. An.
airy).
is?
, . 3:531
mar”?

.. a»...
gm 1%,
n. . .1

‘ ..

- I... .., mu: ,

.. t... x 50-1. .u.s¢4
5.4. a; zanzmfl. h“. "flu“... .21»%r£~£89 . 1,
air: «.2 3 . 25.... 3... a,

a :5 . "4‘3? ._

.r 2.5m“.

3.9 ~D.. ... :1.

.l. J :i‘.\t.~

.. 1.23;; s
1

 

.24 .54.
= r... t. : ix...
x ..K...l x 3:912: it) P. 3.».
gangs“: ‘Hu :1? :5?" gun
(«In an
.. .33....
.

. 5

‘ .34 .71.»?!

tn I‘IOV.
an I ..
2......Iuarfi:
a. x ‘
wggdfevdhx
30‘i’<)n‘-
l}? K.‘ .
i Ru. 0....
.1 s... .
.54.... I... a».
7 ‘ . :
Lwfiflvuhvfir. !

a

3.1...
1..

b

A
3:...
ta!!!

agar”.

I» 3 II
.. IL». 2:

1-1....

:V ”fr .H
e. .

 

a. 515.5

.31: v Es .

.V‘gnfizr 7
7g.§v.Lo!l:.( ..
10?.

is. 505.500...
Kiwi... ‘4‘}.713:
: .

, \ . .(wuhflha
f .155sz .A

V ‘13.. .. gar-:3: 1..
.55.; .

 

r llBRARY
3% Michigan State 9
University

 

 

This is to certify that the

dissertation entitled

Understanding Intimate Male Violence Against Women:
Contributions of Attachment Orientations,
Sexist Attitudes, and Personality Disorders

presented by
Anne M. Mauricio

has been accepted towards fulfillment
of the requirements for

 

 

  

 

Ph.D. degree in Counseling Psychology
Major professor V f
12/11/2001

Date

 

MS U i: an Affirmative Action/Equal Opportunity Institution 0-12771

 

 

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

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c/CIRCIDatoOuepeb-p. 15

 

 

[NE R
CONTRQ i"

 

 

Dc? lmt‘

UNDERSTANDING INTIMATE MALE VIOLENCE AGAINST WOMEN:
CONTRIBUTIONS OF ATTACHMENT ORIENTATIONS, SEXIST ATTITUDES,
AND PERSONALITY DISORDERS
By

Anne Marie Mauricio

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

DOCTOR OF PHILOSOPHY

Department of Counseling, Educational Psychology, and Special Education

2001

ABSTRACT

UNDERSTANDING INTIMATE MALE VIOLENCE AGAINST WOMEN:
CONTRIBUTIONS OF ATTACHMENT ORIENTATIONS, SEXIST ATTITUDES,
AND PERSONALITY DISORDERS

By
Anne Marie Mauricio

The purpose of this study was to test a model wherein psychological constructs-
adult attachment orientations, borderline personality disorder, and antisocial personality
disorder- and patriarchal attitudes, a construct central to sociopolitical theory, would be
independently and significantly predictive of male violence perpetrated toward female
partners. Participants (N = 239) were heterosexual men who were recruited from court-
mandated batterer intervention programs. It was further hypothesized that multiple
violence patterns distinct in frequency, generality, and severity of psychological, verbal,
and physical violence would be observed, and that adult attachment orientations,
personality disorders, and sexist attitudes toward women would be differentially
important in predicting these patterns. Hypotheses about interrelationships among
attachment orientations, personality disorders, sexist attitudes toward women, and
frequency, generality, and severity of psychological, verbal, and physical violence were
also tested. Specifically, antisocial batterers, borderline batterers, participants indicating
no personality disorder, and participants meeting criteria for both personality disorders
were compared on measures assessing severity and frequency of psychological, verbal,
and physical violence, generality of violence, attachment, and sexist attitudes toward

women.

Results indicated three distinct violence dimensions consistent with hypotheses
and attachment orientations, personality disorders, and sexist attitudes about women were
differentially important in predicting each of violence patterns. One dimension included
predominantly psychological and verbal violence and was partner specific. A second
dimension included minor violence and some severe physical violence, was partner-
specific, and reflected a pattern of violence that was escalating, initiated by the
perpetrator, and resulted in injury to the victim. The third dimension included
predominantly severe physical violence and was not partner-specific. Borderline
personality disorder scores, anxious attachment, and sexist attitudes about women
significantly predicted scores on the first violence dimension. Anxious and avoidant
attachment orientations, as well as sexist attitudes about women, were significant
predictors of the second violence dimension. Antisocial and borderline personality
disorder scores were significant predictors of the final violence dimension. Tests
exploring group differences between antisocial batterers, borderline batterers, participants
indicating no personality disorder, and participants meeting criteria for both personality
disorders suggested that men with greater pathology reported greater frequency and
severity of violence against female partners. Implications for practice and policy are

discussed.

DEDICATION

To all the victims of domestic violence past, present, and fiiture

iv

ACKNOWLEDGEMENTS

I wish to first thank my advisor, Dr. Frederick G. L0pez. I greatly appreciate your
encouragement, assistance, support, and commitment in helping me complete this project.
Also, thank you for introducing me to attachment theory, which has shaped my work as a
scientist-practitioner. I also thank my other committee members - Dr. Linda Forrest, Dr.
Betsy Becker, and Dr. Merry Morash, who each made valuable contributions towards this
project. I particularly appreciated the diversity of perspectives that each of you
contributed, which translated to a higher quality of work.

I am very grateful to the Blue Cross and Blue Shield of Michigan Foundation, the
American Psychological Association, and the Society for the Study of Social Issues for
their helping funding this project. Additionally, I am extremely grateful to Diane Mello-
Goldner for her financial donation to this project, which was amazingly thoughtful in
both the amount and in the spirit in which it was made.

I also want to thank my friends, who provided support and encouragement during
my stay at Michigan State University. A special thank you goes to Tracy Simko, who
always helped me maintain perspective of the “bigger picture” throughout graduate
school, and to Sally Theran, who helped me discover the confidence to complete this
project, when most needed.

Thank you to my family for their inspiration and the valuable lesson that all goals,
even completing a Ph.D., are within reach. I am forever grateful to my parents, Joseph
and Juliette, for their hard work and sacrifice, which made my educational endeavors
possible and worthwhile. Finally, a very special thanks to my partner, James Afremow.

Your eternal optimism provided me with the necessary diligence to complete this project.

TABLE OF CONTENTS

LIST OF TABLES ............................................................................ ix
LIST OF FIGURES ........................................................................... xi
CHAPTER I: INTRODUCTION ......................................................... 1
Problem Statement ................................................................... 7
CHAPTER 11: REVIEW OF THE LITERATURE ....................................... 8
Intimate Violence against Women: Prevalence and Severity ........................... 8
What is Battering? ................................................................... 9
Etiologic Explanations of Why Men Batter ............................................. 10
Family-level Explanations ......................................................... 11
SociOpolitical Explanations ......................................................... 12
Empirical support for sociopolitical explanations of battering ...... 13
Limitations of studies supporting sociopolitical explanations
of battering .................................................................... 15
Psychological Explanations ......................................................... 17
Attachment. .................................................................. 1 8
Attachment theory. ......................................................... 19
Internal working models ..................................................... 20
Attachment patterns in infancy ............................................ 20
Attachment patterns in adulthood ........................................ 21
Anxious and avoidant attachment orientations ........................ 22
Romantic love as an attachment process ................................. 23
Fear of abandonment and fear of intimacy .............................. 25
Attachment and battering: A review of empirical studies .............. 27
Limitations of research using attachment theory to explain
battering ....................................................................... 29
Personality disorders among batterers .................................... 32
Limitations of studies exploring personality disorders among
Batterers ...................................................................... 34
Moving Toward an Integrative Etiologic Perspective .................................. 35
Multidimensional Theories: The Importance of Batterer Typologies ......... 39
Empirical support for batterer typologies .............................. 41
Batterer typologies and attachment orientations ........................ 44
Batterer typologies and antisocial and borderline personality
disorders. ................................................................... 46
Antisocial and borderline personality disorders and attachment
orientations ................................................................... 47
Batterer typology and differential treatment effectiveness ............ 50
Methodological limitations common to past typology research ...... 50
Theoretical limitations common to past typology research ............ 52
The Problem Statement ...................................................................... 53
Hypotheses ............................................................................ 56

vi

CHAPTER III: METHODOLOGY ......................................................... 60

Participants ........................................................................... 60
Procedures ............................................................................ 62
Measures .............................................................................. 62
Demographic and Participant Background Questionnaire ............. 62
Basic Reading Inventory: Graded Passage-Form ...................... 62
Personality Diagnostic Questionnaire —Revised ........................ 63
Experiences in Close Relationships ....................................... 65
Attitude Toward Women Scale ........................................... 65
Conflict Tactics Scale ...................................................... 67
Psychological Maltreatment of Women Scale ........................... 70
Marlowe-Crowne Social Desirability Scale .............................. 70
Analyses .............................................................................. 71
Sample Size ........................................................................... 75
Design ................................................................................. 76
CHAPTER IV: RESULTS .................................................................. 77
Sample Descriptives ................................................................. 77
Correlational Analyses ............................................................... 81
Group Comparisons between Personality Disorders: Multivariate Analyses
Of Covariance ......................................................................... 88
Between-group differences on adult attachment orientations ......... 88
Between-group differences on violence-related variables ............. 90

Regression Analysis: Contributions of Personality Disorders, Adult
Attachment Orientations, and Sexism to Male Violence Enacted toward
Female Partners ...................................................................... 99
Factor Analysis of Violence Indexes ............................................. 101
Regression Analyses: Differential Contributions of Personality Disorders,
Adult Attachment Orientations, and Sexism to Three Patterns of Violence

......................................................................................... 104
Correlations among three violence patterns and key predictors ...... 105
Personality disorder group comparisons on factor scores ............. 106
Regressions analyses ....................................................... 107

CHAPTER V: DISCUSSION ............................................................... 116

Overview .............................................................................. l 16

Correlational analyses ............................................................... l 17

Group Comparisons between Personality Disorder Groups on Violence and

Attachment ............................................................................ 1 18

Adult Attachment, Personality Disorders, and Sexist Attitudes against

Women in Predicting Male Relationship Violence against Female

Partners .............................................................................. 121

vii

Violence Dimensions ................................................................ 123

Conclusions .................................................................................... 132
Implications for Practice and Policy ............................................. 138
Limitations ............................................................................ 143
Directions for Future Research .................................................... 147

APPENDICES ................................................................................. 151

Appendix A: Invitation to Participate in the Study ....................................... 152

Appendix B: Consent Form ................................................................... 153

Appendix C: Demographic Questionnaire .................................................. 154

Appendix D: Basic Reading Inventory ...................................................... 160

Appendix E: Personality Diagnostic Questionnaire — Revised:

Borderline Personality Disorder Scale ...................................................... 161

Appendix F: Personality Diagnostic Questionnaire — Revised:

Antisocial Personality Disorder Scale ...................................................... 162

Appendix G: Experiences in Close Relationships ........................................ 163

Appendix H: Attitudes Toward Women Scale ............................................ 165

Appendix 1: Five items added to the Attitudes Toward Women Scale ................. 167

Appendix J: Conflict Tactics Scale ......................................................... 168

Appendix K: Psychological Maltreatment against Women Scale ...................... 170

Appendix L: Social Desirability Scale ...................................................... 171

REFERENCES ................................................................................ 172

viii

 

Table l: Intertc
of Social Desrr;
Attachment .....

Table 2:.\lean 5
Groups on An).

Table 3:.\lean 5
Groups on Dept
Assumption...

Table 4:.\lean 5
Groups and So-‘
Sexism ..........

 

Table 5: Factor
Collapsing For:
Social Desirabl

Table 6: Signéf‘.
Violence Score
CTS Verbal A:-

l—ble 7 Summ
COmposite Va:-

Table 8: Rotate
‘ ‘Sessing Viol:

It“? 101 Inter-
1SOrderSCOR, I

mommy Du'

Table 113M€ar
Groups 0“ Thr.

 

3mm in Prod

Table13; ml]
SCXlsml'nP i

\-
.7:

We .

 

LIST OF TABLES

Table 1: Intercorrelations among Selected Demographic Variables and Measures
of Social Desirability, Sexism, Violence, Personality Disorders, and Adult
Attachment ............. p ........................................................................ 85

Table 2:Mean Scores and Standard Deviations for all Four Personality Disorder
Groups on Anxious and Avoidant Attachment Orientations ............................ 90

Table 3:Mean Scores and Standard Deviations for all Four Personality Disorder
Groups on Dependent Variables Meeting Homogeneity of Slopes MAN COVA
Assumption ..................................................................................... 92 ‘

Table 4:Mean Scores and Standard Deviations for all Four Personality Disorder
Groups and Social Desirability Groups on PTWS EmotionaWerbal Scores and
Sexism .......................................................................................... 94

Table 5: Factors and Cell sizes for New Sixteen-level Factor Created by
Collapsing Four-level Personality Disorder Group Variable and F our-level
Social Desirability Group Variable ......................................................... 96

Table 6: Significant Tukey post-hoc Comparisons on CTS Severe Physical
Violence Scores, Number of Participant Intimate Violent Relationships, and
CT S Verbal Aggression Score ............................................................... 98

Table 7: Summary of Hierarchical Regression Analysis for Variables Predicting
Composite Variable Violence ............................................................... 102

Table 8: Rotated Factor Matrix and Factor Loadings of Subscales and Item
Assessing Violence ........................................................................... 103

Table 9.Intercorrelations between Violence Patterns and Key Predictors ............. 105
Table 10: Intercorrelations between Violence Patterns and Antisocial Personality
Disorder Scores Controlling for Social Desirability Response Bias and Borderline

Personality Disorder Scores .................................................................. 107

Table lleean Scores and Standard Deviations for Four Personality Disorder
Groups on Three Violence Patterns ......................................................... 108

Table 12: Regression of Personality Disorders, Attachment Orientations, and
Sexism in Predicting Partner-Intimidation Violence ..................................... 109

Table 13: Regression of Personality Disorders, Attachment Orientations, and
Sexism in Predicting Partner — Specific Physical Violence .............................. 111

ix

Table
Physic

Table
Sensr

 

Table 14: Significant Tukey post-hoe Comparisons on Partner-specific
Physical Violence ........................................................................... 113

Table 15: Regression of Personality Disorders, Attachment Orientations, and
Sexism in Predicting the Third Pattern of Violence .................................... 115

 

Tigue 1:
Group an
Violence

Figure 2:
Group ar
Pbrsrcal

Firure 3:
Group ar
Paniclpa

Tlgre 4:
nPredle

LIST OF FIGURES

Figure 1: Interaction for Two-Way Analysis of Variance with Personality Disorder
Group and Social Desirability Group as Independent Variables and CTS Verbal
Violence Scores as Outcome ....................................................................... 95

Figure 2: Interaction for Two-Way Analysis of Variance with Personality Disorder
Group and Social Desirability Group as Independent Variables and CT S Severe
Physical Violence Scores as Outcome ........................................................... 95

Figure 3: Interaction for Two-Way Analysis of Variance with Personality Disorder
Group and Social Desirability Group as Independent Variables and Number of
Participant Violent Intimate Relationships as Outcome ...................................... 96

Figure 4: Interaction between Sexism and Adult Anxious Attachment Orientations
in Predicting Partner-specific Physical Violence ............................................. 1 12

xi

 

 

Studies €\i
{at it is a serious :
C1135 (see National
one in 3qu some
their life (Browne.
violence against fe
Frieze and Brown:
513% of all female

Although \
significant disage
recently. theoretic.
Panuers have beer
findings demonstr
CO'é'lllll'e. and em

madequate and Sui
l

 

INTRODUCTION

Studies exploring intimate violence against women by men in the US. indicate
that it is a serious and pervasive problem that permeates every race, religion, and social
class (see National Institute of Justice, 1998). Some studies predict that approximately
one in three women will report assault by their intimate male partner at some point in
their life (Browne, 1993; Koss, 1990; Straus & Gelles, 1986). The lethality of male
violence against female partners is equally alarming as its prevalence. For example,
Frieze and Browne (1989) found that an intimate male partner committed approximately
50% of all female murders.

Although widespread damage due to domestic violence is widely evident, there is
significant disagreement regarding the etiology of this phenomenon. In addition, until
recently, theoretical models explaining origins of male violence against intimate female
partners have been unidimensional (O’ Neil & Harway, 1997). However, research
findings demonstrating that batterers are a heterogeneous group with diverse behavioral,
cognitive, and emotional characteristics suggest that unidimensional theories are
inadequate and support arguments for etiologic explanations drawing from multiple
theories (see Holtzworth-Munroe & Stuart, 1994). More specifically, current research
findings suggest that there may be subtypes of male batterers with distinct patterns of
violence, and different theoretical assumptions may predict violence specific to different
subtypes (Hamberger, Lohr, Bonge, & Tolin, 1996). In general, typology research
supports a three-group taxonomy that categorizes batterers as family-only aggressors,
antisocial/generally violent, or border]ine/dysphon'c/emotionally volatile (Hamberger et

al., 1996; Holtzworth-Munroe, Meehan, Herron, & Rehman, 2000; see Holtzworth—

 

Munroe & 5mm
n’olence are imp

Sociopoli
frequently {APPIAC
Justice. 1998). 5
partners have bee
women is caused
(Arts, 1992; Bo g
Alternatively. the
differences such :
emotional probler
Starzomski, 1995
.As feminists rejec
sociopolitical con
Dobah & Dobas
contribution ofin
“Onabusice mEn {

These [\K'q

 

the contribution r

1998)

 

' Consequr
thEheUristjc Val.

untfi‘najn' Allhc.

 

Munroe & Stuart, 1994). Frequency and severity of psychological, verbal, and physical
violence are important variables distinguishing the three subgroups.

Sociopolitical and psychological theories are two conceptual lenses that have been
frequently applied to explain male violence against women (see National Institute of
Justice, 1998). Sociopolitical theories of male-perpetrated violence against female
partners have been mainly advanced by feminist scholars who argue that violence against
women is caused by patriarchal attitudes and institutions perpetuating male domination
(Avis, 1992; Bograd, 1988; Dobash & Dobash, 1979; Pagelow, 1992; Pence, 1989).
Alternatively, the premise of psychological theories is that individual-level psychological
differences such as personality disorders, early trauma, developmental arrest, or
emotional problems predict violent behavior (Dutton, 1995; Dutton, Ginkel, &
Starzomski, 1995; Dutton & Starzomski, 1993; Floumoy & Wilson, 1991; Lansky, 1993).
As feminists reject psychological explanations of violence for neglecting the patriarchal
sociopolitical context that has contributed to the victimization of women (Bograd, 1990;
Dobash & Dobash, 1979), feminist perspectives have been criticized for neglecting the
contribution of individual psychological variables that have differentiated batters from
nonabusive men (Hamberger & Hastings, 1991).

These two diverse understandings of why men batter have historically been
viewed as competing theories, and feminist scholars have been reluctant to accept an
etiologic model of battering that emphasizes individual variables, potentially devaluing
the contribution of sexist sociopolitical infrastructures to domestic violence (Heise,
1998). Consequently, few studies have simultaneously drawn from both paradigms, and
the heuristic value of models that integrate sociopolitical and psychological theories is

uncertain. Although each perspective appears to be necessary to explain battering, no one

 

 

 

theory is adequate. '
Studies exploring '4
ofnese diverse Lb:
should continue to
1994'. National In:
conceptual under?
. |
gender and pool er
Nonetheless. beca.
Plal' an important
hjr’muiesis to coll-
explanations of v,
TWO p5} c7:
em”
Amhdlfi partners a:
.l
acai . -
ill“
Dun
0 _
n, 1993), ar.
iglallolrlsh
1p dma:
(Duo
0 .
I] 8‘ Brow:

rel
Bled to the HCE;

CO
UCEPI and idenl

 

 

theory is adequately comprehensive to address all factors that influence domestic assault.
Studies exploring etiologic explanations of battering have demonstrated support for each
of these diverse theoretical conceptualizations, suggesting that each of these perspectives
should continue to be included in the discourse on the etiology of battering (Carden,
1994; National Institute of Justice, 1998). Feminist theory has significantly advanced
conceptual understandings regarding the abuse of women and an analysis concerned with
gender and power continues to be important in understanding relationship violence.
Nonetheless, because research demonstrates that individual-level psychological factors
play an important role in battering (Sugannan & Frankel, 1996), utilizing the feminist
hypothesis to collaborate and not compete with individual-level psychological
explanations of violence may be instrumental in decreasing recidivism among batterers.

Two psychological variables that appear to be predictive of male violence toward
female partners are adult attachment orientations and personality disorders. Insecure
adult attachment has been found to predict variables associated with acts of violence
against women (Dutton, Saunders, Starzomski, & Bartholomew, 1994). In addition, adult
insecure attachment has successfully distinguish abusers from non-abusers (Tweed &
Dutton, 1998), and abusers report more anger than non-abusers to scenes depicting
relationship dynamics (e. g., fear of abandonment) that may explain intimate violence
(Dutton & Browning, 1988). Research has also demonstrated that personality disorders
related to the need to control others (e.g., narcissistic and antisocial) and related to self-
concept and identity (e. g., borderline) are particularly prominent among batterers
(Hamberger & Hastings, 1988). For example, Hastings and Hamberger's (1988) found
that batterers demonstrated more borderline symptomatology than non-batterers, and

Dutton, Starzomski, and Ryan (1996) found that descriptions of batterer personalities and

bebatiors fit the Cl“
Meyer. and O'Lear.
consistently demo?
Pmnalilt' disord:
According
(Holtzworth-Munr
antisocial and bor:
betrs'een batterer t;-
t'an'able that coulc‘
important role in cl
three batterer sub:
follows. Antisocr
disorder, are arm.
the greatest frequo
characteristics cor

reD011 mOderatel
ph

19516211 Violence

of psychological ;
antisocial batterer

I .
“St Senst allllllc

Psychological. ‘ c

as».

 

 

behaviors fit the clinical classifications of borderline personality disorder. Murphy,
Meyer, and O'Leary (1993) found that violent men, as compared to nonviolent men,
consistently demonstrated higher scores on a measure of borderline and antisocial
personality disorder.

According to the proposed three-group batterer taxonomy mentioned earlier
(Holtzworth-Munroe & Stuart, 1994), adult attachment orientations and dimensions of
antisocial and borderline personality disorders play an important role in differentiating
between batterer types. Additionally, sexist attitudes toward women, an individual-level
variable that could be interpreted as a product of patriarchal institutions, also plays an
important role in distinguishing between the three batterer subtypes. Specifically, the
three batterer subgroups, each with their distinct pattern of violence, can be described as
follows. Antisocial batterers have characteristics congruent with antisocial personality
disorder, are avoidantly attached, report very sexist attitudes toward women, and report
the greatest frequency and severity of all types of violence. Borderline batterers exhibit
characteristics congruent with borderline personality disorder, are anxiously attached,
report moderately sexist attitudes toward women, and report less frequent and less severe
physical violence compared to antisocial batterers. However, the frequency and severity
of psychological and verbal abuse reported is comparable to or greater than that of the
antisocial batterer. Family-only aggressors evidence no personality disorder, report the
least sexist attitudes toward women, and report the least frequent and severe
psychological, verbal, and physical violence. As implied, this subgroup generally
restricts its violence to family members. Although it is expected that family-only

aggressors will report more insecure attachment compared to non-abusive men, to date

 

there are no find
orientations for

While ti
and personality
relationships be
predominantly
However, reset
constructs and
Holtzworth-M
parental reject;
abandonment l
partner (Hamb
extreme (leper
anxious attach
nmissistic an
needS- His m.
expefience lm
empathy, deta
batterer reSenj

BONE
disorders that
machinem fig

3 Con
Sequence

there are no findings in the literature about specific patterns of adult attachment
orientations for this subgroup.

While there is extant empirical support for relationships between batterer types
and personality disorders (Holtzworth-Munroe et al., 2000), hypotheses regarding
relationships between adult attachment orientations and batterer subgroups are
predominantly generated from theory (see Holtzworth-Munroe & Stuart, 1994).
However, research findings supporting specific relationships between attachment-related
constructs and batterer subtypes support these theory-generated hypotheses (see
Holtzworth-Munroe & Stuart). For example, the borderline batterer has a history of
parental rejection, reports very high levels of jealousy, and is hypersensitive to
abandonment by his female partner, and is characterized by extreme dependency on his
partner (Hamberger et al., 1996). The hypersensitivity to fear of abandonment and
extreme dependency evidenced by the dysphoric/borderline batterer is consistent with
anxious attachment (Bowlby, 1988). The antisocial batterer has been described as
narcissistic and self-centered, and he views his partner as an object that should meet his
needs. His most distinctive characteristic is his lack of empathy and inability to
experience intimacy with another person (Hamberger et al., 1996). The complete lack of
empathy, detached interpersonal style, and fear of engulfment typical of the antisocial
batterer resembles avoidant attachment (Bowlby, 1988).

Bowlby (1988) noted that anxious attachment produces a vulnerability to
disorders that exaggerate negative affect and distress in order to secure the attention of an
attachment figure. Fear of abandonment, clinging behaviors, and an uncertainty as to
whether the attachment figure will respond accompany overwhelming negative affect. As

a consequence, the anxiously attached individual will display frequent care-seeking

 

behatiors to elicit
attachment makes
disorders that den)
'mdin'duals expect
need for attachmer
borderline traits, in
personality traits.
Follouing
attachment theory
Liresley, Schroed
Specifically PFOVic
bOrderline PerSOn;
Maugham, 1994; F
avoidant attachmc
6‘ Meloy, 1992; C
BecauSe E
disorders and b an
dimensions of att.
relationships be” I
SllngOUps a5 Welj
diSorders Suppon I

a ' .
'ttiSOetal ballerc.

Speqfic to each c.

l'io ‘
lence distiller ’

 

behaviors to elicit the support of the attachment figure. On the contrary, avoidant
attachment makes one prone to behaviors that minimize distress and vulnerable to
disorders that deny the presence of negative affect. Because avoidantly attached
individuals expect to be rebuffed, they become compulsively self-reliant and deny any
need for attachment. Psychopathology characteristic of anxious attachment reflects
borderline traits, while psychopathology of avoidant attachment reflects antisocial
personality traits.

Following Bowlby, many researchers have underscored the importance of
attachment theory in understanding personality disorders (Brennan & Shaver, 1998;
Livesley, Schroeder, & Jackson, 1990; Sheldon & West, 1990). Some findings have
specifically provided evidence of relationships between anxious attachment and
borderline personality disorder (Fonagy et al., 1996; Patrick, Hobson, Castle, Howard, &
Maughan, 1994; Rosenstein & Horowitz, 1996; West, Keller, Links, & Patrick, 1993), or
avoidant attachment and antisocial personality disorder (Gacano & Meloy, 1991; Gacano
& Meloy, 1992; Gacano, Meloy, & Berg, 1992).

Because extant typologies demonstrate relationships between personality
disorders and batterer subtypes, informed hypotheses about relationships between
dimensions of attachment and batterer subtypes can be advanced. Empirically supported
relationships between antisocial and borderline personality disorders and batterer
subgroups as well as relationships between attachment orientations and these personality
disorders support expectations that borderline batterers are anxiously attached and that
antisocial batterers are avoidantly attached. Being able to predict attachment orientations
specific to each of the batterer subgroups could deepen etiologic understandings of

violence distinct to each of the subgroups.

 

W
One purpos
between attachmer
frequency. general
Specifically. group
participants meetir
personality disord:
Verbal. and physic
altitudes toward \V
Another pt
constructs-suCh as
antisocial PCYSOna‘
Sociopotrtrcai the
toward female par
distinct in fiE‘QUe:
consistent With pr

attachment Orient.

be differentiallv it

allillldfls “Ere Up.

|

 

 

Problem Statement

One purpose of this study was to test hypotheses about anticipated relationships
between attachment orientations, personality disorder, sexist attitudes toward women, and
frequency, generality, and severity of psychological, verbal, and physical violence.
Specifically, group differences between antisocial batterers, borderline batterers,
participants meeting criteria for both personality disorders, and participants indicating no
personality disorder on measures assessing severity and frequency of psychological,
verbal, and physical violence, generality of violence, attachment orientations, and sexist
attitudes toward women were explored.

Another purpose of this study was to test a model proposing that psychological
constructs-such as adult attachment orientations, borderline personality disorder, and
antisocial personality disorder- and patriarchal attitudes, a construct representative of
sociopolitical theory, are both independently predictive of male violence perpetrated
toward female partners. Additionally, it was expected that multiple violence patterns
distinct in frequency and severity of psychological, verbal, and physical violence, and
consistent with previous typology research, would be observed. it was expected that adult
attachment orientations, personality disorders, and sexist attitudes toward women would
be differentially important in predicting the three distinct violence patterns. Patriarchal
attitudes were operationalized as sexist attitudes toward women, which were

conceptualized as a product of patriarchal institutions.

 

lnt
Studies cX
that it is a serious
Starts, Gelles. ant
their husband or b
in National lnstitu
by their Spouse or
in three women is
life (Bronte. 199.
ofwomen who ha
34 it (Browne. IE
Demite the pret'a
ll131 these figures
Bureau of Justice
m0“ Ofien Occurs
fear Ofrt‘prisal bf.
The letha'
, |
pretalence. For t
“trimmed appro

L

 

lnSlllme OfJuSll ,»

1938 and 1991 ‘ll

 

REVIEW OF THE LITERATURE
Intimate Violence against Women: Prevalence and Severity

Studies exploring intimate violence against women by men in the US. indicate
that it is a serious and pervasive problem (see National Institute of Justice, 1998, 2000).
Straus, Gelles, and Steinmetz (1980) estimated that 1.8 million women are abused by
their husband or boyfi'iend each year. More recently, in a 1993 national survey (as cited
in National Institute of Justice, 1998), approximately 3.9 million women reported abuse
by their spouse or partner the previous year. Some studies predict that approximately one
in three women will report assault by their intimate male partner at some point in their
life (Browne, 1993; Koss, 1990; Straus & Gelles, 1986). Estimates regarding the percent
of women who have experienced abuse by a husband or boyfriend range fiom l4 % to
34 % (Browne, 1993; National Institute of Justice, 1998, 2000; Straus & Gelles, 1986).
Despite the prevalence of male violence against female intimate partners, it is believed
that these figures underestimate the true prevalence of such violence. According to the
Bureau of Justice Statistics (1994), "family violence is difficult to measure because it
most often occurs in private, and victims may be reluctant to report it because of shame or
fear of reprisal by the offender" (p. 6).

The lethality of male violence against female partners is equally alarming as its
prevalence. For example, Frieze and Browne (1989) found that an intimate male partner
committed approximately 50% of all female murders. Another study revealed that a
husband or boyfi'iend committed 28% of all violent crimes against women (Bachman,
1994), and the Center for Study and Prevention of Violence (as cited by the National
Institute of Justice, 1998) found that the perpetrator of 42% of female murders between

1988 and 1991 was the female's partner. More recently, the National Institute of Justice

 

(200th found that
intimates. Resear
consequences of r
23 9t and 35 970 of
\lOlfllCC (Adams.
pod-traumatic str<
Lewandowslci. l»
by findings that b
l€\€l (Bureau oil
& Gelles, 1986 )_
In the don
assaultive and C01

 

partners" to 3)-

t'erbal. sexual. an
to legitimize cont
The goal of the p
control within the
and their victims
manuscript will 1
Studies eXplorins

ab '.
usite behaviOr

(2000) found that between 26% and 49% of female homicides were committed by
intimates. Research also indicates that the negative physical and mental health
consequences of male violence toward female partners are serious. For example, between
22 % and 35 % of women who seek emergency room care do so as a result of male
violence (Adams, 1990). Battered women also report significantly more depressive and
post-traumatic stress disorder symptomatology than do other women (Campbell &
Lewandowski, 1997). The urgency to attend to domestic violence is further highlighted
by findings that battering permeates every race, religion, social class, and educational
level (Bureau of Justice Statistics, 1995; National Institute of Justice, 1998, 2000; Straus
& Gelles, 1986).

