of the study. First, to provide the contextual background for
examining relationships among the variables of interest, longitudinal descriptive
information on psychological well-being, extent of abuse, level of economic resources,
economic dependence, and involvement with the assailant are presented and discussed.
Second, in order to begin to examine the interrelationships between the 5 study variables
at each of the three time points, zero order correlations are presented and discussed.
Third, to examine the direct effects of abuse and the direct and mediating effects of level
of economic resources on women’s psychological well-being, the results of hierarchical
regression analyses are presented and discussed. Lastly, the results of logistic regression
analyses are presented that examined the moderating effect of economic dependence on
the relationship between economic resources and involvement with the assailant.
Descriptive Information

Descriptive analyses are presented here to provide the background information for
understanding the hypothesized relationships about the role of economic resources in
predicting women’s psychological well-being and decisions about staying with their

assailants. Descriptive information at each of the three time periods (including means,

68

69
standard deviations, and percentages) are provided for women’s psychological well-
being, experience of abuse, economic resources, economic dependence, and involvement
with the assailant.

Women's psychological well-being over time. Descriptive analyses were
conducted to provide information on women’s satisfaction with their quality of life and
level of depression over time. Table 6 shows the mean scores and standard deviations on
the Quality of Life and Depression (C’S-D) measures at Time 1 (3 months post-shelter),
Time 2 (9 months post-shelter), and Time 3 (15 months post-shelter). On average
women reported having mostly satisfied to mixed feelings about their overall quality of
life at all three time periods. Over time, however, women reported a slight increase in
their satisfaction with their quality of life. Women reported some depression at each of
the three time periods, indicating that they felt depressed 1-2 days per week. This finding
was fairly consistent at every time period, although at Time 3 women’s reported level of

depression decreased slightly from Time 2.

70

Table 6

Descriptive Information on chhological Well-Being Over Time

 

 

Variable Means (N=224)
Time 1 Time 2 Time 3
Quality of Life“ 4.82 4.97 4.98
(1.04) (1.03) (0.99)
Depressionb 3.05 3.05 3.10
(0.70) (0.66) (0.69)

 

Note: Numbers in parentheses are the standard deviations; Time 1 = 3 months post-
shelter, Time 2 = 9 months post-shelter, Time 3 = 15 months post-shelter.

‘On a 7-point scale where 1 = terrible and 7 = extremely pleased.

b On a 4-point scale where 1 = most or all of the time and 4 = rarely or never.

Women's exmrience of abuse over time. Descriptive analyses were also
conducted to provide information on the extent of abuse women experienced over time
by their original assailant and by any new relationship they had been involved in since
leaving the shelter. Table 7 shows the mean scores and standard deviations on the level
of physical abuse, injuries, and threats fi'om the original assailant or a new relationship,
and psychological abuse from whoever (original assailant or new relationship) women
were currently involved with at Time 1 (3 months post-shelter), Time 2 (9 months post-
shelter), and Time 3 (15 months post-shelter).

The mean scores indicate that women experienced physical violence from their
original assailants less than once a month at any of the three time periods and that the

amount of physical abuse decreased slightly over time. At Time 1, 44% of the women

experienced some level of physical abuse by their original assailants. Twelve months

71
later (at Time 3), fewer women (37%) were still experiencing physical assaults from their
original assailants, although this number still represents over one-third of the sample
being battered by their original assailants 15 months after leaving the shelter.

Women reported very little physical abuse from a new relationship at any of the
three time periods, with most women reporting no physical abuse at all from a new
relationship. At 3 months post-shelter, 95% of women reported no physical abuse from a
new relationship. However, the mean scores did rise slightly at each time period,
indicating that women reported experiencing more physical abuse from a new
relationship as time went on. Further investigation showed that at Time 2, 10% of the
women reported experiencing some physical abuse and at Time 3, 18% reported

experiencing some physical abuse from a new relationship.

Table 7

72

Descriptive Information on Abuse Over Time

 

Variable

Original Assailant

Physical abuse'l
Injuries"

Threats‘

New Relationship

Physical abuse
Injuriesb

Threats“

Psychological abuse from
current relationship (either
original assailant or new
relationship)c

 

Means (N=224)

Time 1 Time 2 Time 3
1.22 1.21 1.19
(0.43) (0.38) (0.41)
0.60 0.63 0.55
(1.17) (1.17) (1.22)
1.95 1.82 1.73
(1.23) (1.17) (1.18)
1.02 1.03 1.07
(0.09) (0.13) (0.24)
0.04 0.08 0.22
(0.29) (0.37) (0.75)
1.07 1.10 1.18
(0.44) (0.38) (0.63)
1.36 1.40 1.33
(0.57) (0.62) (0.56)

 

Note: Numbers in parentheses are the standard deviations; Time 1 = 3 months post-

shelter, Time 2 = 9 months post-shelter, Time 3 = 15 months post-shelter.
'On a 6-point scale where 1 = never and 6 = more than 4 times a week.

l’Possible range is from 0 = no injuries, to 10 types of injuries.

° On a 4-point scale where 1 = never and 4 = often.

At any of the three time periods, women suffered on average one type of injury

(ranging fi'om 0 to 7 different types of injuries) from their assailant in the time period

73
preceding the interview. However, they suffered slightly more different types of injuries
at Time 2 than at Time 1. From Time 2 to Time 3 the number of types of injuries
decreased. Women reported suffering fewer, between 0 and 1, different types of injuries
from their new relationships (between 0 and 5). In fact, most women (98%) reported not
being injured at all by the new person that they were involved with at Time I. Not
surprisingly, the number of injuries from the new relationship did increase from Time 1
to Time 3 (5 % were injured at Time 2 and 11% suffered injuries at Time 3), coinciding
with the rise in level of physical abuse from the new relationship across the same time
period.

Women experienced threats from their original assailants on average once a
month or less. Over half of the participants were still being threatened by their assailants
3 months after leaving the shelter. Twelve percent of the women reported experiencing
threats at least once or twice a week. Fewer women were threatened in any way by their
original assailants at Time 2 (46%), and fewer still were threatened at Time 3 (40%).
Women reported that they almost never experienced threats from their new relationship.
This finding was fairly consistent across time, although the percentage of women who
reported being threatened by the new person they were involved with increased slightly
over time (from only 4% at Time 1 to 1 1% at Time 3).

The mean scores in Table 7 indicate that women reported “rarely” to “never”
experiencing psychological abuse from their current partner (either the original assailant
or a new relationship). However, over 46% of the participants did report some level of

psychological abuse at Time 1, 54% reported being psychologically abused at Time 2,

74
and 50% reported experiencing psychological abuse at Time 3. This suggests that
approximately half of the participants at each time period continued to endure ridicule,
harassment, and criticism from the person they were involved with.

Women's level of economic resources over time. Descriptive analyses were
conducted to provide information on women’s level of economic resources as measured
by total annual income, poverty status, access to a car, and employment status. Table 2
longitudinally describes these four variables (total annual income, poverty status,
employment status, and access to a car) that composed the economic resources scale.
From the table, it is obvious that there is little variance and little change over time in the
level of economic resources of this sample. Almost three-quarters of the women were
living in poverty (as defined by the US. Bureau of the Census, 1990). The mean annual
income of this sample at each of the three time periods was only slightly higher than
$10,000 a year. Over two-thirds of the women were unemployed at the time of the
interviews. Forty percent of the women had access to a car at Time 1, 46% had access to
a car at Time 2, and 51% had access to a car at Time 3.

Women's level of economic dependence on others over time. Descriptive
analyses were conducted to provide information about women’s level of economic
dependence on others over time. Women were defined as economically independent if
they were responsible for bringing in all of the income that they, and their family (if
applicable), lived on. Women were defined as economically dependent if they depended
on income or resources from another person (often the assailant). This could mean that

another person(s) brought in all of the income or brought in a portion of it. At Time 1,

75
88% of the participants were economically independent; that is, solely responsible for
financially supporting themselves and their family (if applicable). This percentage
remained relatively constant across time. Six months later, at Time 2, 75% of the women
were economically independent only to increase again at Time 3 to 79% who were
economically independent. Less than one-quarter of the sample at any given time period
was economically dependent on another adult.

Many of the women who were economically independent were classified this way
because they brought in all of the income in the form of government assistance (such as
ADC or Food Stamps). Between 71% and 80% of the women in the sample were
receiving government assistance at any of the three time points. Relatedly, between 70%
and 72% of the sample were living in poverty at any of the three time points. For this
sample, financial independence did not necessarily mean that women had an adequate
level of economic resources with which to support themselves and their families (if
applicable). In fact, many of the women in this sample were classified as economically
independent, yet were financially reliant on the government and were also living below
the poverty level.

Of the small percentage of women who did not bring in all of the family income
(between 12% and 25% of the sample at each time period), most anecdotally reported
that they were economically dependent on their assailants for full or partial financial
support although this was not a direct question in the interview. Other sources of income
were mentioned less frequently, but included family members, friends, or a new

relationship.

76

Involvement with the assailant over time. Descriptive analyses were conducted to
provide information on the number of women who were involved with their assailants at
each time period. As time progressed, fewer women remained involved with their
assailants. At Time 1, 36% were still involved with their assailants, at Time 2 30% were
involved with their assailants, and at Time 3 only 25% were still involved with their
assailants.

Summary of descriptive information. The following is a summary of the
descriptive findings that are most important and relevant to this study and that provide
the background for examining bivariate and multivariate relationships among the
variables of interest. Women reported feeling mostly satisfied about their overall quality
of life and did not report experiencing much depression at any of the three time periods
following their shelter stay. Over time, women's psychological well-being generally
improved, although the mean C’S-D scores at the later time periods still reflected some
level of depression. The consistency in the findings across time (although there was a
slight elevation in women's psychological well-being over time) may suggest that
women’s psychological well-being had stabilized post-shelter.

Women reported a decrease in physical abuse from their original assailants and an
increase in physical abuse from their new relationships. However, they reported more
different types of injuries from their assailants 9 months post-shelter (Time 2) than they
did at 3 months post-shelter (Time 1) or 15 months post-shelter (Time 3). Women
reported more injuries fi'om their new relationship with each successive time period. On

average, women reported that the frequency of threats by the original assailants

77
decreased over time but that they experienced threats more often than physical abuse.
The frequency of threats made by a new relationship increased over time. Finally,
women experienced an increase in psychological abuse from their current partner 9
months after leaving the shelter and then reported that the emotional abuse decreased six
months later (15 months post-shelter). Although the mean scores did not indicate that
women were, on average, continuing to experience a high level of abuse, the percentages
of women who were experiencing abuse were substantial and indicate that many women
were still being battered over the twelve months of this study.

With regard to women’s level of economic resources, the descriptive data indicate
that most of the women were poor and that their level of economic resources did not
change much across time. Three-quarters of the women were living in poverty; over two-
thirds of them unemployed and reliant on the government or other people for support.
Whether from employment, governmental aid, or from other people, womens’ annual
income averaged only slightly more than $10,000. For a family of three (the average
family size of this sample), this was barely above the poverty line. Furthering the
difficulty of their lives, between 49% and 60% reported that they did not have access to a
car.

Although there was variance within the sample, it was relative to how the sample
was defined in the first place. That is, the sample was women with abusive partners who
stayed at a domestic violence shelter and typically had few economic resources. Within

that group of impoverished women, there was a fairly normal distribution on economic

78
resources. However, the descriptive data support that most of the participants were still
poor and had few economic resources.

Most of the women (between 75% and 80% at each time period) were
economically independent fiom their assailants or another adult; that is, they were
responsible for bringing in all of the income. At first glance, this may seem like a
positive factor. However, many of the women who were economically independent were
classified this way because they brought in all of the income in the form of government
assistance (such as ADC or Food Stamps). The amount of money allocated to families in
the form of government aid often does not cover the cost of living and it is no
coincidence that, in this sample, between 70% and 72% of the women and their families
were also living in poverty.

The decrease in physical abuse and threats experienced by women from their
original assailants may coincide with the fact that over the twelve months of this study,
more women reported no longer being involved with their original assailants. At 3
months post-shelter, many women (64%) reported not being involved with their
assailants but by 15 months post-shelter, 75% of the women were not involved with their
assailants. It is important to remember, however, that being uninvolved with an assailant
was no guarantee that women would be safe from physical, sexual, or emotional

violence.

79

Intercorrelations Among Pachological Well—Being. Involvement With Assailant Abuse

 

Economic Resources, Economic Dependence, and Experimental Condition

The intercorrelations among the six primary study variables are presented next to
provide further information on the bivariate relationships between psychological well-
being, involvement with assailant, abuse, economic resources, economic dependence, and
experimental condition. The zero-order correlations among the six primary study
variables at Time 1, Time 2, and Time 3 are featured in Table 8. The first column lists
the abbreviated version of the variable names followed by a number indicating whether
the variable was measured at Time 1 (3 months post-shelter), Time 2 (9 months post-
shelter), or Time 3 (15 months post-shelter).

The intercorrelations among these variables were examined to assess the relative
strength of the relationships of the independent variables (abuse, economic resources,
economic dependence, and experimental condition) to the dependent variables
(psychological well-being and involvement with the assailant). The results of the zero-
order correlations among variables are organized in the following way. First, the
intercorrelations among variables across time are presented (i.e., correlations among
abuse at Time 1, abuse at Time 2, and abuse at Time 3). Second, the outcome variables
in this study (psychological well-being and involvement with assailant, respectively) and
their intercorrelations with other study variables are presented. Third, intercorrelations

among the independent variables are presented.

80

._o.::8uN .325..on—

:o_.:.:oU dxm 3:020:35 3.8.8880"— ..Evgnov m__ao_.:o:oooud duo doom ”329:."— .uo>_o>:_ 8:1. .23 .>:_ um. 32.: 83E? 32:920.: 05 .8
mice“. a: 8:026 2:. Abzosmémo: 3.8:. n: m 35. u m Abe—2.38: 2.5.: $ N 2:; u N Abzofiémc: 2.5:. Q _ oEE. u _ adore: 2:: o. .32 0:8: 28:5. 05 a:
v8 05 a 83:5: 2:. .5238 3:95.598 n 5:850.qu 69:353.. 3:858 u .90.:8m 45:88 .23 .:o:.o>_o>:_ u :83? .>:_ 3.5.. .3 .888 n 8.27. .5 .xm
$8.582 2:558 u .8¢.:8m $52133 EEO—osomma u mafia. no. .82 8.5: 28...? 3332.3: 2: 8.93:5 5.8.8.8 =03 .8 «N as. NNN :88”... u Z Kev...

mod. bod. Nod hdd mdd nod d _ d cod am _ d. d _ d mod vod Ed. add. 3 _ d. 5:350 dxm .2
oc. .mmd .de 43d. .fid. 42d. «dd. Ndd Nod. .nmd. odmd. 4N_d. 3d .2d. add. finch—doom.“—
dd._ .dmd .oNd. .nmd. .NNd. oNNd. 43d. 2.0. ode. .wmd. .de. wdd vdd 3d Ndondoom .3

do. add. .NNd. oNNd. add. ddd. .de. .oNd. .nmd. .hmd. odd 5d odd _doQ.:oom .m—

dd._ awed omed .mvd «de .nmd aVNd .de .de on _ d- ndd mod. 2.8.9.53 .>:_ .N_

dd. _ .Mdd .hvd _. _ ed .dmd 9d _ d a _ ad a _ md n. d- _ _d. add. N.=:mm<\3 .>:_ ._ _

dd. _ comd .mmd ._ 3d a _ Nd .cNd .eNd mdd. Ndd ddd .4835». .>:_ d—

2: .dnd .hvd hdd cad .m—d ond. cad. .w_d. woman<uoaxm d

dd. .Nvd Nod. :d utd .1th eemd. .MNd. New??? axm .w

dd. vod .2d .3d o-Nd. cNNd. ind. _ow:n<uo..xm .h

 

8.. .2... .5... :8 .2... 8... names: a

8.. .8... 8... S... 8... «8:88.. .n

8.. 8...- 8.... 8.... ..a:.=8m a.

8.. .8... .8... 83%.. .m

8.. .8... 83.»... .N

8.. . 53.5 ..

n. z m. N. .. o. o m a o n v m N . 285$

 

 

a 035.

81

All five variables were significantly correlated with themselves across all three time
periods. The correlations among psychological well-being at Time 1, Time 2, and Time 3
ranged from .56 to .58, p<.05. Economic resources was significantly related at each time point
(1’s from .59 to .75, p<.05) as was extent of abuse (1’s from .42 to .50, p<.05). Involvement with
assailant was highly correlated with itself across time (r’s from .63 to .65, p<.05) while
economic dependence was less correlated (r’s from .20 to .33, p<.05).

Several of the study variables were correlated with women’s psychological well-being.
The only significant correlation between economic resources and psychological well-being was
found between psychological well-being at Time 2 and economic resources at Time 3 (1:. l 7,
p<.05) suggesting that women reporting higher satisfaction with their psychological well-being 9
months post-shelter also had a higher level of economic resources 15 months post-shelter. Other
than this relationship, economic resources and psychological well-being shared small
correlations. Extent of abuse, however, was significantly correlated with psychological well-
being concurrently and across time (with r’s ranging from -.21 to -.36, p<.05). That is, women
reporting lower levels of abuse also reported greater satisfaction with their psychological well-
being. Involvement with the assailant at Time 3 was significantly related to psychological well-
being at Time 3 (r = -.13, p<.05). That is, women who were involved with their assailants at
Time 3 also reported lower satisfaction with their psychological well-being at Time 3. Women
who reported greater satisfaction with their psychological well-being at Time 2 also were more
likely to be economically dependent at Time 3 (r = -.13, p<.05). Finally, with regard to variables
with significant correlations to psychological well-being, experimental condition was

significantly correlated with psychological well-being at Time 1 (r = -. l 9, p_<.05), indicating that

82
women in the experimental condition also reported greater satisfaction with their psychological
well-being.

Table 8 also shows that several of the study variables were correlated with involvement
with the assailant. As was already mentioned above, women who were involved with their
assailants at Time 3 reported lower satisfaction with their psychological well-being at Time 3.
Women who were involved with their assailants, however, had more economic resources (r’s
ranging from .19 to .31, p<.05) at all three time periods. There was a fairly consistent
relationship between economic dependence and involvement with assailant such that women
who reported being economically dependent, also reported being involved with their assailant
(L’s ranging fiom -.16 to -.35, p_<.05). This relationship was true within and across time except
that economic dependence at Time 1 did not significantly correlate with involvement with
assailant at Time 3. Finally, women who were involved with their assailants also reported
experiencing more abuse. Involvement with the assailant was significantly correlated with
extent of abuse concurrently and across time (r’s ranging from .29 to .61, p<.05).

Many of the independent variables in this study were correlated with one another, yet
none too highly as to suggest a measurement problem. Extent of abuse at Time 1, Time 2, and
Time 3 was correlated with level of economic resources at Time 1 (g’s from .14 to .18, p<.05).
That is, women reporting higher levels of economic resources at Time 1 also reported higher
levels of abuse at each of the three time periods. Extent of abuse at Time 1 was also
significantly related to economic resources at Time 2 (r = .13, p<.05). Participants reporting
more abuse at Time 1 also reported higher levels of economic resources at Time 2. Lastly,

economic resources at Time 2 was related to extent of abuse at Time 3 (r = .15, p<.05)

83
suggesting that women with higher levels of economic resources at Time 2 also experienced
higher levels of abuse at Time 3.

Women who reported being economically dependent on someone for their financial
support (most likely the assailant) also reported having a higher level of economic resources (r’s
ranged from -.12 to -.3 8, p<.05) at all three time periods. Economic dependence at Time 1 was
also significantly correlated with extent of abuse at Time 1 (r = -.20, p<.05) suggesting that
women who were economically dependent also reported higher levels of abuse. Economic
dependence at Time 2 was also significantly related to reports of higher levels of abuse at Time 2
(r = -.19, p<.05) and Time 3 (r = -.22, p<.05). Finally, women in the experimental condition also
reported significantly higher levels of abuse at Time 1 (r =-.13, p<.05).

Summary of the intercorrelations among pychological well-being, involvement with

assailant, abuse, economic resources, economic demndence, and experimental condition. The

following is a summary of the zero-order correlations that provide a background for
understanding the results of the regression analyses that follow. As would be expected, there
were high self-intercorrelations of variables at different time periods. That is, psychological
well-being at Time 1 was highly correlated with psychological well-being at Time 2 and at Time
3. Womens level of economic resources, abuse, involvement with assailant, and economic
dependence shared the same pattern of intercorrelation.

Several variables correlated with psychological well-being. First, women’s
’ psychological well-being at Time 2 was significantly related to women’s level of economic
resources at Time 3 and economic dependence at Time 3. That is, women who reported greater

satisfaction with their psychological well-being at Time 2 reported also having a higher level of

84
economic resources and a greater likelihood of being dependent for those resources at Time 3.
Women who reported greater satisfaction with their psychological well-being also reported less
abuse at all three time periods. At Time 3, 15 months post-shelter, women who had higher
psychological well-being also were less likely to still be involved with their assailants. Finally,
women in the experimental condition reported greater satisfaction with their psychological well-
being at the first interview, 3 months post-shelter.

I There were also several significant relationships between study variables and women’s
involvement with the assailant. As was already mentioned above, women who were involved
with their assailants at Time 3 experienced lower psychological well-being at the same time
period. Interestingly and of great relevance to this study, women who were involved with their
assailants had higher economic resources, were more likely to be dependent on someone (most
likely the assailant) financially, and reported higher levels of abuse at all three time periods (the
only exception was that involvement with the assailant at Time 3 did not significantly correlate
with economic dependence at Time 1). I

Finally, there were several significant intercorrelations among the independent variables:
economic resources, abuse, involvement with the assailant, economic dependence, and
experimental condition. There were a series of interesting relationships between level of
economic resources and level of abuse. Women who reported higher levels of economic
resources at Time 1 also reported higher levels of abuse at Time 1, Time 2, and Time 3 and
women who reported higher levels of economic resources at Time 2, reported higher levels of
abuse at Time 3. Women who reported higher levels of abuse at Time 1 reported higher levels of

economic resources at Time 2.

85

Women who were economically dependent had higher economic resources at all three
time periods. Women who were economically dependent at Time 1 also experienced more abuse
at Time 1, and women who were economically dependent at Time 2 reported higher levels of
abuse at Time 2 and Time 3. Lastly, women in the experimental condition experienced higher
levels of abuse at Time 1.
Regmssions With Pachologjcal Well-Being as the Dependent Variable

Hierarchical regression was used to examine the strength and direction of relationships
among the first set of variables of interest in this study; that is, the relationships between abuse,
economic resources, and psychological well-being. The impact of abuse and economic resources
(predictor variables) on women’s psychological well-being were examined in three hypotheses.
The first hypothesis examined the main effects of abuse on women’s psychological well-being,
the second hypothesis examined the main effects of economic resources on women’s
psychological well-being, and the third hypothesis examined the mediating effect of economic
resources on the relationship between level of abuse and women’s psychological well-being.

Within each hypothesis, multiple analyses were conducted to investigate the hypothesis
within each time period and across time periods as stated in the Research Objectives and
Hypotheses section at the end of the Chapter 1. That is, within hypothesis 1, five separate
regression analyses were performed to investigate the hypothesized relationships within Time 1,
within Time 2, within Time 3, across Time 1 and Time 2, and across Time 2 and Time 3. Within
hypothesis 2, five separate regression analyses were also performed to investigate the
hypothesized relationships within Time 1, within Time 2, within Time 3, across Time 1 and

Time 2, and across Time 2 and Time 3. Finally, within hypothesis 3, six separate regression

86
analyses were performed to investigate the hypothesized relationships within Time 1, within
Time 2, within Time 3, across Time 1 and Time 2, and across Time 2 and Time 3, and across all
three time periods. The results of the three hypotheses are presented next, each in their own
section.

Abuse as the predictor. The first hypothesis in this study argued that abuse from intimate
male partners is a major stressor that affects women’s overall psychological well-being. That is,
it was hypothesized that the experience of abuse would have a direct negative effect on the
psychological well-being of women. The more severe the abuse, the lower women’s
psychological well-being.

Table 9 shows a summary of the five hierarchical regression analyses that were used to
test the hypothesis using psychological well-being as the dependent variable and abuse as the
independent or predictor variable. Hypothesis I examined this relationship within time (i.e.,
abuse at Time 1 predicts psychological well-being at Time 1, abuse at Time 2 predicts
psychological well-being at Time 2, and abuse at Time 3 predicts psychological well-being at
Time 3) and across time (i.e, abuse at Time 1 predicts psychological well-being at Time 2 and
abuse at Time 2 predicts psychological well-being at Time 3). Table 9 displays the outcome
variables and predictor variables for each analysis in colmnn one, standardized regression
coefficients (the Beta weights) in column two, the corresponding t-test of the standardized Beta
weight in column three, the step-wise R2 in column four, and the overall multivariate F statistic

in column five.

87

 

 

Table 9
.unrno - rel Han ‘ n unmrl ‘ tn 0 so ‘i'ol 1'
' ”0' A --r' on. 'r o u'In‘r--. 0 0.011.191“. “0 ' 00:1."-
Being
Outcome and Predictors [3 t R2 F

Analysis 1:

Outcome: Psychological Well-Being Time 1

Predictors: Experimental Condition - .18 - 2.32‘ .03 7 .94"
Extent of Abuse Time 1 - .40 - 4.52“ .12 14.53“

Analysis 2:

Outcome: Psychological Well-Being Time 2

Predictors: Experimental Condition .02 .34 .01 1.65
Psychological Well-Being Time 1 .51 9.73" .34 56.99“
Extent of Abuse Time 2 - .29 -4.43* .40 47.79"

Analysis 3:

Outcome: Psychological Well-Being Time 3

Predictors: Experimental Condition .02 .27 .00 .73
Psychological Well-Being Time2 .51 9.35‘ .31 49.79‘
Extent of Abuse Time 3 -.27 -4.29* .37 41.99‘

Analysis 4:

Outcome: Psychological Well-Being Time2

Predictors: Experimental Condition .03 .42 .01 1.65
Psychological Well-Being Tirnel .55 9.86‘ .34 56.99“
Extent of Abuse Time 1 - .07 - .85 .34 3819‘

Analysis 5:

Outcome: Psychological Well-Being Time3

Predictors: Experimental Condition - .01 - .l l .00 .73
Psychological Well-Being Time2 .54 9.18“ .31 49.79"
Extent of Abuse Time 2 - .02 - .21 .31 33.06"

 

N = 221-224 for each analysis (due to missing data).

88

The results suggested that after controlling for experimental condition and prior
level of psychological well-being, the standardized regression coefficient for concurrent
abuse was significant. The following specific relationships were supported: the
experience of abuse at Time 1 had a direct negative effect on the psychological well-
being of women at Time 1 ([3 = -.40, p<.05), the experience of abuse at Time 2 had a
direct negative effect on the psychological well-being of women at Time 2 ([3 = -.29,
p<.05), and the experience of abuse at Time 3 had a direct negative effect on the
psychological well-being of women at Time 3 ([3 = -.27, p<.05). The standardized
regression coefficients for across time relationships (abuse at Time 1 predicting
psychological well-being at Time 2 and abuse at Time 2 predicting psychological well-
being at Time 3) were not significant.

It is important and interesting to note that the standardized regression coefficients
for the control variables, experimental condition and prior level of psychological well-
being, were also significant in some of the analyses. The standardized regression
coefficient for experimental condition, entered first as a control variable, was significant
in the first analysis (abuse at Time 1 predicting psychological well-being at Time 1)

(B = -.18, p<.05), but not in any of the other analyses. This suggests that receiving the
services of an advocate for 10 weeks (experimental condition) predicted that women
would report greater satisfaction with their overall psychological well-being 3 months
after leaving the shelter (which corresponded with the end of their intervention), but not

at any later time period.

89

The other control variable, prior level of psychological well-being, was a
significant predictor of later psychological well-being in analyses 2 through 5. This
finding is not surprising given the high intercorrelations among women’s reported
psychological well-being at every time period. Summary Table 9, analysis 2, shows that
women’s psychological well-being at Time 1 had a direct effect on women’s
psychological well-being at Time 2 (B = .51, p<.05). In analysis 3, women’s
psychological well-being at Time 2 had a significant effect on women’s psychological
well-being at Time 3 (B = .51, p<.05). Table 9 further displays that the relationships
were similar in analyses 4 and 5. In analysis 4, women’s psychological well-being at
Time 1 had a direct impact on women’s psychological well-being at Time 2 (B = .5 5,
p<.05) and in analysis 5, women’s psychological well-being at Time 2 had a direct effect
on women’s psychological well-being at Time 3 (B = .54, p<.05).

Column four (the step-wise R2) of Table 9 shows the amount of variance
accounted for by each of the independent (or predictor) variables and column five reports
the significance of the overall multivariate F statistic. In analysis 1, both experimental
condition and extent of abuse at Time 1 accounted for a significant proportion of the
variance in predicting psychological well-being at Time 1. Experimental condition
accounted for 3% of the variance in predicting psychological well-being at Time 1 and
9% of the variance in psychological well-being at Time 1 was explained by the main
effect of abuse at Time 1.

For analyses two through five, prior level of psychological well-being (entered

second) accounted for the most variance (between 31% and 33%) in each equation.

90

Experimental condition did not account for a significant proportion of the variance in any
of these hypotheses. However, after controlling for experimental condition and prior
level of psychological well-being, extent of abuse also accounted for a significant amount
of the variance in every equation. This further supported the hypothesis that extent of
abuse had a direct effect on women’s psychological well-being.

As Table 9 (analysis 2) shows, psychological well-being at Time 1 accounted for
a significant (33%) proportion of the variance in predicting psychological well-being at
Time 2. Extent of abuse at Time 2 accounted for an additional 6% of the variance in the
equation. Similarly, in analysis 3, psychological well-being at Time 2 accounted for 31%
of the variance in psychological well-being at Time 3 and an additional 6% of the
variance was significantly explained by the direct effect of abuse at Time 3.

Although the standardized regression coefficients for extent of abuse in the
across-time analyses (analyses 4 and 5) were not significant, the overall multivariate F
statistics were. This was not due to the direct effects of extent of abuse but due, instead,
to the contribution of the control variables (experimental condition in analysis 1 and prior
level of psychological well-being in analyses 2 through 5). In analysis 4, psychological
well-being at Time 1 accounted for 33% of the variance and extent of abuse at Time 1 did
not account for any additional variance. The same was true in analysis 5; that is, extent
of abuse at Time 2 did not account for any more of the variance in predicting women’s
psychological well-being at Time 3 than the 31% accounted for by prior level of

psychological well-being.