What is Battering?

In the domestic violence literature, battering is defined as "... a pattern of
assaultive and coercive behaviors, including physical, verbal, sexual, and psychological
attacks, as well as economic coercion, that adults or adolescents use against their intimate
partners" (p. 3). Assaultive and coercive behaviors are defined as physical, emotional,
verbal, sexual, and economic violence, as well as intimidation (i.e., using male privilege
to legitimize controlling behaviors), threats to the victim and her children, and isolation.
The goal of the perpetrator’s assaultive and coercive behaviors is to gain power and
control within the relationship. Because the majority of batterers are heterosexual men
and their victims are their female partners (National Institute of Justice, 1998; 2000), this
manuscript will focus on male perpetrated heterosexual violence. Moreover, among
studies exploring the etiology of domestic violence, research investigating the man's

abusive behavior has proven to be a fruitful line of inquiry with significant implications

 

for the prevention
regarding male Vll
ET

Although
regarding the etic-I

theoretical model:

 

 

have been unidtrr.

   
 

stewed as compet
demonstrating ll‘: "
and emotional ch.
SUpport argument
HOlW'Orth-hlun
there may be sub.
Iiii‘oretical assur
Lohr, Bonge, 8;
taxonomy mm C
tiolenL 0r bord
Holtzu'onh- M L
Stuart, 1994 )
important Van";
antisodalgene
by The bOrderh

x.
1016110: is lea

for the prevention and treatment of battering. This study builds on extant literature
regarding male violence toward female partners.
ETIOLOGIC EXPLANATIONS OF WHY MEN BATTER

Although widespread damage due to domestic violence is clearly evident, findings
regarding the etiology of this phenomenon are inconclusive. In addition, until recently,
theoretical models explaining origins of male violence against intimate female partners
have been unidimensional and diverse understandings of why men batter have been
viewed as competing theories (0’ Neil & Harway, 1997). However, research findings
demonstrating that batterers are a heterogeneous group with diverse behavioral, cognitive,
and emotional characteristics suggest that unidimensional theories are inadequate and
. support arguments for etiologic explanations drawing from multiple theories (see
Holtzworth-Munroe & Stuart, 1994). Specifically, current research findings suggest that
there may be subtypes of male batterers with distinct patterns of violence, and different
theoretical assumptions may predict violence specific to different subtypes (Hamberger,
Lohr, Bonge, & Tolin, 1996). In general, typology research supports a three-group
taxonomy that categorizes batterers as family-only aggressors, antisocial/ generally
violent, or border]ine/dysphoric/emotionally volatile (Hamberger et al. 1996;
Holtzworth-Munroe, Meehan, Herron, & Rehman 2000; see Holtzworth-Munroe &
Stuart, 1994). Frequency and severity of psychological, verbal, and physical violence are
important variables distinguishing the three subgroups. Violence enacted by the
antisocial/generally violent group is generally physical and severe, and violence enacted
by the borderline group is predominantly psychological. The family-only aggressor’s

violence is least severe and least frequent.

10

 

Theories 0
or psvchological e

to the interperson.

 

is adescription of
psychological and
perspectives are c
constructs central
specific to the thr

Theoretic.
as rooted within 1
ineffective confit
this PeT'SPCCtiVe,
resFOIISlble for ll
her 0““ "lCIimiz'
betrachr' Morec
such 35 "iOlence
interactiona] part

etiological expla

Theories of relationship violence have generally emphasized either sociopolitical
or psychological explanations of male violence against women, or attributed the violence
to the interpersonal family dynamics (see National Institute of Justice, 1998). Following
is a description of all three theoretical perspectives. However, only empirical support for
psychological and sociopolitical explanations of battering are reviewed, as only these two
perspectives are central to this study’s hypotheses. That is, only relationships between
constructs central to psychological and sociopolitical theories and violence patterns
specific to the three batterer subtypes identified above were explored.

_F_amilv-level Explanations

 

Theoretical perspectives attributing violence to family dynamics describe violence
as rooted within family interactional patterns that include poor communication and
ineffective conflict-resolution skills (Straus, Gelles, & Steinmetz, 1980). According to
this perspective, all persons involved in the conflict are contributors to the violence and
responsible for the escalation of the violence. The victim is perceived as contributing to
her own victimization by engaging in interactions that facilitate her partner’s abusive
behavior. Moreover, this theory’s premise is that the manifestation of problem behaviors
such as violence is indicative of a dysfimctional family system, and changing maladaptive
interactional patterns will result in the deterrence of violence (Giles-Sims, 1983). This
etiological explanation of violence is critiqued for neglecting the non-neutrality of power
dynamics present in the family system, and, in essence, colluding with the batterer
(Bograd, 1984). In addition, because family-based understandings of violence do not
distinguish between the perpetrator and victim of violence and advocate couples or family

counseling, this theory is also critiqued for failing to hold the batterer accountable for his

11

 

behaxior. This a:
promotes victim t
Sociomlitical Ex

Sociopoli'
been mainly ad\ 3.
caused by patri LII.
Begrad, 1988. 1v
‘0 [his PerSpectix
Contributes to an
condoning male
Perceived as C Or:
men, nor “'Omei:
IEldllOIlSlllps‘ be
Female Perpetra: I
means ofSClfidLI
identify with thol
ofPOwer is eSScI
an analysis suggl

behavior that Cr I

 

behavior. This approach is controversial because it further endangers the woman and
promotes victim blaming (Avis, 1992; Bograd, 1992).
Sociomlitical Explanations

Sociopolitical theories of male-perpetrated violence against female partners have
been mainly advanced by feminist scholars who argue that violence against women is
caused by patriarchal attitudes and institutions perpetuating male domination (Avis, 1992;
Bograd, 1988, 1990; Dobash & Dobash, 1979; Pagelow, 1992; Pence, 1989). According
to this perspective, the power differential between men and women in our society
contributes to and perpetuates the abuse of women. Furthermore, patriarchal attitudes
condoning male violence and maintaining the status quo of male domination are
perceived as contributing to violence against women. Feminists also assert that because
men, not women, are almost always the perpetrators of violence in heterosexual
relationships, battering must be understood as resulting from normal male socialization.
Female perpetrated violence in heterosexual or lesbian relationships is explained as a
means of self-defense or behavior motivated by feelings of vulnerability and a desire to
identify with the aggressor (Macchietto, 1992; Walker, 1996). As such, a gender analysis
of power is essential to the feminist interpretation of woman abuse (Bograd, 1990). Such
an analysis suggests that male dominance is legitimized and battering is perceived as a
behavior that creates and maintains an imbalance of power between the man and his
female partner. Feminists propose that aggression toward female partners has been
tolerated as an acceptable means of conflict resolution and, historically, legal sanctions
have not been imposed on such perpetrators of abuse. In addition, sexist political, social,
and economic infrastructures hinder women from leaving abusive relationships, further

acknowledging tolerance of woman abuse.

12

F eminis
explain why all
\iolence agains
individual psyc?
nonabusive met
attending to ind

The fem
aT‘Proaches in u
Treatment adhc:
iiolent. control}
eliminating the 1
treatment mOdzil
mdll’ldua], PS)'C‘
feminists argUe ‘

balm” in demi

 

 

Feminist perspectives of violence against women has been criticized for failing to
explain why all men do not batter and emphasizing sociopolitical contributions of male
violence against women. In addition, the theory does not consider the contribution of
individual psychological variables that have been found to differentiate batters from
nonabusive men (Hamberger & Hastings, 1991). Dutton (1994a) argued that without
attending to individual level predictors, it is impossible to predict which men will batter.

The feminist model is currently the dominant perspective guiding community
approaches in understanding and treating battering behavior (Tolrnan & Edleson, 1995).
Treatment adhering to this paradigm emphasizes holding the batterer accountable for his
violent, controlling behaviors. The profeminist model assumes that challenging and
eliminating the man's sexist attitudes are essential to ending the violence. Moreover, this
treatment modality suggests that an etiologic explanation of battering that includes
individual, psychological differences of the batterer exonerates male violence. Some
feminists argue that addressing the batterer's psychological motivations colludes with the
batterer in denying responsibility for his behavior (Adams & McCormick, 1982).
Research investigating the effectiveness of these “one-size-fits-all” batterer intervention
programs is inconsistent (National Institute of Justice, 1998; Tolrnan & Edleson, 1995).
One possible explanation for these inconsistent findings is that the heterogeneity of
batterers suggests the need for less generic programs. In fact, patriarchal attitudes differ
among the three batterer subtypes generally identified in batterer typology research, with
the antisocial/generally violent batterer indicating the most conservative and sexist
attitudes toward women (see Holtzworth-Munore & Stuart, 1994).

E_rnpirical support for sociopolitical explanations of battering. Smith's (1990)

telephone survey study is one empirical work that directly investigated relationships

13

 

between patriarch.

assault their fema‘:

 

familial patriarch}
patriarchal attitud
positively related
In anotherI
the ability of an a
of attitudes tows;
Their findings st...
nonviolent men,

distinguishing be

 

Hal-Yahi
and interpersona
The aulhor {Oling

allll‘udgs IO‘VKI'Q

who endorsed tr

tWM women

between patriarchal attitudes and battering. The author investigated whether men who
assault their female partners, compared with men who do not, adhere to values supporting
familial patriarchy and condoning violence against their spouses. Findings indicated that
patriarchal attitudes and male approval of violence toward female partners were
positively related with woman assault.

In another study, Eisikovits, Edleson, Guttrnan, and Sela-Amit (1995) examined
the ability of an aggregated set of cognitive and attitudinal measures, including a measure
of attitudes toward wife beating, to distinguish between violent and nonviolent men.
Their findings suggested that violent men are more tolerant of wife beating compared to
nonviolent men. In fact, attitudes toward wife beating were the strongest factor in
distinguishing between violent and nonviolent men in their sample.

Haj-Yahia (1997) tested a model that integrated sociopolitical, social-leaming,
and interpersonal understandings of battering within a sample of 434 engaged Arab men.
The author found relationships among beliefs about wife beating, sex role attitudes,
attitudes toward women, and maniage role expectations. Men in Haj-Yahia’s sample
who endorsed more masculine sex role stereotypes, negative and conservative attitudes
toward women, and patriarchal expectations of marriage were more likely to support
attitudes that justify wife beating.

Firm (1986), among a sample of 300 college undergraduates, found a positive
relationship between sex role stereotypes and attitudes supporting physical violence in a
marital relationship. In addition, Finn found that men supported more traditional sex role
attitudes than women and were more likely to endorse the use of force in marriage. Using
regression analysis, Finn’s findings also supported the sociopolitical hypothesis in that,

afier controlling for sex, traditional sex role attitudes were the strongest predictor of

14

 

positive attitude
might be more 1
male dominatio

Limitatt
investigating thr
common limitat
limitation threat
partner or the at
interviewer (Sm
threatening the r
marital violence
fat'orably and "l
violent b€ha\‘ior
threaten the \‘alit
collected retrosp
Put CVems QUt
repon abuse thn
“maimed beCa:

Anolher

relationship bet

 

 

positive attitudes toward using physical force in marriage. Finn speculated that men
might be more supportive of sex stereotypes, because gender role proscriptions promote
male domination and the subordination of women.

Limitations in studies supporting sociomlitical explanations of battering. Studies
investigating the relationship between patriarchal attitudes and battering have some
common limitations. First, data are generally collected via a self-report survey. This
limitation threatens the internal and external validity of studies that collect data from the
partner or the abuser. Because abused women often do not reveal their abuse to their
interviewer (Smith, 1994), a significant number of women may not report their abuse,
threatening the validity of those studies (e.g., Smith, 1990) that collect data regarding
marital violence fiom battered women. Because abusers’ tend to present themselves
favorably and “fake good” (Arias & Beach, 1987), particularly if they are aware that their
violent behavior is being studied and is wrong, they may provide dishonest responses and
threaten the validity of batterer studies. In addition, self-report data are generally
collected retrospectively, threatening the accuracy of the data due to problems recalling
past events. Questions regarding the representativeness of both men and partners who do
report abuse threaten the generalizabilityiof the study's findings. Generalizability is also
threatened because men generally included in these studies are court-referred.

Another limitation is that findings from empirical research investigating the
relationship between patriarchy and battering are inconsistent. Whereas the studies
reviewed above have demonstrated support for the sociopolitical hypothesis, other
reviews have not supported the hypothesis that gender role socialization and patriarchal
cultures contribute to battering. For example, Hotaling and Sugarrnan (1990) found that

male traditional sex role expectations was not a consistent risk marker for male violence

15

 

against intimate :'
~urtpon their h}?
attitudes towards
not possess a mi.

Quantita‘
adequate comps:
relationship berm.
that nonviolent :‘
comparison gm.
hfifteen violent

.A final r.
and Operational;
again“ Women .
Social. and ecor.
Patriarchal instii
Patriarchal amt-pl
and that any m:
llt‘éOry‘s emphz; I
operationalize T
attitudes) 0m), F
indies that bar
hat'e {mind that

T r
Elreg’1991)0.|

 

Patriarchal bet-1. l

against intimate female partners. In another study, Sugarman and Frankel (1996) failed to
support their hypothesis that violent husbands have significantly more conservative
attitudes towards women than do nonviolent husbands. Moreover, violent husbands did
not possess a more traditional masculine gender orientation than nonassaultive husbands.

Quantitative studies exploring the etiology of woman assault also rarely use
adequate comparison groups. This limits conclusions that can be made regarding the
relationship between patriarchal beliefs and battering. For example, we cannot be certain
that nonviolent men do not also hold patriarchal attitudes. When studies do include
comparison groups, demographic differences (i.e., education, socioeconomic status, etc.)
between violent and nonviolent men threaten the validity of findings.

A final noteworthy limitation is that patriarchy is a difficult construct to define
and operationalize. Most studies testing sociopolitical explanations of male violence
against women use individual patriarchal attitudes as a measure of a patriarchal political,
social, and economic structures, with the expectation that attitudes prominent within
patriarchal institutions will trickle down to individuals. Feminist theorists define a
patriarchal attitude as adhering to cultural prescriptions that women should be dominated
and that any means to that end is acceptable (Bograd, 1992). However, given feminist
theory’s emphasis on systemic contributions to domestic violence, studies that
operationalize patriarchy as an individual-level variable (i.e., individual patriarchal
attitudes) only provide a limited test of the feminist hypothesis of battering. The few
studies that have compared rates of violence among more and less patriarchal societies
have found that abuse rates of women are lower in more patriarchal societies (Sorenson &
Telles, 1991) or found insignificant correlations between structural inequality and

patriarchal beliefs (Yllo, 1983). However, these studies also demonstrated that battering

16

 

l

|
l
increases as md- I
couples and soci.
|
political power
proposed that \‘2
power loss outs:
power d)narnics
Psvcholoeical E

Accordzr
individual level
Russell, 1988).
Pally tEmma de
behatior. Regs,
and nonabusive
Psr‘sonality dis,
8: Starzomski, l
Moreover, rem
imTallel’SOna] Ci
Goldstein 3; Re:
These comm“
with his female

Hisrom

we“, . I

 

increases as traditional power differentials decrease. For example, among male-dominant
couples and societies, as women transfer increased earning potential and social and
political power to relational dynamics, they are at greater risk for violence. These authors
proposed that violence might be compensatory behavior used by the man to make-up for
power loss outside the home and to communicate to his female partner that change in
power dynamics outside the home will not alter relational power dynamics.
Psycholog’cal Explanations

According to psychological theories, the root of violence can be found in
individual level psychological variables (Bowlby, 1984; Dutton, 1995; Lansky, 1993;
Russell, 1988). For example, psychological differences such as personality disorders,
early trauma, developmental arrest, or emotional problems are assumed to predict violent
behavior. Research supporting this perspective has explored differences between abusive
and nonabusive men has found that abusive men exhibit more psychopathology and
personality disorders than nonabusive men (Dutton & Starzomski, 1993; Dutton, Ginkel,
& Starzomski, 1995; Floumoy & Wilson, 1991; Murphy, Lee-Meyer, & O' Leary, 1993).
Moreover, research has consistently demonstrated that batterers share some common
intmpersonal characteristics and interpersonal behaviors (Dutton & Strachan, 1987;
Goldstein & Rosenbaum, 1985; Holtzworth-Munroe, 1992; Strachan & Dutton, 1992).
These commonalities are perceived as risk markers that predispose a man to be violent
with his female partner (Hotaling & Sugarman, 1986).

Historically, feminists have rebuffed psychological explanations of violence as
advocating victim blaming and allowing the perpetrator to forego accepting responsibility
for the violence. Feminists also critiqued psychological explanations of battering for

viewing violent behavior independent of the violent individual's sociopolitical context

l7

 

and neglecting I
Bowl 1990; I
interest in P533

for these “plan

_a

(Dutton, 1995 ;

Munroe, 1992; .
6'; Zeanah, 198“:
in domestic Viol
advancing attaci
batter (see Natic
that abusiveness
For example, E1
critical. Elbou
may result in b;

the mother was

 

recently, Dutto:
abUSlVe behavi.
‘i'llh acts of vie;-
Emachmsnt has
1998), and abu

Browns 19$

 

 

and neglecting to highlight the societal nature of the problem (Avis, 1992; Bograd, 1984;
Bograd, 1990; Dobash & Dobash, 1979). However, there has been a recent surge of
interest in psychological explanations of battering, and accompanying empirical support
for these explanations highlights the necessity of continuing with this line of inquiry
(Dutton, 1995; Dutton & Strachan, 1987; Goldstein & Rosenbaum, 1985; Holtzworth-
Munroe, 1992; Murphy, Lee-Meyer, & O' Leary, 1994; Strachan & Dutton, 1992; Zeanah
& Zeanah, 1989).

Attachment. One individual psychological variable receiving increased attention
in domestic violence research is attachment, and many researchers and clinicians are
advancing attachment theory as a conceptual framework for understanding why men
batter (see National Institute of Justice, 1998). Research investigating battering has found
that abusiveness may have origins in early interactions between the child and caregivers.
For example, Elbow (1977) found that the abuser had parents who were rejecting and
critical. Elbow hypothesized that unresolved anger associated with this paternal rejection
may result in battering. Dutton (1988) found that early humiliation and verbal abuse by
the mother was correlated with later battering behavior in a group of males. More
recently, Dutton (1994b) found that maternal rejection was an important predictor of
abusive behavior. Insecure attachment has also been found to predict variables associated
with acts of violence against women (Dutton et al., 1994). In addition, insecure
attachment has successfully distinguish abusers fi'om non-abusers (Tweed & Dutton,
1998), and abusers report more anger than non-abusers to scenes depicting relationship
dynamics (e.g., fear of abandonment) that may explain intimate violence (Dutton &

Browning, 1988). Attachment has also been theorized as a variable important in

18

 

distinguishing an
exhibiting relatic

51139,;
for the purpose c
close enduring c
with others begi
needs and provii
needs, SeCure an
infant with the 1’
infant win use p
confide“I that t}
YWPOnsit-e 10 in
regarding the c;
experienCe the
much thishc.
Ringworm B‘,

BOW“);
emOthUa] Teac

and detaching:-

 

Despair W1 1] TC
be main feSled
inflinls whose
expressi0ns o“

beha‘iors are

 

 

distinguishing among three batterer subtypes identified earlier, with different subtypes
exhibiting relationship dynamics corresponding to specific attachment patterns.

Attachment theory. Attachment theory (Bowlby, 1978; 1982; 1980) posits that,
for the purpose of safety and survival, humans are innately driven to seek attachments or
close enduring emotional bonds with others. The human disposition to form attachments
with others begins at birth, when infants depend upon primary caregivers to meet basic
needs and provide security. Ifcaregivers are responsive and sensitive to an infant’s
needs, secure attachments between the infant and caregivers will develop, providing the
infant with the necessary security for exploring the environment. A securely attached
infant will use his/her caregiver as a base from which to explore the environment,
confident that the caregiver will be available when he/she returns. If caregivers are not
responsive to infant needs, the infant will not establish the same sense of security
regarding the caregiver’s proximity. Consequently, the insecure infant will not
experience the same freedom as the secure infant to explore his/her environment, as a
much of his/her energy will be devoted to maintaining proximity to the attachment figure
(Ainsworth, Blehar, Waters, & Wall, 1978).

Bowlby (1988) observed that a human infant goes through a series of predictable
emotional reactions when separated from its mother. The series includes protest, despair,
and detachment. During protest, the infant will cry and actively search for the caregiver.
Despair will result in the infant experiencing passivity and sadness, and detachment will
be manifested as an avoidance of the mother if she returns. During the protest phase,
infants whose caregivers do not respond to their cries may exhibit hostile, overt
expressions of anger in an effort to seek proximity to the caregiver. These hostile

behaviors are a manifestation of the infant’s anger at having an unmet attachment need.

19

 

CGFT

1.09%..

AMA

the:

311th
lE'Sp(

Va.‘ .

“hi.

Internal working models. During early attachment experiences, an infant will
build a cognitive representation or internalized set of beliefs about self and other, defined
by Bowlby (1973; 1988) as “internal working models.” The working model of self
influences the infant’s perceptions about his/her self worth, competence, and lovability,
whereas the working model of other is responsible for expectations about the availability,
and trustworthiness of primary caregivers. If primary caregivers are available and
responsive to a child’s needs, the child will internalize a model of self as lovable and
competent and a model of others as trustworthy and consistently available. Alternatively,
if the primary caregiver is rejecting, unavailable, and unresponsive to the child’s needs,
the child will internalize a model of self as unlovable and a model of others as rejecting,
untrustworthy, and unavailable. Bowlby suggested that these models of self and other
function as relatively stable and enduring expectations, which regulate one’s interpersonal
behaviors throughout life.

moment pattegs in @cv. Distinct infant attachment patterns were initially
identified by Ainsworth et a1. (1978) in the classical Strange Situation studies. Based on
infants’ responses to separation from and subsequent reunion with caretakers, Ainsworth
identified three principal patterns of infant attachment: secure, anxious-ambivalent, and
anxious-avoidant. Infants categorized as secure exhibited the least amount of distress in
response to the caregiver’s absence, were able to engage in exploratory behavior with or
without the caregiver present, and were most easily comforted by the caregiver’s return.
Ainsworth concluded that these infants had a model of self that was worthy and
competent and a model of others as reliable and responsive. Infants categorized as
anxious-ambivalent exhibited great distress in the absence of their caregiver and were

unable to engage in exploratory behavior in the caregiver’s absence. These infants were

20

 

not easily coming
reconnection. p.
perceives the car
unworthy. lnia:
caregiver upon 3
infants appearckf
behavior was it.
appears to haw-
“llh a COHESPC:
mlmlSllt'onh)..
Allhoiig

consistent Mm

 

rePlicaticmg hat

8010mm. 199I_i

 

 

combination of

f)
V

and exhibit “C
“ “Du-0%), ar
%

many reSEarch

not easily comforted upon reunion with the caregiver and appeared ambivalent about this
reconnection, perhaps fearful of a future absence. The anxious-ambivalent infant
perceives the caregiver as inconsistently available and intemalizes a view of self as
unworthy. Infants categorized as anxious-avoidant did not seek proximity to the
caregiver upon her return and rebuffed the caregiver’s attempts to reconnect. These
infants appeared less concerned with the mother’s proximity and their exploratory
behavior was not affected by the caregiver’s proximity. The anxious-avoidant infant
appears to have deactivated attachment needs and adopts a compulsive self-sufficiency,
with a corresponding model of self as unwanted and model of others as rejecting and
untrustworthy.

Although some replications of the classical Strange Situation have borne results
consistent with the original study (Egeland & Sroufe, 1981; Main, 1990), other
replications have identified an additional disorganized/disoriented pattern (Main &
Solomon, 1990). During the Strange Situation, disorganized infants evidence a
combination of avoidant and anxious responses. They simultaneously avoid proximity
and exhibit needy behavior upon the caregiver’s return. Disorganized infants have an
internalized model of other as threatening and rejecting and an internalized model of self
as unworthy and unlovable.

Attachment patterns in adulthood. Bartholomew and Horowitz (1991) are two of
many researchers who have extended attachment theory into adulthood. Although some
researchers investigating adult attachment have employed a 3-category model (e.g.,
secure, ambivalent, and avoidant), Bartholomew and Horowitz proposed a 4-category
model of adult attachment that dichotomizes a person's image of self and a person's image

of other into positively and negatively valenced categories. The internalization of a

21

 

positive or neg.
The intemaliza'.
which the indit
attachment st}
mepicd. lie.
self-worth and l

mistivorthy. T}

 

 

vith a positive

other-model mi
person expener
rejecting. This
feelings of ten;
rejection. The
With an expec
accompanied

and attachme]

detailed “31'

Am
C
OHCepmaliZ
folngroup [4'
d

positive or negative model of self depends upon the individual's experienced self—worth.
The internalization of a positive or negative model of other depends upon the degree to
which the individual expects the other to be accepting and trustworthy. The four adult
attachment styles yielding from Bartholomew and Horowitz's model are secure,
preoccupied, fearful, and dismissive. The secure person possesses a positive sense of
self-worth and has an internalized expectation that the other will be accepting and
trustworthy. The preoccupied person possesses a'sense of unworthiness accompanied
with a positive view of others. This combination of a negative self-model and a positive
other-model may lead the person to continuously seek the approval of others. The fearful
person experiences the self as unworthy and expects the other to be untrustworthy and
rejecting. This may result in a person who desires connection with others to alleviate
feelings of tenuous self-worth but avoids interacting with others due to expectations of
rejection. The dismissive person possesses a sense of self-worth that is positive coupled
with an expectation that others are untrustworthy and unreliable. A positive self-model
accompanied with a negative self-model may result in an individual who finds intimacy
and attachment threatening. As such, the dismissive person may be prone to manifest a
detached style of relating.

Anxious and avoidant attachment orientations. Attachment has also been

 

conceptualized as a two-dimensional structure. These dimensions are anxious and
avoidant attachment (Brennan, Clark, & Shaver, 1998). Recent research supported the
conceptualization of attachment as a two-dimensional structure rather than a three or
four-group taxonomy (Lopez & Brennan, 2000). Research with adults has also
demonstrated that four attachment categories can be conceptualized as two dimensions,

comparable to the two dimensions underlying infant attachment patterns (Bartholomew &

22

Horowitz, \90 1

 

discomfort wit?
close relations?"
ltars abandonr:
dimensions up,
Anxious attac't:
exP‘frienced as
pattern 0f exce
Partner‘s Unav t
lttsecur-‘lty an d
mtgame mOde
anxiety of “mi
measuremem fl
attained items
They {Ound 1“.

dimensions alii

COHCCPIUaliZm

attachment du-

 

attachIrlent be
Ham“ and Sh
Smiles based 0

aduhhmd an ;

 

Horowitz, 1991; Brennan & Shaver, 1995; Simpson, 1990). These two dimensions are
discomfort with closeness, or the extent to which an individual feels uncomfortable in
close relationships, and anxiety over abandonment, or the extent to which an individual
fears abandonment fi'om romantic partners. As with infant attachment categories, the two
dimensions underlying adult attachment styles reflect anxious and avoidant attachment.
Anxious attachment incorporates both fear of abandonment and anger at partners who are
experienced as insufficiently available and responsive. Avoidant attachment embodies a
pattern of excessive self-reliance developed as a defensive strategy to cope with the
partner’s unavailability. Avoidant attachment protects the individual from feelings of
insecurity and vulnerability. Whereas the avoidant dimension is closely related with a
negative model of others, a negative model of self is associated with abandomnent
anxiety or anxious attachment (Brennan et al., 1998). In an effort to develop a common
measurement for assessing adult romantic attachment styles, Brennan et al. factor-
analyzed items from virtually all extant self-report adult romantic attachment measures.
They found two independent factors corresponding to the anxious and avoidant
dimensions already proposed by other researchers.

Romantic love as an attachment process. Hazan and Shaver (1987)
conceptualized romantic relationships as an attachment process and proposed that the
attachment dynamics between partners in a love relationship might parallel the
attachment behaviors of individuals in early relationships with caregivers. Specifically,
Hazan and Shaver developed a self-report categorical measure of three adult attachment
styles based on the three attachment patterns found in infancy. Hazan and Shaver
demonstrated that the same three types of infancy attachment patterns existed in

adulthood and that the relative prevalence of the three styles in infancy and adulthood is

23

 

roughly equal. 1
of internal wotl
caregivers. Til;
experienced the
happiness, acce
insecure adults

love relationshi

 

 

 

accept and trust
their partner‘ 5 ‘
sight, desired g
range of emoti
Many I
romantic relat
activated by t‘
attachment f1
Participants
Style. They 1
lnlelCan] di
Significantly
Emattired par
p artiCiPants
CPIlsistent vi
ll‘leir Selfies

pannets. g

roughly equal. They conjectured that the continuity of patterns is due to the persistence
of internal working models of self and other that developed in relationship to early
caregivers. Their findings also indicated that adults with different attachment orientations
experienced their love relationships differently. Whereas secure adults indicated that
happiness, acceptance, and trust were common experiences in their love relationships,
insecure adults reported greater dissatisfaction and less happiness and acceptance in their
love relationships. Specifically, avoidant adults indicated that it was difficult for them to
accept and trust others, reported greater fear of intimacy, and were less likely to accept
their partner’s faults. Anxious-ambivalent adults were more likely to fall in love at first
sight, desired greater reciprocation from their intimate partners, and experienced a greater
range of emotional fluctuation in their relationships.

Many other researchers have subsequently applied attachment theory to adult
romantic relationships. For example, motivated by the premise that jealousy is a response
activated by the attachment system in reaction to a loss or a perceived loss of the
attachment figure, Sharpsteen and Kirkpatrick (1997) expected to find differences in
participants’ experiences of jealousy in romantic relationships based on their attachment
style. They found that attachment style was predictive of quantitative and qualitative
individual differences in jealousy experiences. Anxiously attached participants’ reported
significantly more jealousy than securely attached participants, and although securely
attached participants were able to express anger at a straying partner, anxiously attached
participants were less likely to confront their partners and express their anger. Moreover,
consistent with attachment theory, avoidantly attached participants worked to maintain
their self-esteem in a jealousy-provoking situation, further distancing them from their

partners. Simpson, Rholes, and Nelligan (1992) explored the moderating effect of

24

 

attachment sill
relationship is it
revealed that ur
support from p"
offer their stres
so. In a longitt
gender. a secur
interdependenc
With an insecu
Bartholomew r
51.1716 predicted
month period.
Constructive”
Were Unable rrx

 

their Partner‘s
aPP1}1ng anaC

attachment th

 

relationship it

 

 

F€\1rg
findings, it is I
to Unmet alta,

””aVallabirjt. i

attachment style on support seeking and support giving between partners in a romantic
relationship when confronted with an anxiety-provoking situation. Their findings
revealed that under anxiety-arousing circumstances secure women tend to seek out
support from partners, whereas avoidant women do not. Moreover, secure men tend to
offer their stressed partners more support, whereas avoidant men are less inclined to do
so. In a longitudinal study of 144 couples, Simpson (1990) found that, irrespective of
gender, a secure attachment style was associated with greater relationship
interdependence, commitment, trust, and satisfaction when compared with participants
with an insecure attachment style. Consistent with Simpson’s findings, Scharfe and
Bartholomew (1995) found that, after controlling for relationship satisfaction, attachment
style predicted how one would respond to a partner’s “destructive behavior” over an 8-
month period. Specifically, although secure individuals were able to respond
constructively to their partner’s destructive behavior, individuals with fearful attachment
were unable to do so. Couple participants with a secure attachment style were more able
to use effective problem solving and strategies such as open discussion in response to
their partner’s potentially destructive behaviors. Taken together, this sample of studies
applying attachment theory to the study of adult romantic relationships demonstrate that
attachment theory provides a valuable conceptual fiamework for understanding
relationship functioning.

Fear of abandonment and fear of im Given Hazan and Shaver’s seminal
findings, it is possible that individuals within a love relationship might continue to react
to unmet attachment needs in the same manner that they responded to unmet attachment
needs with their primary caregivers. As such, an infant who responded to the

unavailability or inconsistency of the primary caregiver with hostile, overt expressions of

25

 

anger might res.
separation or fe
attached persor.
capable of intir.
proximity sect.
WChment Sty.
car 89"65. Fe.
Validate their 1;
With lhe pl’OSp
are most like];
their partner
The art
PTCOCcupied a?
maintaining er
angry prOteSI it

MayseleSS. (1 S

 

I

(If1

rebumng the
Unsuccessful a
due to fEars 01

theorists who

 

inf;
dm'". "lav

d

512).

 

 

anger might result in an adult prone to react to his lover with anger when a threat of
separation or fear of abandonment exists (Dutton et al., 1994). Given that securely
attached persons have internal working models of others as dependable, trustworthy, and
capable of intimacy and models of self as worthy of love, it is probable that aggressive
proximity seeking behaviors are more likely to be used by persons with insecure
attachment styles who experienced untrustworthy, unpredictable, and/or rejecting primary
caregivers. Fearful and preoccupied individuals who seek approval from others to
validate their tenuous self-worth may be most likely to feel vulnerable when confronted
with the prospects of abandonment. As such, it is expected that these attachment patterns
are most likely to manifest angry protest behavior in an effort to maintain proximity to
their partner.

The attachment needs of the dismissive person may differ from those of the
preoccupied and fearful person. Dismissive individuals maintain a positive self-image by
maintaining emotional distance from others. The disrrrissive individual may demonstrate
angry protest behavior in response to his partner's attempts for increased intimacy.
Mayseless' (1991) theoretical exploration of adult attachment patterns and courtship
violence suggested that dismissive persons maintain power in a romantic relationship by
rebuffing the partner's attempts to become intimate. However, if the dismissive male is
unsuccessful at rebuffrng his partner's attempts, he may become hostile and aggressive
due to fears of increased intimacy. This is consistent with the work of attachment
theorists who suggest that a primary caregiver who intrudes or "... forces affection on the
infant..." may result in the infant's overt expressions of anger (Hazan & Shaver, 1987, p.

512)

26

 

  
   
   
  
  
  

exploring the u“
support to the l:
Dutton et al. (I
constellation 0'.
attachment “'er
attachment Walt
that the assault
previously ass(
maltreatment E
attachment “.8
Chronic child'r
"intimacy“,E
In anc
behavim “as
and higher p:
scored Siam
macinherit.
SigmfiCantly
and mama
Dullr
assaultive, \-
three videon

conflict in It

Attachment and battering: A review of empirical studies. Extant empirical work
exploring the utility of attachment theory in explaining the etiology of battering lends
support to the hypothesized relationship between attachment orientation and battering.
Dutton et al. (1994) investigated whether attachment style was correlated with a
constellation of variables previously associated with male batterers. Whereas secure
attachment was negatively and significantly correlated with these variables, anxious
attachment was positively and significantly correlated with them. Findings also indicated
that the assaultive group demonstrated higher scores on the constellation of variables
previously associated with battering and indicated higher levels of psychological
maltreatment against their female partners. The authors also found that fearful
attachment was a significant predictor of psychological abuse. Dutton et al. proposed that
chronic childhood fi'ustration of attachment needs may dispose adults to express extreme
"intimacy-anger" when adult attachment needs are unmet.

In another study, Dutton, Starzomski, and Ryan (1996) found that abusive
behavior was significantly correlated with early recollections of lower parental warmth
and higher parental rejection. In addition, men with a history of assaultive behavior
scored significantly higher than the nonassaultive control group on a measure of fearfirl
attachment. The authors also found that measures of paternal and maternal rejection were
significantly positively correlated with fearful attachment, whereas measures of paternal
and maternal warmth were significantly negatively correlated with fearful attachment.

Dutton and Browning (1988) compared three groups of men — physically
assaultive, verbally assaultive, and nonassualtive- on expressed anger toward a series of
three videotaped scenarios depicting conflict between a man and his female partner. The

conflict in the first scene depicted abandonment (e. g., verbalized expression of more

27

 

  
  
    

autonomy), ant
request for hin:
between a man
Physically ass;
reported the le.
anger to the ab

physically assq

 

was reflective
|
assaultive mal
abandonment.
abandOnmem
suggested [ha
nonviolent tr.
\Voil
lDClustD 0f
Stile, and gc
dimCUlty re
gain C01ltro
hetfiogexU;
COnraimng
Bo.

Who repel-1

Satisfactio

autonomy), and the conflict in the second scene depicted engulfment (e.g., verbalized
request for him to spend more time with her). The third scene included a conflict
between a man and his female partner but depicted neither abandonment nor engulfinent.
Physically assaultive men reported the most anger, whereas the nonassaultive group
reported the least anger. Moreover, the physically assaultive group reported the most
anger to the abandonment scenario as compared with the other two groups. The
physically assaultive group was also most likely to indicate that the abandonment scene
was reflective of their actual relationship. This study’s findings demonstrated that
assaultive males express more anger in response to relationship conflict and
abandonment. Given that attachment theory supports a relationship between fear of
abandonment and anxious attachment (Mayseless, 1991), this study’s findings also
suggested that physically assaultive men may have more attachment anxiety than do
nonviolent men.

Woike, Osier, and Candela (1996) investigated the relationship between the
inclusion of violent imagery in Thematic Apperception Test (TAT) stories, attachment
style, and gender. The authors proposed that insecure attachment styles might have
difficulty regulating affect associated with the loss of their lover and may use violence to
gain control over their lover and maintain her in close proximity. Findings indicated that
heterosexual men with preoccupied attachment styles were most likely to report stories
containing violent imagery with a male as the perpetrator and a female as the victim.

Bookwala and Zdaniuk (1998) compared the attachment patterns of individuals
who reported being in non—aggressive dating relationships to those who reported being in
mutually aggressive relationships. They found that, after controlling for relationship

satisfaction and length of relationship, participants involved in mutually aggressive

28

 

   
  

preoccupied a:
Ruben
predicted vio\e
also predicted
individual‘s a
Partner was ;
ahanQQmm
\tolence m.
emotiOna]
Q
abQVe fin.
Usefui in
DUtton e
g 011p. }
again“ 1
disr‘m C ti
earn; 0 1'1
florewoi

th

relationships reported more preoccupied and fearful-avoidant attachment styles and

acknowledged more interpersonal problems than did participants in non-aggressive

relationships. Controlling for interpersonal problems, only the relationship between
preoccupied attachment and aggression was significant.

Roberts and Noller’s (1998) findings indicated that abandonment anxiety
predicted violence perpetrated against one’s partner but male anxiety about abandonment
also predicted whether or not theywere victims of relationship violence. However, an
individual’s abandonment anxiety was only associated with the use of violence if the
partner was also uncomfortable with closeness. This significant interaction suggests that
abandonment anxiety may be exacerbated by a partner’s fear of intimacy and that
violence may be the tactic used by the anxious partner to maintain the other in close
emotional and/or physical proximity.

Limitations of research using attachment theory to enlain baflering. Although the

 

above findings are pivotal in supporting the hypothesis that attachment theory may be
useful in explaining battering behavior, each study has some significant limitations.
Dutton et al. (1994) used a control group that was significantly smaller than the treatment
group. Furthermore, the control group reported an average of 1.34 acts of violence
against their partner the previous year. This finding raises questions about the
distinctiveness of the two groups. Second, the self-report nature of the study warrants
caution in interpreting findings as men may have underreported abuse. A final
noteworthy limitation is the nonrepresentativeness of the treatment group. Because only
the participants' psychological maltreatment against their partners was measured, findings

cannot be generalized to men who physically abuse their partners.

29

 

  
 
   

Dutton.
via self-report
retrospective d
mandatory ass;
may have felt c
and dishonestl
thus threatenir

assaultive got

 

control group ,
Duttor
Small sample
rt’plicated w“
Phl’Sically as,
limitation is
dlmmics, SL
did not sqef
phl'SiCally a
may be lndi
cognitive, a
A150. altho‘
Sitcomc “C
Finally, all]

abusive be]

Dutton, Starzomski, and Ryan’s data (1996) were also retrospective and collected
via self-report being subject to self-report bias and other general limitations of
retrospective data. In addition, because study measures were presented as part of a
mandatory assessment, completion rates were close to 100%. Alternatively, participants
may have felt coerced into participation and consequently responded more haphazardly
and dishonestly than if they had perceived greater choice in whether or not to participate,
thus threatening the data’s validity. Next, although the control group matched the
assaultive group on demographic characteristics, there were only 45 participants in the
control group and 140 participants in the assaultive group.

Dutton and Browning (1988) noted that because of within-group variance and
small sample sizes the study’s statistical tests lacked power and that this study must be
replicated with a larger sample to provide stronger support for the hypothesis that
physically assaultive males react more strongly to abandonment scenarios. Another
limitation is that the use of videotaped scenarios as mechanisms for priming attachment
dynamics, such as fear of abandonment, need to be further validated. The authors also
did not screen out men who are violent outside their romantic relationship from the
physically assaultive group. Violence enacted toward persons in the general population
may be indicative of a broader learned pattern of violence rather than prompted by
cognitive, affective, or behavioral triggers that prompt relationship-specific violence.
Also, although the author’s hypotheses were specific to perpetrators of relationship-
specific violence, this study’s findings can only be generalized to generally violent men.
Finally, although Dutton and Browning demonstrated a relationship between physically

abusive behavior and fear of abandonment, they did not demonstrate a direct relationship

30

 

   
  

betweeninsecu
and abusive he.

Asign;

sizes threatenel
attachment sty
methodologic
included as a
An in
attachmems
Precludes in
members or
anafhment
Additional?
Collected‘ l
defensive l
in a relatig
their Previ
telarjODSh].
relationshi
resaid

ing-

aggTESSior.

Ro

Infereflces

between insecure attachment and fear of abandonment or between insecure attachment
and abusive behavior.

A significant limitation of Woike et al.'s study was the small number of
participants reporting violent imagery within each attachment category. The small cell
sizes threatened the validity of significant between group differences for the different
attachment styles. The manner in which violent imagery was categorized is a final
methodological issue worth noting. For example, it is unclear why fatal disease was
included as a category of violent imagery used to predict domestic violence.

An important limitation of Bookwala and Zdaniuk’s (1998) study is that data on
attachment style were only collected from one member of the relationship. This limitation
precludes investigation of the effect of a mismatch of attachment styles between couple
members on the occurrence of relationship aggression as well as whether self and partner
attachment styles were differentially related to the occurrence of aggression.
Additionally, no data regarding precipitants to the violence reported by participants were
collected, making it impossible to determine whether the violence may have been self-
defensive in response to experienced abuse. Moreover, 25% of the participants were not
in a relationship during the time of the study and reported on relationship aggression from
their previous relationship. Perhaps these participants elected to terminate the
relationship because of the violence and had no previous experience in a violent
relationship. Finally, the nature of their cross-sectional design precludes certainty that
regarding the causal relationship between insecure attachment and relationship
aggression.

Roberts and Noller’s (1998) findings are correlational in nature and prohibit

inferences regarding causal relationships between attachment patterns and relationship

31

 

violence. To r
patterns and re
attachment on
limited by the
interactions of
violence, the v
given the perp

Person

 

understanding
lmPOflant role

[fie-Meyer. &

   
  
   
 
 
 
 
 

(‘38-. narcissi
are Particular
Hastings and
53mptomatolt
descriptions
borderline pc-
vlolem mm,
a ”lemme O l.

Hasti
more borderI

ll
0P~batrerep

 

violence. To more accurately ascertain the nature of associations between attachment
patterns and relationship violence, longitudinal studies exploring effects of adult
attachment on couples’ use of violence are needed. Roberts and Noller’s findings are also
limited by the use of self-report measures, which are unable to fully assess the conflictual
interactions of violent couples. Moreover, particularly for non-reciprocal couple
violence, the validity of self-report data regarding couple violence is subject to question
given the perpetrator’s tendency to underreport acts of violence.

Bergman“! disorders gong batterers. Research has demonstrated that
understanding relationships between personality disorders and violence may play an
important role in improving current etiologic explanations of woman assault (Murphy,
Lee-Meyer, & O' Leary, 1993). Personality disorders related to the need to control others
(e.g., narcissistic and antisocial) and related to self-concept and identity (e. g., borderline)
are particularly prominent among batterers (Hamberger & Hastings, 1988). For example,
Hastings and Hamberger’s (1988) found that batterers demonstrated more borderline
symptomatology than non-batterers, and Dutton, Starzomski, and Ryan (1996) found that
descriptions of batterer personalities and behaviors fit the clinical classifications of
borderline personality disorder. Murphy, Lee-Meyer, and O'Leary (1993) found that
violent men, as compared to nonviolent men, consistently demonstrated higher scores on
a measure of borderline and antisocial personality disorder.

Hastings and Hamberger's (1988) findings indicated that batterers demonstrated
more borderline symptomatology and negativistic, passive-aggressive tendencies than
non-batterers did, but not more narcissistic and antisocial symptomatology. Although
there were no group differences on the antisocial factor, batterers did score lower on

measures of non-conformity than nonbatterers did.

32

 

Furthe

Ryan {1996).

behaviors fit I
personality orl
personality dis
issues and use
borderline pe;
individual‘s i
individual‘ 5 ‘
between abu
influential if
Foelsch. l9€
batterers an
‘0 fills Teal
Mu
hjsmr)’ of 5
Persona“,
they exPlc
lhat Viola-1

anllsgcia]

Further support for the borderline batterer is provided by Dutton, Starzomski, and
Ryan (1996). Their findings indicated that descriptions of batterer personalities and
behaviors fit the clinical classifications of borderline personality disorder and borderline
personality organization (BPO). Although BPO may be less severe than borderline
personality disorder, both diagnoses reflect personality problems characterized by identity
issues and use of primitive defenses such as splitting, which are often reflective of
borderline personality disorder. Splitting is a defensive mechanism characterized by the
individual’s inability to integrate good and bad images of significant persons in the
individual’s life (Kernberg, 1975). Dutton et al.’s findings also suggested a relationship
between abusiveness and early recollections of parental rejection, a factor believed to be
influential in the development of borderline personality (Sack, Sperling, Fagen, &
Foelsch, 1996). Specifically, it is argued that fear of abandonment is central to both
batterers and persons with borderline personality disorder. The batterer, however, reacts
to this real or perceived abandonment with violence.

Murphy, Lee-Meyer, and O'Leary (1993) found that batterers who experienced a
history of severe physical abuse exhibited more evidence of psychopathology, including
personality disorders, and were more violent toward their current partners. Moreover,
they explored differences between violent and nonviolent men on pathology. They found
that violent men consistently demonstrated higher scores on measures of borderline and
antisocial personality disorder.

Other research using the MMPI has also supported a subgroup of batterers with
antisocial tendencies. F loumy and Wilson (1991) found that a subgroup of batterers in
their sample had profiles suggesting antisocial characteristics as well as a tendency to

extemalize responsibility for behavior. In another study, 75% of a male sample

33

expressing concern for their violent behavior toward their female partner exhibited a
profile indicating antisocial personality and a tendency to externalize responsibility for
problem behaviors (Hale, Zimostrad, Duckworth, & Nicholas, 1988). Floumy and
Wilson called for research that matches different batterer subgroups with different
treatment approaches. Effectiveness of the different treatment approaches could be
compared across batterer subgroups.

Dutton (1995) also concluded that there are subgroups of batterers, and behavioral
and emotional characteristics of these subgroups reflected criteria for borderline and
antisocial personality disorders. The emotionally volatile subgroup enacts violence as a
response to fear of abandonment and intense anxiety about intimacy. Characteristics of
this subgroup resemble borderline personality disorder. The psychopathic batterer,
exhibiting characteristics reflective of antisocial personality disorder, is unempathic and
perceives woman as responsible for her abuse. These men fear engulfinent and use
violence to regulate intimacy. Dutton also conjectured that some batterers may not
evidence pathology indicative of a personality disorder. Dutton’s conclusions are
consistent with batterer typology research that categorizes batterer subtypes along three
dimensions — antisocial personality disorder, borderline personality disorder, and no
personality disorder. Further exploration of relationships between batterer subtypes and
personality disorders appears necessary to increase understandings of male violence
against intimate female partners.

Limitations of studies exploring personality disorders gmong bptterers. One
common limitation is that control samples, which may also include subgroups of men
exhibiting personality disorders, are rarely employed. When control samples are

included, often there are group differences on demographic variables between violent and

34

 

   
  
 

nonviolent in
higher rates 0
1988). Such
relationships
Anot‘;
are generally
Batterers pan|
personality p
sclf-respomr
retroSpectiv
develOpme,
Ont
Scales Usec
true began
amsodal
Conseq Ue
Clisorder )
menalf
may; avg

be 1W.e1

nonviolent men. Specifically, batterers report more unemployment, less education, and
higher rates of violence and divorce in their families of origin (Hastings & Hamberger,
1988). Such differences limit interpretations regarding research findings suggesting
relationships between personality disorders and batterers.

Another limitation associated with sampling is that participants in these studies
are generally recruited from treatment programs to which they have been court-referred.
Batterers participating in a treatment program of their own volition may exhibit different
personality profiles. Also, because of the nature of the referrals, modest sample sizes and
self-response bias plague these studies. Because reports of one's childhood history are
retrospective in nature, studies reporting relationships between abusiveness and
developmental histories common to persons with personality disorders are also limited.

One final noteworthy limitation is that, among batterers, average scores on clinical
scales used to identify psychopathology do not always exceed cut-off scores. This may be
true because batterers are a heterogeneous group. Although some may experience more
antisocial psychopathology, others may experience more borderline symptomatology.
Consequently, average scores on specific clinical sales (i.e., borderline personality
disorder) may not reflect the true level of pathology (i.e., associated with borderline
personality disorder) in the batterer subgroups. For example, whereas borderline batterers
may average scores higher than the cut-off score, the average score for all batterers may
be lower than the average score.

MOVING TOWARD AN INTEGRATIVE ETIOLOGIC PERSPECTIVE

Diverse understandings of why men batter have historically been viewed as
competing theories. Researchers have variously argued the importance of sociopolitical

and psychological variables. More recently, different tenets of the theoretical models

35

 

have been a;
assault. and
multidimen
Mauiro, 19
no one the

assault. l
suggestec
factors tl
addition
Perpetra

(it. no,

becaus
empire
Socio;
fOUnd
aclmc

equc

Prim.
atlag‘
5‘1ny
am,

adxa

have been appreciated as valuable contributions to etiologic explanations of woman
assault, and practitioners as well as researchers are beginning to embrace
multidimensional models (Ganley, 1989; Heise, 1998; Miller, 1994; Rosenbaurn &
Mauiro, 1989). Although each perspective appears to be necessary to explain battering,
no one theory is adequately comprehensive to address all factors that influence domestic
assault. In a review of studies exploring the etiology of battering, Dutton (1994a)
suggested that unidimensional explanations of battering are inadequate in identifying all
factors that contribute to male-perpetrated violence against their female partners. In
addition, he proposed that increased research investigating moderating variables from the
perpetrators developmental history"... are necessary to complete the predictive picture"
(p. 170).

Integration of psychological and sociopolitical theories has been particularly slow
because feminist scholars are reluctant to accept an etiologic model of battering that
emphasizes individual variables, potentially devaluing the contribution of sexist
sociopolitical infrastructures to domestic violence (Heise, 1998). Such resistance is well-
founded “. .. in the context of a discourse on violence that has been traditionally slow to
acknowledge the significance of gender inequalities and power differentials in the
etiology of violence directed at women” (Heise, p. 263).

As recent findings suggest that abuse has its origins in early interactions with the
primary caregiver, it seems appropriate to explore how individual differences in
attachment may influence battering. Theoretical arguments that attachment dynamics,
such as fear of abandonment and fear of intimacy, may explain battering (Dutton &
Browning, 1988; Mayseless, 1991) further underscore the utility of attachment theory in

advancing etiologic understandings of battering. Other findings that antisocial and

36

 

borderline
partners St
mullidimc
concemec
relations}
psycholo

fiMmm

SOCiOpo'
diverse
Pfimpec
C onseq
HOWex-
researg
paradig

1989;]

multip‘
Studies
into 511'
present
theOretj
machm
DiffgrEr

borderline personality disorders are prominent among men who abuse their female
partners suggests that such disorders may be important variables to include in
multidimensional etiologic models of woman assault. However, a feminist analysis
concerned with gender and power significantly contributes to the explanation of violent
relationship dynamics, and it should not be dismissed. Rather, feminist theory and
psychological theories should be concurrently examined to advance etiologic
explanations of battering.

As evidenced by this study’s literature review, there is support for both
sociopolitical and psychological conceptualizations of battering. However, because these
diverse understandings of violence have historically been viewed as competing
perspectives, few studies have simultaneously drawn from both paradigms.
Consequently, the heuristic value of models that draw upon theories is uncertain.
However, because unidimensional models have proven to be limited, practitioners and
researchers are beginning to embrace multidimensional models that draw upon these two
paradigms to understand the origins of male violence toward female partners (Ganley,
1989; Heise, 1998; Miller, 1994; Rosenbaum & Mauiro, 1989).

Following, the need for etiological explanations of battering that draw from
multiple theories will be further highlighted by research supporting batterer typologies.
Studies demonstrating that batterers are a heterogeneous group that may be categorized
into subtypes with distinct emotional, behavioral, and cognitive characteristics will be
presented. The different batterer subtypes will be reviewed, and empirical as well as
theoretical support for relationships among the different subtypes, anxious and avoidant
attachment orientations, and borderline and antisocial personality disorders is discussed.

Differences among batterer subgroups on sexist and/or negative attitudes toward women,

37

 

 

Take

the generall
only batten
attachme“
appears I(
With gem
The gem
a“fi3gci
Charam

genera

511ng

“Sun

a construct pertinent to sociopolitical explanations of battering, will also be presented.
Arguments that inconsistent findings regarding the efficacy of batterer intervention
programs may be explained by “one-size-fit-all” interventions that do not accommodate
for differences among batterers will be addressed. Research supporting the argument that
different types of intervention may be more or less effective depending upon the
batterer’s distinct characteristics will be presented (Saunders, 1996).

Taken together, empirical and theoretical support suggest three batterer subtypes-
the generally violent/antisocial batter, the berderline/dysphoric batterer, and the family
only batterer. Each subtype has distinct attachment dynamics and corresponding
attachment orientations and exhibits a unique violence pattern. Batterer subtype also
appears to be predictive of a batterer’s negative and/or sexist attitudes toward women,
with generally violent/antisocial batterers reporting more sexist and/or negative attitudes.
The generally violent antisocial batterer exhibits personality characteristics reflective of
antisocial personality disorder, whereas the borderline batterer exhibits personality
characteristics reflective of borderline personality disorder. The family only batterer
generally does not report pathology reflective of any personality disorder.

Given extant typology research, a reasonable conjecture is that different batterer
subtypes will enact different patterns of violence and that different theoretical
assumptions and corresponding constructs will predict the different violence patterns
associated with each of the subtypes. Support for this hypothesis implies that one-size-
fits all interventions modeled on single theories may need to be replaced with
interventions modeled on multiple theories. Another possibility is that multiple types of
interventions with different interventions modeled on different theories are necessary.

Although interventions motivated by psychological paradigms may prove efficacious in

38

 

 

treating boni-
decreasing r
Res
programs,
National
inconsistc
inten'em
Justice ).
$0610le
D01 all, ‘.
among '
intent}
01' no '11
altem a

be tail.

treating borderline batterers, interventions modeled on feminist theory may be effective in
decreasing recidivism for the antisocial batterer.
Multidimensional Theories and the Importance of Bfiatterer Typologies

Research frndings investigating the effectiveness of current batterer intervention
programs, most of which incorporate a gendered analysis of violence, are inconsistent
(National Institute of Justice, 1998; Tolrnan & Edleson, 1995). One explanation for these
inconsistent findings is that batterers are not a homogenous group of men, and current
interventions do not adequately address the needs of this population (National Institute of
Justice). The intervention paradigm predominantly employed (i.e., profeminist,
sociopolitical model) may be successful in decreasing violent behavior among some, but
not all, batterers. Because recidivism studies do not account for within group differences
among batterers, the success rate for the group of batterers impacted by existing
interventions may be underestimated. Moreover, if current programs are having minimal
or no impact for the other groups of batters, it may be a worthwhile endeavor to explore
alternative treatment paradigms. More specifically, alternative interventions may need to
be tailored to accommodate the needs of different batterer types.

The hypothesis that inconclusive treatment effects may be due to different types of
batterers is supported by recent empirical findings that confirm batterer typologies
(Gondolf, 1988; Hamberger et al., 1996; Holtzworth—Munroe & Stuart, 1994; Saunders,
1992). Holtzworth-Munroe and Stuart reviewed extant batterer typologies. They
concluded that batterer types could be differentiated based upon severity of violence,
generality of violence, and psychopathology or personality disorders. Using these three
dimensions, the authors proposed a batterer typology consisting of three subtypes: family

only, dysphoric/borderline, and violent/antisocial. Family only batterers were described

39

as the least likely to engage in psychological or sexual abuse and severe marital violence,
or have legal problems. Their violence is generally restricted to their family, and they
have little evidence of psychopathology. Dysphoric/borderline batterers were described
as engaging in moderate to severe violence, with their violence also likely to be directed
toward their family. They may demonstrate borderline or schizoidal personality
characteristics and alcohol or drug abuse. Holtzworth-Munroe and Stuart also described
this group as the most psychologically distressed, dysphoric, and emotionally volatile.
The antisocial cluster was described as engaging in moderate to severe violence and most
likely to be psychologically or sexually abusive. They are also most likely to have a
history of legal involvement and violence outside of their family. They also exhibit
personality characteristics congruent with antisocial personality disorder.

Next, Holtzworth-Munroe and Stuart addressed different developmental variables
that contributed to the violence perpetrated by the three different subtypes. Attitudes
toward women and attachment orientations were two of the variables included in their
developmental model. They found that the antisocial group consistently demonstrated the
most conservative attitudes toward women. The family only batterer had the most liberal
attitudes about women. Findings regarding the dysphoric/borderline batterer’s attitudes
were inconsistent. Holtzworth-Munroe and Stuart found no studies that directly
examined the link between attachment orientations and batterer type. However, they
were able to locate studies that explored relationships between batterer type and
constructs related to attachment. For example, dysphoric/ borderline batterers appear to
have an anxious attachment orientation. This group was the most pathologically
dependent on were most likely to view their partners as an extension of themselves.

Generally violent/antisocial batterers appear to have an avoidant attachment orientation.

4O

They view their partners as objects, demonstrate no or minimal empathy for their partner,
and appear to be self-absorbed in intimate relationships. Although family-only batterers
appear to be overly dependent on their wives, they appear to have the fewest attachment-
related problems. Holtzworth-Munroe and Stuart were also able to find studies
examining the relationship between attachment in battering in general. Findings from
these studies seem to indicate that a fear of rejection or abandonment, pathological
dependency, and jealousy- situations that can be interpreted as a threat to one's
relationship and elicit attachment behaviors- are risk markers for violence. Moreover,
violent men as compared to nonviolent men are more likely to have unresolved
attachment issues and be more dependent on their partners.

Holtzworth-Munroe and Stuart's developmental analysis was significant because it
integrated many of the available theories explaining violence toward female partners.
Specifically, the authors suggested that different theoretical assumptions might predict
violent behavior specific to the different subtypes. Although the authors argued a need to
replicate research demonstrating the existence of batter typologies, a more important line
of inquiry is to investigate risk markers specific to the different subtypes. Such inquiry
would elucidate the different factors contributing to the violence perpetrated by each
specific type and may inform us how to differentially intervene with each of the subtypes.

Empirical support for batterer typologig. Holtzworth-Munroe and Stuart's (1994)
review was a critical piece of work because it presented a typology, inclusive of all past
typologies, that could be empirically validated. Hamberger et al. (1996) tested the
theoretical model proposed by Holtzworth-Munroe and Stuart. They found three main
personality clusters- nonpathological, antisocial, and negativistic/dependent. The

nonpathological group showed no evidence of personality disorders, indicated the lowest

41

depression and lowest violence severity and frequency, and was the least prone to anger.
The antisocial group exhibited antisocial characteristics and was most prone to anger and
drug abuse. As expected, men in the antisocial group also indicated the most
extrafamilial violence and had the most extensive arrest records. Lastly, the borderline
group indicated borderline symptomatology and had significantly higher depression as
compared to Clusters 1 and 2. Overall, Hamberger et al.’s findings are consistent with
the theoretical model proposed by Holtzworth-Munroe and Stuart.

In another study, Holtzworth-Munroe et al. (2000) identified four clusters of men
who are violent towards their wives. Three of the clusters resembled Holtzworth-Munore
and Stuart’s (1994) proposed typology. There was a generally violent subgroup that
reported hostile attitudes toward women and attitudes tolerant of violence. This group
had a history of criminal behavior and reported the highest levels of general and marital
violence, as well as the highest levels of dismissing attachment. A borderline group also
emerged, and men in this group reported significantly more fear of abandonment, anxious
attachment, jealousy, and emotional dependency than did men in the other three
subgroups. This group also reported the highest scores on a measure of borderline
personality organization. The third subgroup was a family only batterer subgroup, and,
consistent with the proposed typology, this subgroup reported the lowest levels of marital
and general violence and indicated the least pathology. The fourth unexpected group was
labeled the “low-level antisocial” group. Although this group’s antisocial scores
resembled the two more extreme violent groups, their scores on other predictors of
violence fell in between the family only subgroup and the two more extreme violent
subgroups. Holtzworth-Munore et al. suggested that this group is comparable to the

family-only groups that have emerged in other typology studies that recruited men from

42

 

 

on most m

SUbgoup‘
“AI“lSOCi;
Significa
Consists]

dimensi

“‘0 Su‘.
PC1301]
bOrdeI
more

r31301-1
gr 011p
Howe
attach
need 1

mainn

intervention programs, whereas the family only subgroup in this study represent men who
were recruited from the community. The low-level antisocial group in this study was
more similar to family-only subgroups from other typology studies, whereas the family
only subgroup in this study is a new subgroup evidenced in this study due to the inclusion
of community as well as clinical samples. In fact, the family-only subgroup in this study
is less violent and less pathological than family-only subgroups found in other studies
Holtzworth-Munore et al. (2000) noted that the low level antisocial group’s scores
on most measures fell intermediate to the family only subgroup and the generally violent
subgroup, suggesting that these three subgroups follow along a continuum of
“antisociality.” Alternatively, the borderline group forms a distinct subgroup scoring
significantly higher than the other three subgroups on a set of measures theoretically
consistent with borderline personality. As such, the authors argued that an antisocial
dimension and a borderline dimension are necessary to conceptualize batterer typologies.
Tweed and Dutton’s (1998) study supported this conceptualization. They found
two subgroups, with one group scoring significantly higher on a measure of antisocial
personality and the second subgroup scoring significantly higher on a measure of
borderline personality. As expected, the antisocial group reported more frequent and
more severe violence than did the borderline group, whereas the borderline group
reported a more symptoms consistent with a history of trauma than did the antisocial
group. Also as expected, the borderline group reported the most fearful attachment.
However, unexpectedly, both the antisocial and borderline group reported preoccupied
attachment. Tweed and Dutton explained this finding as due to the antisocial batterer’s
need for dominance and control over his partner. Specifically, his preoccupation with

maintaining his attachment to his partner is motivated by his need to control his partner.