91

In summary, the results supported that the experience of abuse did have a direct
negative effect on women’s psychological well-being when both variables were measured
at Time 1, Time 2, and Time 3. That is, the more severe the abuse, the lower women’s
psychological well-being.

As was expected, prior level of psychological well-being predicted psychological
well-being in each of the analyses and accounted for a significant proportion of the
variance. However, the fact that extent of abuse had a significant R2 over and above that
of prior level of psychological well-being attests to the strength of abuse as a concurrent
predictor of psychological well-being.

Abuse at an earlier time period (e.g., Time 1), however, did not predict how
satisfied women were with their psychological well-being 6 months later (e.g., Time 2).
These longitudinal data provide additional information that women who are abused are
resilient and do not necessarily have a lower level of psychological well-being 6 months
after experiencing abuse.

W. The second hypothesis in this study argued
that level of economic resources impacts women’s overall psychological well-being and
that the lower women’s economic resources, the lower women’s psychological well-
being. That is, it was hypothesized that level of economic resources would have a direct
positive effect on the psychological well-being of women.

A summary of the results of the five hierarchical regression analyses that were
used to test the hypothesis using psychological well-being as the dependent variable and

economic resources as the independent or predictor variable are displayed in Table 10.

92

Hypothesis 2 examined this relationship within time (i.e., economic resources at Time 1
predicts psychological well-being at Time 1, economic resources at Time 2 predicts
psychological well-being at Time 2, and economic resources at Time 3 predicts
psychological well-being at Time 3) and across time (i.e., economic resources at Time 1
predicts psychological well-being at Time 2 and economic resources at Time 2 predicts
psychological well-being at Time 3). Identical to the previous hierarchical regression
summary, Table 10 displays the outcome variables and predictor variables for each
analysis in column one, standardized regression coefficients (the Beta weights) in coltunn
two, the corresponding t-test of the standardized Beta weight in column three, the step-

wise R2 in column four, and the overall multivariate F statistic in column five.

93

 

 

Table 10
.nnaao .1‘.‘r-,ro.i'u‘ or“: ‘ 0 can ”Nil" 11 o
, “.. ,-:-',. ”U ... . “by“ ”....”LHB. -v- . , “...
82in:
Outcome and Predictors B t R2 F
Analysis 1:
Outcome: Psychological Well-Being Timel
Predictors: Experimental Condition - .23 - 2.79" .03 794*
Economic Resources Time 1 - .05 - .81 .04 4.29“
Analysis 2:
Outcome: Psychological Well-Being Time2
Predictors: Experimental Condition .02 .29 .01 1.65
Psychological Well-BeingTimel .56 10.48” .34 5699“
Economic Resources Time 2 .06 1.33 .35 3872*
Analysis 3:
Outcome: Psychological Well-Being Time3
Predictors: Experimental Condition -.01 -.ll .00 .73
Psychological Well-BeingTime2 .55 9.75‘ .31 4979“
Economic Resources Time 3 2.08 .01 .31 3304"
Analysis 4:
Outcome: Psychological Well-Being Time2
Predictors: Experimental Condition .02 .35 .01 1.65
Psychological Well-BeingTimel .57 10.55" .34 5699“
Economic Resources Time 1 .02 .45 .34 3792‘
Analysis 5:
Outcome: Psychological Well-Being Time3
Predictors: Experimental Condition - .01 - .10 .00 .73
Psychological Well-BeingTime2 .55 9.86‘ .31 4979‘
Economic Resources Time 2 - .01 - .16 .31 33.05‘

 

N=221-224 for each analysis (due to missing data).

94

The results suggested that after controlling for experimental condition and prior
level of psychological well-being, the standardized regression coefficients for economic
resources were not significant in any of the concurrent or across time analyses.
Therefore, the hypothesis that level of economic resources would have a direct positive
effect on the psychological well-being of women within time and across time (Time 1 to
Time 2 and Time 2 to Time 3) was not supported. The standardized regression
coefficients for the control variables, experimental condition and prior level of
psychological well-being, however, were significant in some of the analyses. The
standardized regression coefficient for experimental condition, entered first as a control
variable, was significant in the first analysis (economic resources at Time 1 predicting
psychological well-being at Time 1) (B = -.23, p<.05), but not in any of the other
analyses. Again, this suggested that receiving the services of an advocate for 10 weeks
post—shelter (experimental condition) predicted that women would report greater
satisfaction with their overall psychological well-being 3 months afier leaving the shelter
(which corresponded with the end of their intervention), but not at any later time period.

The other control variable, prior level of psychological well-being, was a
significant predictor of later psychological well-being in analyses 2 through 5. Summary
Table 10 (analysis 2) shows that women’s psychological well-being at Time 1 had a
significant effect on women’s psychological well-being at Time 2 (B = .56, p<.05). In
analysis 3, women’s psychological well-being at Time 2 had a direct effect on women’s

psychological well-being at Time 3 (B = .55, p<.05). In analysis 4, women’s

95

psychological well-being at Time 1 had a direct impact on women’s psychological well-
being at Time 2 (B = .57, p<.05) and in analysis 5, women’s psychological well-being at
Time 2 had a direct effect on women’s psychological well-being at Time 3 (B = .55,
p<.05).

Column four (the step-wise R2) of Table 10 shows the amount of variance
accounted for by each of the independent (or predictor) variables and column five reports
the significance of the overall multivariate F statistic. Although none of the standardized
regression coefficients for level of economic resources were significant, the overall
multivariate F statistics were. This was not due to the direct effects of level of economic
resources but due, instead, to the contribution of the control variables (experimental
condition in analysis 1 and prior level of psychological well-being in analyses 2 through
5). In analysis 2, psychological well-being at Time 1 accounted for 33% of the variance
and economic resources at Time 2 only accounted for an additional 1% of the variance.
In analysis 3, psychological well-being at Time 2 accounted for 31% of the variance and
economic resources at Time 3 did not account for any additional variance. The same was
true in analyses 4 and 5. That is, psychological well-being at Time 1 accounted for 33%
of the variance and level of economic resources at Time 1 did not account for any
additional variance in analysis 4. In analysis 5, the level of economic resources at Time 2
did not account for any more of the variance in predicting women’s psychological well-
being at Time 3 than the 31% accounted for by prior level of psychological well-being.

In summary, the results did not support the hypothesis (either within or across

time) predicting that women’s level of economic resources would have a direct positive

96

effect on the psychological well-being of women. The results did not support previous
research which suggested that level of economic resources would impact women’s overall
psychological well-being such that women with lower economic resources would also
experience lower quality of life and more depression. The lack of variance in the sample
(almost all of the women were poor) was a serious confound that may explain why this
relationship was not supported.

As in the first hypothesis examining the direct effects of abuse on women’s
psychological well-being, the results of this set of analyses provide further evidence for
the importance of controlling for prior level of psychological well-being and, therefore,
looking at change in psychological well-being over time. Prior level of psychological
well-being was a significant predictor of psychological well-being and accounted for a
significant amount of the variance in all five hierarchical regression analyses.

W. The third hypothesis in this study argued
that level of economic resources would mediate, or help to explain why or how women’s
experience of abuse affected their psychological well-being. Results of a series of
regression models using psychological well-being as the outcome, abuse as the
independent variable, and economic resources as the mediator within and across time are
displayed in Tables 11 - 16. Each table displays the series of regression analyses used to
test the hypothesis within, or across, different time periods. For example, Table 11 shows
the analyses used to examine the mediating relationship within Time 1, Table 12 shows
the analyses used to examine the mediating relationship within Time 2, and so on. Three

regression equations, recommended by Baron & Kenny (1986), are displayed in each

97

table to test the mediating relationships. The format of each of these tables is similar to
the preceding two hierarchical regression tables and also display the outcome and
predictor variables in column one, standardized regression coefficients (the Beta weights)
in column two, the corresponding t-test of the standardized Beta weight in column three,
the step-wise R2 in column four, and the overall multivariate F statistic in column five.
As a reminder, the third hypothesis was tested both within time and across time.
The first analysis, displayed in Table 11, examined whether economic resources at Time
1 mediates the relationship between abuse at Time 1 and psychological well-being at
Time 1. The second analysis, shown in table 12, examined whether economic resources
at Time 2 mediates the relationship between abuse at Time 2 and psychological well-
being at Time 2. The third within time analysis examine whether economic resources at
Time 3 mediates the relationship between abuse at Time 3 and psychological well-being
at Time 3, as shown in Table 13. The other analyses examined the mediating relationship
proposed in the third hypothesis across time. The fourth analysis (Table 14) examined
the linkages of the mediational model for economic resources at Time 1 mediating the
effect extent of abuse at Time 1 had on psychological well-being at Time 2. Table 15
displays the fifth analysis that examines whether economic resources at Time 2 mediates
the effect extent of abuse at Time 2 had on psychological well-being at Time 3. The last
analysis of hypothesis 3 examined whether economic resources at Time 2 mediated the
relationship between abuse at Time 1 and psychological well-being at Time 3 and is

shown in Table 16.

98

After controlling for experimental condition and prior level of psychological well-
being, the linkages in each of the mediational models were tested. The following criteria
were used to test each mediational hypothesis (Baron & Kenny, 1986). Perfect mediation
holds when the independent variable (abuse) has no effect when the mediator (economic
resources) is controlled. Therefore, abuse must affect economic resources in the first
equation, abuse must affect psychological well-being in the second equation, and
economic resources must affect psychological well-being in the third equation. The
effect of abuse on psychological well-being must be less in the third equation than in the
second equation. The results of the regression equations used to test hypothesis 3 within
and across time are presented below, followed by an overall summary of findings.

Table 11 displays a summary of the three regression equations used to test the
linkages of the mediational model for economic resources at Time 1 mediating the effect
extent of abuse at Time 1 has on psychological well-being at Time 1. In equation one,
afier controlling for experimental condition, extent of abuse emerges as a significant
predictor of economic resources (B = .21, p<.05). However, the direction of the
relationship was opposite that which was predicted. That is, the results showed that the
more abuse women experienced, the higher their level of economic resources. The
change in R2 associated with the step involving extent of abuse was not statistically
significant.

In equation two, after controlling for experimental condition, extent of abuse was
related to psychological well-being (B = -.40, p<.05) in the predicted direction. That is,

the more abuse women reported, the lower their overall psychological well-being. In this

99

equation, the regression weight for experimental condition was also statistically
significant (B = -.18, p<.05) suggesting that women in the experimental condition
experienced more abuse. The change in R2 associated with the step involving
experimental condition and extent of abuse was also statistically significant; that is, both
experimental condition and extent of abuse at Time 1 accounted for a significant
proportion of the variance in predicting psychological well-being at Time 1.
Experimental condition accounted for 3% of the variance in predicting psychological
well-being at Time 1 and 9% of the variance in psychological well-being at Time 1 was
explained by the main effect of abuse at Time 1.

In equation three, economic resources did not significantly affect psychological
well-being, thereby suggesting that economic resources does not mediate the relationship
between abuse and psychological well-being at Time 1. Although economic resources
did not predict psychological well-being, the regression weights for experimental
condition (B = -.18, p<.05) and extent of abuse (B = -.39, p<.05) at Time 1 were
statistically significant. Again, experimental condition significantly accounted for 3% of
the variance in the equation and extent of abuse at Time 1 and economic resources at

Time 1 together accounted for 9% of the variance.

100

 

 

Tablell
nuts 0 .1‘.‘-.-, 1° -. L"‘ In 51'. ‘ 0 S 1100." ',. Hm L' t. ‘ . .u‘
u‘ter'r' i‘ '10 i0.S‘.-_ u‘ -:.H' 1.00'- --;-,7-, u‘
Outcome and Predictors B t R2 F
Equation]:

Outcome: Economic Resources Tl
Predictors: Experimental Condition .02 .25 .00 .28
Extent of Abuse T1 .21 2.06“ .02 2.26

Equation 2:

Outcome: Psychological Well-Being T1

Predictors: Experimental Condition -. 18 -2.32* .03 7.94“
Extent of Abuse T1 -.40 -4.52"‘ .12 14.53“

Equation 3:

Outcome: Psychological Well-Being Tl

Predictors: Experimental Condition -. 18 -2.31* .03 7.94"
Extent of Abuse T1 -.39 -4.44"
Economic Resources T1 .55 - .23 .12 9.66"

 

N=224; T1 = 3 months post-shelter, T2 = 9 months post-shelter, T3 = 15 months post-shelter .

Table 12 displays a summary of the three regression equations used to test the
linkages of the mediational model for economic resources at Time 2 mediating the effect
extent of abuse at Time 2 has on psychological well-being at Time 2. In equation one,
after controlling for experimental condition, extent of abuse does not significantly affect
economic resources.

In equation two, after controlling for experimental condition and Time 1

psychological well-being, extent of abuse was related to psychological well-being (B = -

101

.29, p<.05). That is, the more abuse women experienced at Time 2, the lower their
psychological well-being at Time 2. Extent of abuse accounted for a significant portion
of the variance in this equation (6%), even though prior level of psychological well being
was also a significant predictor (B = .51, p<.05) and accounted for a large, and
significant, amount of the variance (33%).

In equation three, however, after controlling for experimental condition and Time
1 psychological well-being, economic resources did not significantly predict
psychological well-being. The regression weights for prior level of psychological well-
being (B = .51, p<.05) and extent of abuse (B = -.31, p<.05) were significant. Prior level
of psychological well-being was a significant predictor and accounted for 33% of the
variance, with extent of abuse and economic resources together adding another 7%. Both
control variables (experimental condition and prior level of psychological well-being) and
both predictors (extent of abuse and economic resources) accounted for 41% of the
variance. Again, because the results of equation one and equation three did not satisfy
the mediational criteria as proposed by Baron & Kenny (1986), economic resources did

not mediate the relationship between abuse and psychological well-being at Time 2.

102

 

 

Table 12
JIM-,1. -r'o‘ra. 1'2. ;... urn, oe mu: ' u-u tw- '{ '21::
U~.":_.~A.- i- ,-. . ;.,-. u‘ .1! ...9', ,‘-:-1,-; '..~
Outcome and Predictors B t R2 F
Equation 1:
Outcome: Economic Resources T2
Predictors: Experimental Condition .04 .42 .00 .23
Extent of Abuse T2 .16 1.66 .01 1.50
Equation 2:
Outcome: Psychological Well-Being T‘2
Predictors: Experimental Condition .02 .34 .01 1.65
Psychological Well-Being T1 .51 9.73“ .34 56.99”
Extent of Abuse T2 -.29 -4.44* .40 47.79"
Equation 3:
Outcome: Psychological Well-Being T2
Predictors: Experimental Condition .02 .24 .01 1.65
Psychological Well-Being T1 .51 9.59“ .34 56.99"
Extent of Abuse T2 -.31 -4.69*
Economic Resources 12 .09 2.00 .41 37.33"

 

N = 222-223 for each analysis (due to missing data); T1 = 3 months post-shelter, T2 = 9 months post-

shelter, T3 = 15 months post-shelter.