43

 

San:

   

mint. and

batterer Sul
\lolence It
on measm.
m0“?- “bel
activity ’u
activity '1
dE’pende
@010?
dQTlVed
dettm
35 iItsn
COHSls‘.

amlSm

adult -

Saunders (1992, 1993) found three distinct batterer types- family only, generally
violent, and emotionally volatile aggressors. F amily-only aggressors reported low levels
of childhood victimization, anger, jealousy, or depression. As expected, they were also
the group least likely to be violent with others outside their family. Generally violent
batterers were less likely to respond with anger to intimacy conflicts, had rigid role
expectations of women, and were most likely to engage in acts of violence outside their
home. They were also most likely to have an arrest record. The emotionally volatile
group included men that Were depressed, suicidal, angry, and fearful of abandonment.

Gottrnan et al. (1995) explored differences in physiological responses among two
batterer subtypes evidenced in their study. Type I batterers reported greater levels of
violence toward persons other than their partners and significantly more elevated scales
on measures of antisocial and sadistic aggression. Type I batterers were also described as
more “belligerent and contemptuous.” Type I batterers also evidenced lowered heart rate
activity in response to marital conflict. Type II batterers evidenced increased heart rate
activity in response to marital conflict and scored significantly higher on a measure of
dependency. Gottrnan et al.’s findings are the first to suggest a physiologically based
typology of male batterers, which is consistent with previously theorized and empirically
derived batterer typologies. Moreover, the authors proposed that Type I’s heart rate may
decrease in response to violent activity, because the Type I batterer perceives the violence
as instrumental in seeking purposeful control of his partner. This supposition is
consistent with Tweed and Dutton’s (1998) explanation for the unexpected finding that
antisocial batterers reported preoccupied attachment.

merer typologies and attachment orientations; Although there is evidence that

adult attachment orientations may be an important predictor of violence among batters

44

 

 

(seeNation.

 

I
Holtzworth-

maybe asse

  
    
 

that each be.
example, ll

avoidant an

generally vi
the dFSpho

Th4
rePMS hig
relationsh
hWCTSEns
to m an jpl
hlpfirsen
dBSthon
mathm (

i

that int}
altitudeE
narcissi
most di:

““11 an.

(see National Institute of Justice, 1998), there is minimal empirical support (see
Holtzworth-Munore et al., 2000) for the hypothesis that different attachment orientations
may be associated with different batterer types. However, recently, it has been purported
that each batterer type may be characterized by a distinct pattern of attachment. For
example, Holtzworth-Munroe et al. found that generally violent batterers reported
avoidant attachment, whereas borderline batterers reported anxious attachment. Research
findings of specific relationships between attachment-related constructs and batterer
subtypes support the hypothesis that avoidant attachment may be descriptive of the
generally violent antisocial batterer, whereas anxious attachment may be descriptive of
the dysphoric/borderline batterer (Holtzworth-Munroe & Stuart, 1994).

The dysphoric/borderline batterer has a history of parental neglect and rejection,
reports high levels of jealousy and marital dissatisfaction, is ambivalent about his
relationship, and is characterized by extreme dependency on his partner. He is
hypersensitive to abandonment and rejection by his female partner and may use violence
to manipulate his partner into maintaining close proximity (Hamberger et al., 1996). The
hypersensitivity to fear of abandonment and extreme dependency evidenced by the
dysphoric/borderline batterer is consistent with Bowlby's (1988) description of anxious
attachment.

The generally violent/antisocial batterer has a childhood history of victimization
that includes experiencing and witnessing extreme violence. He has extremely negative
attitudes toward women and supports the use of violence. He has been described as
narcissistic and self-centered, viewing his partner as an object to meet his needs. His
most distinctive characteristic is his lack of empathy and inability to experience intimacy

with another person (Hamberger et al., 1996). Although violence toward his partner may

45

 

just be ano:
may also be

1991). The

 

 

 

engulhnent

 

rejection at"

a
k

men. In ad‘
aggressors

this battere

 

COUCSpond
dissatisfaci

ii
hi the IE\'
dlfierenti.
Saunders
bOrdetlir
C-Onsigte.
0i the 3.
(listing
helm.“

Conela

batterE

1988).

just be another behavioral manifestation of his generally violent and criminal behavior, it
may also be precipitated by his partner's attempts for increased intimacy (Mayseless,
1991). The complete lack of empathy, detached interpersonal style, and fear of
engulfinent typical of the antisocial batterer resembles avoidant attachment.

The family-only batterer may experience some hypersensitivity to threats of
rejection and may exhibit some attachment dynamics consistent with anxiously attached
men. In addition, it is expected that, compared to non-violent men, family-only
aggressors will report more overall insecure attachment. However, it is anticipated that
this batterer subgroup will have the fewest attachment-related problems and no specific
corresponding attachment orientation. Generally, family-only batterers report less marital
dissatisfaction compared to the two other batterer subgroups.

B‘atterer typologies MtisociaL and borderline personality disordeg Evidenced
by the review of batterer taxonomies, personality disorders play an important role in
differentiating among batterer types. For example, personality characteristics of
Saunders’ (1992) emotionally volatile batterer were consistent with traits of the
borderline personality disorder, and characteristics of Saunders’ antisocial batterer were
consistent with traits of the antisocial personality disorder. Hamberger et al.'s (1996) use
of the MCMI-I to cluster batterers highlights the importance of personality disorders in
distinguishing among batterer types, and Saunders (1993) found that emotional traits and
behaviors specific to batterers classified as generally violent or emotionally volatile were
correlated in expected directions with expected MCMI-I scales.

Elbow's (1977) early typology based on clinical observations also identified
batterer subtypes consistent with personality disorders (see Hamberger & Hastings,

1988). For example, three of Elbow's categories overlapped with specific personality

46

 

disorders
his need.
batterer
charact:
Will res
ieel ec
Chara
perso
his p

mm

80'

11"!

ir.

disorders. The first type, the controller, uses violence to manipulate his partner to meet
his needs. He reacts with violence when the partner does not respond as expected. This
batterer type is described as antisocial. The second type, the incorporator, has
characteristics consistent with borderline personality. He fears his partner's abandonment
will result in a loss of self and requires "... a symbiotic relationship with the partner to
feel complete (p. 765)”, and he uses violence to maintain his partner in close proximity.
Characteristics of Elbow's third type, the approval seeker, are consistent with narcissistic
personality disorder. He needs to receive recognition and external validation and expects
his partner to provide constant validation. Her failure to do so is interpreted narcissistic
injury, and violence is his defense to such injury.

AntisocialJand borderline gersonalitv disorders angLagtachment orientations.
Bowlby (1988) argued that the quality of one's early attachments determine internal
representations of self and other. In turn, these mental representations influence
interpersonal functioning and impact later psychological health. Secure attachments early
in life contribute to healthy psychological development, whereas insecure attachments
make one vulnerable to psychopathology. Specifically, anxious attachment produces a
vulnerability to disorders that exaggerate negative affect and distress in an effort to get
the attention of an attachment figure. Fear of abandonment, clinging behaviors, and an
uncertainty as to whether the attachment figure will respond accompany overwhelming
negative affects. As a consequence, the anxiously attached individual will display
frequent care-seeking behaviors to elicit the support of the attachment figure. On the
contrary, avoidant attachment makes one prone to behaviors that minimize distress and
vulnerable to disorders that deny the presence of negative affect. Because avoidantly

attached individuals expect to be rebuffed, they become compulsively self-reliant and

47

 

 

 

deny intim;

 

personality
l996; Sabl
antisocial l
Meloy, 8;

Sa
bQTderlim
Clients. 5
“COndirio
pJOXimitj
Claim m;
elem (i.<
Hem.“
ai'mdam
digihkm
Home“

[Won

deny intimacy needs. Psychopathology of anxious attachment reflects borderline traits,
whereas psychopathology of avoidant attachment reflects antisocial personality traits. As
such, attachment theory provides a rich conceptual framework for understanding
borderline and antisocial personality disorders.

Following Bowlby, many researchers have underscored the importance of
attachment in understanding personality disorders (Brennan & Shaver, 1998; Livesley,
Schroeder, & Jackson, 1990; Sheldon & West; 1990). Some findings have specifically
provided evidence of relationships between anxious attachment and borderline
personality disorder (F onagy et al., 1996; Patrick et al., 1994; Rosenstein & Horowitz,
1996; Sable, 1997; West, Keller, Links, & Patrick; 1993), or avoidant attachment and
antisocial personality disorder (Gacano & Meloy, 1991; Gacano & Meloy, 1992; Gacano,
Meloy, & Berg, 1992).

Sable (1997) used attachment theory to expand understanding of the etiology of
borderline personality disorder, as well as a framework for clinical work with borderline
clients. She proposed that borderline personality disorder should be conceptualized as a
“condition of profound insecure attachment, with extreme vacillations between desire for
proximity. . .and a dread of engagement” (p. 171). She highlighted Bowlby’s (1979)
claim that borderline personality disorder may result in response to trauma or a disruptive
event (i.e., bereavement) that makes the primary caregiver unavailable. Rosenstein and
Horowitz's (1996) findings suggested that antisocial behavior was associated with
avoidant dimensions of attachment, whereas affective disorders such as depression and
dysthymia were associated with anxious dimensions of attachment. Rosenstein and
Horowitz's findings are consistent with batterer typology research indicating that

borderline batterers are more likely to report depression (Holtzworth-Munroe & Stuart's,

48

 

W94). A ll-
to personal I
mwwb
between he

and behax'i

also hallm'.

 

borderline
than Chane
borderline
Because f

number 0

aVoidam
between
some su
mOIher-
Wong l
"afieerj
appear!
SOme E
More r
Damn
Mario

We Q

1994). A limitation of Rosenstein and Horowitz's study is that personality traits specific
to personality disorders assessed by the MCMI—I are not mutually exclusive, and one trait
may apply to more than one disorder. West et al. (1993) empirically confirmed links
between borderline personality disorder and anxious attachment. Attachment dynamics
and behavior patterns that were characteristic of borderline outpatients in their sample are
also hallmarks of anxious attachment. Fonagy et al. (1996) found that the proportion of
borderline patients with anxious attachment among their sample was significantly greater
than chance. Possible comorbidity of diagnoses and inadequate numbers of males with
borderline personality disorders were limitations common to West et al. and Fonagy et al.
Because findings from this line of inquiry are generalized to male batterers, the small
number of borderline men is especially problematic.

Empirical support for the relationship between antisocial personality disorder and
avoidant attachment is not as extensive as the extant empirical support for the relationship
between borderline personality disorder and anxious attachment. Nonetheless, there is
some support for this relationship. In an early study, Bowlby (1944) studied the "broken
mother-child relation" among a group of 44 juvenile thieves. Bowlby observed that these
young thieves were emotionally withdrawn and isolated, and he described them as
"affectionless psychopaths." The relational style of Bowlby’s delinquent juveniles
appears congruent with avoidant attachment. Thus, Bowlby's early observations provide
some evidence of the association between antisocial behavior and avoidant attachment.
More recently, Gacano et al. (1992) found that participants diagnosed with antisocial
personality disorder demonstrated a diminished capacity to form attachments and
relationships with others and an increased likelihood to devalue others. The relational

Style of their antisocial participants parallels avoidant attachment.

49

 

 

As

 

patterns ar

   

between a:'

antisocial §
bettt'een bt
borderline
E
fairly net
the relati
Ofientati
COglittit'
batterer
traits Vt
compa
inrem
again:
begall
deCre

SlZe-1

As noted earlier, direct empirical support for relationships between attachment
patterns and batterer subtypes is lacking. However, given extant support for relationships
between attachment patterns and personality disorders, informed hypotheses can be made
about relationships between personality disorders and batterer subtypes. Associations
between antisocial personality disorder and avoidant attachment suggest that the
antisocial/generally violent batterer subtype may be avoidantly attached. Similarly, links
between borderline personality disorder and anxious attachment suggest that the
borderline/dysphoric batterer may be anxiously attached.

Bitterer typology_and differential treatment effectiveness. Typology research is a
fairly new line of inquiry, and only one study (Saunders, 1996) has directly investigated
the relationship between batterer type and treatment outcome for different therapy
orientations. Saunders found that the antisocial batterers completing a feminist-
cognitive-behavioral intervention had more favorable outcomes than did antisocial
batterers completing a psychodynamic intervention. In addition, batterers with dependent
traits who completed the psychodynamic intervention reported lower recidivism
compared to dependent batterers who completed the feminist-cognitive-behavioral
intervention. Because a no-treatment control group was not used, Saunders cautioned
against interpreting outcome as conclusively related to intervention. For example,
because participants were court-referred, arrest may have been an alternative motive for
decreased recidivism. Despite limitations, Saunders' (1996) study suggests that "one-
size—fit-all" interventions may not meet the needs of all batter types.

Methodological limitationwmmon to past typology research. Sampling issues

are a predominant methodological weakness in the typology studies discussed above.

Specifically, these studies rely on samples recruited from court-mandated interventions.

50

 

 

   
  
   
   
  

\i arianee on se\ -
be sufficiently 1.
both communit}
it difficult to cor
generalizabihty
community tie.
become involve.
threatening gene
addition, the nu
threatening the
Another notev
the violence,
inclined to t.
Rudy has IT
Character-1's
Regarding

Whether .

Variance on severity, frequency, and generality of violence among these samples may not
be sufficiently large. As evidenced by Holtzworth-Munore et al.’s (2000) study, using
both community samples as well as clinical samples could alter study findings and make
it difficult to compare such studies with those that only use clinical samples. The
generalizability of findings from studies relying on clinical samples to violent men in the
community (i.e., adolescents, men who rely mostly on psychological abuse and never
become involved with the criminal justice or other social systems) is limited. Also
threatening generalizability is the absence of random sampling in typology research. In
addition, the number of men categorized into batterer subgroups is often small,
threatening the power of analyses conducted to make comparisons across subgroups.
Another noteworthy problem is the likelihood that batterers may, in an effort to minimize
the violence, under-report violence. Additionally, some batterer subtypes may be more
inclined to respond in a socially desirable manner than are others. Next, no typology
study has included a control group to ensure that conclusions about clusters of personality
characteristics evidenced by the different types do not generalize to nonviolent men.
Regarding the measurement of constructs employed in typology studies, it is not clear
whether there is conceptual agreement about indicators of abuse (Gondolf, 1988). For
example, are researchers conceptualizing frequency and severity of abuse on the same
continuum or the same scale? Moreover, although there is agreement as to what types of
abuse comprise battering (National Institute of Justice, 1998), typology studies do not
account for all these types of abuse in their measurement of this construct. For example,
although economic abuse is a behavior identified as battering, no efforts have been made

to include a measure of this behavior in typology research.

51

 

 

   
  
   
   

[hm
lloltztt'orth-hlu
based on theor}.
proposed ttpoh
ope description
batter subtypes.
tiolence should

all lipolOgt' stut
N595. Howeee
Variables are \‘
lGondolf, 195
hpes of VlOk
Monolity c‘

W)4) and t

Theoreticgl limitationsgommon to past typology research. First, although
Holtzworth-Munroe and Stuart (1994) proposed a developmental model of batterer types
based on theory, typology studies, to date, have been predominantly descriptive. If
proposed typologies are to have any clinical utility, this line of inquiry must move beyond
type description and formulate hypotheses regarding the developmental trajectories of the
batter subtypes. Investigating hypotheses regarding why the different types enact
violence should prove fruitful toward this end. A second theoretical limitation is that not
all typology studies include both behavioral and personality variables to predict batterer
types. However, Holtzworth-Munroe and Stuart's review demonstrated that both types of
variables are valuable in developing batterer typologies. In addition, some researchers
(Gondolf, 1988) have suggested that different phases of violence rather than different
types of violent men may explain the evidenced heterogeneity of batterers. However,

personality disorders are stable characteristics (Diagnostic and Statistical Manual-IV,

1994) and have been shown to differentiate among batterer types (Dutton, 1988).

52

 

 

 

   
  

(Hamberger et .
l99l). Accords.
sexist attitudes

disorders play a

 

Frequency, sett
also important
SUb‘éTOUp will
generality o f
The t
[8pm Chara
Wong Sexis
Wilmer (6. !
gleategr fr
illSlory ill
Qongluer
Roam
"Erhal ;
balleye.
altitud
and p}

‘iOier

THE PROBLEM STATEMENT

Research investigating subtypes of male batterers supports a three-group
taxonomy that categorizes batterers as family-only aggressors, antisocial, or borderline
(Hamberger et al. 1996; Holtzworth-Munroe et al. 2000; Holtzworth-Munroe & Stuart,
1994). According to the proposed three-group taxonomy, adult attachment orientations,
sexist attitudes toward women, and dimensions of antisocial and borderline personality
disorders play an important-role in differentiating among the types of batterers.
Frequency, severity, and generality of psychological, verbal, and physical violence are
also important variables distinguishing the three subgroups. It is expected that each
subgroup will display distinct patterns of violence with respect to frequency, severity, and
generality of the various types of abuse.

The three batterer subgroups can be described as follows. Antisocial batterers
report characteristics congruent with antisocial personality disorder, avoidant attachment,
strong sexist attitudes toward women, and violence toward persons other than their
partner (e.g., friends, strangers, other family members, and police). They also report the
greatest frequency and severity of physical violence and indicate a more extensive arrest
history in comparison to the other subtypes. Borderline batterers indicate characteristics
congruent with borderline personality disorder, are anxiously attached, report moderately
sexist attitudes toward women, and report the greatest frequency of psychological and
verbal abuse but report less frequent and less severe physical violence than antisocial
batterers. F amily-only aggressors evidence no personality disorder, report the least sexist
attitudes toward women, and report the least frequent and severe psychological, verbal,
and physical violence. As implied by its label, this subgroup generally restricts its

violence to family members. Although it is expected that family-only aggressors report

53

 

 

more in:
assump‘

onenta‘

among
tostar.
physi
frequ
attac

batte

Sllg

10]

In

more insecure attachment compared to that of non-abusive men, to date, there are no
assumptions in the literature about specific relationships between adult attachment
orientations and this subgroup.

One purpose of this study is to test hypotheses about anticipated interrelationships
among anxious and avoidant attachment orientations, personality disorder, sexist attitudes
toward women, and frequency, generality, and severity of psychological, verbal, and
physical violence. Specifically, group differences on measures assessing severity and
frequency of psychological, verbal, and physical violence, generality of violence,
attachment, and sexist attitudes toward women among antisocial batterers, borderline
batterers, and participants indicating no personality disorder will be explored.

Given typology research’s support for different batterer subtypes, it has also been
suggested that different theoretical assumptions may predict patterns of violence specific
to the different subtypes (see Holtzworth & Munroe, 1994). The individual and society
are two levels of analysis dividing assumptions regarding factors that motivate male
heterosexual relationship violence (see National Institute of Justice, 1998). An individual
analysis suggests that individual psychological or personality variables are predictive of
male violence toward female partners, and a social-level analysis proposes that patriarchal
social, political, and economic structures promote sexist and/or negative attitudes toward
women that contribute to violence. Although extant research supports both sociopolitical
and psychological conceptualizations of battering, these diverse understandings of
violence have historically been viewed as competing perspectives (Miller, 1994).
Consequently, few studies have simultaneously drawn from both paradigms, and the
heuristic value of models that integrate sociopolitical and psychological theories is

uncertain. However, limitations of unidimensional models have recently been

54

 

aduonh
heongt

Rosana

concm
haso
ai,l
aha

int

acknowledged, and multidimensional models that draw upon both theories to understand
the origins of battering are being embraced (Ganley, 1989; Heise, 1998; Miller, 1994;
Rosenbaum & Mauiro, 1989).

Attachment theory is one psychological perspective that has been advanced as a
conceptual framework for understanding battering. Research indicating that abusiveness
has origins in relationships with early caregivers (Elbow, 1977; Dutton, 1988; Dutton et
al., 1994) and theoretical arguments that attachment dynamics, such as fear of
abandonment, explain battering (Mayseless, 1991) suggest the utility of attachment theory
in furthering etiologic understandings of battering. Research has also demonstrated that
personality disorders related to the need to control others (e. g., narcissistic and antisocial)
and related to self-concept and identity (e. g., borderline) are particularly prominent
among batterers (Hastings & Hamberger, 1988). Further understanding of relationships
among personality disorders, adult attachment orientations, and pattems of male violence

enacted toward female partners may enrich current etiologic explanations of woman

assault.

Another purpose of this study was to test a model wherein constructs central to
both sociopolitical (e.g., patriarchal attitudes) and psychological theories (e. g., adult
attachment orientation; borderline and antisocial personality disorders) are considered
unique, significant predictors of violence among a sample of men attending court-referred
batterer intervention programs. In addition, the model will test the hypothesis that
interactions representing patriarchal attitudes and the psychological constructs will also

be unique, significant predictors of violence. Patriarchal attitudes are operationalized as

Sexist attitudes toward women, understood to be the product of patriarchal institutions.

55

 

 

Git en this
indit'idual
Fir
distinct in
tiolence.
importzm
Purpose
Persona
tiolenet

m
This di

relatec

p03in

Given this understanding, it is important to note that measuring patriarchal attitudes at the
individual-level provides a very limited test of the feminist hypothesis.

Finally, it was expected that the data would reveal multiple violence patterns
distinct in frequency, severity, and generality of psychological, verbal, and physical
violence. Given the expectation that different theoretical assumptions are differentially
important in understanding the etiology of the distinct patterns of violence, the primary
purpose of this study was to explore differential effects of adult attachment orientations,
personality disorders, and sexist attitudes toward women in predicting patterns of
violence.

Hypotheses

This dissertation tested the following specific hypotheses.

1. Adult attachment dimensions and personality disorder scores will be significantly
related. More specifically:

1a. Antisocial personality disorder scores will have a significant, positive

relationship with avoidant attachment orientation scores.

1b. Borderline personality disorder scores will have a significant, positive

relationship with anxious attachment orientation scores.

1c. Antisocial personality disorder scores will be unrelated or

negatively related with anxious attachment orientation scores.
1d. Borderline personality disorder scores will be unrelated or
negatively related with avoidant attachment orientation scores.

2 - Both antisocial and borderline personality disorder scores will have significant

Positive relationships with sexist or negative attitudes toward women.

56

2a. The relationship between antisocial personality disorder scores and

negative attitudes toward women will be significantly greater than the

relationship between borderline personality disorder scores and negative

attitudes toward women.

3. Both anxious and avoidant adult attachment orientations will have significant

positive relationships with sexist or negative attitudes toward women.

3a. Given the previously hypothesized relationship between antisocial

personality disorder and avoidant attachment, the relationship between

avoidant attachment and negative attitudes toward women will be

significantly greater than the relationship between anxious attachment and

negative attitudes toward women.

4. Relative to their counterparts categorized as having borderline personality

disorder, participants categorized as having antisocial personality disorder will score
significantly higher on measures of severe physical violence, generality of violence,
frequency with which participant escalates and initiates violence, level of injury partner

Sustained due to violence, sexist attitudes against women, avoidant attachment, and
number of assault-related arrests and general arrests.

5-

Participants categorized as having borderline personality disorder will score
Si gnificantly higher than participants categorized as having antisocial personality disorder
011 measures of psychological and verbal violence, as well as anxious attachment.

6 ~ Constructs central to psychological (e. g., anxious and avoidant adult attachment

311d antisocial and borderline personality disorders) and sociopolitical (e. g., sexist

attitudes toward women) understandings of male perpetrated violence toward women will

57

beinde

olnda

f.)

andsc
ands:
Spec

ueh

 

 

be independently and significantly related in a positive direction with men's self-reports

of relationship violence.

7. After controlling for main effects, interactions between attachment orientations
and sexist attitudes toward women as well as interactions between personality disorders
and sexist attitudes toward women will significantly enhance the prediction of violence.
Specifically, among participants with high scores on anxious or avoidant attachment as
well as among participants with high scores on borderline or antisocial personality
disorders, the relationship between sexist attitudes and violence will be greater than
among participants with low scores on measures of insecure attachment or personality
disorders.
8. The data will reveal multiple patterns of violence among men arrested
for violence against their female intimate partners. More specifically, three distinct
patterns are anticipated:

(a) violence that targets predominantly the partner, although other family members
may also be targeted, includes psychological violence that is moderate in
fi'equency, and minimal or no physical violence,

(b) violence that is partner-specific, includes fi'equent psychological and verbal abuse,
and physical violence that is only minor in severity, and

(c) violence that is not partner-specific (e. g., general violence), is predominantly
physical violence, which is more severe than the violence exhibited in the other
two patterns, and moderate psychological violence.

9~ Contributions of anxious and avoidant attachment, antisocial and borderline
personality disorders, and sexist attitudes toward women to relationship violence will

Vary depending upon the violence dimension assessed. Specifically:

58

9a.

9b.

9c.

Anxious and avoidant attachment as well as sexist attitudes toward women
will be positively and significantly important predictors of pattern 8a,
while personality disorders will not be significant.

Anxious attachment, borderline personality disorder, and sexist attitudes
toward women will be positively and significantly important predictors of
pattern 8b.

Avoidant attachment, antisocial personality disorder, and sexist attitudes
toward women will be positively and significantly important predictors of

pattern 8c.

59

 

Panic

tho it
panic
perce
rt’pQr
Font

least

and
div
\t‘it

Ca

METHODOLOGY
Participants
Participants were 239 heterosexual men with at least an 8th grade reading level
who were enrolled in a batterer intervention program. Eighty-five percent of the
participants had been referred to the intervention program by court mandate. Thirteen
percent reported that they were self-referred to the intervention program, and another 2%
reported that they enrolled in the intervention program upon their partner’s request.
Forty percent of the sample had a high school diploma or graduate equivalent, 29% had at
least one year of college, 15% had at least one year of high school, 8% had a college
degree, 5% had only a junior-high school education, and 3% had a graduate degree.
Thirty-nine percent of the sample self-reported that they were married, 22% were single
and not living with a girlfriend, 20% were single and living with a girlfriend, 10% were
divorced, and 9% were separated. The age of participants ranged from 18 to 57 years old,
with a mean of 33 (SD = 8.97). The racial ethnic breakdown of the sample was 68%
Caucasian, 15% African-American, 10% Hispanic, 2% Native American, 1% Asian, and
4% self-identified as “other”. Participants’ reported ages ranged from 18 to 54 (M =
33 .24; _S_I; = 8.97).

Date were collected from an additional 67 participants who were not included in
the data analyses for the following reason. Initially, CTS directions asked participants to
identify violent behavior in the one-year period prior to their participation in the study,
and PMWS directions asked participants to identify psychologically abusive behavior in

the Six-month period prior to their participation in the study. However, after data had
been collected from 67 men, a batterer intervention program coordinator facilitating data

cone(Etion suggested that the nature of the CTS and PMWS directions could result in men

60

 

under

mand

crimi

less ‘

“fit

the

Sub

\‘io

Oil

underreporting acts of violence. Specifically, she highlighted that men who are court-
mandated to attend batterer intervention programs are also being monitored by the
criminal justice system. As such, while participating in programs, men are committing
less frequent and less severe violence than prior to their arrest, or than they would if they
were not being monitored. The coordinator recommended that I also ask men to report on
the one-year (for CTS items) or six-month (for PMWS items) period prior to their arrest.
Subsequent data collections asked participants (11 =239) to answer questions about acts of
violence committed from the perspective of two time points. Time point one was the
one-year or six-month time period before their participation in the study, and time point
two was the one-year or six-month time period before their arrest. Once all data were
collected, the initial 67 participants were compared with the 239 participants included in
the final sample on key variables (i.e., attachment orientations, personality disorders,
arrest information, social desirability, and all violence-related variables). For those
participants providing data from two time perspectives, scores for CTS minor and severe
physical violence subscales, the CTS verbal aggression subscale, and both PMWS
subscales were computed by averaging data fiom time points 1 and 2. For the initial 67
participants, scores on these subscales were reports of violence in the time period directly
before the study. Results from a MANOVA indicated that the initial 67 participants
(group 1) were significantly different from the 239 participants included in the final
Salnple (group 2), If (17, 288) = 1.84, p < .05. Univariate tests demonstrated significant
grOUp effects for verbal aggression, E (l, 304) = 8.08, p < .01, and PMWS
emOtionallverbal violence subscale scores, E (1, 304) = 4.16, p < .05, with group 1
SCOI‘ing significantly higher than group2. Given the significant differences between

groups 1 and 2, data collected from the first 67 participants were not analyzed.

61

Procedures
To facilitate data collection, this researcher established collaborative relationships
with nine batterer intervention programs participating in the Batterer Intervention
Services Coalition of Michigan. Intervention program participants were invited to
participate in this study by the investigator during one of their weekly group sessions.
One week prior to the invitation to participate, their group counselor informed men about
the investigator’s impending visit. Participants were informed that their participation in
this study was voluntary and anonymous, and that fifteen dollars would be applied toward
intervention program fees for those men choosing to participate. The counselor also
informed the men that an 8th grade reading level was required for participation, and about
the confidential and anonymous nature of the study. After being invited to participate in
the study, men interested in participating first completed informed consent forms and
were then given a packet of self-report questionnaires. The order in which the
questionnaires were administered was varied fi'om participant to participant.
Participation time was approximately 45 minutes. Men not interested in participating or
not meeting requirements for participation met with their counselor and worked on an
alternative assignment.

Measures

 

Demogpaphic and participant background questionnaire (see Ammndix CL This
questionnaire was developed to gather information regarding the participants’ ethnicity,
age, educational level, relationship history, arrest history, history of services received due

to Violence toward a partner, and number of weeks in the current treatment program.
33$ Reading Inventory: GLaded Passage-Form B. (BRI; Johns, 1997; see

Appendix D). The BRI is an individually administered informal reading test used to

62

 

 

rm

assess the respondent's reading level. There are five forms (A, B, C, LN, LE). Form B
should be used for silent reading. Each form is comprised of a graded word list and a

graded passage. Participants in this study were given the graded passage. After reading

the 8th grade level passage, participants were asked to respond in writing to 10

comprehension questions (i.e., “What is this passage about?”). If the participant responds

inaccurately to more than 3 questions, his comprehension of the material is below the 8th
grade level. The Basic Reading Inventory has been in publication since 1978, and, since
that time, it has been used extensively by various types of professionals in education and
psychology.
Personality Diagnostic Questionnaire —Revig_d. (PDQ-R; Hyler, Rieder,
Williams, Spitzer, Hendler, & Lyons, 1988; see Appendix E and F). The PDQ-R is a
self-report measure that screens for personality disorders described in the Diagnostic and
Statistical Manual (DSM-III-R; APA, 1987). Scores can be computed as both continuous
and categorical variables. Categorical scores can be compared with cut-off scores and
used to determine whether a person meets criteria for a specific personality disorder.
Participants in this study only completed borderline (Appendix E) and antisocial
(Appendix F) subscales. “Some people consider me a drifter” and “lying comes easily to
me and I often do it” are representative of items on the PDQ-R antisocial subscale. “I feel
empty or bored much of the time” and “my feelings toward another person can often
change drastically” are representative of items on the PDQ-R borderline subscale. The
PIDQ-R has been shown to have high negative predictive power, but low positive
Predictive power (Hyler, Skodol, Oldham, Kellman, & Doidge, 1992; Hyler, Skodol,

Oldham, Kellman, Oldham, & Rosnick, 1990). That is, the PDQ-R has been shown to

63

 

 

gener:
positi
respo

but a.

per
bet
to

Sit

generate a relatively low rate of false-negative diagnoses but a generally high rate of false
positive diagnoses. Given that the population to be sampled in this study is likely to
respond in a socially desirable manner (Arias & Beach, 1997), this bias was not optimal
but acceptable.

The PDQ-R is valid for use with both clinical and nonclinical samples (Hyler et
al., 1990; Hyler et al., 1992). Among an inpatient sample, the PDQ-R was shown to
differentiate persons with high likelihood for personality disturbance from those with a
low likelihood, suggesting that the PDQ-R is efficient in screening inpatients for DSM-III
personality disorders (Hyler et al., 1990). Among a group of outpatient clients, agreement
between PDQ-R subscales and ratings by two independent psychiatrists was comparable
to agreement between the two psychiatrists (Hyler et al., 1992). In a nonclinical college
student sample, participants scoring above the clinical threshold on PDQ-R subscales also
had elevated scores on the SCL-90, a self-report measure of psychopathology. In
addition, participants scoring above the clinical threshold on the PDQ-R had SCL-90
scores comparable to clinical outpatient samples (Johnson & Bomstein, 1992). In another

nonclinical sample, Wierzbicki and Gorman, (1995) found that MCMI-II and PDQ-R
borderline subscales were significantly correlated .57, and MCMI-II and PDQ-R
antisocial subscales were significantly correlated .37, suggesting construct validity of the
PDQ-R borderline and antisocial subscales. In the present study, Cronbach alpha
Coefficients of .66 and .58 were obtained for the borderline and antisocial subscales,
respectively, indicating that the PDQ-R is a marginally reliable measure of these 2
PerSonality disorders for the population sampled in this study. F ossati et al. (1998) found
soInewhat comparable internal consistency coefficients, with an alpha equal to .70 for the

borderline scale and .63 for the antisocial scale.

64

 

Appenc
orienta
onenu
worry
repres
not It
Pam
attac
disa

\bh

mo

\ll'lr

l8

si

Experiences in Close Religioashins. (ECR; Brennan, Clark, & Shaver, 1998; see
Appendix G). The ECR-short form is a 36-item measure of adult romantic attachment
orientation comprised of 2 scales: orientation toward anxious adult attachment and
orientation toward avoidant adult attachment. Each subscale is comprised of 18 items. “I
worry about being abandoned [and] I worry a lot about my relationships” are
representative of items comprising the anxious attachment orientation subscale. “I prefer
not to show a partner how I feel deep down [and] I get uncomfortable when a romantic
partner wants to be very close” are representative of items comprising the avoidant
attachment orientation subscale. On a scale of l to 7 ranging from agree strongly to
disagree strongly, respondents were asked to rate their level of agreement for each item.
When participants were clustered into four groups based on their scores on avoidant and
anxious attachment subscales, the groups corresponded to Bartholomew’s four-category
model of adult attachment, demonstrating that the ECR is a valid measure of dimensions
underlying attachment styles. Moreover, a correlational analysis including the avoidant
and anxious attachment orientation subscales and 60 subscales assessing attachment-

related constructs revealed that the anxious attachment orientation subscale was
significantly correlated with anticipated measures (i.e., jealousy, romantic anxiety, need
for approval) and the avoidant attachment orientation subscale was significantly
Correlated with anticipated measures (i.e., avoidance, dismissiveness, defensiveness).
Alpha coefficients for the anxious and avoidant scales are .94 and .91, respectively,
indicating good internal consistency. In the present study, Cronbach alpha coefficients
Were .90 and .85 for the anxious and avoidant subscales, respectively.
Aflitppde Toward Women Scal_e.. (AWS; Spence & Helmrich, 1972; see Appendix

H)- The AWS-short form (Spence, Helmrich, & Stapp, 1973) is a 15-item measure

65

describing the rights, roles, and privileges of women in society. The AWS provides
considerable variability in scores, especially among men (Spence & Hahn, 1997). Items
on the AWS are accompanied by a 4-point response ranging from agree strongly to
disagree strongly. “Swearing and obscenity are more repulsive in the speech of a woman
than a man [and] a woman should be as free as a man to propose marriage” are
representative of AWS items. Factor analysis of the AWS yielded only one major factor
(Spence et al.). Separate reliability analyses for females and male subjects yielded
Cronbach alphas of .86 and .90, respectively (Smith & Bradley, 1980). Daugherty and
Dambrot (1986) reported Spearman-Brown split half reliability coefficients of .86.
Yoder, Rice, Adams, Priest, and Prince (1982) have reported test-retest reliability
coefficients of .74 for males and .80 for females over a ten-week period. Construct
validity was established using a criterion groups method on several samples, with results
always in the predicted direction (Spence & Helmrich, 1972). In the present study, a
Cronbach alpha coefficient of .73 indicated that the AWS is a reliable measure of
negative attitudes against women for the population sampled in this study.

Although the AWS has been criticized as “outdated” (see McHugh & Frieze,
1997) due to a trend toward more egalitarian attitudes over the last few decades, Swim
and Cohen (1997) found that AWS scores, like scores on their “Modern Sexism Scale”,
were correlated with more conservative attitudes toward women. Moreover, study
findings suggesting a trend toward liberalism and egalitarianism have been predominantly
conducted with college samples, a group in our society that is most likely to have changed
conservative attitudes about women to more liberal attitudes. Given the nature of the
sample in this study, it is expected that the AWS is still a valid measure in predicting

conservative attitudes toward women. Nonetheless, due to criticism that the AWS does

66

 

 

not \'

to re:

sexis

[her

an.“

thre

afra

Cl'Ei

cha

AV

I16

1'8}

Vt

not validly measure contemporary sexist attitudes against women, participants were asked
to respond to 5 additional questions, which were developed by this researcher, assessing
sexist attitudes toward female intimate partners (see Appendix 1, questions 28-32).

These questions were accompanied by the same 4-point response rating as the AWS.
“The only way to get my needs met in my relationship with my girlfiiend or wife is to
threaten to hit, slap, or push her” and “The worst thing in a relationship is if a woman is
afraid of a man” are examples of these additional items. These items were aggregated to
create the subscale “chauvinism”, with a Cronbach alpha coefficient of .82. Because this
chauvinism subscale was significantly correlated (g = .31, p < .001) with AWS scores,
AWS scores and chauvinism subscale scores were standardized and summed to create a
new composite variable “sexism.” This new composite variable will, from this point on,
represent the sociopolitical construct — sexist attitudes against women — in all analyses.
The Cronbach alpha coefficient for the new composite scale sexism was .73.

Conflict Tactics Scale. (CTS; Straus, 1979; 1990a; see Appendix J). The CTS is
an 18-item self-report inventory consisting of a list of actions the respondent might take
in a conflict with his partner. The modes of conflict are scored on 3 subscales: reasoning,
verbal aggression, and physical violence. “Got information to back up my side of things”
is representative of items on the reasoning subscale. “Insulted or swore at her’ ’ is
representative of items on the verbal subscale. “Slapped her [and] threatened her with a
knife or gun” are representative of items on the physical violence subscale. Participants
were asked to indicate their use of violent tactics in the 12-month period prior to their
arrest or other motivating factor that precipitated their beginning counseling.

As a measure of husband-to-wife violence, the CTS has demonstrated high

internal consistency for the subscales verbal aggression and physical violence with alpha

67

coefficients of .80 and .83, respectively. However, an alpha coefficient of .50 for the
reasoning subscale suggests inadequate reliability. A comparison of student reports and
parent reports of mother-to-father violence demonstrated that father and student reports
for verbal aggression and physical violence were strongly correlated, with a correlation
coefficient of .51 for the verbal aggression subscale and .64 for the physical violence
subscale. Mother—student reports for the verbal aggression and physical violence
subscales were also significantly correlated with correlation coefficients of .43 and .33,
respectively. Correlation coefficients of .19 for father-student reports and -.12 for
mother-student reports for the reasoning subscale suggested that parent-student reports
were not strongly correlated for this subscale. These findings demonstrate evidence of
concurrent validity for the CTS verbal aggression and physical violence subscales, but not
the reasoning subscale. The high rates of verbal aggression and physical violence found
with the CTS subscales are consistent with previous in-depth interviews, providing
additional evidence of the construct validity of the CTS verbal aggression and physical
violence subscales. Further evidence of CTS’ validity is demonstrated by anticipated
relationships between CTS violence subscales and variables predictive of male violent
behavior toward female partners (i.e., the lower a husband’s economic and prestige
resources, the greater his tendency to use violence to maintain a male-dominant power
position; Straus, 1979; 1990a).

The CTS has been criticized for neglecting to consider the context in which
partner violence occurs as well as the impact or negative consequences experienced as a
result of the violence (Straus, 1990b). For example, the CTS does not assess who
initiated the violence or the injury sustained by partners as a result of relationship

violence. These criticisms are concerning because they highlight the CTS’ limitation in

68

assessing violence severity and whether the perpetrator is using violence to maintain
control of the victim (see Straus, 1990b). In response to criticism regarding the CTS’
inability to assess violence severity, Straus (1990c) developed a coding scheme that
distinguishes assaults that are “minor [or] less dangerous” from those assaults which are
severe [and] have a greater likelihood of causing injury. “Slapped her” is an example of
physical violence that Straus categorizes as minor violence, and “threatened her with a
knife or gun” is an example of a physical act of violence that Straus categorizes as severe
violence. As directed by Straus (1990c), subscales assessing minor and severe violence
were created. Due to limitations of the CT S, participants also answered questions
regarding the following — whether they initiated the violence (see Appendix C,
Demographic and Background Questionnaire question 16); injuries partner sustained
from past violence (see Appendix C, Demographic and Background Questionnaire
questions 23a —23c, 25, 26, and 27); whether they escalated violence between them and
their partners (see Appendix C, Demographic and Background Questionnaire questions
19-21); and whether they were violent toward general others as well as their partner (see
Appendix C, Demographic and Background Questionnaire questions 18a —18f). Items
assessing the last 3 violence characteristics were aggregated creating the following
subscales -generality of violence, injury, and violence escalation. Cronbach alpha
coefficients for these subscales were .67, .67, and .5 8, respectively. In the present study,
Cronbach alpha coefficients for reasoning, verbal, and physical violence subscales were
.49, .86, and .69, respectively. Given its low reliability in the present sample as well as
the absence of previous reliability and validity data, the CTS reasoning subscale was
dropped from the analyses. Cronbach alpha coefficients for minor and severe physical

violence subscales were .86 and .69, respectively.

69

Psychological Maltreatment of Women Scale. (PMWS; Tolrnan, 1989, 1999; see
Appendix K). The PMWS is a measure of the psychological maltreatment of women. It
contains 14 items that the respondents were asked to rate on a Likert scale ranging from
(1) never to (5) very frequently. For this study, respondents were asked to rate the
frequency of acts of psychological abuse 6 months prior to their arrest or other event
motivating them to commence a batterer intervention program. A factor analysis of the
long form revealed two subscales: emotional/verbal abuse and dominance/isolation.
Emotional/verbal abuse items include withholding emotional support, verbal attacks, and
humiliation. “I called my partner names [and] I treated my partner like an inferior” are
representative of items in the emotional/verbal abuse domain. Dominance/isolation items
tap the respondent's demands for subservience. “I interfered in my partner’s relationships
with other family members [and] I restricted my partner’s use of the telephone” are
representative of dominance/isolation items.

To increase content validity of the short-form, items from the two subscales on the
short version were purposively selected from their parent scales on the long version.
Cronbach alpha coefficients for the dominance/isolation and emotional/verbal short
subscales were .88 and .92, respectively. A factor analysis of the short form revealed that
the 14 items on the short form loaded on the same factor as when included in a factor
analysis of the 58 items on the long form. In the present study, Cronbach alpha
coefficients of .85 and.75 were obtained for the emotional/verbal abuse and
dominance/isolation items, respectively.

_M_arlowe-Crowne Social Desirability 8%, (SDS; Crowne & Marlowe, 1964; see
Appendix L). The SDS is the instrument most often used in domestic violence research to

measure a socially desirable response bias. This study used a short-form of the SDS that

70

 

contains 10 items. “I am quick to admit making a mistake” is representative of items on
the short form of the SDS. Participants were asked to respond on a 7-point Likert scale
the extent to which they agree or disagree with inventory items. The correlation between
the long and short form of the SDS is .87 (Greenwald & Satow, 1970).

The Cronbach alpha coefficient for SDS scores within a sample of 232
undergraduate students in a Canadian university was .72 (Loo & Thorpe, 2000). An
administration of the SDS to a sample of 236 undergraduate students in an American
university revealed a Cronbach alpha coefficient of .77 (Crino, Svoboda, Rubenfeld &
White, 1983). A one-month test-retest reliability coefficient for two administrations of
the SDS to 60 undergraduate students was .86 (Crino et al.). In another study, factor
analysis of the SDS revealed one factor, with an internal consistency coefficient of .70
(Ballard, 1992). To determine whether SDS items are sensitive enough to assess the
construct it is intended to measure, Ballard, Crino, and Rubenfeld (1988) asked a sample
of undergraduate students to rate the degree of social desirability of specific items and the
degree to which they would be willing to respond affirmatively in public to these items.
Findings indicated that desirability ratings and willingness to disclose ratings were
correlated in expected directions. Moreover, Crino et al. found that the Marlowe Crowne
SDS and the Edwards’ Social Desirability Scale were correlated .39, and then .35 at a
second administration of the two scales one month later. In the present study, a Cronbach
alpha coefficient of .75 was obtained.

Analfies

First, intercorrelations of all key variables were computed. It was expected that

social desirability would be correlated with some violence measures. Correlation

coefficients describing relationships between personality disorder scores and adult

71

 

attachment orientations, personality disorder scores and sexist attitudes toward women,

and adult attachment orientations and sexist attitudes toward women were computed. T2,
a modification of Hotelling's statistic-T1 (see Steiger, 1980) was computed to test for

significant differences as predicted among these pairs of correlation coefficients. In
interpreting the significance of correlation coefficients, Bonferroni corrections were made
to control for Type I error.

Next, using the PDQ-R as a categorical measure, men in the sample were
categorized into four groups — those meeting criteria for antisocial personality disorder,
those meeting criteria for borderline personality disorder, those meeting criteria for both
antisocial and borderline personality disorder, and those meeting criteria for no
personality disorder. Differences among these four groups on all violence-related
measures and attachment orientations were explored by conducting two one-way
multivariate analyses of covariances (MANCOVA). The independent variable for both
MANCOVAs was personality disorder and it had four levels- antisocial, borderline, both
antisocial and borderline, and none. SDS scores were the covariate for both
MANCOVAs. The dependent variables in the first MANCOVA were ECR anxious and
avoidant attachment orientation scores. The dependent variables in the second
MANCOVA were all violence-related constructs. This included CTS minor violence,
severe violence, and verbal abuse subscales, PMWS verbal/psychological abuse and
dominance/isolation subscales, injury partner sustained as a result of relationship
violence, generality of violence, frequency with which participant initiates and escalates
violence, number of participant relationships with violence, sexist attitudes against
women, and number of general and assault-related arrests. Given that a significant

multivariate F for both MANCOVAs was expected, a series of planned, orthogonal

72

contrasts were conducted to determine whether specific group differences between
participants categorized as antisocial and participants categorized as borderline are
consistent with the hypotheses 4 and 5 outlined above. Type I error for the contrasts was
controlled by adjusting .05 alpha by the number of contrasts conducted. Bonferroni post
hoc comparisons were used to explore group differences between all four personality
disorder groups on the dependent variables, given that there were no specific hypotheses
about differences between the four groups on the dependent variables.

It is important to note that the group of participants who met criteria for both
borderline and antisocial personality disorders was an unexpected finding. It was
anticipated that the data would reveal three groups — participants meeting criteria for
borderline personality disorder, antisocial personality disorder, or no personality disorder,
with men meeting criteria for only one category and the pr0portion of men meeting
criteria for each of the three categories as relatively equal. The initial data analyses plan
included computing two one-way MAN COVAs with three levels, allowing for specific
comparisons between participants meeting criteria for borderline personality disorder and
participants categorized as having antisocial personality disorder (see hypotheses 4 and
5). Although the unexpected fourth group allows for 2 x 2 MANCOVAs as a means to
explore group differences on all key variables, one-way MANCOVAs were computed as
planned because study hypotheses only targeted group differences between participants
categorized as borderline and participants categorized as antisocial. There were no a
priori hypotheses regarding group differences between participants meeting criteria for
both personality disorders or for participants meeting criteria for no personality disorder.
However, because the number of men meeting criteria for no personality disorder was

55% of the overall sample, Bonferroni post-hoc comparisons were computed to explore

73

 

ho“

ho

eri

be

$3

how this group differed from the other personality disorder groups. Additionally,
Bonferroni post-hoe comparisons were computed to explore how participants meeting
criteria for no personality disorder differed from the other personality disorder groups, to
better understand patterns of relationships among key variables evidenced in this study’s
sample.

Next, a hierarchical regression was conducted to test whether constructs central to
psychological explanations (e. g., attachment and personality disorders) and constructs
central to sociopolitical explanations (e. g., sexist attitudes against women) of male
violence toward female partners were independent, positive, significant predictors of
violence. Violence was a composite variable created by aggregating measures assessing
frequency of violence (i.e., PMWS and CTS scores), after standardizing PMWS and CTS
scores. The significance of interaction effects, after controlling for main effects, in
predicting violence was also tested.

Because past research supporting batterer typologies (see Holtzworth-Munroe &
Stuart, 1994) suggested that batterer subgroups evidence distinct patterns of violence, a
principal components exploratory factor analysis was computed. Items assessing
frequency with which participants initiate and escalate violence, items assessing the
generality of participant violence, items assessing injury experienced by participant
partners, CTS verbal abuse, reasoning abuse, and minor and severe physical abuse
subscales, and PMW S subscales were included in the factor analysis.

As expected, the factor analysis yielded more than one pattern of violence, and a
series of hierarchical regression analyses with the different dimensions of violence as the
outcomes were conducted. Results of these analyses indicated whether attachment

Orientations, sexist attitudes toward women, and borderline and antisocial personality

74

 

 

disorders were differentially important predictors of the distinct violence patterns that
emerged fiom the exploratory factor analysis. Hypotheses specific to each analysis
outlined above were tested.
Sample Size

According to Cohen (1969, & 1992), to detect a medium effect size for all
analyses, while maintaining power equal to .80 and alpha set at .05, the one-way
MANCOVAs with 4 groups required 45 participants in each group. For a priori
contrasts, a minimum of 64 participants in each group to be included in the a priori
contrast was required. The multivariate hierarchical regression analysis with nine
independent variables and one covariate required a total of 119 participants, and 85
participants were needed to test for the significance of computed correlation coefficients.
Given that 9 variables were entered into the exploratory factor analysis, approximately
45-90 participants were needed for this analysis, given recommendations that 5-10
persons are included in the factor analysis for each variable in the model (Cone & Foster,

1993; Tabachnick & Fidell, 1983). Finally, Steiger (1980) reported that T2, a

modification 0f Hotelling's statistic-T1, can be used “with confidence on sample sizes as
small as 20 (p. 250).”

The obtained sample was adequate to meet power requirements for most analyses.
However, in the interest of maintaining power equal to .80, the number of men
categorized as antisocial (< 64) and the number of men categorized as borderline (< 64)
did not allow a priori contrasts to test for medium effect sizes of between group
differences at alpha equal to .05. Although risking an increased possibility of Type 11
error, group sizes were adequate to detect large effect sizes of between group differences

at alpha equal to .10. Again in the interest of maintaining power equal to .80, for the

75

 

 

MANCOVAs, the number of men categorized as antisocial ( < 45) precluded tests for
medium effect sizes at alpha equal to .05 and power equal to .80. However, the number
of men categorized as antisocial did meet requirements for tests of large effect sizes with
alpha equal to .05 and power equal to .80. Type II error for hypotheses tested by
MANCOVAs was more than would be expected if the number of men categorized as
antisocial had been adequate to discern medium effect sizes.
2mm

This study employed a passive correlational design. There was no manipulation
of variables or group assignment via random or nonrandom methods. Because this type
of design only describes relationships among variables in the study, no causal inferences
can be made about the relationships among the variables in this study. A convenience
sample was used; thus, the generalizability of the results is limited. However, field
designs tend to have high external validity as compared to laboratory studies (Heppner,

Kivlighan, & Wampold, 1992)

76

RESULTS

The following section provides a detailed description of all data analyses and
findings. First, sample descriptive statistics are provided. Next, a correlation matrix of
all key variables is presented, and Hotelling’s T2 statistics are computed to assess whether
predicted differences among correlation coefficients are supported. Then, results from
MANCOVAS testing for group differences between personality disorder groups on
violence-related measures as well as measures of adult attachment orientations are
presented, along with results from planned, orthogonal contrasts testing specific
hypotheses regarding group differences on these variables between antisocial and
borderline participants. Univariate test results exploring group differences between all
the personality disorder groups, as well as follow-up Bonferroni post-hoe comparisons
are then presented. Next, a hierarchical regression is computed to test whether constructs
central to both sociopolitical (e.g., “sexism”) and psychological (e.g., adult romantic
attachment orientations and personality disorders) theories independently and
significantly predict variance in a composite measure of violence enacted toward female
romantic partners. Following this, a factor analysis was conducted to determine whether
the data indicated more than one pattern of violence enacted toward female intimate
partners. Finally, given that more than one pattern of violence emerged, findings from a
series of hierarchical regressions with the different patterns of violence as outcomes are
presented. These regressions explored whether sexism, anxious and avoidant adult
attachment, and borderline and antisocial personality disorders were differentially
predictive of the distinct patterns of violence.

Sample Descriptives

 

Number of significant relationships participant had in the last year. Eighty-five
percent of the sample reported having only 1 significant relationship in the past year. Six
percent reported having 2 significant relationships, whereas another remaining 6%

reported having none. The remaining participants reported three or more significant

77

_-— p up- .

 

relationships in the past year. Fifty—eight of the participants reported that their most
significant relationship lasted 12 months or longer, and 11% of the participants reported
that their most significant relationship lasted 1 month or less. Sixty-six percent of the
participants reported that they were still involved with the woman they had been fighting
with at the time of their arrest.

Number of participant relationships with physical violence. Participants were
asked to report the number of relationships in which they engaged in minor (e.g.,
slapping, hitting, or pushing) or severe (e.g., burning, choking, or cutting) physical
violence the year prior to data collection. Sixty percent of the sample indicated that they
used minor violence in only one of their relationships in the past year, whereas 4%
indicated that they used minor violence in 2 or more of their relationships during the past
year. Thirty-four percent denied any physical violence in any of their relationships
during the past year. Eighty-eight percent of the sample denied any severe physical
violence in any of their relationships during the past year. Twelve percent reported using
severe physical violence in only one of their relationships during the past year.

Arrest history. Only 12% of the sample reported no arrest history, whereas 29%

 

of the sample reported no assault-related arrest history. Twenty percent of the
participants reported only one arrest in their lifetime; 17% reported 2 arrests; 50%
reported 3 or more arrests in their lifetime. Fifty percent of the sample reported only 1
assault-related arrest in their lifetime; 16% reported 2 or more assault-related arrests; 10%
reported 3 or more assaulted-related arrests in their lifetime. Mean number of general
arrests for the sample was 3.9 @Q = 5.0), and mean number of assault-related arrests was
1.2 (SD = 1.28).

Number of weeks participant had been at the intervention progpam. The number
of weeks participants had been at the intervention program ranged from 1 to 52 (M = 9.6;
S_D = 7.26), and 29% of the participants had been in counseling before due to being

violent toward their female intimate partner.

78

Initiation and escalation of violence. Forty-four percent of the sample indicated

 

that they had “never” been responsible for initiating physical violence toward their
female partner. Thirty percent reported that they had only initiated violence “once”, 21%
reported that they had initiated violence a few of the times, and 5% reported that they had
initiated violence at least “half of the time”. Fifty-nine percent of the sample reported
that they had hit, slapped, or pushed their partner after yelling at and/or calling her names,
and 18% of the sample reported that they had burned, punched, or cut their partner after
yelling at and/or calling her names.

Generality of violence. Seventeen percent of the sample reported that they had

 

been violent toward a sibling in the past year, and only 8% reported that they had been
violent toward a parent in the past year. Thirteen percent of the participants reported that
they had been violent toward another relative or fiiend in the past year. Twenty-nine
percent reported that they had been violent toward a stranger in the past year, and 6%
reported that they had been violent toward a police officer in the past year.

In'Lury experienced by participant partners. Fifty percent of the participants

 

reported that their partner had never experienced any injury, including a minor cut, bruise
or swelling, as a result of relationship violence. Thirty-six percent of the sample reported
that their partner had experienced a minor, cut, or bruise as a result of relationship
violence only once and 13% of the sample reported their partner had experienced a minor
cut, bruise, or swelling as a result of relationship violence two or more times. Fifteen
percent of the participants reported that their partner had experienced a burn, major cut or
bruise, or fracture as a result of relationship violence, and 2% reported that their partner
had experienced major bleeding or wounds or had been knocked out as a result of
relationship violence. ' Ten percent of the participants reported that their partner had
needed hospital care as a result of relationship violence in the past year, and 12% reported

that she had needed some other medical attention. Finally, 16% of the participants

79

reported that their partner missed her regular activities (e.g., shopping, family/friend
social activities) as a result of relationship violence.

Violence: CTS subscales. Frequency with which participant reported physical

 

violence toward a female intimate partner in the year before his commencing the
intervention program ranged from 0-79 M = 4.7; _S_D = 7.9.). The reported frequency of
acts of severe physical violence ranged from 0 —25 (M = 1.3; _S__Q = 3.7), and the
frequency of acts of minor physical violence ranged from 0-55 (M = 3.4; _S_D_ = 6.6).
Frequency of acts of verbal aggression during that time period ranged fi'om 0-150 Q1 =
34.8; SD = 35.2).1

Violence: PMWS subscales. The frequency of acts comprising the

 

emotional/verbal abuse subscale committed by participants 6 months prior to
commencing the intervention program ranged from 1- never to 5- very frequently (M =
2.15; SD = .79). The frequency of acts comprising the dominance/isolation abuse
subscale reported by participants ranged from 1- never to 4- frequently (M = 1.8; SD =
.7 1).

Personality disorders: PDQ -R antisocial and borderline personality disorder
subscales. F ifty-five percent (11 = 132) of the participants did not meet criteria for any
personality disorder. Nineteen percent (n = 45) met criteria for borderline personality.
Seven percent (n = 17) of the sample met criteria for antisocial personality disorder.
Another 19% (n = 45) met criteria for both antisocial and borderline personality
disorders.

Sexist attitudes against women. Range of responses for individual AWS items

 

was 0 (disagree strongly) to 3 (agree strongly), with high scores reflecting more
traditional attitudes against women (M = 13.11; SD = 6.45). Range for total AWS scores

was 0-35, with approximately 50% of the sample reporting a score of 12 or higher. This

 

1 Observed scatterplots for all dependent measures in the MANOVA or regression analyses were linear
and/or normal to meet necessary data assumptions for these analyses. Maximum frequencies were outliers.

8O

 

finding suggests that the AWS had significant within-group variability to validly assess
sexist attitudes against women among this study’s sample.

Correlational Analyses

 

Intercorrelations among selected demographic variables, as well as measures of
social desirability, violence, negative attitudes against women, attachment orientations,
and borderline and antisocial personality disorders were computed. Table 1 presents a
correlation matrix demonstrating relationships between selected demographic variables
and other key variables. A brief summary of Table 1 follows. Given the number of '
correlations computed: Bonferroni corrections will be made to control for Type I error.
As such, the significance of all correlation coefficients was tested at p < .001 . Specific
correlation coefficients will be discussed for bivariate relationships testing hypotheses
identified earlier.

Social desirability scores. Social desirability scores were significantly and

 

negatively correlated with most measures of violence as well as items and subscales
assessing additional background information regarding participants’ history of intimate
violence. Unexpectedly, social desirability scores were not correlated with CTS minor
and severe physical violence subscales, injury experienced by partner, number of
participant relationships with severe violence, or the composite variable sexism. Social
desirability scores were significantly and negatively correlated with both measures of
personality disorders and anxious attachment, but not avoidant attachment.

PDQ-R personality disorder scores. As expected, PDQ-R antisocial personality

 

disorder scores were significantly and positively correlated with avoidant attachment Q =
.29, p < .001), and PDQ-R borderline personality disorder scores were significantly and
positively correlated with anxious attachment Q = .57, p < .001). Unexpectedly, PDQ-R
antisocial scores were also significantly and positively correlated with anxious
attachment, and PDQ-R borderline scores were also significantly and positively

correlated with avoidant attachment. However, given the significant intercorrelation

81

between the two PDQ-R scales Q = .57, p < .001), partial correlations were computed to
assess relationships between PDQ scales and attachment orientations and test hypotheses
la-ld. After controlling for PDQ-R antisocial personality disorder scores, borderline
personality disorder scores were significantly and positively correlated with anxious
attachment Q = .45, p < .001), but unrelated to avoidant attachment. After controlling for
PDQ-R borderline personality disorder scores, antisocial personality disorder scores were
unrelated to both avoidant and anxious attachment. These findings support hypothesis 1b
but fail to support hypotheses la, 1c, and 1d. I

PDQ-R borderline personality disorder scores were also significantly and
positively correlated with number of relationships with severe physical violence,
generality of violence subscale scores, escalation of violence, both PMWS subscales,
CTS minor physical violence subscale, CTS verbal aggression subscale, frequency with
which participants initiated violence, and number of assault-related arrests. PDQ-R
antisocial personality disorder scores were significantly and positively correlated with
generality of violence, escalation of violence, both PMWS subscales, CTS minor physical
violence and verbal abuse subscales, frequency with which participant initiated violence,
number of assault-related arrests, number of general assaults, and number of violent
intimate relationships participant had in the past year. Both PDQ-R borderline and
antisocial personality disorder scores were significantly and negatively correlated with
age. As expected, the data suggested significant positive correlations between antisocial
personality disorder scores and general as well as severe violence. However,
unexpectedly, the data also suggested significant positive correlations between borderline
personality disorder scores and generality of violence as well as severe violence.

Adult attachment orientations. Adult avoidant attachment orientations were

 

significantly and positively related with generality of violence and escalation of violence,
suggesting that persons who score high on avoidant attachment are also likely to score

high on measures assessing generality of violence and participant responsibility for

82

 

escalating violence. Contrary to expectation, CTS subscale scores assessing severe
violence were not related to avoidant attachment. Anxious adult attachment orientations
were significantly and positively related to escalation of violence, PMWS subscales, CTS
minor and verbal aggression subscales, and frequency with which participant initiated
violence. As expected, persons reporting anxious attachment were more likely to report
psychological and verbal violence.

m Partial correlations were again computed to test relationships between
personality disorders and sexist attitudes and to test hypotheses 2 and 2a, given the
correlation between the two PDQ-R scales. After controlling for PDQ-R antisocial
personality disorder scores, borderline personality disorder scores were significantly and
positively correlated with sexist attitudes Q = .19, p < .001). After controlling for PDQ-R
borderline personality disorder scores, antisocial personality disorder scores were
unrelated to sexist attitudes. These findings offer partial support for hypothesis 2 and fail
to support hypothesis 2a.