Table 13 shows that the pattern of results for the three regression equations used

to test the linkages of the mediational model for economic resources at Time 3 mediating

the effect extent of abuse at Time 3 has on psychological well-being at Time 3 are

identical to those within Time 2 (see Table 12). After controlling for experimental

103

condition in the first equation, the results showed that extent of abuse did not
significantly affect economic resources.

In equation two, after controlling for experimental condition and Time 2
psychological well-being, extent of abuse was related to psychological well-being
(B = -.27, p<.05). That is, the more abuse women experienced at Time 2, the lower their
psychological well-being at Time 3. Extent of abuse accounted for a significant portion
of the variance in this equation (6%), even though prior level of psychological well being
was also a significant predictor (B = .51, p<.05) and accounted for a substantial amount
of the variance (31%).

As the third equation in Table 13 shows (after controlling for experimental
condition and Time 2 psychological well-being), however, economic resources did not
significantly predict psychological well-being. The regression weights for prior level of
psychological well-being (B = .50, p<.05) and extent of abuse (B = -.27, p<.05) were
significant. Both control variables (experimental condition and prior level of
psychological well-being) and both predictors (extent of abuse and economic resources)
accounted for 37% of the variance. Psychological well-being significantly accounted for
31% of the variance and extent of abuse and economic resources together accounted for
an additional 6% of the variance. However, economic resources did not mediate the
relationship between abuse and psychological well-being at Time 3 because the results of
equation one and equation three did not satisfy the mediational criteria as proposed by

Baron & Kenny (1986).

104

 

 

Table 13
Jon-... ',.1'o't¢.-,ll:,i"4‘ urn-.1 ‘ o ‘ ormr 2 01011170. ‘ ._ ,n‘
U"11'.1' i‘ 1‘! LaL‘Z 11' 11-01,. h"C-. l‘-:‘r'-. .11‘
Outcome and Predictors B t R2 F
Equationl:

Outcome: Economic Resources T3
Predictors: Experimental Condition .14 1.43 .01 2.33
Extent of Abuse T3 .09 .93 .01 1.60

Equation 2:

Outcome: Psychological Well-Being T3

Predictors: Experimental Condition .02 .27 .00 .73
Psychological Well-Being T2 .51 9.35" .31 49.79“
Extent of Abuse T3 -.27 -4.29"' .37 41 .99“
Equation 3:

Outcome: Psychological Well-Being T3

Predictors: Experimental Condition .01 .22 .00 .73
Psychological Well-Being T2 .50 9.08‘ .31 49.79“
Extent of Abuse T3 -.27 -4.3 l *
Economic Resources T3 .02 .44 .37 31.42‘

 

N = 221-224 for each analysis (due to missing data); T1 = 3 months post-shelter, T2 = 9 months post-
shelter, T3 = 15 months post-shelter.

Similar to the preceding analyses that examined the within time mediating
relationships, none of the across time mediating relationships proposed in hypothesis 3
supported a mediating relationship. Table 14 displays a summary of the regression
analyses used to test whether economic resources at Time 1 mediates the effect extent of

abuse at Time 1 has on psychological well-being 6 months later (Time 2). Although

105

abuse was a significant predictor of economic resources (B = .21, p<.05) in equation one,
abuse did not significantly relate to psychological well-being in equation two, nor did
economic resources significantly predict psychological well-being in equation 3.
Therefore, Baron & Kenny’s (1986) mediational criteria were not met and it can be
concluded that economic resources at Time 1 did not mediate the relationship between
abuse at Time 1 and psychological well-being at Time 2.

However, prior level of psychological well being was a significant predictor of
psychological well-being in equations 2 and 3 (B = .55, p<.05) and significantly
accounted for 33% of the variance in both equations. In the third equation, extent of
abuse at Time 1 also had a direct negative effect on psychological well-being at Time 2
(B = -.07, p<.05). Psychological well-being accounted for 33% of the variance in
equation 2 and equation 3. Extent of abuse and economic resources only accounted for

an additional 1% of the variance in equation 3.

106

 

 

Tab1e14
.111121 0 .170'111. 11‘2“ 11 a1: .' 1 it. 11.1“! ' 1,1111 ’t‘ 1. 'Se 11‘
U"tl'.1' ,i‘, ‘1 0 it ': '11‘ gain,’ 10 0': W -,,=‘1°-. -11‘
Outcome and Predictors B t R2 F
Equationl:

Outcome: Economic Resources Tl
Predictors: Experimental Condition .02 .25 .00 .28
Extent of Abuse T1 .21 2.06“ .02 2.26

Equation 2:

Outcome: Psychological Well-Being T2

Predictors: Experimental Condition .03 .42 .01 1.65
Psychological Well-Being T1 .55 9.86“ .34 56.99“
Extent of Abuse T1 -.07 - .85 .34 38.19‘
Equation 3:

Outcome: Psychological Well-Being T2

Predictors: Experimental Condition .03 .41 .01 1.65
Psychological Well-Being T1 .55 9.85“ .34 56.99"
Extent of Abuse T1 -.07 - .92"
Economic Resources T1 .03 .56 .35 28.63"

 

N=222-224 for each analysis (due to missing data); T1 = 3 months post-shelter, T2 = 9 months post-
shelter, T3 = 15 months post-shelter.

Table 15 illustrates the mediational model for economic resources at Time 2
mediating the effect extent of abuse at Time 2 has on psychological well-being at Time 3.
After controlling for experimental condition and prior level of psychological well-being,
none of the equations contain the significant relationships that were necessary to support

this mediational analysis. That is, abuse at Time 2 did not have a direct negative impact

107

on economic resources at Time 2, abuse at Time 2 did not have a direct negative impact
on psychological well-being at Time 3, and economic resources at Time 2 did not have a
direct positive effect on psychological well-being at Time 3.

Consistent with other findings, the regression weight for prior level of
psychological well-being in equations two and three were significant (B = .54 and .55,
respectively, p<.05) and responsible for the significant overall F statistic. In both
equations, psychological well-being accounted for 31% of the variance which was also
the total variance accounted for in each of the equations. That is, extent of abuse (in
equation 2) and extent of abuse and economic resources (in equation 3) did not account

for any additional variance.

108

 

 

TablelS
1111-1 1 ..1‘o‘r-u 1' 1. i“‘ 11‘- ‘ 1 ‘ .1111‘1 ' 11111 i“ 11 ‘ e 11'
11‘1'21'1 i‘ ‘1 1 a1-‘-. 11' 1.411’ 111': ‘-='1,'-. 11‘
Outcome and Predictors B t R2 F
Equationl:

Outcome: Economic Resources T2
Predictors: Experimental Condition .04 .42 .00 .23
Extent of Abuse T2 .16 1.66 .01 1.50

Equation 2:

Outcome: Psychological Well-Being T3

Predictors: Experimental Condition -.01 -.11 .00 .73
Psychological Well-Being T2 .54 9.18“ .31 49.79“
Extent of Abuse T2 -.02 - .21 .31 33.06“
Equation 3:

Outcome: Psychological Well-Being T3

Predictors: Experimental Condition -.01 -. 10 .00 .73
Psychological Well-Being T2 .55 9.06“ .31 49.79"
Extent of Abuse T2 -.01 - .19‘
Economic Resources T2 -.01 -.13 .31 24.69‘

 

N = 221-223 for each analysis (due to missing data); T1 = 3 months post-shelter, T2 = 9 months post-
shelter, T3 = 15 months post-shelter.

Table 16 displays the final set of equations used to test the mediational model
across all three time points (economic resources at Time 2 mediating the effect extent of
abuse at Time 1 has on psychological well-being at Time 3). In equation one, extent of
abuse at Time 1 was significantly related to economic resources at Time 2 (B = .21,

p<.05) but did not account for a significant portion of the variance in the equation (2%).

109

The results that were necessary to support the mediational relationship in the second two
equations were not significant. Therefore, economic resources at Time 2 did not mediate
the relationship between extent of abuse at Time 1 and psychological well-being at Time
3.

However, the regression weight for psychological well-being at Time 2 was
significant (B = .53, p<.05) in both equation two and equation three and prior level of
psychological well-being accounted for 31% of the variance. The regression weight for
extent of abuse at Time 1 in equation three was also significant (B = -.12, p<.05),
although extent of abuse and economic resources only accounted for an additional 1% of
the variance. The overall multivariate F statistic was significant but only because of the

influence of controlling for prior level of psychological well-being.

110

 

 

Tablel6
.11112.t0.1'I'I’-__1o.i‘°‘ ..;.._ ‘ 1 ‘ 1111111‘ ' 1 111, 1'1- ‘ , .11‘
11‘1111'1‘ i' ‘11 i1_‘o. 11‘ 11-1 ' 11111:..Wl--.=‘n'.t 11‘
Outcome and Predictors B t R2 F
Equationl:

Outcome: Economic Resources T2
Predictors: Experimental Condition .02 .23 .00 .23

Extent of Abuse T1 .21 1.95" .02 2.02

Equation 2:

Outcome: Psychological Well-Being T3

Predictors: Experimental Condition .00 .07 .00 .73
Psychological Well-Being T2 .53 9.41 * .31 49.79"
Extent of Abuse T1 -. 12 -1.56 .32 34.22“
Equation 3:

Outcome: Psychological Well-Being T3

Predictors: Experimental Condition .00 .06 .00 .73
Psychological Well-Being T2 .53 9.28‘ .31 49.79"
Extent of Abuse T1 -.12 -1.55"‘
Economic Resources 12 .00 .09 .32 25.55"

 

N = 221-223 for each analysis (due to missing data); T1 = 3 months post-shelter, T2 = 9 months post-
shelter, T3 = 15 months post-shelter.

In summary, the within time and across time analyses of the hypothesis that
women’s level of economic resources would mediate the relationship between abuse and
psychological well-being were not substantiated. Economic resources did not mediate
this relationship at any of the time points. Interestingly, the direction of the hypothesized

relationship between abuse and economic resources was wrong. The results suggest that

111

women who experience higher levels of abuse at Time 1 also have higher levels of
resources at Time 1 and Time 2. This is contrary to hypothesis 3 that proposed that
women who experience higher levels of abuse would have lower levels of resources
because assailants would undermine their employment by harassing them at work or
injuring them to the point of missing work, isolate them from community resources, and
drain the family resources. Instead, for the women in this sample, there is a relationship

between higher levels of abuse and higher level of economic resources.

 

Logistic regression was used to examine the strength and direction of
relationships among the second set of variables of interest in this study; that is, the
relationships between women’s level of economic resources and involvement with the
assailant as moderated by economic dependence. The literature and previous research
suggest that women’s level of economic resources are related to concurrent and future
involvement with the assailant. That is, women stay with, or return to, their assailants
because they lack the economic and community resources to maintain the same standard
of living on their own. Relatedly, previous research also suggests that women stay with
or return to their assailants because they are economically dependent on them. Therefore,
the relationship between economic resources and involvement with the assailant was
hypothesized to be moderated by whether women were economically dependent. That is,
women’s level of economic resources would affect whether they were involved with their

assailant only if they were economically dependent for those resources.

112

The ability of economic dependence to moderate the relationship between
economic resources and involvement with assailant was hypothesis 4 in the Research
Objectives and Hypotheses section of Chapter 1. This frnal hypothesis was tested within
the three time periods, across Time 1 and Time 2, and across Time 2 and Time 3 (5
separate analyses). Logistic regression was used to predict involvement with the assailant
because it was a dichotomous variable. Of primary interest was whether the interaction
between economic resources and economic dependence was significant.

A summary of the results of each of the five logistic regression equations are
featured in Table 17. The first column displays the outcome variable and the predictor
variables. The second column (with the heading B) features the logistic coefficient which
can be interpreted as the amount of change in the log odds of assailant involvement that is
associated with a one-unit change in the independent variable. The third column is
labeled Exp(B) and is the relative effect on odds ratio. The fourth column presents the
Wald statistic which is a z-test on the regression weight (B) and has a chi-square
distribution. The fifih column, R, features the partial correlation between each
independent variable and assailant involvement. The last line of each hypothesis in Table
17 features the Model-Chi Square which is the test for overall R2. It explains whether the

group of predictors significantly predicts involvement with assailant.

113

Table 17

111141 1' i".' 11;“ ‘ 1 ‘ .1111“ 11111 h‘1r'11‘1 ‘-..o.u11‘ro_1

E '1! E 11'111111-5'

 

Outcome and Predictors B Exp(B) Wald R
Statistic

Analysis 1:

Outcome: Involvement with Assailant T1

Predictors: Experimental Condition .17 1 . 19 .3 5 .00
Economic Resources T1 .35 1.42 .48 .00
Economic Dependence T1 -1.1 1 .33 4.28“ -.09*
Econ. Res. T1 X Econ. Dep. T1 .38 1.46 .45 .00

Model Chi-Square: 2005"

Analysis 2:

Outcome: Involvement with Assailant T2

Predictors: Experimental Condition -.05 .94 .03 .00
Economic Resources '12 2.00 7.39 13.41". .21“
Economic Dependence T2 -.96 .38 5.01 "' -.1 1“
Econ. Res. T2 X Econ. Dep. T2 -2.07 .13 10.31" -.18"‘

Model Chi-Square: 46.99"

Analysis 3:

Outcome: Involvement with Assailant T3

Predictors: Experimental Condition . 17 1 . 18 .24 .00
Economic Resources T3 1.65 5.20 10.33" .19*
Economic Dependence T3 .07 1.07 .01 .00
Econ. Res. T3 X Econ. Dep. T3 -1.56 .21 6.88“ -.14"'

Model Chi-Square: 24.07‘

1 14
Table 17 continued

.1111: 1.,1'111“: ‘ 1 ‘ .1111‘4. , 111119‘1‘1‘1‘1.-.1111‘r-_1 1

E 'E E ll'lllIII-E'

 

Outcome and Predictors B Exp(B) Wald R
Statistic

 

Analysis 4:

Outcome: Involvement with Assailant '12

Predictors: Experimental Condition .12 1 .13 . l 5 .00
Economic Resources T1 .71 2.03 1.57 .00
Economic Dependence T1 -.90 .41 2.69 -.05
Econ. Res. T1 X Econ. Dep. T1 .19 1.21 .10 .00

Model Chi-Square:24.25“

Analysis 5:

Outcome: Involvement with Assailant T3

Predictors: Experimental Condition .1 5 I . 16 . 19 .00
Economic Resources T2 1.32 3.76 9.32“ .17‘
Economic Dependence T2 -.56 .57 1.67 .00
Econ. Res. T2 X Econ. Dep. T2 -1.11 .33 4.07‘ -.09"

Model Chi-Square:26.74"

 

N = 221-224 for each analysis (due to missing data); T1 = 3 months post-shelter, T2 = 9 months post-
shelter, T3 = 15 months post-shelter; p<.05.

The first analysis within hypothesis 4 stated that economic dependence at Time 1
moderates the relationship between level of economic resources at Time I and
involvement with the assailant at Time 1. After controlling for experimental condition,
there was a main effect for economic dependence at Time 1 predicting involvement with
the assailant (B = -1.11; Wald = 4.28, p<.05) such that women who were economically

dependent were more likely to also be involved with their assailants. However, there was

115

no interaction between economic resources and economic dependence. Therefore, the
results did not provide support for the moderational hypothesis.

The Model Chi-Square was significant (Model x2 = 20.05, p<.05) in this equation
meaning that this group of predictors (experimental condition, economic resources at
Time 1, economic dependence at Time 1, and the interaction of economic resources at
Time 1 by economic dependence at Time 1) were able to significantly predict
involvement with assailant at Time 1. In addition, this group of predictors was overall
able to correctly classify the involvement status of 68.3% of the participants.

The second analysis in hypothesis 4 stated that economic dependence at Time 2
moderates the relationship between level of economic resources at Time 2 and
involvement with the assailant at Time 2. After controlling for experimental condition,
there was a significant main effect of economic resources at Time 2 on Involvement with
Assailant at Time 2 (B = 2.00; Wald = 13.41, p<.05). That is, women with higher
economic resources were more likely to be involved with their assailants. Economic
dependence, too, directly impacted involvement with assailant (B = -.96; Wald = 5.01,
p<.05) such that women who were economically independent were more likely to be
involved with their assailants.

Interestingly, the interaction between economic resources at Time 2 and economic
dependence at Time 2 was significant (B = -2.07; Wald = 10.31, p<.05) providing
evidence for the moderational hypothesis. To help understand this moderating effect,
Figure 5 shows a plot of economic dependence at Time 2 as a moderator of the

relationship between level of economic resources at Time 2 and the log odds of

ll6

involvement with the assailant at Time 2. Women’s standardized scores on level of
economic resources are plotted on the x-axis. The log odds that women would be
involved with their assailants are plotted on the y-axis. The probability scores were
converted to logits (1n (1t(x)/ 1 - 1t(x)) before they were plotted in order to obtain a linear
plot. As the legend shows, the line with shorter dashes represents women who were
economically independent and the line with longer dashes represents women who were

economically dependent.

 

 

 

 

FigureS
, 11111 |‘1‘11‘1‘o. 11‘ -_- 111117111 1 .1‘ {3111 11=‘h“1 ‘ ‘ 1
11111 {‘1._‘ -. 11‘ 1.11.1‘ 1°.11 1 11 111111. 1
Time2
6
fl
//
41
\ J
3 ,’
a 2. ’0'
,0
0
g 01 /
‘8 . “I
g -2“, "m" """""" Economic Dependenc
: _Econ. Independent
g -4 o Econ. Dependent
-1 0 ‘1 2

Economic Resources

Figure 5 confirms the hypothesis and illustrates that when women were
economically dependent at Time 2, the relationship between economic resources at Time
2 and involvement with assailant at Time 2 was positive. When women were

economically independent (i.e., not dependent on anyone else for the money they live

117

on), the relationship between economic resources at Time 2 and involvement with
assailant at Time 2 was near-zero. As in the first analysis, the Model Chi-Square was
significant (Model x2 = 46.99, p<.05) in this equation meaning that this group of
predictors (experimental condition, economic resources at Time 2, economic dependence
at Time 2, and the interaction of economic resources at Time 2 by economic dependence
at Time 2) were able to significantly predict involvement with assailant at Time 2. This
group of predictors was able to correctly classify the involvement status of 78.8% of the
participants.

The third analysis of hypothesis 4 explored whether economic dependence at
Time 3 moderates the relationship between level of economic resources at Time 3 and
involvement with the assailant at Time 3. Afier controlling for experimental condition,
there was a significant main effect of economic resources at Time 3 on Involvement with
Assailant at Time 3 (B = 1.65; Wald = 10.33, p<.05). That is, women with higher
economic resources were more likely to be involved with their assailants. Economic
dependence, in this equation, did not independently impact involvement with assailant.
The interaction between economic resources at Time 3 and economic dependence at Time
3, however, was significant (B = -1.56; Wald = 6.88, p<.05) providing evidence for the
moderational hypothesis.

To help understand this moderating effect, Figure 6 shows a plot of economic
dependence at Time 3 as a moderator of the relationship between level of economic

resources at Time 3 and the probability of involvement with the assailant at Time 3.

118

The plot of economic dependence at Time 3 in Figure 6 is very similar to the plot of
economic dependence at Time 2 in Figure 5. Figure 6 confirms the hypothesis and shows
that when women were economically dependent at Time 3, the relationship between
economic resources at Time 3 and involvement with assailant at Time 3 was positive.
When women were economically independent (i.e., not dependent on anyone else for the
money they live on), the relationship between economic resources at Time 3 and

involvement with assailant at Time 3 was not affected.

Figure6
.11111 .‘11‘11‘1‘2 .11‘ 1.111117111 1 .1‘1‘111113‘11‘1 ‘ ‘ 1
111111 1511‘ -. 11‘ 0.1.. 11‘ 1' .11 1 11V'11‘1 1-1-1‘; .2. 1.1 1

 

'9 L9
\
\
\
g

‘51.

1'9
°9

Economic Depende

: _Econ. Independent
0 Econ. Dependent

 

Li;

 

IogOddsofInvolvenent w/Assailant
éj

 

I
—l

' 0 1 2 3 4

Economic Resources

The Model Chi-Square was significant (Model x2 = 24.07, p<.05) in this equation
meaning that this group of predictors (experimental condition, economic resources at

Time 3, economic dependence at Time 3, and the interaction of economic resources at

119

Time 3 by economic dependence at Time 3) were able to significantly predict
involvement with assailant at Time 3. This group of predictors was able to correctly
classify the involvement status of 78.7% of the participants.

The fourth analysis of hypothesis 4 examined whether economic dependence at
Time 1 moderates the relationship between level of economic resources at Time 1 and
involvement with the assailant at Time 2. After controlling for experimental condition,
there were no statistically significant main effects for economic resources at Time 1 or
economic dependence at Time 1 and no interaction effect between economic resources at
Time 1 and economic dependence at Time 1. Therefore, the results did not provide
support for the moderational hypothesis. Once again, however, this group of predictors
(experimental condition, economic resources at Time 1, economic dependence at Time 1,
and the interaction of economic resources at Time 1 by economic dependence at Time I)
were able to significantly predict involvement with assailant at Time 2 (Model x2 =
24.25, p<.05) in this analysis meaning that this group of predictors were able to
significantly predict involvment with assailant at Time 2. This group of predictors was
able to correctly classify the involvement status of 74.4% of the participants.

The fifth, and last, analysis of hypothesis 4 examined whether economic
dependence at Time 2 moderates the relationship between level of economic resources at
Time 2 and involvement with the assailant at Time 3. After controlling for experimental
condition, there was a significant main effect of economic resources at Time 2 on
Involvement with Assailant at Time 3 (B = 1.32; Wald = 9.32, p<.05). That is, women

with higher economic resources were more likely to be involved with their assailants.

120

Economic dependence, in this equation, did not impact involvement with assailant. The
interaction between economic resources at Time 2 and economic dependence at Time 2,
however, was significant (B = -l.11; Wald = 4.07, p<.05) providing evidence for the
moderational hypothesis.

Figure 7 displays a plot of economic dependence at Time 2 as a moderator of the
relationship between level of economic resources at Time 2 and the probability of
involvement with the assailant at Time 3 to help describe this interaction effect. The plot
of economic dependence in Figure 7 displays the same pattern of results as the plots of
economic dependence in Figures 5 and 6. Figure 7 confirms the hypothesis and shows
that when women were economically dependent at Time 2, the relationship between
economic resources at Time 2 and involvement with assailant at Time 3 was positive.
When women were economically independent (i.e., not dependent on anyone else for the
money they live on), the relationship between economic resources at Time 2 and
involvement with assailant at Time 3 was near-zero. The Model Chi-Square was
significant (Model x2 = 26.74, p<.05) in this last equation also, meaning that this group of
predictors (experimental condition, economic resources at Time 2, economic dependence
at Time 2, and the interaction of economic resources at Time 2 by economic dependence
at Time 2) were able to significantly predict involvement with assailant at Time 3. This
group of predictors was able to correctly classify the involvement status of 78.8% of the

participants.

121

 

 

 

 

Figure7
1111 .‘11‘11‘1‘1 11‘ 1111113111 1 .1‘1‘111113‘ “1 ‘V‘ 1
1,1111 1‘ 1.. ‘ -. 11‘ 1111‘ 1' .11. 1 ,1V1V1‘11‘1 -1 .1 1'11 - ,11
3.
4
31 ’0‘
\ /
a 21 //
IO
/
E 1. 4’3
9°
E o. 3
("5' -11 ------ 4
g , 4W" W- Economic Depende
“2‘4: ° Econ. Independent
E.” -3 o Econ. Dependent
-1 0 1 2

Economic Resources

A significant number of findings were significant in the logistic regression
analyses used to test hypothesis 4. First, economic dependence moderated the
relationship between economic resources and involvement with the assailant at Time 2.
Second, economic dependence moderated the relationship between economic resources
and involvement with the assailant at Time 3. And third, economic dependence at Time 2
moderated the relationship between economic resources at Time 2 and involvement with
the assailant at Time 3. These interactions suggest that when women were economically
dependent on someone for the money they live on, their level of economic resources had
a positive relationship with the probability that they would be involved with their

assailant. In other words, the probability of women’s involvement with their assailant

122

increased as their levels of economic resources increased. When women were
economically independent there was no relationship between level of economic resources
and the probability that women would be involved with their assailants. That is, the
probability of women’s involvement with their assailant did not change based on their

level of economic resources.

DISCUSSION

The primary objective of the present research was to explore how level of
economic resources and economic dependence impacts women who have been battered
and stayed at a domestic violence shelter. More specifically, to look at how both abuse
and level of economic resources affects women’s psychological well-being and to
investigate the role of economic dependence and level of economic resources in
predicting involvement with the assailant. The results of this investigation support
previous research suggesting that being abused by an intimate male partner negatively
impacts women’s psychological well-being at concurrent time periods (Gelles & Harrop,
1989; Goodman, Koss, Russo, 1993; Koss, 1990). This study did not, however, support
the predicted positive relationship between women’s level of economic resources and
women’s psychological well-being either within or across time. Women’s level of
economic resources also did not appear to explain, or mediate, the relationship between
level of abuse and psychological well-being.

Economic dependence overwhelmed the effects of economic resources in the
current study and emerged as a critical factor. This study showed that women who were
economically dependent (i.e., either partially or fully reliant on another person for the
money they needed to live on) had a higher level of economic resources than women
who were economically independent. However, this study also intimates that economic
dependence increased the odds that women would be involved with their assailants (1’5

123

124

ranging from -.16 to -.35, p<.05). Women who were involved with their assailants
experienced higher levels of continued abuse than women who were no longer involved
with their assailants (r’s ranging from .29 to .61, p<.05). Finally, results showed strong
direct effects of concurrent abuse on women’s psychological well-being; that is, women
who experienced further abuse reported greater dissatisfaction with their overall
psychological well-being (r’s ranging from -.21 to -.36, p<.05).

Economic dependence moderated the relationship between women’s level of
economic resources and involvement with the assailant at 9 months post-shelter, 15
months post-shelter, and across these two time periods. The interaction effects suggest
that it was not the level of economic resources that women had, but being economically
dependent on someone else for financial stability that kept women trapped in abusive
relationships. That is, if women were economically independent, their level of economic
resources did nothing to predict involvement with assailant. If women were
economically dependent, however, their level of economic resources did affect their odds
of being involved with their assailant. Women who were dependent and had a high level

of economic resources were more likely to be involved with their assailants than women

who were dependent and had a low level of economic resources.

 

Participants in this study were originally recruited because they had been
physically battered and stayed at a domestic violence shelter. Although, on average,
women did not report continuing to experience high levels of abuse, the percentages of

women who were still experiencing abuse 3, 9, and 15 months post-shelter were

125
substantial and indicate that many women were still being battered over the fifteen month
period after they left the shelter.

The results of this investigation affirrn previous studies suggesting that the
experience of abuse negatively impacts women’s psychological well-being (e. g., Astin et
al., 1993; Cascardi & O'Leary, 1992; Follingstad et al., 1991; Gelles & Harrop, 1989;
Houskamp & F oy, 1991; Kemp et al., 1991; Mitchell & Hodson, 1983; Sato & Heiby,
1992). Women who were continually physically and emotionally abused by their
partners or ex-partners were more likely to have lower psychological well-being at 3, 9,
and 15 months post-shelter. Experience of abuse, however, did not predict how women
felt about their psychological well-being six months in the future. The six month time lag
in- between interviews may have been too long to detect a causal relationship.

- 1 . ..." {g - ,.. 101171, ' .. 0'1: ‘- -:-.o

It was hypothesized, because poverty itself is a stressor and brings with it
persistent undesirable conditions such as inadequate housing, inadequate nutrition,
dangerous neighborhoods, poor medical care, and financial uncertainties (Belle, 1990;
Belle et a1, 1982; Dohrenwend, 1978), that women with lower economic resources would
also have lower psychological well-being. However, the results of this study did not
substantiate a relationship between level of economic resources and the psychological
well-being of women with abusive partners.

The failure to find significant direct effects of economic resources on women’s
psychological well-being within or across time may be due to several factors. Even

though there was a fairly normal distribution on economic resources, the distribution was

126

within a very restricted range. That is, most of the women were economically
impoverished and had few economic resources so that the group differences were
between women who were somewhat poor, women who were poor, and women who were
really poor. This lack of variance is not surprising because women who use battered
women’s shelters tend to be without other economic and/or social resources (Okun, 1986;
Sullivan & Rumptz, 1994; Sullivan et al., 1992), but it is a limitation of current study that
may have influenced some of the findings.

Relatedly, this lack of variance in women’s level of economic resources was fairly
constant over time. Some research suggests that it may be the change, or relative drop, in
the standard of living that women face when leaving an abusive partner that may be the
critical dimension of economic resources that would affect their well-being (Greaves,
Heapy, & Wylie, 1988). F ollow-up analyses in the present study looking at change in
level of economic resources across time as a predictor of psychological well-being did
not produce different results than looking at absolute level of economic resources. Again,
this most likely was true because there were no significant drops, on average, in
women’s standard of living so change in economic resources was not possible to assess.
Because women in this sample had a low level of economic resources to begin with (most
were living in poverty), there was less room for change in level of economic resources.

The failure to find significant direct effects of economic resources on women’s
psychological well-being may also be due to the measure used to assess economic
resources. The Economic Resources scale was a new measure with previously untested

reliability and validity. The Economic Resources scale used in this study was composed

127
of four items assessing absolute level of resources and included income, poverty level,
employment, and access to a car. Absolute measures of level of economic resources,
such as this, may be systematically misleading in assessing the vulnerability and
ambivalence battered women feel about their situation (Greaves, Heapy, & Wylie, 1988).
Perhaps more critical to assess would be a broader and more applicable range of
economic resource variables (e.g., child care, health care, transportation), and the degree
to which each of those economic resources were tied to the assailant.

Another reason for the lack of relationship between women’s level of economic
resources and psychological well-being within and across time may be that the direct
effects of economic resources were overwhelmed by other interrelated factors such as
economic dependence.

101011: 9.. ' -.. -. U'Qs 0 0 .1' .\ 1-01.10:'h"1.1'..|'l'1' 0. $0. '

The fact that there was little variance on economic factors within this sample may
also account for why economic resources did not mediate the relationship between extent
of abuse and psychological well-being. For the mediational relationship to hold true, it
was hypothesized that not only would extent of abuse directly affect psychological well-
being, but also that extent of abuse would affect level of economic resources, and level of
economic resources would in turn affect women’s psychological well-being. A

discussion of each of the components necessary to indicate a mediational model are

presented below to try to understand why the predicted relationships did not hold true.