As expected, the composite variable sexism was significantly and positively
correlated with anxious Q = . 19, p < .001) and avoidant attachment Q = .27, p < .001)
orientations. This finding supports hypothesis 3. As expected, the relationship between
avoidant attachment and sexism was greater than the relationship between anxious
attachment and sexism. Using the formula presented by Steiger (1980) to compute I,
revealed that L equals 1.102 (df = 236) and is smaller than L (1.645, df = 236). As such,
although the relationship between avoidant attachment and sexism is greater than the
relationship between anxious attachment and sexism, the difference between the
correlation coefficients is not significant. This finding failed to support hypothesis 3a.

The composite variable sexism was also significantly and positively correlated
with generality of violence subscale scores, PMWS dominance/isolation subscale scores,
CTS minor violence subscale scores, frequency with which participant initiates violence,

and number of relationships participant had in the past year. Unexpectedly, the

83

composite variable sexism was not significantly correlated with CTS severe physical
violence subscale scores, PMWS emotional/verbal subscale scores, or CTS verbal

aggression subscale scores.

84

so. was ”202

 

co; .2
am _. 2: a
sun. 1mm. 2: w
.3». .3. Lb. co; m
awn. .2. sum. im. 2: .o
*3. PM. com. 12. LN. co; m
3. 3. E. 2. :3. oo. co; v
2. aom. .Nm. .3. 2. .mm. o_. co; .m
2. 3. 2. 2. 4mm. 2. Ln. .5. 2: .N
NH. mH. nom. mo. swm. 2. wow. amm. 3N. as. F .—
8533 8:22.»
“58583 258533 8:203 8:203 .858 Ego“. 5:5
324% “529..” 0:53ch Boom?” mo coca—8mm 8&wa .05?“ .«o rag—83% 5:» £222.22.“ 338332.“

 

EoEfimt< «33.. can £8985
bzacofiom .8565 .Emmxom 5238600 Eoom mo 8.5982 Ea fling; oEmeoEoQ vogue—om waofiw meoufloboocoafi

_ 2an

85

:5. mm. ”202

 

2.- 8.. 2. Z. 3. .8. .Nm. .2. ac. .3. .3. .mm.- .2
«co.- Nor oo. 2. mo. 2. 2. 2. we. 2. E. 2.- .m
.3”.- 2.- 2. .mm. .2. .3. .3. .mm. 2. .5. .On. .2..- .w
.2: 3.- .2. .bm. in. .mm. .3. .mm. LN. .em. .3. .3..- <-
oo. mo. no. 5.- 5.- .mm. .3. .mm. .5. .mm. .mm. .2.- .o
2.- 3.- .2. 8.- mo. .2. co. .2. 2. .2. we. 2.- .w
2.- .2. mo. 3. mo. .Om. .2. 1m. .om. .om. .5. 2.- .2.
.mm.- we: .2. .2. .2. 2. .2. .mm. 2. .2. no. .2.- .m
8.. .vm. .2. no. 2. .3. ac. 2. 2. .2. 2. .2.- .N
2 .- 2. 2. mo. 2. .5. we. so. no. #2. ho. mot .—
80» “man 5
3:80:22 8?» :2:ch 8:203 coco—02> 2833 2333
#628. uvdq 2: 5 $85 3:20? 8:203 3393a .8?an cows—mi 39:5
.«o 2285 83823.8 3.55% :3me c8 .8 82? SEE 238 $23588 \ancofim 2:29.63
a». we a co . .6 a 385 no a 5538 ”who go ”...-6 ”was: ”was; 38.
.mm .mm ._m .om .2 .3 .2 .2 .3 .2 .2 .:

 

68388 _ 2.3

86

8. wm. ”2oz

 

 

o2 .mm
.2. co; .ov. .mm
no; mo. 2: 2m
2.- 3.. S. .om
mo: 8.. 2. .3. oo._ .2
mo; 2. Z. No. :. 82 .2
8.- oo. 2. 8.- S.- 8: 2
.om: S. 3.- .w _. co. .vm. co; .2
2.- 3.. S.- 2. 2. .on. .3. 2: .2
no; 2.. mo. 2. 2. .2. .2. .3. cc; .3
2.- 8. mo. .3. .2. .2. .3. .3. .mm. 82 .2
2..- mo. 2. 2. 8. .mm. .8. .mm. .2. .8. co; .2
m —. mo. ab; 2 .- 2.- .2; .2..- m 2.- b? lb.- .2..- so; .:
Box 3.3 5
5:85:22 so» .383 85355 8:20? 8:203 2838 2833
“825. 38 05 5 38.5 .8 8:20? ooze—o3 3525a .33an 858—95 .255
.3 5808 358552 338% 335 mo it? .858 233. 328583 28588.2 5.5588
8. no a .6 a a0 a co . 3538 ”m5 ”m8 ”who ”maze ”was“. 38...
.mm .mm ._m .om .2 .2 .2 .2 .vr .2 .2 .:

 

98888 _ 2.3

87

Group Comparisons between Personality Disorders: Multivariate Analyses of Covariance

Two one-way multivariate analyses of covariances (MAN COVA) were computed
with personality disorder as the independent variable. For this analysis, the PDQ-R was
used as a categorical measure. The independent variable had four levels- antisocial
personality disorder, borderline personality disorder, no personality disorder, and persons
with cut-off scores for both antisocial and borderline personality disorders. It was not
anticipated that participants might meet criteria for both antisocial and borderline
personality disorders. Nonetheless, this fourth group was included in the MANCOVAs
on the basis that doing so might reveal additional information regarding specific
differences between the personality disorder groups.

Only 17 participants in the sample had cut-off scores indicating antisocial
personality disorder. Forty-five participants met criteria for borderline personality
disorders, and another 45 participants met criteria for both antisocial and borderline
personality disorders. One hundred and thirty-two participants did not meet criteria for
any personality disorder. These group sizes are only nearly adequate to discern a large
effect size with alpha equal to .05 and power equal to .80, given a 4-group design. This
suggests that unless there is a large effect, between group differences will not be detected
and there is a risk for Type 11 error (see Cohen, 1992).

Between-group differences on adult attachment orientations. Dependent variables
in the first MANCOVA model were ECR anxious and avoidant adult attachment
orientation scores. SDS scores were entered as the covariate. When the model was run
with the interaction between social desirability and personality disorder groups included,
The MAN COVA did not produce a significant multivariate effect for personality disorder
group. However, because the interaction term was not significant, meeting the
MAN COVA assumption of homogeneity of slopes, the model was run a second time
without the interaction term included in the analyses. MAN COVA assumptions

regarding normality, independence, and homogeneity of variances were met. Results

88

indicated a significant multivariate effect for personality disorder group, Wilks’ E (6,
466) = 8.23, p 5 .001.

Following the significant multivariate F, planned orthogonal contrasts were
conducted to determine if group differences between participants meeting criteria for
antisocial and borderline personality disorder on the dependent variables are consistent
with hypotheses 4 and 5 identified in an earlier section. Given that the MAN COVA
tested for group differences on 2 dependent variables, Type I error in testing the group
differences between the two personality disorders was controlled by Bonferroni
corrections (.05/2) and alpha was equal to .025. Contrast results indicated no significant
group differences between participants meeting criteria for antisocial and borderline
personality disorders on adult anxious, Q (235) = 1.45, p > .025), or avoidant attachment
Q (23 5) = .13, p > .025), orientations. Given this finding, the expectation that participants
meeting criteria for antisocial personality disorder would score significantly higher on
adult avoidant attachment orientation than participants meeting criteria for borderline
personality disorder was not supported. The expectation that participants meeting criteria
for borderline personality disorder would score significantly higher on adult anxious
attachment orientation than participants meeting criteria for antisocial personality
disorder was also not supported (see hypotheses 4 and 5).

However, univariate F tests indicated that the four groups of the independent
variable in the MANCOVA differed significantly on both anxious, E (3, 234) = 14.26, p g
.001, and avoidant, E (3, 234) = 3.91, p 5 .01 , attachment dimensions (see Table 2). As
such, Bonferroni post-hoc analyses were conducted to explore these group differences.
Post-hoe comparisons indicated that participants categorized as borderline scored
significantly higher than participants not meeting criteria for any personality disorder on
ECR adult avoidant attachment orientation scores, Q(235) = 1.81, p < .05), and ECR
anxious attachment orientation scores, (_t (235) = 2.24, p < .001). Participants meeting

criteria for both personality disorders also scored significantly higher than participants

89

not meeting criteria for any personality disorder on ECR anxious attachment orientation
scores, (1 (235) = 2.56, p < .001).

Table 2

Adjusted Mean Scores and Standard Deviations for all Four PersonalitLDisorder Groups

on Anxious and Avoidant Attachment Orientations

 

Anxious Attachment Avoidant Attachment
Group
E M S} M §2

No Personality Disorder 132 p 3.18 1.15 2.76 .97
Borderline Personality 45 4.38 1.03 3.27 1.12
Disorder

Antisocial Personality 17 3.06 1.21 3.34 1.03
Disorder

Both Personality 45 3.72 1.28 3.27 .93
Disorders

 

 

 

Between-group differences on violence-related variables. Dependent variables in
the second MANCOVA model were sexism, PMWS subscales — dominance/isolation and
emotional/verbal violence, CTS subscales -verba1 aggression, minor physical violence,
and severe physical violence, as well as additional items and subscales assessing
participant violence. These additional items and subscales included generality of
violence, frequency with which participant initiates and escalates violence, injury partner
sustained as a result of violence, number of participant intimate relationships with severe
or minor physical violence, number of general arrests, and number of assault-related
arrests. SDS scores were again entered as the covariate. However, when the model was
run with the interaction between social desirability and personality disorder groups
included, results indicated that the interaction between the covariate and the independent
variable was significant, Wilks’ E (42, 657) = 1.81, p 5 .01, violating a key assumption of
the MAN COVA model - homogeneity of slopes. However, univariate E statistics
indicated that this assumption had only been violated for five of the dependent variables

entered in the model - sexism, CTS verbal aggression subscale scores, PMWS

90

 

emotional/verbal subscale scores, CTS severe violence subscale scores, and number of
participant intimate relationships with minor violence. Moreover, MANCOVA
assumptions regarding normality, independence, and homogeneity of variances for
variables meeting the homogeneity of slopes assumption were met. Given this
observation, main effects of personality disorder group for those variables for which the
homogeneity of slopes assumption was not violated were interpreted (see Table 3).
Univariate E statistics for these variables indicated no significant differences between the
four personality disorder groups, so no post-hoe comparisons or a priori contrasts were
computed. Expected group differences between participants categorized as antisocial and
participants categorized as borderline on violence-related variables were not supported
(see hypotheses 4 and 5).

Between-group differences for the five dependent variables not meeting the
homogeneity of slopes assumption were assessed via an alternative procedure (see
Huitema, 1980). The covariate was transformed into a categorical variable, with
participants categorized into one of four groups. Participants scoring in the 25‘h percentile
on the covariate were categorized as low SDS. Participants scoring between the 25th and
50"I percentile on the covariate were categorized as moderate SDS. Participants scoring
between the 50'” and the 75th percentile on the covariate were categorized as moderate-
high SDS. Participants scoring between the 75'” and the 100th percentile on the covariate
were categorized as high SDS. Then, a two-way multivariate analysis of variance
(MAN OVA) was computed. Personality disorder groups and social desirability groups
were entered as the two factors and an interaction of these two factors was included in the
model. Sexism, CTS verbal aggression subscale scores, PMWS emotional/verbal
subscale scores, CTS severe violence subscale scores, and number of participant intimate
relationships with minor violence were the dependent variables in the model. MAN OVA
assumptions regarding normality and independence were met and homogeneity of

variances across the groups was met for PTWS emotional/verbal subscale scores. Given

91

the violation of equal error variances across the groups for four of the five dependent
variables, Pillai’s V was used to test for significant multivariate main and interaction
effects due to its robustness under violations of MANOVA assumptions (Haase & Ellis,
1987)

Table 3

Adjusted Mean Scores and Standard Deviations for all Four Personality Disorder Groups

on Dependent Measures Meeting Homogeneity of Slopes MANCOVA Assumption

 

Dependent Groups
Variable No Personality BorcTerIine Antisocial Both PersonaIrTy
Disorder Personality Personality Disorders
Disorder Disorder

(N= r32) (N=45) (N=17) (N=45)
& £12 & a). M S_D M §2

 

How often
Participant
Initiated .65 .88 1.09 .90 1.18 1.29 1.32 1.18
Violence?
PTW S
Dominance/ 10.84 3.63 13.93 4.61 12.88 6.98 16.34 5.46
Isolation Scores
CTS Minor
Physical 1.81 3.65 3.91 4.06 6.29 8.90 6.41 1 1.49
Violence Scores
How often
Participant
Escalated 1.11 1.58 2.40 2.16 1.88 2.06 2.32 2.49
Violence?
Generality of
Violence .98 1.75 2.07 3.24 .58 .87 2.57 3.34
Number of
Assault-related .93 .95 1.51 1.67 1.00 1.27 1.52 1.55
Arrests
Number of
Arrests
Frequency with
which Partner .80 1.15 1.47 1.71 1.29 1.57 1.29 1.73
Sustained Injury
Number of
Participant
Relationships
with Severe .64 .59 .84 .64 1.12 2.34 .82 .15
Physical
Violence Last
Year

2.70 3.80 4.20 4.40 7.47 8.38 5.70 5.74

 

92

Results indicated a significant multivariate main effect for personality disorder
group, E (15, 663) = 3.89, p 5 .001, and social desirability group, E (15, 663) = 3.82, p 5
.001, as well as a significant interaction effect, E (45, 1115) = 2.89, p 5 .001. Tests of
univariate effects demonstrated a significant interaction effect for CTS verbal aggression
scores, E (9, 223) = 2.29, p 5 .05, CTS severe physical violence scores, E (9, 223) = 2.45,
p 5 .01, and number of participant intimate relationships with violence, E (9, 223) = 8.79,
p 5 .001, precluding interpretation of main effects for these dependent variables.
Univariate tests of main effects for sexism and PTWS emotional/verbal violence scores
indicated significant personality disorder group, E (3, 223) = 7.30, p 5 .001, and social
desirability group differences, E (3, 223) = 5.62, p 5 .001, on PTWS emotional/verbal
violence scores only (see Table 4). A priori contrast results exploring differences
between antisocial and borderline participant groups on PTWS emotional/verbal violence
scores indicated no significant between-group differences, failing to support hypotheses 4
and 5. Bonferroni post-hoe comparisons exploring between-group differences for the
four personality disorder groups on PTWS emotional/verbal violence scores indicated
that the borderline group, Q (235) = 5.01 , p < .001), antisocial group, Q (235) = 3.03, p <
.05), as well as participants meeting criteria for both personality disorders, Q (235) =
8.48, p < .001), scored significantly higher than participants not meeting criteria for any
personality disorder on PTWS emotional/verbal violence scores. Bonferroni post-hoc
comparisons exploring between-group differences for the four social desirability groups
on PTWS emotional/verbal violence scores indicated that participants reporting low
social desirability scored significantly higher on PTWS emotional/verbal scores than
participants reporting moderate, Q (23 5) = 3.48, p < .001), moderately high, Q (235) =
5.65, p < .001), and high, Q (235) = 7.79, p < .001), social desirability. Additionally,
participants reporting moderate social desirability scored significantly higher on PTWS
emotional/verbal scores than participants reporting high social desirability disorder, Q
(235) = 8.47, p < .001).

93

 

lid—_4‘

 

Table 4

Mean Scores and Standard Deviations for all Four Personality Disorder Groups and

Social Desirability Groups on PTWS EmotionaWerbal Scores and Sexism

 

 

 

 

PTWS Sexism
G EmotionaWerbal
roup -N- Scores
M SD

M. §2 — —
No Personality Disorder 132 12.73 4.44 -.22 1.47
Borderline Personality 45 16.71 5.25 .37 1.82
Disorder
Antisocial Personality 17 16.36 5.99 -.50 1.65
Disorder
Both Personality 45 19.52 4.91 .48 1.72
Disorders
Low SDS 56 18.79 5.56 .26 1.86
Moderate SDS 64 15.36 5.0 .27 1.54
Moderate-High SDS 61 13.95 4.77 -.03 1.66
High SDS 58 12.07 4.48 -.53 1.29

For the dependent variables verbal aggression, severe physical violence, and

number of participant intimate relationships with violence, the significant interaction

effect between personality disorder group and social desirability group was plotted as

recommended by Shavelson (1988; see Figures 1-3).

Next, given that there were no predetermined hypotheses regarding differences

between cell means, the two way MANOVA was arranged as an one-way MAN OVA, as

recommended by Shavelson (1988). A new variable with 16 levels (i.e., 4 personality

disorder groups x 4 social desirability groups) was created and Tukey’s post-hoe

comparisons were conducted to explore differences between all possible pairs of means

(see Table 5).

94

 

Figure 1: Interaction for Two-Way Analysis of Variance:
Personality Disorder Group and Social Desirability Group
as Independent Variables and CT S Verbal
Violence Scam as Outcome

 

 

 

 

 

 

 

 

 

 

 

80
7O .
8 —o—normonality
5 60 t disorder
E 50 . +b0rderline
E 40 - —A—antisocia1
3 3o -
> _ .x_ both personality
20 “ disorders
10 .
0
Social Desirability
Low Moderate High
Figure 2: Interaction for Two-Way Analysis of Variance:
Personality Disorder Group and Social Desirability Group
as Independent Variables and CT S Severe Physical
Violence Scores as Outcome
0 5 ._._no personality
3 4-5 — disorder
g 4 ~ +borderline
> 3.5 .
.3 3 T _‘_antisocial
O
z. 2.5 1 .
a. 2 .. .x .. both personality
2 1 disorders
0 1.5 .
>
0
m 1 .
0.5 -
0
Social Desirability
Low Moderate High

95

Figure 3: Interaction for Two-Way Analysis of Variance:

Personality Disorder Group and Social Desirability Group as
Independent Variables and Number of Particpant Violent

Intimate Relationships as Outcome

 

 

—-o—no personality
disorder

+ borderline / \

_‘—antisocial /

_ .x _ both personality
disorders / \

 

 

 

 

o 6
O
s:
03
3.2 5*
.§>
15:-‘2 4.
:3
Om
5.3- 3-
0
£5
E:
=.§ 2-
22
0
a: r.
0
Table5

Social Desirability

Low Moderate

High

Factors and Cell Sizes for New Sixteen-Level Factor Created by Collapsing Four-level

Personality Disorder Group Variable and Four-level Social Desirability Group Variable

 

Personality Disorder Groups

 

 

19491 Kurrqwsacr reroos

 

 

 

 

 

No Personality Borderline Antisocial Both
Disorder Personality Personality Personality
Disorder Disorder Disorders
18%: “:9 n=18 n=6 n=23
gigglemte n =29 n =10 n =7 n =18
Moderately n =47 _ _ =
High SDS n —10 n —2 n
High SDS n =47 n = n = n =

 

 

 

 

 

96

 

Pillai’s V was again used to test for multivariate effects due to the violation of the
assumption of homogeneity of variance across groups. This variance was expected given
the variance in number of participants per cell. Results of the multivariate test
demonstrated significant group differences, E (45, 669) = 4.72, p 5 .001. Tests of
univariate effects indicated that there were significant group differences for CTS verbal
aggression scores, E (15, 223) = 7.06, p 5 .001, CTS severe physical violence scores, E
(15, 223) = 2.12, p 5 .01, and number of participant relationships with violent intimate
relationships, E (15, 223) = 6.03, p 5 .001. Tukey post-hoc comparisons revealed only
one significant mean difference for CTS severe physical violence scores. Participants
meeting criteria for both personality disorders and reporting moderate social desirability
scored significantly higher than participants meeting criteria for no personality disorder
and reporting moderately-high social desirability, Q (223) = 3.6, p < .05). For number of
participant intimate relationships with violence, participants categorized as antisocial and
reporting moderately high social desirability had significantly more violent intimate
relationships than all other groups. With respect to CTS verbal aggression scores,
participants meeting criteria for borderline personality disorder and reporting low social
desirability indicated greater frequency of verbal aggression compared with participants
not meeting criteria for any personality disorder. Additionally, participants meeting
criteria for both personality disorders and reporting low social desirability reported
greater frequency of verbal aggression compared with participants categorized as
borderline and reporting moderately high social desirability as well as compared with

participants not meeting criteria for any personality disorder (see Table 6).

97

Table 6

Significant Tukey post-hoe Comparisons on CTS Severe Physical Violence and Verbal

Aggression Scores, and Number of Partimant Intimate Violent Relationships

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

high sns

 

 

 

 

Dependent Reference
Variable Group Comparison Group Mean s1; t 9
Difference
CTS severe No personality disorder/
physical moderately high SDS
violence Both personality disorders]
subscale scores “10(36th SDS 3.58 .99 3.62 .05
# of partrcrpant Antrsocral/ moderately No personality disorders]
violent intimate high SDS low SDS 4.72 .58 8.14 .001
relationships
No personality disorders]
moderate SDS 4.84 .54 8.96 .001
No personality disorders]
moderately high SDS 4.82 .53 7.92 .001
No personality disorders]
high SDS 4.93 .53 9.3 .001
Borderlrne personality
disorder/ low SDS 4'61 '52 8'87 '00]
Borderline personality
disorder/ moderate SDS 4'50 '57 7'89 ‘001
Borderline personality
disorder/ moderately high 5135 4'30 '57 8'4 '0‘"
Borderline personality
disorder/high SDS 4.79 .59 8.11 .001
Antisocial personality
disorder/ low SDS 4'83 '60 8'05 '00]
Antrsocral personality
disorder/ moderate SDS 5'07 '59 8'59 ’00]
Antisocial personamy
disorder/ high SDS 5'0 '74 6'76 '00]
Both personality disorders]
low SDS 4.67 .55 8.49 .001
Both personality (fisordersl
moderate SDS 4.67 .55 8.49 .001
Bothersonalrty drsorfisT
moderately-high SDS 5'0 '74 6'76 '00]
Both personality disorders]
high SDS 5.0 .74 6.76 .001
CTS verbal Hordefline personality
aggression disorder/ low SDS No personality disorders/ 41.94 12.23 3.43 .05
scores low SDS
No personality drsorders/
moderate SDS 47.14 8.9 5.3 .001
No personality disordersF
No personality drsorders/ 55.68 8.301 6.7 .001

 

98

 

Table 6 (continued)
Significant Tukey post-hoc Comparisons on CTS Severe Physical Violence and Verbal

Aggression Scores, and Number of Participant Intimate Violent Relationships

 

 

 

 

 

 

 

Borderlme personality
disorder/ moderately high SDS “'83 l 1'82 3'89 '01
Both personality Borderline personality
disorders/ low sos disorder/ moderately high 3135 42'9 10:97 3'91 '05
Antisocral personality No personafity disorder
disorders/ low SDS high sns ”'97 ”'33 4'21 '0'
Both personality No personality disorders]
disorders/ low SDS moderate SDS 44‘“ 8'37 5’33 '001
1% personality drsorEFersf
moderately high SDS 38.54 7.63 5.05 .001
No personafity disorders]
high SDS 53.16 7.63 6.97 .001
Both personality 1% personality drsordErs7
disorders/ moderate high SDS 39.3 8.31 4.73 .001
SDS

 

 

 

 

 

 

 

 

 

Regression Analysis: Contributions of Personality Disorders, Adult Attachment
Orientations, and Sexism to Male Violence Enacted toward Female Partners

A hierarchical regression was conducted to test the hypothesis that measures of
psychological constructs and the composite variable sexism, a construct central to
sociopolitical theory, each makes independent and significant contributions to the
prediction of male violence perpetrated toward female partners, once social desirability
response bias is controlled. In addition, the hierarchical regression tested the hypothesis
that, after controlling for main effects, interactions representing patriarchal attitudes and
the psychological constructs will also be unique, significant predictors of violence.
Violence was a composite variable created by aggregating the two measures assessing
frequency of violence (e.g., PMWS and CTS overall scores), after standardizing PMWS
and CTS overall scores. Specifically, all CTS items were summed and then standardized.
Next, all PMWS items were summed and then standardized. The two standardized scores
were then summed. SDS scores were entered as a covariate at the first step.
Psychological constructs central to psychological theories explaining violence, that is

measures of anxious and avoidant adult attachment as well as measures of antisocial and

99

measures of anxious and avoidant adult attachment as well as measures of antisocial and
borderline personality disorders, were entered as a block at the second step. The
composite variable sexism was entered at the next step. Four interaction terms
representing interactions between measures of personality disorders and sexism as well as
measures of adult romantic orientations and sexism (i.e., antisocial * sexism; borderline "'
sexism; anxious attachment * sexism; avoidant attachment * sexism) were entered at the
final step. As recommended by Aiken and West (1991), measures of sexism, attachment,
and personality disorders were centered prior to creating the interaction terms and testing
their significance. Data met regression assumptions regarding normality, linearity, and
homoscedasticity.

Table 7 presents the results of these analyses. These results indicated that social
desirability response bias was a unique, significant predictor of the composite variable
violence, explaining 19% of the overall variance (adjusted 32 = .192, p < .001).

Measures of psychological constructs also made significant independent contributions to
the prediction of violence collectively accounting for 16% of the additional overall
variance in the composite variable violence (adjusted AB} = .162, p < .001). However,
while measures of antisocial personality disorder, Q (238) = 2.43, p < .05), borderline
personality disorder, (t (23 8) = 3.75, p < .001), and anxious attachment, Q (238) = 2.26, p
< .05), were significantly and positively predictive of violence, avoidant attachment was
not significant. The subsequent entry of the composite independent variable sexism was
not significant in enhancing the prediction of violencez. Not one of the four interaction
terms entered at the final step were significant predictors of the composite dependent
variable violence. These findings fail to support hypotheses 6 and 7 in that only
constructs central psychological understandings of male violence toward women were

significant in enhancing the prediction of the composite variable violence in this

 

2 Sexism was not a significant predictor of the composite variable violence irrespective if its entry at Step
2, before entering the psychological variables, or Step 3, after entering the psychological variables.

100

regression analysis. Moreover, after controlling for main effects, none of the interaction
terms were significant.

Factor Analysis of Violence Indexes

Because past research supporting batterer typologies (see Holtzworth-Munroe & Stuart,
1994) identified batterer subgroups with distinct patterns of violence, a second- order
exploratory factor analysis was computed. A principal components analysis with varimax
rotation of the item assessing frequency with which participant initiated violence,
subscale assessing frequency with which participant escalated violence (violence
escalation), subscale assessing injury experienced by participants’ partners (injury), CTS
verbal, severe, and minor violence subscales, and PMWS emotional/verbal and
dominance/isolation subscales yielded three major factors with eigenvalues greater than
1.0. These 3 factors collectively accounted for 67.3% of the shared variance among all
subscales or items assessing violence that were included in the factor analysis. The first,
second, and third factors independently accounted for 42.9%, 13.19%, and 11.21% of the
total variance, respectively. Table 8 presents the loadings of each of the measures or
items assessing violence on the three factors. The CTS verbal aggression and PMWS
emotional/verbal abuse and dominance/isolation subscales emerged as the variables most
representative of the first factor. However, the CTS minor physical violence subscale
also loaded on the first factor, suggesting that this variable is also important in the
interpretation of the first factor. The injury and violence escalation subscales as well as
the item assessing participants’ responsibility in initiating violence emerged as the
variables most representative of the second factor. The CTS minor and severe physical
violence subscales also demonstrated moderate loadings on the second factor. Finally,
the CTS severe and minor physical violence subscales and the generality of violence

subscale emerged as the variables most representative of the third factor.

101

Table 7

Summary of Hierarchical Regression Analysis for Variables Predicting Composite

 

Variable Violence (N = 239)

 

 

 

Variable __B_ _SE E 13

Step 1

social desirability -7.68E-02 .010 -.439**
Step 2

anxious attachment .200 .088 .147*

avoidant attachment -3.15E-02 .094 -.018

borderline personality .235 .063 .279"

antisocial personality .150 .062 .163*
Step 3

sexism -.1 10 .060 -.102
Step 4

sexism * anxious attachment 1.216E-02 .055 .013

sexism * avoidant attachment 1.910E-02 .070 .018

sexism * borderline personality 2.430E-02 .038 .047

sexism * antisocial personality -5.49E-03 .034 -.011

 

Note: Adjusted R} = .192 for Step 1 (p < .001); A Adjusted £3 = .162 for Step 2 (p <

.001); A Adjusted RE = .009 for Step 3 (ns); A Adjusted R_2 = .003 for Step 4 (ns).

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

102

Table 8

Rotated Factor Matrix and Factor Loadings of Subscales and Item Assessing Violence

 

 

 

Factor
Measure I II III
PMWS: emotional/verbal abuse .884
CTS: verbal aggression subscale .790
PMWS: dominance/isolation .763
Violence escalation .832
Injury .781
How often did you hit, slap, or push first? .754
CTS: minor physical violence .535 .407 .488
CTS: severe physical violence .331 .489
Generality of Violence .874
Eigenvalues 3.86 1.19 1.01
%Variance 42.9% 13.2% 1 1.2%

 

Given the pattern of factor loadings, there appear to be three distinct patterns of
violence. The first pattern of violence includes verbal and emotional abuse that is high in
frequency, as well as acts of minor physical violence that are moderate in frequency. The
second pattern of violence is one in which the participant is responsible for initiating as
well as escalating the violence, and one that can result in the partner’s injury. Although
this second pattern of violence does not include emotional or verbal aggression, acts of
minor physical violence do contribute to this factor’s variance. The third pattern of
violence includes predominantly physical violence, both minor and severe, toward his

partner, as well as physical violence toward non-partners (e.g., family, strangers, friends,

103

and police). The results of this factor analysis lend support to hypothesis 8 in that three
patterns of violence emerged, and these patterns of violence were reflective of the
anticipated patterns. The first factor appears to be tapping partner intimidation through
the use of verbal and psychological abuse, as well as minor physical violence, indicating
a violence pattern wherein participants may be purposefully using violence as a means 'to
intimidate and control their partner. This factor that will be labeled “partner-intimidation
violence” is consistent with the hypothesis that a pattern of violence predominantly
psychological and verbal in nature and targeting participant’s partner would emerge.
Subscale loadings on the second factor indicated a pattern in which participants generally
initiated the violence and engaged in escalating aggression that included both minor and
severe physical violence. Emotional and verbal abuse did not load on this factor.
Moreover, because this violence pattern is partner specific, the second factor will be
labeled “partner-specific physical violence”. This factor is not consistent with
hypotheses regarding anticipated violence patterns present in the data. Factor three
appears to be assessing a more generalized form of aggression with acts of violence that
are more severe than the first two violence patterns and do not specifically target partners.
This final factor that will be labeled “general violence” is consistent with the hypothesis
that the data would suggest a pattern of severe physical violence that includes persons
other than participant’s partner.
Regression Analyses: Differential Contributions of Personality Disorders, Adult
Attachment Orientations, and Sexism to Three Patterns of Violence

Given that the data suggested multiple patterns of violence, a series of three
additional hierarchical regression analyses were conducted. For each of the three
hierarchical regressions, a different dimension of violence was used as the dependent
variable. Results of these analyses indicated whether attachment orientations, sexist
attitudes toward women, and borderline and antisocial personality disorders were

differentially significant predictors of the distinct violence patterns that emerged from the

104

exploratory factor analysis. Preliminary data analyses prior to running the three
hierarchical regressions included computing correlation coefficients between violence
patterns and key predictors as well as exploring personality disorder group differences on
the factor scores.