128

The first relationship that was required to hold true in order to support the
mediational model was that women who experienced more abuse would also report
having fewer resources. The literature supported the hypothesis that abusive men
directly and indirectly negatively affect women’s economic standing. That is, abusive
men may make women miss work or harass them at work to the point of being fired,
thereby negatively affecting their employment status. Women who are battered may be
physically injured or depressed, thus affecting their ability to go to work or to a social
service agency (if they are on government assistance). Abusive men may run up
household bills that are in the women’s name as punishment or break household items
and tear clothing in violent attacks which also negatively affects women’s level of
economic resources. Additionally, assailants may take the car and leave women stranded
or purposefully sabotage women’s attempts to return to school.

However, this study did not provide evidence that women who experienced more
abuse would also report having fewer resources. In fact, at 3 months post-shelter the
opposite was true. That is, women who experienced more abuse also reported having
more economic resources. Conceptually, this finding can be explained in light of some of
the findings from this research. In this study, women who had more economic resources
were also more likely to be economically dependent (most likely on their assailant) for
those resources and were more likely to be involved with the assailant, therefore being
more likely to experience further abuse. Therefore, it is possible that because of the high

correlations between economic resources and involvement with assailant, that

129

involvement with the assailant was really responsible for the finding that women who
reported more abuse also reported having more economic resources.

The second relationship that needed to be significant in order to support the
mediational model was that women who experienced more abuse would also report
having lower psychological well-being. Examined concurrently, extent of abuse did
predict women’s level of psychological well-being such that women who reported higher
levels of abuse at either 3, 9, or 15 months post-shelter also reported lower satisfaction
with their psychological well-being at the same time period. As previously discussed,
however, this relationship did not hold up across time periods.

The final component necessary to support economic resources as a mediator
between extent of abuse and psychological well-being was that economic resources
needed to have a direct, positive effect on psychological well-being. Again, as was
discussed earlier, this relationship was not substantiated possibly due to the lack of
variance and change over time in the level of economic resources in the sample,
measurement problems, and the possible confound of economic dependence. Due to the
fact that several components necessary to establish a mediational relationship were not
found, it was clear that economic resources did not mediate the relationship between
abuse and psychological well-being either concurrently or across time.

|.- 111.. 1,..." {-1_.-.__11_11111 I-.-.._-.-... ...-1’1-‘1.
One of the most interesting findings in this study was that economic dependence

moderated the relationship between economic resources and involvement with assailant 9

130

months post-shelter, 15 months post-shelter, and across these two time periods. That is,
women’s level of economic resources affected whether they were involved with their
assailants only if they were economically dependent at the time. If women were not
economically dependent, their level of economic resources did little to predict whether
they were involved with their assailants. For example, if women brought in all of the
money that they lived on (regardless of whether it was $200.00 a month or $2,000.00 a
month), their level of economic resources did not influence whether they would be
involved with their assailant. However, if women were economically dependent for all
or part of their income, their level of economic resources did influence whether they
would be involved with their assailant.

That is, women who were economically dependent and had a higher level of
economic resources also had greater odds of being involved with their assailant. Women
who were economically dependent and had lower levels of economic resources had lower
odds of being involved with their assailant. In simple terms, dependent women with
higher levels of economic resources would be more likely to be involved with their
assailants because they would have more to lose if they left. Dependent women with
lower levels of economic resources would have less to lose. Therefore, this study does
not support that economic entrapment is most severe for women with fewer economic
resources. The results are contrary to the widespread presumption that the more
resources women have (such as employment, income, material goods and services that

money can buy, and level of education) the more likely they are to leave their assailant.

131

This research lends evidence that the relationship between economic resources and
involvement with the assailant is more complicated than that.

This study suggests (and replicated at different time periods) that it is not level of
economic resources, but economic dependence, that keeps women trapped in abusive
relationships. In a sample where almost three-quarters of the women were living in
poverty, any extra money that women received from an outside source (such as the
batterer or family member) was very important to preserve. When women were reliant on
the assailant for some or all of the income, it is understandable why the decision to end
the relationship would be a difficult one. The decision to endure continued abuse versus
the potential for a reduction in already precious economic resources places women in
double jeopardy; a situation in which society will blame them for making either of these
difficult choices.

Other factors also enter into the decision making process. Economic dependence
is often reinforced by considerations of the impact of the separation on children
(Greaves, Heapy, & Wylie, 1988). Over three-quarters of this sample had children living
with them. When women were economically dependent, the decision to leave and
experience a significant negative change in standard of living becomes more difficult
when it involves the disruption of the children, and is further exacerbated the more
children women have.

Unfortunately, when economically dependent women make the choice to preserve
their economic status and continue involvement with their assailant, it may lead to further

abuse and, therefore, lower overall psychological well-being. This study found that

132

women who were involved with their assailants had higher economic resources, were
more likely to be dependent on someone (most likely the assailant) financially, and also
experienced higher levels of abuse at all three time periods (the only exception was that
involvement with the assailant 15 months post-shelter did not significantly correlate with
economic dependence 3 months post-shelter). Fifteen months after leaving the shelter,
women were involved with their assailants also reported lower psychological well-being.

It is of importance to note the possible implications of the measure of economic
resources in these analyses. The concern with adequate variance is especially relevant in
these analyses where it could be hypothesized that the variance was different for
economically dependent women and economically independent women. That is, there
may be more variance in the level of economic resources among women who were
economically dependent, whereas most of the women who were economically
independent were on government assistance (a fixed, low level of economic resources).
However, it may speak to the strength of the relationships in these analyses that
significant relationships were still detected regardless of the restriction of range in the
sample.

The findings of the present study provide support for an economic dependence
analysis of woman battering, as suggested by feminist theory. The results lend empirical
support to the feminist theory on woman battering which posits that the inequitable
distribution of economic resources between men and women contributes to women’s
economic dependence on men, limiting women’s options and often forcing them to

remain in abusive relationships (Corcoran, Duncan, & Hill, 1984; Dobash & Dobash,

133

1992; Edwards, 1985; Hofeller, 1982; Okun, 1986). Illustrations of the sexist economic
and occupational structure in the United States are plentiful. Research has shown that
women who are employed in the same jobs as men get paid less, women have less access
to higher education, women receive less economic rettu‘n for their education than men,
and women are vastly under represented at the top levels of power and authority. At the
same time, women are more likely than men to be living in poverty (Wilson, 1987). All
of these factors serve to control women’s options, make women economically dependent
on men, and trap them in abusive relationships.
I . . . E l C S 1

Although some of the methodological limitations of prior studies were remedied
in this investigation, this study still has several shortcomings in common with past
research. Because of the difficulties in recruiting women with abusive partners to
participate in research, most of the research to date, including this study, has relied on
samples of women staying in battered women's shelters or seeking emergency/crisis
assistance at a community-based agency. This sampling is problematic because studies
using these convenient samples are restrictive in terms of the overall conclusions one can
draw about woman battering since women who use shelters or community-based agencies
are not representative of all women who are battered (McDonald, 1989). Women who
were abused by their partners and were able to leave their partners, at least temporarily,
and seek shelter or emergency assistance may be different from women who are battered
and do not seek those services. Therefore, the sample may be biased toward women who

have experienced more severe violence, women who might have experienced negative

134

responses from other sources of support, women who share certain demographic or
economic characteristics (e. g., poor, living in an urban area, etc.), and women who were
able to leave and undergo the potentially positive outcomes of leaving and receiving help
at a shelter. Generalizations of the findings of this study, therefore, are limited to women
who have resided at shelters. More studies need to be conducted that use samples who
were not recruited from a battered women’s shelter. It is important to challenge ourselves
to begin to examine the issues of intimate male violence in the general population even
though participant recruitment may be more difficult.

The limitations of using a shelter sample tie in with another limitation of the
current study; that is, the lack of variability in the economic resources of the sample. In
order to more fully examine the impact of economic resources and dependence, research
must be conducted with a sample represented by women with various levels of economic
resources. A sample more representative of all women who have experienced intimate
male violence would allow researchers to assess the degree to which women’s level of
economic resources were tied to their assailant as well as the change in level of economic
resources when women leave their assailant.

Another limitation of the current study is that data are based on self-reports.

Thus, response bias and other sources of shared method variance may have contributed to
the correlations or lack of relationships among the variables. Multiple methods of
measuring outcomes are needed. The correlational nature of the data are also of

concern, despite using a longitudinal, multivariate design, because it still leaves the

135

results open to validity threats from competing hypotheses. It also raises the issue of the
existence of potentially bidirectional relationships.

A further limitation of this study was that it measured economic dependence
differently than other studies. The current study assessed economic dependence on
“others” which usually included the original assailant but also may have included other
people (e.g., a new partner, family member, friend). All of the other studies that assessed
economic dependence focused specifically on economic dependence from the assailant.
Therefore, the findings of this study regarding economic dependence are not directly
comparable to the findings of past research.

Finally, the findings from this study may have been limited because the six month
time periods between interviews may have been too long for causal mechanisms to be
evident. For example, psychological well-being may be influenced by more immediate
causal processes. There is a need to share this information so that we can begin to
determine the effects and implications of various time lags used in documenting causal
effects in related areas of research.

WWII:

The present study adds to what we know about the process involved for women
after their exit from a shelter for women with abusive partners. More specifically, the
present study contributes to our understanding of the impact of economic resources and
economic dependence on battered women’s psychological well-being and decisions about
being involved with their assailant over the fifteen months after leaving a battered

women’s shelter. This research supports earlier work that highlights economic

136

dependence as a critical factor that limits women’s choices and keeps women trapped in
abusive relationships. The current study goes further to illustrate a general pattern of
interconnected relationships that are rather disturbing. That is, women who have higher
economic resources are more likely to be dependent and involved with their assailants
and more likely to be experiencing abuse and have lower psychological well-being.
W. The present study clearly indicates the need for abuse prevention
and intervention programs. Many women in this sample continued to experience abuse
15 months after having stayed at a shelter for women with abusive partners and this abuse
negatively impacted their psychological well-being. Interventions that bring the
community together to form a comprehensive response to batterers and their victims are
necessary in order to see a decrease in the level of intimate male violence against women.
Clearly, the present study also illustrates that economic dependence is a critical
variable that indirectly negatively impacts the lives of women who have stayed at a
domestic violence shelter. We need to design interventions at various levels that move
women toward economic independence. Advocacy strategies and other interventions that
move women and girls toward economic independency would be effective starts. Such
programs should not be targeted only toward women who had already been battered but
should be for all women and girls. Programs that allow women to get an education or
stay in school are critical as are efforts to provide women with access to good, decent
paying jobs with adequate child care and health care. At the same time, we need to do
early intervention with girls in school (to keep girls in school and support the education

of girls).

137

We also need to educate community-based service providers (including health
care providers, police, social service workers, judges, prosecutors) about the impact of
economic dependence so that they would better understand the complexities involved in
women’s lives. For example, sometimes leaving an abusive partner may mean choosing
a much lower standard of living (or even poverty) for women and their children which
would have important implications for community-based service providers (Sullivan et
al., 1994). It is important for service providers to understand all of the factors that are
relevant to women’s decision making-process and overall well-being in order to most
effectively serve women’s needs.

At a broader level, it is also crucial to simultaneously focus on examining and
influencing the economic systems that contribute to women’s dependence on men.

One of the most obvious examples of economic inequality between men and women is
the disparity in pay rates for equal work. Currently, women earn less money than men for
performing the same job responsibilities. As long as this is accepted, we are making a
statement that women are worth less than men and are placing women in the vulnerable,
and sometimes dangerous, situation of being economically dependent on men. Policy
changes implementing equal pay for equal work are essential for remedying this
inequality that leads to dependence.

Another example of how we can intervene at a broader level to influence the
economic systems that contribute to women’s dependence on men is through welfare
reform. Current proposed changes in welfare reform may have devastating implications

for battered women’s lives if they go through. Many women need to apply for

138

government assistance when fleeing an abusive partner. We do not know exactly what
portion of the welfare rolls is made up of survivors of domestic violence but
knowledgeable estimates suggest that it may be half to three-quarters of the female
welfare recipients that are also survivors of woman abuse. This suggests that the cycle of
dependency that needs to be stopped is not so much one of the dependence on
government assistance as one of dependence on abusive men.

Reforms that make welfare harder to get and worth less when you get it will only
serve to increase women’s dependence on men and trap them in abusive relationships.
Welfare benefits have been shrinking for the past two decades, to a level (a little less than
$400 a month per family, on average) that forces many recipients into financial reliance
on any man who can help pay for the groceries or clothes for the children. Residency
requirements effectively bar women from fleeing their abusers from one state to another
and work requirements will discourage women who cannot afford child care from
escaping her assailant.

Crafting welfare policy as if woman battering doesn’t exist will prevent many
women from leaving abusive relationships. The present study demonstrates that
economic dependence is a critical factor in keeping women trapped and we must work
against any policies that would increase women’s dependence rather than decrease
women’s dependence.

Rem. Empirical and theoretical evidence point to the importance of
examining the interconnection between level of economic resources and economic

dependence and physical and emotional battering of women in future research. The

139

design of this research study and the complex nature of the relationships suggest many
directions for intervention and firture research. Specifically, firture research should build
from the significant relationships from this study to suggest a model of the factors that
trap women in abusive relationships and negatively impact their psychological well-
being.

Future research should include further analysis of the relationship between
economic resources and abuse by collecting information from women who are more
representative (represent various economic levels) on the impact of economic resources,
dependence, and abuse on women’s lives. Need to do research with women of higher
economic resources because this study suggested that women who are economically
dependent and have a higher level of economic resources have the greatest odds of
staying with their assailants and experiencing firrther abuse. It would also be interesting
to understand the impact of economic dependence on all women’s decisions about staying
in or leaving a relationship regardless of whether they had been abused.

A great deal of work must also be done to refine measures to assess the constructs
of economic resources and economic dependence. The economic resources scale used in
this study was new and had previously untested reliability and validity. The lack of
findings with regard to economic resources may be due to the items used to assess the
construct. Historically, women’s level of economic resources has been linked with men’s
level of economic resources. Future studies should consider assessing a broader range of

items that may be more applicable to women’s lives and critical to determining women’s

140

level of economic resources. For example, child care, transportation, and health care may
be relevant dimensions for women’s level of economic resources.

In addition to assessing a broader range of items that may be more applicable to
women’s lives, it would be important to assess to what degree access to these resources
were tied to the assailant. This information in turn could be used to develop a related
economic indicator assessing relative change in economic resources. Future research
should also assess whether it is economic dependence on anyone or economic
dependence on the assailant that is important.

More research must be done to refine measures to assess the construct of abuse. It
may be important, in future research, to focus on the nature of the relationship of the
abuser. In the present study, abuse by anyone was assessed, either an original assailant or
a new relationship. Most past research has only looked at physical abuse by the original
assailant; however, abuse in a new relationship may be important to assess as well when
measuring outcomes like psychological well-being.

Future research may also benefit from examining psychological abuse from the
original assailant or a new relationship. In the present study, only women who were
currently in a relationship were asked about the frequency of psychological abuse they
experienced from whoever they were currently involved with. Women anecdotally
reported, however, that they continued to experience psychological abuse from their
original assailants even if they were no longer involved with them. Again, when
assessing outcomes such as psychological well-being it may be important to capture any

emotional abuse from anyone that may impact women’s psychological well-being.

141

It is also critical that the field of woman battering engage in dialogue about what
defines woman abuse. Previous work suggests that our definitions need to be more
comprehensive than just physical battering and that the effects of psychological and
sexual abuse are just as devastating to women. It is important that future research is
specific about the definition of abuse that is operationalized and that it be as
comprehensive as possible.

It would also be important for future research to investigate other ways to assess
the reliability of scales that tap “life events” sorts of constructs such as abuse since
internal consistency is inconsistent with the notion of independently occurring life! events
(e. g., perhaps comprehensiveness and relevance of life event inventories are more
important than internal consistency) (Cleary, 1981). The problem with finding a
satisfactory, comprehensive list of items is that scoring high on one does not necessarily
mean that an individual would score high on another even though the items seem to
measure the same construct (i.e., a woman who is slapped is not necessarily likely to also
have been raped or kicked). Clearly more work needs to be done to determine how to
best evaluate the reliability of woman abuse measures.

Ihepry. Finally, the present study has implications for the application of feminist
theory to economic dependence and woman battering. By further application of feminist
theory, a better understanding of abusive relationships will be attained. This study
provides empirical evidence for firrther examining the impact of multiple oppressions on
women’s lives since all are rooted in the patriarchal social structure. It provides evidence

that intimate male violence against women should be categorized with other forms of

142

violence under capitalism in the United States such as classism and racism. For example,
this study suggests the importance of research that examines the role of marriage in
keeping women economically dependent on their intimate male partners and of research
that investigates potential race differences in level of economic dependence and level of
economic resources. Feminist theory also provides support for future cross-cultural
research that investigates the impact of economic dependence on women with abusive

partners.

 

APPENDICES

APPENDIX A

Items That Compose the Economic Resources Scale

143

ITEMS THAT COMPOSE THE ECONOMIC RESOURCES SCALE

1. Would you mind telling me how much money you live on each month, including
ADC, food stamps, and rent money if you receive government aid? (Monthly
income is multiplied by 12 to calculate annual income)

(WRITE EXACT AMOUNT) ............................................................

2. Are you employed right now?

YES .......................................................................................................... 1

NO ............................................................................................................ 2

(not applicable) ......................................................................................... 8
3. Do you have regular access to a car?

YES .......................................................................................................... 1

NO ............................................................................................................ 2

(not applicable) ......................................................................................... 8

4. Is participant living above or below the federal poverty line? This is calculated by
assessing total annual income and the number of people in the family including
the woman herself, the assailant (if the income supports him), and any children
living with the family (U .8. Census, 1990). (*We did not ask participants this
question; instead we calculated it ourselves)

ABOVE FEDERAL POVERTY LINE ..................................................... 1
BELOW FEDERAL POVERTY LINE .................................................... 2

*All of these items were standardized (z-scored).

APPENDIX B

Revised Conflict-Tactics Scale

144

REVISED CONFLICT-TACTICS SCALE

Now I have a list of different types of violence that women have experienced from their
partners and ex-partners. I wonder if you could tell me how many times in the last 6
months has done any of the following things to you: (USE PINK CARD #3)
l = NEVER ‘

2 = ONCE A MONTH OR LESS

3 = 2 OR 3 TIMES A MONTH

4 = ONCE OR TWICE A WEEK

5=3 OR4TTMESAWEEK

6 = MORE THAN 4 TIMES A WEEK

8 = (not applicable); 9 = (refused to answer)

How often did he break your glasses or tear your clothing ....................................
Pushed or shoved you .............................................................................................
Grabbed you ............................................................................................................
Slapped you with an open hand ..............................................................................
Hit you with a fist ....................................................................................................
Kicked you ..............................................................................................................
Threw something at you .........................................................................................
Aside from throwing, how often did he hit you with an object ..............................
Tried to hit you with an object ................................................................................
Drove recklessly, so that you felt scared or endangered .........................................
Choked you .............................................................................................................
Burned you .............................................................................................................
Tied you up or physically restrained you in some way ..........................................
Beat you up .............................................................................................................
Forced sexual activity .............................................................................................
Threatened you with a gun or knife ........................................................................

Used a gun or knife on you .....................................................................................

APPENDIX C

Injuries Measure

145

INJURIES MEASURE

Now I am going to go through a list of injuries and ask you yes or no if you sustained
these injuries in the last six months. Did you sustain:

YES NO N/A

Cuts, scrapes, bruises ................................................... 1 .................... 2 ..................... 8
Soreness without bruises .............................................. 1 .................... 2 ..................... 8
Burns, including rug burns ............................................ l .................... 2 ..................... 8
Loose or broken teeth ................................................... 1 .................... 2 ..................... 8
Broken bones or fractures ............................................. 1 .................... 2 ..................... 8
Internal injuries ............................................................ 1 .................... 2 ..................... 8

Strains or sprains .......................................................... 1 .................... 2 ..................... 8

Dislocated joints ........................................................... 1 .................... 2 ..................... 8
Pregnancy complications or miscarriage ...................... 1 .................... 2 ..................... 8

(NO PREGNANCY = 8)

Knife or gunshot wound ................................................ 1 .................... 2 .................... 8
Permanent scarring ....................................................... 1 ..................... 2 .................... 8
Any other injmies I haven’t mentioned (Specify) ........ 1 .................... 2 ..................... 8

APPENDIX D

Threats Measure

146

THREATS MEASURE

How many times in the last impmhs has threatened you in any
way? By that I mean said or did things that made you feel scared or in danger, whether in
person, over the phone, through the mail, or through other people. (SHOW PINK CARD
#3)

 

l = NEVER

2 = ONCE A MONTH OR LESS

3 = 2 OR 3 TIMES A MONTH

4 = ONCE OR TWICE A WEEK

5 = THREE OR FOUR TIMES A WEEK

6 = MORE THAN FOUR TIMES A WEEK
8 = (not applicable)

9 = refused to answer

 

APPENDIX E

Index of Psychological Abuse (IPA)

147

INDEX OF PSYCHOLOGICAL ABUSE (IPA)

Now I’d like to go through a list of some things some men do to annoy or hurt their
partners and ex-partners. These are more the emotional things. Using this card (SHOW
YELLOW CARD #2) could you tell me, to the best of your recollection, how many times
in the last 6 months has done any of these things to annoy or hurt you?

 

1 = NEVER

2 = RARELY

3 = SOMETIMES

4 = OFTEN

8 = not applicable (i.e., no children, no pets)
9 = refused to answer

How often has he refused to talk to you ...................................................................
How often has he accused you of having or wanting other sexual relationships .....
Told you about other sexual relationships he wanted or was having in order

to hurt you ..........................................................................................................
Refused to do things with you that you wanted to do ..............................................
Forbid you to go out without him ............................................................................
Tried to control your money ....................................................................................
Tried to control your activities ................................................................................
Withheld approval, appreciation, or affection as punishment .................................
Lied to you or deliberately misled you ...................................................................
Made contradictory demands or requests of you ....................................................
Called you names ...................................................................................................
Tried to humiliate you ............................................................................................
Ignored or made light of your anger .......................................................................
Ignored or made light of your other feelings ..........................................................
Ridiculed or criticized you in public ......................................................................
Ridiculed or insulted your most valued beliefs .......................................................
Ridiculed or insulted your religion, race, heritage, or class ....................................
Ridiculed or insulted women as a group .................................................................
Criticized your strengths, or those parts of yourself which you are or once

were proud of .....................................................................................................
Criticized your intelligence .....................................................................................
Criticized your physical appearance and/or sexual attractiveness ..........................
Criticized your family or friends to you .................................................................
Harassed your family or friends in some way .........................................................

Discouraged your contact with family or friends ....................................................

148

Threatened to hurt your family or friends .............................................................
Broken or destroyed something important to you .................................................
Abused or threatened to abuse pets to hurt you (N O PETS =8) ............................

Punished or deprived the children when he was angry at you (N 0 KIDS = 8) .....
Threatened to take the children away from you (N 0 KIDS = 8) ...........................
Left you somewhere with no way to get home ......................................................

Threatened to end the relationship if you didn’t do what he wanted ......................
Tried to force you to leave your home ...................................................................

Threatened to commit suicide when he was angry at you ......................................

148

Threatened to hurt your family or friends .............................................................
Broken or destroyed something important to you .................................................
Abused or threatened to abuse pets to hurt you (N O PETS =8) ............................

Punished or deprived the children when he was angry at you (N 0 KIDS = 8) .....
Threatened to take the children away from you (NO KIDS = 8) ...........................
Left you somewhere with no way to get home ......................................................

Threatened to end the relationship if you didn’t do what he wanted ......................
Tried to force you to leave your home ...................................................................

Threatened to commit suicide when he was angry at you ......................................

APPENDIX F

Quality of Life Measure

149

W

In this section of the interview, I want to find out how you feel about various parts of
your life. Please tell me the feelings you have now -- taking into account what has
happened in the last 6 months, or since about , and what you expect in
the near future. (HAND PARTICIPANT GREEN RESPONSE CARD #1)

On this card are the answers that I’d like you to give me. I’ll be asking you about a list of
things. After I ask you each question, please tell me what phrase on this card gives the
best summary of how you feel: either “EXTREMELY PLEASED,” “PLEASED,”
“MOSTLY SATISFIED,” “EQUALLY DISSATISFIED AND SATISFIED,” “MOSTLY
DISSATISFIED,” “UNHAPPY,” OR “TERRIBLE,” depending on how you feel about
that part of your life.

If you feel that a question doesn’t apply to you, just tell me.

1 = EXTREMELY PLEASED

2 = PLEASED

3 = MOSTLY SATISFIED

4 = MIXED (MOSTLY SATISFIED AND DISSATISFIED)
5 = MOSTLY DISSATISFIED

6 = UNHAPPY

7 = TERRIBLE

8 = NOT APPLICABLE

9 = NO ANSWER (EXPLAIN WHY!)

I. First, a very general question. How do you feel about your life overall?...
2. In general, how do you feel about yourself? ................................................
3. How do you feel about your personal safety? ..............................................
4. How do you feel about the amount of fun and enjoyment you have? .........
5. How do you feel about the responsibilities you have for members of

your family? ................................................................................................
6. How do you feel about what you are accomplishing in your life? ..............
7. How do you feel about your independence or freedom —- that is, how

free you feel to live the kind of life you want? ...........................................
8. How do you feel about your emotional and psychological well-being? ......

9. How do you feel about the way you spend your spare time? ......................

APPENDIX G

Revised CES-D (Depression Measure)

150

REVISED CES-D (DEPRESSION MEASURE)

Below is a list of the ways you might have felt or behaved in the last week. We would
like to know how often you have felt any of these ways 111W. Please circle
the number that best describes how often you felt this way.

1 = RARELY OR NEVER (LESS THAN ONE DAY)

2 = SOME OR A LITTLE (1-2 DAYS)

3 = OCCASIONALLY (3-4 DAYS)

4 = MOST OR ALL THE TIME (5-7 DAYS)

W'

1. I was bothered by things that usually don't bother me ............ l ..... 2 ..... 3 ..... 4
2. I did not feel like eating; my appetite was poor ...................... 1 ..... 2 ..... 3 ..... 4
3. I felt that I could not shake off the blues even with

help from my family or friends ............................................... l ..... 2 ..... 3 ..... 4

4. I had trouble keeping my mind on what I was doing ............... 1 ..... 2 ..... 3 ..... 4

5. I felt depressed ......................................................................... l ..... 2 ..... 3 ..... 4

6. I felt that everything I did was an effort .................................. l ..... 2 ..... 3 ..... 4

7. I thought my life had been a failure ........................................ 1 ..... 2 ..... 3 ..... 4

8. I felt fearful ............................................................................. l ..... 2 ..... 3 ..... 4

9. My sleep was restless .............................................................. l ..... 2 ..... 3 ..... 4

10. I talked less than usual ........................................................... 1 ..... 2 ..... 3 ..... 4

1 1. I felt lonely ............................................................................ 1 ..... 2 ..... 3 ..... 4

12. People were unfriendly ......................................................... 1 ..... 2 ..... 3 ..... 4

13. I had crying spells ................................................................. 1 ..... 2 ..... 3 ..... 4

I4. I felt sad ................................................................................ 1 ..... 2 ..... 3 ..... 4

15. I felt that people dislike me ................................................... 1 ..... 2 ..... 3 ..... 4

16. I could not "get going" .......................................................... 1 ..... 2 ..... 3 ..... 4

APPENDIX H

Economic Dependence Item

151

ECONOMIC DEPENDENCE ITEM

How much of this income do you directly bring into the home, rather than another adult?
This includes ADC and any other government aid you may get. Would you say:

NONE OF THE INCOME ................................................................................... 1
1/2 OF THE INCOME OR LESS .......................................................................... 2
OVER '/2 BUT LESS THAN ALL OF THE INCOME ....................................... 3

ALL OF THE INCOME ...................................................................................... 4

APPENDIX I

Involvement With Assailant Item

152

INVOLVEMENT WITH ASSAILANT ITEM

 

Are you currently involved with (initial assailant)?
YES ...................................................................................................................... 1
NO ....................................................................................................................... 2

(Assailant deceased) ............................................................................................ 0

 

LIST OF REFERENCES

153

LIST OF REFERENCES

Aguirre, BE. (1985). Why do they return? Abused wives in shelters. W, 30, 350-354.

Allen, L. & Britt, D. W. (1984). Black women in American society. In A. Rickel, M. Gerrard &
I..,ISCOC(EdS) -_ 11. 1 11 1'1: 1 11111 1 W111 ' ',‘V‘ 11 __11 1.,
mwp. 61 -.79) New York: McGraw Hill.

American Psychiatric Association (1987). W
(3rd ed., rev.). Washington, DC: Author.

Andrews, F- & Withcy, S. (1976). W
M1111. NY: Plenum Press.

Astin, M.C., Lawrence, K.J., & Foy, D.W. (1993). Posttraumatic Stress Disorder among
battered women: Risk and resiliency factors. W5, 8, 17-28.

Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social
psychological research: Conceptual, strategic, and statistical considerations. 194mm

RemnalimaniSmialfimhnlm 51(6) 1173- 1182

Becerra, R. M. & Zambrana, R. E. (1985). Methodological approaches to research on Hispanics.
mmmmm 21(2) 42-49

Belle, D. (1990). Poverty and women's mental health. W, 45, 385-389.
Belle, D. Dill, D., Feld, E. ,,Greywolf E., Reese, M. F. ,& Steele, E. (1982). Mental health

problems and their treatment. In D. Belle (Ed)141e51n§m§5_flom§n_and_depmsipn.
Sage: Beverly Hills.

Birmcy,V Harkell G &Nix0n J (1981) WWW
WW. London: Women's Aid Federation.

Bowker, L.H. (1983). W. Lexington, MA: Lexington.
Browne, A. (1987). WWII. New York: Macmillan.

Browne, A. (1993). Violence against women by male partners: Prevalence, outcomes, and

policy implications. WW, 48(10), 1077-1087.

154

Browne, A. & Williams, KR. (1993). Gender, intimacy, and lethal violence: Trends from 1976 -
1987. Wear, 1, 78-98.

Bureau of Census (1990). WISE Washington, DC: US.
Department of Commerce.

Campbell, J .C. (1989). Women’s response to sexual abuse in intimate relationships. Health
Wetland, 11, 335-346.