Correlations among three violence patterns and key predictors. Interrelationships
between the three patterns of violence and the key predictors were computed to explore
bivariate relationships among each of the predictors and patterns of violence. Results of
this analysis are presented in Table 9. All three patterns of violence are positively related
with antisocial personality disorder scores. Unexpectedly, the relationship between
partner-intimidation violence and antisocial personality disorder scores appears to be
Table 9

Intercorrelations between Violence Patterns and KeyPredictors

 

 

1. 2. 3. 4. 5. 6. 7. 8. 9.
1. Partner-
intirnidation 1.00 .00 .00 .41" .50" .10 .42" -.42" .01
violence
2. Partner-specific 1.00 .00 .21" .21" .19“ .24" -.13* .22"
physical violence
3. General violence 1.00 .22" .24" .17* .05 -.04 .27”
4. Antisocial
Personality 1.00 .57" .23" .40” -.49"”" .19‘
disorder scores
5. Borderline
personality 1.00 .26" .56" -.51" .26”
disorder scores
6. Avoidant
Attachment 1.00 .19“ -. 13 .27" ‘
7. Anxious
Attachment 1.00 -.40** .19"
8. Social
Desirability 1 .00 -. 16
9. Sexism 1.0

 

Note: * p < .01; *" p < .001

105

stronger than the relationship between general violence and antisocial personality
disorder scores. Given the very strong relationships between antisocial personality
disorder scores and social desirability response bias as well between antisocial
personality disorder scores and borderline personality disorder scores, relationships
among the three violence patterns and antisocial personality disorder scores, controlling
for social desirability and borderline personality disorder scores, were computed. Partial
correlation coefficients indicated that the relationship between general violence and
antisocial personality disorder scores remained significant, while relationships between
antisocial personality disorder scores and the other two violence patterns were no longer
significant (see Table 10).

All three patterns of violence were also positively correlated with borderline
personality disorder scores, with partner-intimidation violence and borderline personality
disorder scores demonstrating the strongest relationship. While partner-intimidation
violence and partner-specific physical violence are positively correlated with anxious
attachment, partner-specific physical violence and general violence are positively
correlated with avoidant attachment. Only partner-specific physical violence and general
violence were positively correlated with the composite variable sexism. Both partner-
intimidation and partner-specific physical violence patterns were negatively correlated
with social desirability response bias.

Personality disorder group comparisons on factor scores. A series of three one-
way AN OVAs were computed to assess personality disorder group differences on factor
scores. Results from the first ANOVA revealed significant group differences on partner-
intimidation violence, E (3, 235) = 25.43, p 5 .001. Scheffe’s post-hoe comparisons
indicated that participants with no personality disorder scored significantly lower than
participants categorized as having borderline personality disorder, Q (23 5) = 4.93, p 5
.001), antisocial personality disorder, Q (235) = 3.16, p 5 .05), or both personality
disorders, Q (235) = 8.13, p 5 .001). Results from the second ANOVA revealed

106

Table 10
Intercorrelations between Violence Patterns and Antisocial Personality Disorder Scores

Controlling for Social Desirability and Borderline Personality Disorder Scores

 

 

 

l 2 3. 4

l. Antisocial personality 1.00 .12 .1 1 .13 "
disorder scores

2. Partner-intimidation 1.00 -. 13 * -. 12
violence

3. Partner-specific 1 .00 -05
violence

4. General violence 1.00

Note: *9 5 .05

significant group differences on partner-specific physical violence, E (3, 235) = 3.07, p 5
.05. However, Scheffe’s post-hoe comparisons indicated that specific between group
differences did not reach statistical significance, after adjusting alpha levels to control for
Type I error. Results from the third ANOVA revealed significant group differences on the
general violence pattern, E (3, 235) = 3.8, p 5 .01. Scheffe’s post-hoe comparisons
indicated that participants with no personality disorder scored significantly lower than
participants categorized as having both personality disorders, 1 (235) = 3.08, p 5 .05 (see
Table 11).

Regression analyses. For each of the following regressions, SDS scores were

 

entered as a covariate at the first step. Psychological constructs central to psychological
theories explaining violence (i.e., anxious and avoidant adult attachment as well as
measures of antisocial and borderline personality disorders) were entered as a block at the
second step. The composite variable sexism was entered at the next step, and the four
interaction terms were entered at the final step. Data met regression assumptions

regarding normality, linearity, and homoscedasticity for each of the three regressions.

107

Table 11

Mean Scores and Standard Deviations for Four Personality Disorder Groups on Three

Violence Patterns

 

 

Partner- Partner-specific General Violence
Intimidation Physical Violence
Group E Violence
M .31.). _M. 5.12 M. §2
No Personality Disorder 132 -.42 .80 -.17 .78 -.17 .67
Borderline Personality
Disorder 45 .32 .98 .24 1.09 . 19 1.17
Antisocial Personality 17 .29 1.34 .26 1 .03 -.06 .70
Disorder
45 .80 .77 .17 1.35 .34 1.5

Both Personality Disorders

 

 

 

 

Results of the first regression using partner-intimidation violence as the dependent

variable are presented in Table 12. These results indicated that social desirability

response bias was a unique, significant predictor of this violence pattern, explaining 18%

of the overall variance (adjusted _R_2 = .179, p < .05). Social desirability was negatively

related to the dependent variable, Q (23 8) = -7.18, p < .05). Measures of psychological

constructs also made significant independent contributions to the prediction of the first

pattern of violence collectively accounting for 14% of the additional overall variance in

this pattern of violence (adjusted A R} = .135, p < .05). However, only anxious

attachment,Q (23 8) = 2.58, p < .05), and borderline personality disorder, Q (238) = 3.21, p

< .05), were significantly and positively predictive of partner-intimidation violence. The

composite independent variable sexism entered at the third step was also a significant

predictor of partner-intimidation violence (adjusted A_R2 = .085, p < .05), explaining an

additional 9% of the overall variance. However, the relationship between sexism and the

dependent variable was negative, Q (23 8) = - 2.53, p < .05). The subsequent entry of the

four interaction terms at the final step was not significant in enhancing the prediction of

the first pattern of violence.

108

Table 12

Regression of Personality Disorders, Attachment Orientations, and Sexism in Predicting

Partner-Intimidation Violence 01 = 239)

 

 

 

Variable E S_E l_3_ 13
Step 1

social desirability -4.23E-02 .006 -.423*
Step 2

anxious attachment .134 .052 .178”

avoidant attachment -4.72E-02 .055 -.048

borderline personality .119 .037 .246*

antisocial personality 6.621E-02 .036 .126
Step 3

sexism -8.86E-02 .035 —.144*
Step 4

sexism * anxious attachment -2.77E-02 .032 -.052

sexism * avoidant attachment 3.278E-02 .041 .053

sexism * borderline personality 6.3 1013-02 .022 .022

sexism * antisocial personality 1.542E-03 .020 .005

 

Note: Adjusted _RE = .179 for Step 1 (p 5 .05); A Adjusted IE = .132 for Step 2 (p 5 .05);

A Adjusted 33 = .018 for Step 3 (p 5 .05); A Adjusted RE = .005 for Step 4 (ns).

*p5.05.

109

Results of the second regression with partner-specific physical violence as the
dependent variable are presented in Table 13. These results indicated that social
desirability response bias was a unique, significant predictor of scores on this violence
dimension, explaining 2% of the overall variance (adjusted R2 = .017, p < .05). Once
again, social desirability was negatively related to the dependent variable, (t (23 8) = -
2.05, p < .05). Measures of psychological constructs also made significant independent
contributions to the prediction of the first pattern of violence collectively accounting for
7% of the additional overall variance in the dependent variable (adjusted A R2 = .073, p <
.05). However, only measures of anxious, (t (23 8) = 2.10, p < .05), and avoidant, (t (238)
= 1.96, p < .05), attachment were significantly and positively predictive of partner-
specific physical violence. The subsequent entry of the composite independent variable
sexism at the third step also made a significant independent contribution to the second
pattern of violence, accounting for an additional 2% of the overall remaining variance in
the dependent variable (adjusted A R2 = .017, p < .05). The four interaction terms entered
at the final step also significantly contributed to variation in the dependent variable,
accounting for an additional 4% of the remaining variance (adjusted A R2 = .042, p <.05).
However, only the interaction of anxious attachment and sexism was the only variable
accounting for explained variance at this step, (t (23 8) = 3.31, p < .05).

The significant interaction effect was plotted as advised by Aiken and West
(1991) to help interpret the interaction (see Figure 4). Whereas the relationship between
partner-specific physical violence and anxious attachment is linear and positive for
participants reporting moderate sexism, this is not true for participants reporting low or
high sexism. For participants with low sexism scores, the relationship between partner-
specific physical violence and adult anxious attachment orientations is positive for those
participants reporting low to moderate ECR adult anxious attachment scores. However,
for those participants reporting moderate to high ECR adult anxious attachment

orientation scores, the relationship between partner-specific physical violence and adult

110

Table 13
Regression of Personality DisorderngAttachment Orientations, and Sexism in Predicting

Partner — Specific Physical Violence (N = 239)

 

 

Variable E §_E_ E 13
Step 1

social desirability -1.32E-02 .006 -.132*
Step 2

anxious attachment .126 .060 .161*

avoidant attachment .125 .064 .128“

borderline personality 1.83 5E-02 .043 .038

antisocial personality 5.146E-02 .042 .1 03
Step 3

sexism 8.584E-02 .041 .139“
Step 4

sexism * anxious attachment .114 .036 .214*

sexism * avoidant attachment 2.195E-03 .047 .004

sexism * borderline personality 5.963E-04 .025 .002

sexism * antisocial personality -5.18E-03 .023 -.018

 

Note: Adjusted 133 = .017 for Step 1 (p 5 .05); A Adjusted R_2 = .073 for Step 2 (p 5 .05);
A Adjusted _R_2 = .017 for Step 3 (p 5 .05); A Adjusted 33 = .042 for Step 4 (p 5 .05).
* p 5 .05

111

anxious attachment orientations is diminishing or negative. For participants reporting
high sexism, the relationship between partner-specific physical violence and adult
anxious attachment orientations appears to be enhanced for those participants reporting
moderate to high ECR adult anxious attachment orientation scores. As advised by Aiken
and West (1991), t-tests to assess whether the slopes representing the anxious attachment
orientation-partner—specific relationship were different from zero at high, moderate, and
low sexism were conducted. Slopes were significantly different from zero when sexism
was moderate, (t (237) =2.30, p < .025), or high, Q (237) =2.99, p < .005), but not when
sexism was low. This finding suggests a significant relationship between partner-specific
violence and adult anxious attachment scores at sexism equal to high and moderate

values, but not low values.

Figure 4: Interaction between Sexism and Adult
Anxious Attachment Orientations in Predicting
Partner-specific Physical Violence

 

+ Low Sexism ’ /
— .g _ Moderate Sexism ,/

_._m@&mm '

Partner-specific
Physical Violence

 

 

 

 

Anxious Attachment
Low Moderate High

112

For further interpretation of the significant interaction term, a new variable with
nine levels (i.e., low, moderate, high sexism x low, moderate, high ECR adult anxious
attachment orientation scores) was created, and Tukey’s post-hoe comparisons were
conducted to explore all possible between group differences. Results supported the visual
interpretation of the interaction effect. There were significant between-group differences
on partner-specific physical violence, E (8, 230) = 4.23, p < .001. In general, participants
with high ECR adult anxious attachment orientation scores and high scores on sexism
scored significantly higher than the other groups (see Table 14).

Table 14

Significant Tukey post-hoe Comparisons on Partner-specific Physical Violence

 

 

 

Reference
Group Comparison Group Mean §E t 9
Difference
High anxious low anxious attachment/
lattachment/ high sexism low sexism 1'19 “269 4’42 '00]
1°“ mm“ amcme’m’ 1.31 .232 4.65 .001

moderate sexism
modemtelgr‘ryxgqisisagtachment/ 1.08 .275 3.93 .01
massagesrthmm” 264 -°°'
h1g5 31:13:: :33“me 1.24 .322 3.85 .01

 

 

 

 

 

 

 

 

 

 

 

Results of the third and final regression, using the violence pattern labeled general
violence as the dependent variable, are presented in Table 15. These results indicated that
social desirability response bias was not a unique, significant predictor of this violence
pattern. However, measures of psychological constructs entered at the second step
(adjusted A 3’ = .10, p < .05) and the composite variable sexism entered at the third step
(adjusted A E” = .04, p < .05) did significantly and positively enhance the prediction of
the third pattern of violence. Psychological constructs entered as a block at the second

step collectively accounted for 10% of the overall variance in the third pattern of

113

violence, whereas the composite variable sexism explained an additional 4% of the
overall remaining variance. Of the 4 psychological predictors of violence entered at the
second step, both PDQ-R antisocial violence, Q (23 8) = 1.98, p < .05), and borderline
violence, Q (23 8) = 3.01 , p < .02), personality disorder scores were significantly and
positively predictive of the third pattern of violence. Entry of the four interaction terms at
the final step did not significantly enhance the prediction of the third pattern of violence.
Overall, findings from the series of regressions using the three patterns of
violence that emerged from the factor analysis as outcomes lend partial support to
hypothesis 9. Specifically, hypothesis 9a was supported in that borderline personality
disorder, anxious attachment, and sexism proved to be independently and significantly
positive predictors of partner-intimidation violence. It is difficult to discern whether
hypothesis 9b was met, given the inconsistency in the anticipated second violence pattern
and the actual pattern reflected in the data. Nonetheless, both adult attachment
orientations and sexism, as well as the interaction between anxious attachment and
sexism, were significant and positive predictors of partner-specific physical violence.
Hypothesis 9c was also partially supported in that antisocial personality disorder scores
and sexism were independently and significantly positive predictors of general violence.
However, unexpectedly, avoidant attachment was not significant and borderline

personality disorder scores were significant.

114

Table 15

Regression of Personality Disorders, Attachment Orientations, and Sexism in Predicting

General Violence (N = 239)

 

 

Variable E _S_E E 13
Step 1

social desirability -4.37E-03 .006 -.044
Swp2

anxious attachment -.105 .060 -.134

avoidant attachment 9.916E-02 .063 .102

borderline personality .127 .042 .264*

antisocial personality 8.232E—02 .042 .157"'
Step 3

sexism .128 .040 208*
Step 4

sexism * anxious attachment -2.04E-02 .036 -.038

sexism * avoidant attachment 3.085E-02 .047 .050

sexism * borderline personality 2.617E-02 .025 .089

sexism * antisocial personality 6.869E-03 .022 .024

 

Note: Adjusted 33 = .002 for Step 1 (ns); A Adjusted R_2 = .10 for Step 2 (p 5 .05); A

Adjusted E3 = .038 for Step 3 (p 5 .05); A Adjusted 133 = .014 for Step 4 (ns).
p 5 .05

115

DISCUSSION
Overview

This chapter summarizes and interprets this study’s key findings. Ways in which
this study’s findings advance inquiries regarding causes of heterosexual relationship
violence against women are then highlighted. Where data did not support study
hypotheses, possible explanations drawing upon extant research and theory are offered.
In addition, implications of this study’s findings for intervention as well as policy are
addressed. Lastly, study limitations and recommendations for future research are
discussed.

The purpose of this study was premised on the assumption that multiple and
diverse theories are needed to understand the etiology of male violence perpetrated
against female partners. This study specifically examined whether a model integrating
constructs derived from attachment theory and feminist perspectives, compiled with
constructs from general personality theories made important and unique contributions to
explanations of male intimate violence that could enhance current understandings of
battering. Multidimensional theories of battering are gaining increasing support among
practitioners as well as researchers in the field of violence, predominantly due to
limitations of unidimensional theories in explaining multiple patterns of violence
evidenced among the heterogeneous population of men who batter their female intimates.

First, this study tested empirically or theoretically derived hypotheses regarding
relationships among anxious and avoidant adult attachment orientations, antisocial and
borderline personality disorders, sexist attitudes toward women, and frequency and
severity of psychological, verbal, and physical violence. Next, a regression model
indicating that constructs central to both feminist (e.g., sexist attitudes against women)
and psychological theories (e.g., adult attachment orientation; borderline and antisocial
personality disorders) are unique, significant predictors of violence among men attending

court-referred batterer intervention programs was tested. Consistent with the perspective

116

that men who are violent exhibit multiple, distinct patterns of violence, it was also
hypothesized that the data would reveal multiple violence patterns that were distinctive in
frequency and severity of psychological, verbal, and physical violence. Moreover, it was
anticipated that adult attachment orientations, personality disorders, sexist attitudes, and
interaction terms included in the model would be differentially predictive of the distinct
patterns of violence. These hypotheses were motivated by research demonstrating that
there are subtypes of batterers each with diverse behavioral, cognitive, and emotional
characteristics as well as with violence patterns distinct in frequency and severity of
psychological, verbal, and physical violence. The heuristic value of different theories
may depend upon batterer subtype, and understandings of the unique developmental
characteristics of batterer subgroups may be an important step toward the construction of
multidimensional etiologic models of battering as well as differential treatment
interventions.

Correlational Analyses

 

Although borderline personality disorder scores were positively correlated with
anxious attachment after controlling for antisocial personality disorder scores, antisocial
personality disorder scores were unrelated to avoidant attachment after controlling for
borderline personality disorder scores. Moreover, whereas sexist attitudes about women
were correlated with borderline personality disorder as anticipated, sexism scores were
unexpectedly unrelated to antisocial personality disorder. As expected, anxious and
avoidant attachment was related to sexism; however, contrary to hypothesis, the
relationship between avoidant attachment and sexism was not significantly greater than
the relationship between anxious attachment and sexism.

Because both antisocial and borderline personality disorders are significantly
related in the same direction with a number of key variables and with each other (see
Table 1), some partial correlations were computed post-hoc to increase understanding of

relationships between both personality disorders and key variables. Afier controlling for

117

borderline personality disorder scores, antisocial personality disorder scores were only
positively and significantly related to minor physical violence scores and number of
general arrests. After controlling for antisocial personality disorder scores, borderline
personality disorder scores were positively and significantly related to violence escalation
scores, PTWS emotional/verbal and dominance/isolation subscale scores, CTS verbal
abuse subscale scores, and general violence scores. All of these significant relationships
were consistent with expectations, with the exception of the association between
borderline personality disorder and general violence.

Relationships between attachment orientations and other variables entered into the
correlation matrix were anticipated. For example, participants scoring high on avoidant
attachment also reported a greater frequency of general violence towards persons other
than their partner, and participants with high scores on anxious attachment also indicated
a greater frequency of verbal and psychological violence. However, the expectation that
avoidant attachment would be related to severe physical violence was not substantiated.
Finally, although both avoidant and anxious attachment were significantly correlated with
participant responsibility for escalating violence, only anxious attachment was related
with participant responsibility for initiating the violence.

Group Comparisons of Personality Disorder Grorygs on Violence and Attachment

Group comparisons between the four personality disorder groups demonstrated
that batterers meeting criteria for borderline personality disorder did not score
significantly higher than batterers meeting criteria for antisocial personality disorder on
adult anxious attachment orientations, as expected. Additionally, participants meeting
criteria for antisocial personality disorder did not score significantly higher than
borderline batterers on adult avoidant attachment orientations.

This unsubstantiated hypothesis may be due to the possibility that some
borderline and/or antisocial batterers in the sample have fearful attachment, thus their

attachment style reflects both anxious and avoidant attachment. The finding that

118

borderline batterers scored significantly higher on avoidant attachment than participants
not meeting criteria for any personality disorder supports this conjecture. Participants
meeting criteria for both personality disorders also scored significantly higher than
participants not meeting criteria for any personality disorder on anxious attachment.
Moreover, Tweed and Dutton’s (1998) findings suggested that some antisocial batterers
may have preoccupied attachment. Although borderline batterers reported more anxious
attachment than antisocial batterers, this difference may not be significant because some
antisocial batterers may also have anxious attachment (i.e., fearful or preoccupied).

It is also noteworthy that personality disorders were associated with anxious and
avoidant attachment orientations. This is consistent with research demonstrating a
positive relationship between personality disorders and insecure attachment and
consistent with the low prevalence of insecure attachment among batterers indicating no
pathology (Holtzworth-Munore et al., 2000). F ifty-five percent of this study’s sample
indicated no personality disorder, and this group was significantly less anxious and less
avoidant than the other groups, after controlling for social desirability.

Group comparisons among the four personality disorder groups on violence-
related variables revealed that many anticipated group differences were not substantiated.
One possibility for the absence of significant differences is that the number of
participants in each of the personality disorder groups was unequal and/or not large
enough to discern group differences. Given the small and unequal group sizes, it is
possible that tests exploring group differences did not have adequate statistical power to
detect these differences.

One observed group difference was that participants not meeting criteria for any
personality disorder scored significantly lower on a measure of emotional/verbal abuse
than all other groups, indicating an association between personality disorders and
emotional/verbal abuse. Moreover, participants reporting low social desirability scored

significantly higher on a measure of emotional/verbal abuse than participants reporting

119

moderate or high social desirability, supporting the expectation that batterers will
minimize abuse perpetrated in an effort to present themselves in a desirable manner.
Additionally, participants meeting criteria for both personality disorders and reporting
moderate social desirability scored significantly higher on severe physical violence than
participants meeting criteria for no personality disorder and reporting moderately high
social desirability. Antisocial batterers reporting moderately high social desirability
indicated significantly more violent intimate relationships than all other groups. This
finding should be interpreted with caution due to the small and unequal cell sizes.
Nonetheless, as indicated by their moderately high scores on social desirability, antisocial
participants may want to present themselves desirably, but they do not perceive the
presence of violence in their intimate relationships as a threat to their positive self-image.

Participants categorized as borderline batterers and reporting low social
desirability scored significantly higher on verbal aggression than participants not meeting
criteria for any personality disorder, irrespective of their social desirability scores.
Finally, participants meeting criteria for both personality disorders and reporting low
social desirability scored significantly higher on verbal aggression than borderline
batterers with moderately high social desirability and batterers not meeting criteria for
any personality disorder, irrespective of social desirability scores. With the exception of
antisocial batterers, participants meeting criteria for personality disorders who also report
low social desirability indicate greater levels of verbal abuse.

In general, results of group comparisons did not support this study’s hypotheses
regarding anticipated differences between antisocial batterers and borderline batterers. It
was expected that the borderline group would report significantly more psychological
violence than the antisocial group and that the antisocial group would report significantly
more physical violence, particularly severe physical violence, than the borderline group.
It was also expected that the antisocial group would indicate more violence that was not

partner-specific when compared with the borderline group. However, it is important to

120

note that findings offer partial support for expected relationships between social
desirability and self-reports of perpetrated abuse as well as partial support for expected
relationships between personality disorders and perpetrated abuse.

A plausible explanation for the small number of participants meeting criteria for
antisocial or borderline personality disorder is that the PDQ-R may have erroneously
classified participants as having both antisocial and borderline personality disorders. A
limitation of the PDQ-R is that, due to its low predictive positive power, false positives
are a possibility (Hyler et al., 1990; Hyler et al., 1992). That is, research has shown that
the PDQ-R may indicate a diagnosis when no personality disorder is present. Given that
it is characteristic of the batterers to fake good, false positives were not expected to
plague this study’s findings. However, the classification of participants as both antisocial
and borderline personality disorders may be attributable to the PDQ-R’s poor
discriminative validity. Moreover, the finding that participants indicating no personality
disorder scored higher on social desirability than the other three personality disorder
groups suggests that participants in this sample who did not meet criteria for any
personality disorder may have responded more defensively. As such, antisocial and
borderline personality disorders may in fact be more prevalent among participants
classified with no personality disorder than PDQ-R scores suggest.

Adult Attachment, Personality Disorders, and Sexist Attitudes against Women in
Predicting Male Relationship Violence against Female Partners

The data did not support a model integrating constructs from psychological and
feminist theories. Afier controlling for social desirability, the psychological variables
included in the model significantly enhanced the prediction of the composite variable
violence. Among the psychological variables, all but avoidant attachment made
independent contributions to explained variance. However, sexist attitudes made no
significant contributions in explaining variation in the outcome measure after controlling

for the psychological variables. Given that contributions of sexist attitudes toward the

121

enactment of male violence against female intimate partners are well-documented
(Eisikovits, Edleson, Guttman, & Sela-Amit, 1995, Finn, 1986; Haj-Yahia, 1997), the
absence of support for this hypothesis was unexpected. The absence of a relationship
between sexist attitudes toward women and male violence against intimate female
partners may also be attributable to a restricted range of variance on CTS minor and
severe physical violence subscales. That is, maximum frequencies reported were outliers
and visual distributions of the data suggested that, with the exception of these outliers,
CTS minor and severe physical violence subscales had a restricted range of scores. The
restricted range of scores on these measures may have reduced the ability of the
regression analysis to detect significant effects. Additionally, it may be that sexism is
related to some of the dimensions of violence included in the composite variable, yet
unrelated to others. Once violence measures were aggregated, the relationship between
the composite variable violence and sexism was no longer significant.

The finding that avoidant attachment was not a significant psychological predictor
warrants comment. This finding is consistent with the body of research examining
relationships between battering and insecure attachment. That is, although findings
emerging from this line of inquiry generally support relationships between anxious
attachment and domestic violence (Dutton & Browning, 1988; Dutton, Ginkel, &
Starzomski, 1995; Dutton et al., 1994; Dutton, Starzomski, & Ryan, 1992; Tweed &
Dutton, 1998), relationships between domestic violence and avoidant attachment were
theorized but generally unsupported. Although attachment theory suggests that persons
with dismissing attachment may exhibit a more general hostile manner of relating (Troy
& Sroufe, 1987), there is no empirical support suggesting that such hostile relating is
partner-specific. Moreover, although attachment theory conjectured that relationship
dynamics common to avoidantly attached persons (e.g., fear of engulfrnent) may lead
such persons to commit violence against a partner seeking intimacy (Mayseless, 1991),

this expectation has not yet been empirically supported.

122

Violence Dimensions

 

Results of the factor analysis supported the hypothesis that there are multiple
patterns of violence among men who have been arrested for domestic violence and court-
mandated to a batterer intervention program. As hypothesized, three distinct patterns of
violence emerged. Whereas the general violence pattern and the partner-intimidation
pattern were consistent with this study’s hypotheses, the partner-specific physical
violence pattern was not. It was anticipated that there would be two dimensions of
violence that would be predominantly partner-specific. However, it was expected that
this second partner-specific violence pattern would include minimal or no physical
violence with moderate amounts of verbal and psychological violence and might include
some violence directed at family members other than his partner. However, results
indicated that the second partner-specific violence dimension included both minor and
severe physical violence that was almost equal in frequency to the general violence
pattern. In addition, the factor loading for the variable assessing violence directed at
persons other than his partner was close to zero, suggesting that this violence is partner-
specific and does not include other family members. Partner injury and violence that is
escalating and initiated by the batterer is also characteristic of this violence dimension.

The general violence pattern that was observed in this study appears to be
consistent with the generally violent or antisocial batterer subtype that has been identified
in batterer typology research. The partner-intimidation pattern appears to describe the
violence typical of the borderline/dysphoric batterer. The third hypothesized violence
pattern was expected to be enacted by the family-only batterer subtype (Holtzworth-
Munore & Stuart, 1994; Holtzworth et al., 2000). The third pattern that emerged,
however, does not describe the violent behavior typical of the family-only batterer.

It is probable that men exhibiting partner-specific physical violence were
psychologically and verbally violent in earlier stages of the relationship, and at the time

of their arrest, the violence had escalated to acts of physical violence, often resulting in

123

injury to the woman. As with partner-intimidation violence, this pattern may be
instrumental in that the objective of the violence is to maintain power and control over
the victim. When acts of psychological violence are no longer effective in meeting the
perpetrator’s objective, physical violence is used to yield power over the victim. .

Parallels between the partner-specific physical violence and the partner-
intimidation violence patterns suggest that that these two violence patterns may be
subsets of behaviors on a continuum of male relationship violence against women. It
may be that the two distinct patterns do not suggest distinct subtypes of men, each
enacting a different pattern of violence, but distinct phases of battering through which
most batterers progress. Another possibility is that a third, less severe violence pattern did
not emerge from our data because the sample was comprised of men recruited from court-
mandated batterer intervention programs. If the present sample had included men
recruited from the community as well as from batterer intervention programs, a third less
severe violence pattern might have emerged, given the expectation that community
recruits would evidence less pathology and less frequent and severe violent behavior.

Accepting the argument that violent behavior is a continuum and that violent
behaviors specific to the subtypes reflect to some extent different stages of the continuum
does not exclude the possibility that there are subtypes of batterers. While the partner-
intimidation violence and partner-specific violence patters of violence observed in this
study may reflect stages of a continuum of partner-specific violence, it may be that an
antisocial batterer subtype enacting violent behaviors consistent with the general violence
pattern observed in this study exists. There may only be two distinct violence patterns -
generally violent and partner-specific-physical violence and the third pattern may be a
phase of violence within the partner-specific violence pattern. This conceptualization is
supported by studies that have found only two distinct batterer subtypes (Gottman et al.,
1995; Tweed & Dutton, 1998), with one subtype described as generally violent antisocial

batterer and the second subtype described as a dysphoric/borderline whose violence is

124

partner-specific. In addition, Holtzworth et al. (2000) proposed that batterers could be
classified along two dimensions - antisocial and borderline tendencies and that
classifying batterers along these dimensions may be a useful way to categorize batterers.
Longitudinal research exploring associations between length of intimate relationship and
frequency as well as severity of physical and psychological violence could help answer
questions about the validity of batterer subtypes. If such research demonstrated more
severe violence and increased injury to the partner with the progression of the
relationship, the hypothesis that batterer subtypes reflect a continuum of behaviors would
be supported.

Gondolf (1988) questioned whether research identifying batterer subtypes
provided support for distinct subtypes with different causal explanations for their violent
behavior or whether these findings only provide support for battering as a continuum of
violent behaviors “ . . .motivated by escalating male control and degradation of women”
(p. 198). To some extent, this study provides some direction in answering this question.
The series of hierarchical regressions conducted to explore differential contributions of
attachment, personality disorders, and sexist attitudes in predicting each of three violence
patterns observed in this study provides information as to whether different factors
motivate the different violent behaviors representative of the three patterns.

Anxious attachment, borderline personality disorder scores, and sexist attitudes
toward women were significant predictors of partner-intimidation violence. However,
whereas anxious attachment and borderline personality disorder scores were positively
related to partner-intimidation violence, the relationship between this pattern of violence
and sexist attitudes was negative. The partner-intimidation violence pattern observed in
this study is reflective of the violence exhibited by the borderline/dysphoric batterer, who
is generally described as anxiously attached, jealous and possessive of his partner, and
emotionally dependent upon his partner (Hamberger et al. 1996; Holtzworth-Munroe et

al., 2000). Research and theory suggest that this batterer’s violence may be motivated by

125

fear of abandonment (Holtzworth et al.; Mayseless, 1991). The borderline batterer also
reports characteristics consistent with borderline personality diagnosis (Gottman et al.,
1995; Holtzworth et al.; Tweed & Dutton, 1998). To summarize, the predictors generally
descriptive of the borderline batterer were significant in explaining variation in the
partner-intimidation violence pattern, and the borderline batterer’s violence and this
violence pattern are comparable. Moreover, as group comparisons on factor scores
suggested, participants categorized as borderline had higher scores on partner-
intimidation violence, when compared with participants categorized as antisocial
personality disorder or with no personality disorder3. As such, it is conceivable that the
partner-intimidation violence pattern is attributable to the subset of borderline batterers in
this sample.