Campbell, R. & Fischer, K. (under review). Defining domestic violence: Dimensions of abuse.

Campbell, R., Sullivan, C.M., & Davidson, W.S. (1995). Women who use domestic violence

shelters: Changes in depression over time. My, 12, 237-
255.

Carlson, B.E. (1977). Battered women and their assailants. $95M, 22(6), 455-460.

Carlson, B. E. (1984). Causes and maintenance of domestic violence: An ecological analysis.

Wm Dmmlm 569- 587.

Cascardi, M. & O’ Leary, K. D. (1992). Depressive symptomatology, self-esteem, and self-blame
in battered women. Wm, Z, 249- 259.

Cleary, P.J. (1981). Problems of internal consistency and scaling in life events schedules.

Wank 25(4), 309-320.

Corcoran, M., Duncan, G.J., & Hill, MS. (1984). The economic fortunes of women and

children: Lessons from the panel study of income dynamics. Sign5; 1191111121 pf 119mm in
Wicca, 1110), 232-248.

Davis, A..Y (1985).

' - - ' 11.. Kitchen
Table: Women of Color Press: NY.

 

Dill, B.T. (1983). Race, class, and gender: Prospects for an all-inclusive sisterhood. mm
males, 2(Spring), 131-150.

Dobash, R.E. & Dobash, RP. (1979). Mlenmgainstmmmmnmhmmamhx

New York: Free Press.

Dobash, R.E. & Dobash, RP. (1984). The nature and antecedents of violent events. Bdrm;
lmrmaLQfStimianng 251(3), 269-288-

155

Dobash, R.E. & Dobash, RP. (1992). W. New York:
Routledge, Chapman, and Hall.

Dobash, R. E., Dobash, R. P. ,& Cavanagh, K. (1985). The contact between battered women and

social and medical agencies. In J. Pahl, (Ed), WM
1“. 0 .1! ‘0 on 1-“. .1 ‘ 0H ‘ 0, .1‘ 0.0. ‘ (pp. 142- 165). London.

Routledge & Kegan Paul.

Dohrenwend, B. S. (1978). Social stress and community psychology. WW
Whom 6, 1-14

Dutton, D. G. (1985). An ecologically nested theory of male violence toward intimates.
WWW 8(4) 404-413

Dutton, D. (1987). The criminal justice response to wife assault. MW .11,
189-206.

Dutton,D.G.(1992). um 'H' 210.1%. 1' .1' 021' '0. 1011-.1'a1H!‘ 0 0.. ' ”'1 :1!
W. New York: Springer.

Edwards, SS. (1985). A socio-legal evaluation of gender ideologies in domestic assault and

Spousal homicides. WWW 19(1-4), 186-205.

Finkelhor, D. & Yllo, K. (1983). Rape mmarriage: A sociological view. InD. Finkelhor, R.J.

Gelles G T Hotaling, & M A Straus (EdS) WW
me. 119-130). Beverly Hills, CA: Sage Publications.

Follingstad, D. R., Rutledge, L. L, Berg, B. J., & Hause, E. S. (1990). The role of emotional abuse
in physically abusive relationships. Wag, 5, 107-120.

Follingstaad, D.R., Brennan, A.F., Hause, E.S., Polek, D.S., & Rutledge, L. (1991). Factors
moderating physical and psychological symptoms of battered women. Wfiamfly

2191mm, 6(1), 31-95-

Frieze, I.H. (1983). Investigating the causes and consequences of marital rape. Signs, 8,, 532-
552.

Frieze, I.H. & Browne, A. (1989). Violence in marriage. In L. Ohlin & M. Tonry (Eds), Family
21919119; (pp. 163-218). Chicago, IL: University of Chicago Press.

Frieze, I.H., Knoble, J ., Washbum, C., & Zomnir, G. (1980, March). W911.
Paper presented at the meeting of the Annual Research Conference of the Association for
Women in Psychology. Santa Monica, CA.

156

Gayford, J. (1975). Wife-battering: A preliminary study of 100 cases. MW,
1, 194-197.

Gelles, R.J. (1976). Abused wives: Why do they stay? mm, 18,
659-668.

Gelles, R. (1979). W. Beverly Hills, CA: Sage.

Gelles. R.J. & Harrop, J .W. (1989). Violence, battering, and psychological distress among
women. loumalnflntemersonalliolenee, 5(4), 400-420.

Gleason, WJ. (1993). Mental disorders in battered women: An empirical study. Viglgnc; and
W9 89 53'68'

Gondolf, E..W (1988). .. ‘ - - ' ,. ... ‘
mums; Lexington, MA: Lexington Books.

 

Goodman, L. A, Koss, M. P, Fitzgerald, L. F., Russo, N. F, & Keita, G. P. (1993). Male violence
against women: Current research and future directions. MW, 18,
1054-1058.

Goodman, L. A., Koss, M. P, & Russo, N. F. (1993a). Violence against women: Physical and

mental health effects. Part1: Research findings. WM, 2,
79- 89.

Greaves, L. ,Heapy, N., & Wylie, A. (1988). Advocacy services. Reassessing the profile and
needs of battered women CanadianluumaloflCommuninLMemaLHealth 1(2) 39- 51

Hanneke, C.R., Shields, N.M., & McCall, G.J. (1986). Assessing the prevalence of marital rape.
WWW. 1(3), 350-362-

Hilberman, E. (1980). Overview: The "wife-beater's wife" reconsidered. W
W. 131, 1336-1347.

Hilberman, E. & Munson, K. (1977-78). Sixty battered women. Yietimology An International
1911:1131, 2(3-4), 460-470.

Hilbert, J. & Hilbert, H. (1984). Battered women leaving shelter: Which way do they go? 111$.
WW 3(2) 292- 297

Hofeller, KR (1982). WW1: Palo Alto,
CA: R & E Research Associates, Inc.

157

Hoff, LA. (1990). Wm. New York: Routledge.

Hoffman, P. (1984). Psychological abuse of women by spouses and live-in lovers. mm
1112.122!- 3., 3747-

Horton, A.L., Simonidis, K.M., & Simonidis, LL. (1987). Legal remedies for spouse abuse:

Victim characteristics, expectations, and satisfaction. WW 2(3),
265-279.

Hotaling, G.T. & Sugarman, DB. (1986). An analysis of risk markers in husband to wife
violence: The current state of knowledge. WM 1(2), 101-124.

Houskamp, B.M. & Foy, D.W. (1991). The assessment of posttraumatic stress disorder in
battered women. WWW, 5(3), 367-375.

Jaffe, P., Wolfe, D., Wilson, 8., & Zak, L. (1986). Similarities in behavioral and social
maladjustment among child victims and witnesses to family violence. Am;[i9_an_.[9_umal_

W. 5.6.. 142-146.

Jones A & Sohoohtor S (1992) W
W. New York. Harper Collins.

Kemp, A., Rawlings, E.I., & Green, BL. (1991). Post-traumatic stress disorder (PTSD) in
battered women: A shelter sample. Was, 4(1), 137-148.

Kilpatrick, D.G., Saunders, B.E., Veronen, L.J., Best, C.L., & Von, J .M. (1987). Criminal
victimization: Lifetime prevalence, reporting to police, and psychological impact. £11m;
andDelmquenex, 33(4), 479-489.

King, L. S. (1981). Responding to spouse abuse. The mental health profession. mm
W Mame, 6-9.

King, M.M. & Bohom, J .S. (1993). Social construction of gender: Implications for domestic
violence. Paper presented at the Association for Women in Psychology, Atlanta, GA.

Koss, M. P. (1990). The women's mental health research agenda: Violence against women.
Ameneanfisxgholomst .45. 374-3 80

Kurz, D. (1993). Social science perspectives on wife abuse: Current debates and future

directions. In P.B. Bart & E.G. Moran (Eds), WW
footprints, pp. 252-269. CA: Sage.

Martin, D. (1976). W. San Francisco: Glide.

158

McDonald, P. L. (1989). Helping with the termination of an assaultive relationship. In B.

Pressman G Cameron & M Rothory (EdS) WWW
Wm. NJ: Lawrence Erlbaum Associates, pp. 93- 110.

Mitchell, R. & Hodson, C. (1983). Coping with domestic violence: Social support and

psychological health among battered women. WW
Wm 1.1. 629- 654

NiCarthy, G- (1986). mmmmmmmmmmmmmm Seattle.
Washington: The Seal Press.

Okun, L. (1986). W; Beverly Hills, CA: Sage.
Pagelow, M- (1981). WWW Beverly Hi118.CA:Sago-
Pagelow, MD. (1984). W. New York: Praeger.

— Pagelow, MD. (1992). Adult victims of domestic violence: Battered women. W
W. 1(1). 87-120.

Radloff, L. S. (1977). The CES- D Scale: A self report depression scale for research 1n the general
Impulation Amfiedlmhologicalemman 1(3) 385-401

, Randall, T. (1990). Domestic violence begets other problems of which physicians must be aware
to be effective WWW 2E6) 940 943- 944

Roy, M- (Ed) (1977). W New York:
Van Nostrand Reinhold.

RumotZ, M H & Sullivan C M (1994) W
W Michigan State University. E Lansing. MI.

Rumptz, M., Sullivan, C.M., Davidson 11, W.S., & Basta, J. (1991). Brief research report: An

ecological approach to tracking battered women over time. Wm, 5(3),
237-244.

Russell, D.E.H. (1990). W (2nd edition). New York: Macmillan.

Russo, N. F. (1990). Overview. Forging research priorities for women’s mental health.

Amencanfiwhclogist 45. 368- 373

159

Saltzman, L. E. (1990). Battering during pregnancy: A role for physicians. W, E,
45—48.

Sato, R. A. & Heiby, E. M (1992). Correlates of depressive symptoms among battered women.
Winner. 1(3) 229- 245

Schechter, S. (1982). Wow: 1. ,1 '
W. Boston, MA: South End Press.

 

Shepard, M. & Pence, E. (1988). The effect of battering on the employment status of women.
Aflilia. 3(2). 55-61-

Shields, N. M. & Hanneke, C. R. (1983). Battered wives' reactions to marital rape. In D.
Finkelhor, R.J. Gelles, G..T Hotaling, & M..A Straus,Ih;_da:k_§1d;_foamm;s_._ant_
mummwp. 131-148). Beverly Hills, CA: Sage Publications.

Sidel,R.(l986). 011‘. n! 1111‘: -.. ' '1' ' '10 '00 011'1_1;ii-‘1 411‘! .
New York: Viking.

Snyder, D. K. & Scheer, N. S. (1981). Predicting adjustment following brief residence at a shelter
for battered women WW 2. 559-566

Stark, E. & F11tcraft A. (1982). Medical therapy as repression: The case of the battered woman.
HealthandMedicine Sumerian 27- 32.

“ Stark, E. & Flitcraft, A. (1988). Violence among intimates: An epidemiological review. In V. B.
Van Hasselt, R. L. Morrison, A. S. Bellack, & M. Hersen (Eds.), W11);
13219119; (pp. 293- 317). New York: Plenum Press.

Stark, E., Flitcraft, A., Zuckennan, D., Grey, A., Robison, J. & Frazier, W. (1981). mm
. - (Domestic Violence
Monograph series No. 7). Rockville, MD: National Clearinghouse on Domestic
Violence.

 

Straus, M. A. (1977). A sociological perspective on the prevention and treatment of wifebeating.
InM..,R0y(Ed) --1 1 A." 1' 40 no 11 ._ 11.0. 0 Nu ° (pp.
194- 239). New York: Van Nostrand Reinhold.

Straus, M. A. (1979). Measuring intrafamily conflict and violence: The Conflict Tactics (CT)
Scales WWI): 41(1) 75- 88

160

Straus, M. & Gelles, R. (1986). Societal change and change 1n family violence from 1975 to

1985 as revealed by two national surveys. MW 48, 465-
479.

Straus, M.A. & Gelles, R.J. (1987). The cost of family violence. MW, 102(6),
638-641.

Straus, M. A. & Gelles, R. J. (1988). How violent are American Families? Estimates from the
National Family Violence Survey and other studies. In G. T. Hotaling, D. F 1nkelhor J. T.

Kirkpatrick & M A Straus (EdS).Eam11Lab.use_and_tseensequenees._I\_le_w_d1reenenm
mach. Newbury Park, CA: Sage Publications, Inc.

Straus MA &Gclles KL (1990) WWW
W New Brunswick NJ: Transaction

Straus, M..,A Gelles, R..,J & Steinmetz, J. K. (1980). W
W. Garden City, NJ: Anchor/Doubleday.

Strube, M.J. (1988). The decision to leave an abusive relationship: Empirical evidence and

theoretical issues. WWW, 153(2), 236-250.

Strube, M. & Barbour, L. (1983). The decision to leave an abusive relationship. Economic

dependence and psychological commitment. W131.
Member. 7 85 793

Strube, M. J. & Barbour, L. S. (1984). Factors related to the decision to leave an abusive

relationship W1! 46. 837- 844
Sullivan. CM (1989). WWW Unpublished

manuscript, East Lansing: Michigan State University.

Sullivan, CM. (1991). The provision of advocacy services to women leaving abusive partners:

An oKoloratory study. leumaleflntemersenaulielenee. 6(1). 41-54.

Sullivan, C. M. ,Basta, J, Tan, C., & Davidson 11, W. S. (1992). Brief Report: After the crisis. A
needs assessment of women leaving a domestic violence shelter. MW,
1(3), 267- 275.

Sullivan, C. M. ,,Campbell R., Angelique, H. E,by, K. K., & Davidson II, W. S. (1994). An
advocacy intervention program for women with abusive partners. Six-month follow-up.

AmerieanlmrmaleLQQnununimfisxehelegx 22(1),101-122

 

161

Sullivan, C.M. & Davidson, W.S. (1991). The provision of advocacy services to women leaving
abusive partners: The examination of short-term effects. MW
251911919211. 12(6). 953-960-

Sullivan, C.M., Parisian, J .A., & Davidson II, W.S. (1991). Index of psychological abuse:
Development of a measure. Paper presented at the 99th Annual Convention of The
American Psychological Association, San Francisco, CA.

Sullivan, C.M., Tan, C., Basta, J., Rumptz, M., & Davidson 11, W.S. (1992). An advocacy
intervention program for women with abusive partners: Initial evaluation. mm

Whey. 29(3). 309-332-

Sullivan, C.M. & Rumptz, M.H. (1994). The adjustment and needs of African-American women
who utilized a domestic violence shelter. W5, 2(3).

Walker, LE. (1978). Battered women and learned helplessness. mm, 2(3-4), 525-534.

Walker, LE. (1979). W. New York: Harper & Row.

Walker, L. E. (1983). Victimology and the psychological perspectives of battered women.
MW, 8(1-2). 82-104

Walker, L. (1984). Wm. New York: Springer.

Walker, L. (1985). Psychological impact of the criminalization of domestic violence on victims.

Mietimeleu:.An.Intematienal.leumd.19(1-4). 281M300
Williams, L. (1981). Violence against women. W, 12(1), 18-24.

Wilson, J .B. (1987). Women and poverty: A demographic overview. W, 12,
21-40.

Zambrana, R.E. (1987). A research agenda on issues affecting poor and minority women: A
model for understanding their health needs. W111, 12, 137-160.

 

HICHIGAN STATE UNIV. LIBRARIES
111|111111111111111111111111111111111
31293014101855