One inconsistency between current findings and previous research is that the latter
has demonstrated that borderline batterers generally express hostile attitudes toward
women, whereas this study found that sexist attitudes toward women were negatively
associated with the partner-intimidation violence pattern. However, this finding may be
related to the anxious attachment of the subgroup of batterers enacting this violence
pattern. That is, the four-group taxonomy of adult attachment styles proposed by
Bartholomew and Horowitz (1991) suggests that the adult with a preoccupied attachment
style is comparable to the anxious-ambivalent pattern initially identified by Ainsworth et
a1. (1978). Moreover, the adult with preoccupied attachment has internalized a positive
model of other and a negative model of self. The negative relationship between sexist
attitudes toward women and partner-intimidation violence may be due to the batterer’s
negative concept of self and positive internalized model of his partner. Given this
interpersonal dynamic and the expectation that this batterer’s violence is motivated by a

desire to maintain his partner in close proximity (Dutton et al., 1994; Gottrnan et al.,

 

3 Although borderline batterers did score higher on this factor in comparison to participants categorized
with antisocial personality disorder or with no personality disorder, these mean differences did not reach
statistical significance.

126

1995; Mayseless, 1991), he may not support sexist attitudes toward women, if doing so
incurs the threat of abandonment.

For partner-specific physical violence, only anxious and avoidant attachment
orientations were significant psychological predictors. Sexist attitudes also significantly
enhanced the prediction of partner-specific physical violence. Moreover, the interaction
between sexist attitudes and anxious attachment was also significant. This significant
interaction suggests that etiologic models of battering that integrate constructs from
psychological theories and feminist theory may be necessary to advance understandings
of at least some patterns of male violence against female partners. As suggested by this
finding, it appears that anxiously attached men who concurrently have very sexist
attitudes about women are the most likely to enact relationship violence.

Typology research has generally supported the finding that one of the batterer
subtypes, the family-only batterer, indicates less psychopathology than the other subtypes
(Hamberger et al., 1996; Holtzworth-Munroe & Stuart, 1994; Holtzworth-Munroe et al.,
2000). This study’s findings are consistent with typology research in that neither
antisocial nor borderline personality disorder scores significantly enhanced the prediction
of the partner-specific physical violence when adult attachment orientations were
controlled. However, this body of research also suggests that the family only batterer
exhibited the least severe violent behavior against his partner. In this study, the pattern
that was not predicted by personality disorder scores did include severe physical violence
as well as minor physical violence. This discrepancy suggests that this partner-specific
violence pattern is not the constellation of behaviors typical of the family-only batterer.

The finding that neither personality disorder significantly predicted variance in the
partner-specific physical violence pattern sheds doubt on whether the partner-intimidation
and the partner-specific physical violence patterns are two distinct phases of one common
violence pattern. Personality disorders are stable characteristics that are not subject to

change (Diagnostic and Statistical Manual of Mental Disorders -DSM IV, 1994) without

127

many years of intensive psychotherapy (Gunderson, 1984). If it is expected that the same
type of batterer is responsible for both partner-intimidation and partner-specific physical
violence, then personality disorders should have been a significant predictor for both
violence patterns. However, it is important to note that, group comparisons between
personality disorder groups on factor scores suggested that participants categorized as
borderline or antisocial had higher scores than participants with no personality disorder or
categorized as having both personality disorders on partner-specific physical violence,
although these differences were not statistically significant. The small number of
participants meeting criteria for antisocial personality disorder and borderline personality
disorder may have contributed to less than statistically significant differences on group
comparisons. Nonetheless, participants categorized as both personality disorders may be
qualitatively different from participants categorized as either antisocial or borderline
personality disorder, and, if participants with both antisocial and borderline
characteristics were eliminated from the analysis, borderline and antisocial personality
disorder scores may have been significant predictors of partner-specific physical violence.
Additionally, it is conceivable that personality disorder scores were not significant
predictors of partner-specific violence due to limitations of the measure used to assess
personality disorders, and that partner-specific physical violence is a subset of violent
behaviors in a more comprehensive violence pattern. Moreover, the restricted range of
scores on the physical violence measures (i.e., CTS minor and severe physical violence
subscales) may have reduced the power of the regression analysis to determine whether
borderline and antisocial personality disorder scores were statistically significant
predictors of partner-specific physical violence.

Alternatively, the finding that personality disorders were not significantly
predictive of partner-specific physical violence may suggest that this pattern is a truly
distinct pattern. However, the partner-specific physical violence pattern found in this

study may not be reflective of the violence typically enacted by the type that is generally

128

described as the family-only batterer. As suggested by Holtzworth-Munore et al. (2000),
it may be that batterers recruited from court-mandated intervention programs exhibit
greater frequency and severity of violence in comparison to batterers recruited from the
community and that the family only batterer subtype is only present in samples that
include men recruited from both the community and court-mandated programs. Given
the exclusively clinical sample in this study, it is thus not surprising that the violence
pattern typical of the family only batterer was not observed.

If it is assumed that partner-specific physical violence is a distinct violence pattern
that is not associated with the family only batterer, it would be appropriate to make a
prediction regarding what batterer type would be affiliated with this violence. In this
study, these individuals indicated no pathology but did report attachment orientations that
impeded intimacy and relationship security. Furthermore, the absence of personality
disorders suggested that these batterers are more invested in maintaining violent
behaviors because of rewards reaped from this behavior (e. g., control of their partner), as
opposed to attributing inability to desist violent behavior to a long-standing personality
disorder. That is, for these men, violence could be a planned behavior based on a cost-
benefit analysis, and the benefits of the behavior outweigh the potential costs. The
benefit received fi'om the violence may be control over the partner’s proximity.
Consistent with this conjecture is the finding that partner-specific physical violence is
escalated and initiated by the perpetrator, even at the expense of injuring the victim.

Both anxiously and avoidantly attached men in this category of batterers may use
violence to regulate their partner’s proximity. However, the goal of the violence for the
anxiously attached man in this category may be to incite fear should the women try to
establish some independence or disengage fiom the relationship. On the other hand, the
avoidantly attached batterer’s goal may be to keep the partner at a distance that ensures
his protection from potentially painful relationship-related affect. Consistent with

previous research on attachment and defensive mechanisms (see Fuendeling, 1998),

129

violence may be the avoidant batterer’s strategy of regulating anxiety associated with
both internal and interpersonal processes. The significance of both anxious and avoidant
attachment orientations in predicting partner-specific physical violence suggests that the
batterer perpetrating this violence may have a fearful attachment style, which combines
abandonment anxiety with avoidant behavior (Brennan et al., 1998). Batterers with a
fearful attachment style may elicit approach and/or avoidance behaviors from their
partner, dependent upon internal (i.e., affect) or external (i.e., work-related stress) factors.
Such “disorganized” attachment behaviors may contribute to the victim’s sense of
powerlessness in that there is no consistency between violence and the batterer’s
motivations for violence. This speculation could not be explored in the present study due
to the absence of partner data assessing types of partner behaviors (i.e., intimacy
demands, withdrawal) associated with participants’ violence. Future studies should
include partner data in order to test this hypothesis.

Of the cluster of psychological variables, only both personality disorder scores
were predictive of the general violence pattern. Although it was anticipated that
antisocial personality disorder would be a significant predictor, it was not expected that
borderline personality disorder scores would enhance the prediction of this violence
pattern. Perhaps, participants who were not borderline evidenced clinical profiles
indicating otherwise due to the PDQ-R’s tendency to over diagnosis. The finding that
both personality disorders were significant predictors of a generally violent pattern that
encompasses severe violence towards one’s partner might reflect the reality that greater
levels of psychopathology are associated with more severe violence against one’s partner
(Dutton & Starzomski, 1993; F lourny & Wilson, 1991; Hamberger & Hastings, 1988).

Another possibility is that the general violence pattern is attributable to batterers
with antisocial tendencies as well as to batterers with borderline tendencies. Group
comparisons between the four personality disorder groups on factor scores support this

assumption. Specifically, participants categorized as having both personality disorders

130

had higher scores than participants categorized as antisocial and participants categorized
as borderline had a relatively high mean compared with participants with no or antisocial
personality disorder on the general violence pattern, although these differences were not
statistically significant. Nonetheless, the general and severe physical violence enacted by
antisocial batterers or batterers with characteristics from both personality disorders may
differ in intent and motivation compared to that enacted by the borderline batterer.
Specifically, it is plausible that the general and severe violence of the antisocial batterer
or the battterer with both antisocial and borderline characteristics is reflective of
antisocial tendencies, whereas the borderline batterer’s physical and severe violence is
explained by his desperate realization that abandonment is imminent. Given this threat,
the borderline batterer is not concerned that his excessively violent behaviors may isolate
him from his partner or actualize his fears of abandonment. This speculation is consistent
with research findings suggesting that the batterer’s violence escalates as the likelihood
that he will be separated from his victim increases (National Institute of Justice, 2000).
Moreover, the threat of abandonment may trigger negative affect that the borderline
batterer cannot regulate inducing stress that he manifests through a more generalized
hostility.

Sexist attitudes against women were a strong positive predictor of the general
violence pattern. This is consistent with research suggesting that generally violent
antisocial batterers enacting severe violence against their partners indicate stronger
negative attitudes towards women than do other types (see Holtzworth-Munroe & Stuart,
1994). However, for those batterers enacting this more severe violence pattern in
response to fear of imminent abandonment, the strong relationship between negative
attitudes toward women and this violence pattern might be explained by their use of
splitting to manage this desperate fear of abandonment. Consistent with theory and
research on borderline personality, the batterer’s perception of his female partner changes

from positive to negative. Initially, the batterer’s attempts to maintain proximity to his

131

attachment object were motivated by his internalization of his partner as the “all-good
object.” However, should the batterer perceive abandomnent as more of a reality than a
threat, his partner may then be construed as the “bad object” who would have inevitably
disappointed him (Kemberg, 1975; Millon, 1981).

Interestingly, social desirability scores were not a significant predictor of the
general violence pattern. This was contrary to expectations that all batterers minimize
their violent behavior, deny responsibility for their behavior, and fake good when
providing self-report data regarding their battering behaviors (Arias & Beach, 1987).
Assuming that the generally violent pattern corresponds with the violence enacted by the
antisocial generally violent batterer, it may be that this subtype is distinct from other
batterers and that his violence toward female partners is part of an indiscriminate pattern
of violence. If so, then his battering may be one of many antisocial behaviors in his
repertoire of antisocial behaviors. He may not be concerned about minimizing reports of
his behavior and may not be influenced to respond in a socially desirable manner. Given
that research with antisocial personality disorder persons has shown a lack of concern
regarding potential consequences of antisocial behavior (see Diagnostic and Statistical
Manual of Mental Disorders, APA, 1994), the antisocial batterer may not care about the
consequences of his violent behavior. As a result, he may not experience any empathy
for the victim or corresponding expected motivation to change his violent behavior.

CONCLUSIONS

This study did not support the view that psychological (i.e., attachment,
personality disorders) and sociopolitical (i.e., sexism) constructs independently
contributed to the prediction of relationship violence, without accounting for the distinct
violence patterns evidenced by the data. Only three psychological variables - anxious
attachment, and borderline as well as antisocial personality disorder scores were
important in enhancing explanations of battering among this sample of men when

violence was operationalized as a unidimensional construct. However, this study’s

132

findings revealed three distinct dimensions of violence, and when the differential
contributions of psychological and sociopolitical constructs for the distinct dimensions
were explored, findings suggested that both psychological and sociopolitical
understandings of violence are necessary.

The three violence dimensions that were supported by this study’s data are
partner-intimidation violence, partner-specific physical violence, and general violence.
Partner-intimidation violence is partner-specific and includes verbal and emotional abuse
that is high in frequency, as well as acts of minor physical violence (i.e., “shoved her”)
that are” moderate in frequency. Use of this violence pattern may be motivated by a
purposeful intent to intimidate and control one’s partner. The partner-specific physical
violence dimension includes both minor and severe physical violence that is partner-
specific. Partner injury and violence that is escalating and initiated by the male
perpetrator is also characteristic of this violence dimension. The general violence
dimension includes the most severe physical violence and targets the perpetrator’s partner
as well persons other than his partner.

Psychological constructs of adult attachment orientations and personality
disorders as well as the sociopolitical construct of sexist attitudes about women made
important contributions in enhancing the prediction of all three violence dimensions.
However, while sexist attitudes were an important predictor for all three dimensions, yet
the significance of each psychological variable varied from dimension to dimension. For
example, anxious attachment and borderline personality disorder scores were important in
explaining partner-intimidation violence, but only anxious and avoidant attachment were
important in explaining partner-specific physical violence. Moreover, the interaction
between sexist attitudes and anxious attachment was also significant. This significant
interaction indicated that anxiously attached men who hold strong sexist attitudes about
women are most likely to exhibit relationship violence. Only borderline and antisocial

personality disorder scores were important in explaining the general violence pattern. The

133

finding that both personality disorders were significant predictors of a generally violent
pattern encompassing severe violence might reflect the reality that greater levels of
psychOpathology are associated with more severe relationship violence.

Juxtaposition of the violence patterns observed in this study and previous
typology research findings allows for hypotheses regarding the compatibility of patterns
and types. The partner-intimidation pattern appears to describe the violence typical of the
borderline/dysphoric batterer subtype, whereas the general violence pattern appears to be
associated with the generally violent or antisocial batterer subtype that has been identified
in batterer typology research (Holtzworth-Munroe et al., 2000; Holtzworth-Munroe &
Stuart, 1994). The partner-specific physical violence pattern that actually emerged,
however, does not describe the type of violent behavior expected from the family-only
batterer. This may be because this study’s sample was recruited from court-mandated
interventions, whereas past typology research suggesting a family-only batterer type
included men recruited from both the community and court-mandated intervention
programs. The violent behavior of men recruited from the community is much less
violent in comparison to the behavior of men arrested and mandated to a program because
of their violence (Holtzworth-Munroe et al.). Another possibility is that men exhibiting
partner-specific physical violence may have been psychologically and verbally abusive
earlier in their relationship, and later escalated to physical violence when less severe
abuse was no longer effective in maintaining power and control over their victim.

The batterer who progresses from less to more severe partner-specific physical
violence and the borderline batterer type may be two distinct types. Given that
personality disorders were not important in explaining partner-specific physical violence
and that personality disorders are stable characteristics not subject to change without
intensive psychotherapy, it is unlikely that the borderline batterer who relies on partner-
intimidation violence is at a later time the batterer using partner-specific physical

violence. Nonetheless, it is conceivable that the anxiously attached batterer who initially

134

uses partner-intimidation violence for the purpose of gaining control of his victim is the
anxiously attached batterer who later exercises partner-specific physical violence when
less severe violence has proven ineffective as a control strategy. However, this does not
preclude the possibility that the borderline batterer’s violence is a progressive
phenomenon that escalates from partner-intimidation violence to more severe violence
(i.e., general violence pattern) that, although directed predominantly at his partner, may
also be directed at others.

Persons exercising partner-specific physical violence reported both anxious and
avoidant attachment orientations. Although the partner-specific physical violence of the
anxiously attached batterer may be representative of violence that has progressed from
less to more severe levels, according to this study’s findings, the avoidantly attached
batterer never exhibited partner-intimidation violence. As such, the avoidant batterer’s
partner-specific physical violence is not another level of violence on a more
comprehensive continuum of violence, but his only violence modality. Both anxiously
and avoidantly attached batterers relying on partner-specific physical violence may use
violence to regulate their partner’s proximity. Although the anxiously attached batterer
may use violence to make his partner fearful of abandoning him, the avoidantly attached
batterer may use violence to maintain a proximity with his partner that protects him from
potentially painful relationship-related affect (F uendeling, 1998). Given the effectiveness
of partner-specific physical violence in meeting the attachment needs of the avoidant
batterer, there is no need to progress to a more severe form of physical violence.
Moreover, verbal and psychological abuse may not meet the needs of the avoidantly
attached batterer, given that these modes of violence may be ineffective in maintaining
his partner in the desired proximity. The verbal and psychological violence of the
borderline or anxiously attached batterer is an attempt to keep his partner engaged, but
unlike these batterers, the avoidantly attached batterer wants to maintain sufficient

emotional distance from his partner (see Gottman et al., 1995).

135

Moreover, in addition to anxiously attached batterers and avoidantly attached
batterers enacting partner-specific physical violence, batterers with a fearful attachment
style, which is comprised of both anxious and avoidant attachment dynamics, may also
exhibit this pattern of violence. The violence of the fearful batterer is physical and
partner-specific and initiated as well as escalated by him. Because his violence is
motivated by both fear of abandonment and fear of intimacy, his victim remains unaware
of what triggers his violence making her particularly vulnerable to feelings of
powerlessness. Thus, the fearful batter’s mode of violence if effective in meeting his
attachment needs and maintaining control over his partner, and it is unnecessary to
escalate violence to a very severe level typical of the generally violent pattern.

Overall, the picture painted by this study’s findings is consistent with past
typology research. That is, there appear to be subtypes of batterers and at least two of
these subtypes are the antisocial, generally violent subtype, and the borderline subtype.
Consistent with previous research, this study’s findings also suggest that a pattern of
violence that is general, severe in intensity, and high in frequency can be explained by the
presence of antisocial personality disorder. A pattern of violence that is verbal and
psychological in nature but may include some minor physical violence appears to be
explained by borderline personality disorder and anxious attachment.

Although anxious attachment may be important in explaining violence expected
to be enacted by the borderline batterer because it is the corresponding attachment style
of borderline batterers, it is also possible that batterers who are anxiously attached and
not borderline use this violence pattern against their partners. However, anxiously
attached batterers who did not get their attachment needs met through the use of this
violence pattern may escalate their violence to a more severe violence that might prove
more effective in getting their attachment needs met. These batterers exhibit no clinical

disturbances of self and their violence is always partner specific.

136

The borderline batterer’s violence may also become progressively severe.
However, the progression of violence evidenced by the borderline batterer may be
qualitatively and quantitatively different from the progression of violence demonstrated
by the anxious batterer. Quantitatively, the anxiously attached batterer’s most severe
violence may never be as severe as the most severe violence enacted by the borderline
batterer. Qualitatively, the borderline batterer may only escalate his violence when
perceived threats of abandonment become real and he fears he will lose his self due to the
loss of his attachment object. At this point of desperation, the borderline batterer may
progress to a more severe physical violence against his partner. It may be that this
intrapsychic process explains why victims are at a much greater risk for harm and
potential murder when they try to leave their batterer. Conceivably, the batterer cannot
tolerate abandonment from his attachment object (Kemberg, 1975) and will employ any
tactic to ensure that such abandonment does not occur. Moreover, the threat of
abandonment may trigger negative affect that the borderline batterer cannot regulate, thus
inducing stress that contributes to a more generalized hostility.

Given that the violence of the borderline batterer and anxious batterer are
quantitatively and qualitatively different, it is possible that the anxious batterer who is not
concurrently borderline represents a distinct batterer subtype. Avoidantly attached
batterers may also be a distinct group responsible for initiating and escalating violence in
their intimate relationships, and whose violence is limited to a pattern that includes minor
and severe physical violence. However, the severity of their violence is not comparable
to that of the antisocial batterer or the desperate borderline batterer prior to imminent
abandonment by his partner. Finally, fearful batterers with both anxious and avoidant
attachment dynamics and corresponding behaviors represent yet another category of
batterers. Fearful batterers enact violence in response to fear of abandonment and/or fear

of intimacy, and the frequency, intensity, and modality of their violence is always within

137

the range of partner-specific physical violence. Sexist attitudes about women make
significant contributions to all violence patterns, but not in the same direction.

This study’s findings contribute to the dialogue among practitioners and
researchers in the domestic violence arena regarding whether research documenting types
of batterers suggests that there are in fact distinct subtypes of batterers or whether types
reflect developmentally progressive phases of relationship violence (Gondolf, 1988). This
study’s findings imply that there are different batterer types each with a corresponding
pattern of violence, and the pattern of violence affiliated with some subtypes might
include progression through different stages of violence that advance in severity.
Moreover, at varying times, different batterer subtypes may use similar violence patterns.

Implications for Practice and Policy

 

Present efforts by batterer intervention programs to deter male violence are only
moderately effective (Schmidt & Sherman, 1996; Tolman & Edleson, 1995). One
explanation for this outcome is that unidimensional conceptualizations of violence are
inadequate, and intervention programs founded on these theoretical models cannot
successfully decrease recidivism among all types of batterers. In addition, recent
empirical support has emerged that interventions founded on unidimensional
understandings of violence may not adequately meet the needs of all batter types and that
interventions may need to be tailored to accommodate the needs of different batterer
types (Saunders, 1996). Given that there may be subtypes of male batterers with different
etiologies of violence, different types of interventions may be needed. This study’s
findings suggest that different theories may be differentially important in understanding
the different types of violence enacted by batterers and that interventions modeled on
different theories are needed. The effectiveness of batterer intervention may be enhanced
if decisions regarding intervention are based on an understanding of the batterer’s history

of violence and of the factors motivating his violence. The development of less generic

138

and more perpetrator-specific treatment paradigms may ultimately enhance the efficacy
of batterer intervention programs.

One possibility is that different intervention programs can target different types of
violence, and decisions regarding which techniques should be used to reduce violent
behavior should be motivated by specific etiologic explanations of a particular violence
type. For example, whereas some attachment styles may be prone to fear of
abandonment, others may be threatened by fear of engulfmentf Expression of unmet
attachment needs consequential to these two interpersonal dynamics may be qualitatively
distinct. Although both attachment dynamics trigger dysfunctional anger and subsequent
violence, the type of violence corresponding with each attachment dynamic enacted by
the batterer may be different. Interventions can help the batterer identify external stimuli
associated with these dynamics and managing corresponding negative affect. Future
research that more specifically explores the nature of associations between fears of
abandonment vs. engulfrnent and violence may have significant implications for batterer
interventions.

An intervention strategy contingent upon the pattern of violence exhibited could
include separating batterers who are generally violent and more likely to enact severe
violence from those batterers who are predominantly psychologically violent. The
expectation is that techniques successful in motivating the generally violent batterer to
desist violence may not be equally effective with the psychological batterer. Moreover,
the effectiveness of intervention programs that simultaneously service these two types of
batterers may be limited due to varying levels of motivation to change their violent
behavior. For example, the goal of the anxiously attached psychological batterer who
uses violence to control his partner’s proximity due to his fear of abandonment appears to
be relationship security. However, this batterer’s hostile, violent tactics to prevent
abandonment may actually cause his partner to seek distance from him and reinforce his

anxious attachment and fear of abandonment, as well as his need to exercise violence as a

139

means to gain relationship security with future partners. For this batterer, intervention
programs may want to increase awareness of experienced fears of abandonment and skills
in constructively addressing this fear with his partner, along with increasing awareness
that his current tactics may actually contribute to his feared abandonment. Doing so may
lead to increased relationship security and eliminate his need to use violence in response
to his fear. The understanding that participation in an intervention program might be self-
serving could motivate the psychological batterer to consistently attend sessions and
complete his intervention program.

Alternatively, the generally violent batterer whose violence derives from stable
antisocial personality characteristics may not benefit from intervention strategies aimed at
increasing awareness that interpersonal dynamics might lead to violence and contribute to
unwanted relationship patterns. As such, he may experience no motivation to participate
in such an intervention program or to change his violent behavior. Unlike the
psychological batterer, cessation of violence may not be self-serving for the generally
violent batterer. Some practitioners have expressed concern that some batterers are
disruptive and impede the progress of other men in the group (Murphy & Dienemann,
1999). It may be that those men who are less motivated for change are disruptive to those
men who are motivated for change, and an alternative intervention, such as incarceration,
might be more appropriate for the unmotivated, antisocial batterer.

Although some batterers do have personality disorders, a significant number of
batterers serviced by court-mandated batterer intervention programs may not. It is
reasonable to expect that batterers without personality disorders may be more easily
treated than diagnosed batterers and that post-treatment recidivism should be less within
the former group. Moreover, although all batterers should be held accountable for their
behavior, batterers without pathology may be more easily motivated to desist from
violence if they are made to understand that violence will result in criminal sanctions as

well as other legal consequences. Incorporating information regarding the certainty of

140

legal consequences for violence may increase effectiveness of intervention programs for
at least some batterers (Tolman, Edleson, & Fendrich, 1996). However, intervention
programs should ensure that the legal system has the capacity to follow through with
consequences that the batterer can expect in response to his violence before implementing
this strategy. Otherwise, the batterer may learn that consequences are less than certain
leading him to conclude that the benefits of violence outweigh the costs. Moreover, if
battering is planned behavior, policies such as mandatory arrest may be instrumental in
decreasing violence against women in our communities.

Implementation of interventions based on the batterer’s history of violence and
factors motivating his violence would require careful pre-intervention assessment. This
assessment should include measures that could inform practitioners regarding the
batterer’s motivation for violence as well as history of violence. For example,
intervention recommendations for batterers with a history of general violence could be
different from those recommendations made for batterers with a history of violence that
has always been partner-specific. Moreover, batterers evidencing more clinical
disturbances of the self (e.g., Axis II diagnoses) might require longer-term intervention
and treatment modalities that address these disturbances. However, treatment targeting
the pathology should not substitute interventions specifically aimed at cessation of
violent behavior against woman, as there are no data supporting the expectation that
treating the pathology would result in cessation of violence. Dutton, Bodnarchuk, Kropp,
Hart, and Olgoff (1997) found that batterers with some personality disorders, including
antisocial and borderline, had the worst post-treatment success in cessation of violence.
This finding suggests that treatment targeting both the pathology as well as the batterer’s
violence may be necessary to achieve the goal of ending relationship violence against
women. Moreover, according to Murphy and Dienemann (1997), practitioners have
expressed concern that batterers with severe clinical pathology and personality disorders

might impede the progress of less pathological batterers, suggesting that intervention

141

programs combining nonpathological batterers with batterers evidencing clinical
pathology is counterproductive.

Conducting comprehensive assessments of men arrested for battering may also
have some implications for victims of violence. Specifically, information about the
batterer’s frequency, severity, and generality of verbal, psychological, and physical
violence could inform practitioners working with victims of violence. For example,
clinicians working with victims of the antisocial batterer subtype, whose violence is
predominantly physical and severe, may need to inform their clients that treatment may
not be effective in decreasing recidivism for this batterer subtype. Victims of the
generally violent/antisocial batterer may need to be warned that their perpetrator’s
successful follow through with a court mandated batterer intervention program does not
suggest that their relationship will be violence free. Clinicians providing services to
victims may want to work more collaboratively with service providers conducting
assessments and group therapy with batterers, given that clinical data derived from such
assessments and information about the perpetrator’s behavior in group could warn
victims if the perpetrator is unlikely or unwilling to rehabilitate. Additionally, although
typology research suggests that the violence of some batterer subtypes may be less severe
than that others, it is critical that findings from typology research not be interpreted as
suggesting that only some batterers need intervention. Less severe violence frequently
escalates to more severe, and sometimes, lethal violence. Practitioners need to inform
victims that their perpetrator’s verbal and psychological violence has the potential to
escalate to physical violence and ensure victims that this pattern of escalation is beyond
their control.

As with current intervention programs, assessment could be court-mandated as the
first step of an intervention program, and all treatment programs could equip themselves
to satisfy this mandate. Because treatment programs assessing the batterer may not

provide the type of intervention needed by the batterer, collaboration between service

142

providers in a given community and the criminal justice system would be encouraged to
ensure appropriate referral. However, prior to implementation of typology-based
intervention programs, research exploring which interventions are most effective for
eliminating distinct violence patterns is needed.

Because decisions regarding violence intervention are often motivated by legal as
well as public policy, theory-driven research on battering can provide useful information
for guiding the development of domestic violence related policies. Many states have
created domestic violence coalitions comprised of service providers and criminal justice
personnel (National Institute of Justice, 1998), and one purpose of these coalitions is to
create policies that can guide the responses of the criminal justice system and the
treatment community to domestic violence. In some states, these coalitions are
responsible for developing state standards that govern batterer intervention programs.
Because these standards regulate theoretical perspectives guiding treatment as well as
modalities of intervention, it is important that the standards are informed by research.
Given that decisions regarding how to intervene with domestic violence are influenced by
policy, it is important to continue with lines of inquiry that could be instrumental in
shaping such policy.

Limitations

 

The proposed study has some noteworthy limitations. First, the study's design
prohibits statements regarding the causation of battering. Although attachment patterns
and/or attitudes toward women may significantly explain variance in battering, it cannot
be concluded that attachment and/or attitudes toward women cause battering. Next, the
study's design has some inherent threats to external validity. Participants included in this
study were self-selected. The self-selection process may bias the sample in a systematic
way threatening generalizability of findings. That is, batterers choosing to participate
may vary in background characteristics from those choosing not to participate. For

example, batterers representative of a lower socio-economic status may have been more

143

inclined to participate because of the financial incentive. Men included in this sample
were also engaged with the criminal justice system, and findings may not generalize to
batterers who are not being monitored by the criminal justice system. In addition, studies
focusing on batterers generally access samples through court-mandated interventions.
Given that these programs do not service batterers who commit felonies, most batterer
samples do not generalize to those men who have committed felonies against their female
partners.

Another limitation assOciated with data collection procedures and threatening
internal validity is that the self-report nature of the data introduces self-report bias, which
is inherent in all survey research. For example, because of their involvement with the
criminal justice system, participants may have been cautious about reporting the severity
and frequency of enacted violence. In addition, men were asked to report incidents of
violence in the 12-month period prior to their arrest or incident motivating their
participation in batterer intervention. Participants may not have accurately recalled their
history of violence during this 12-month period.

This study also suffers from mono-operation bias in that only one instrument was
used to measure each variable. Cook and Campbell (1979) recommend using several
measures of each construct to increase the study’s validity and reliability. Moreover,
self-report data collection as a sampling method is particularly vulnerable to error when
used with batterers, given their tendency to underreport violence (Arias & Beach, 1987).
Additionally, because all data were collected via self-report and from only one data
source, results are subject to mono-method bias. Gathering data through self-report
methods alone limit the construct validity of each variable measured (Cook & Campbell).

Additional threats to internal validity introduced through sampling procedures
include the following. First, data collection occurred at more than one site. This
sampling procedure may have introduced error variance and may have threatened the

validity of the findings. Second, the decision to pay men to participate may have been

144

effective in increasing sample size; however, some men may have chosen to participate
due to the financial compensation and may not have been motivated to respond honestly
to the questionnaires. Next, the number of participants meeting criteria for antisocial
personality disorder was a very small percentage of the sample. Number of participants
meeting criteria for borderline personality disorder was also a relatively small percentage
of the overall participants. Given the small number of participants in each of these levels
of the independent variable, it is possible that Type 11 error affected this study’s findings
and that significant differences between the two personality disorder groups went
undetected.

The small number of participants meeting criteria for antisocial personality
disorder may be reflective of the nature of the sample. That is, intervention programs
from which this study’s data were collected generally service men that have committed
violence categorized as a misdemeanor. Given that antisocial batterers are prone to more
severe violence, it may be that these batterers are not serviced by court-mandated
intervention programs but by prisons.

Moreover, nineteen percent of the participants met criteria for both antisocial and
borderline personality disorders. This phenomenon might be explained by the cross-
loading of several items on borderline personality disorder and antisocial personality
disorder scales. Another possible explanation for participants meeting criteria for both
personality disorders is the PDQ-R’s tendency to indicate false positives. The PDQ-R
has received support as a dimensional measure of personality disorders. However, given
the PDQ-R’s tendency to indicate a diagnosis when none exists, use of the instrument as
a categorical measure may have been problematic. Future studies categorizing batterers
according to personality disorder should use measures that more reliably classify batterers
according to specific personality disorders (e. g., MCMI-II; Millon, 1987).

In addition, research has shown that personality disorders other than antisocial and

borderline are prevalent among batterers (Floumy & Wilson, 1991; Hamberger &

145

Hastings, 1988; Murphy et al., 1993). This suggests that batterers in this study who did
not meet criteria for antisocial or borderline personality disorder may have met criteria for
some other personality disorder. A limitation of this study is that participants were not
screened for these additional personality disorders.

Finally, there are some conceptual limitations of this study that are worth noting.
First, although conceptualization and operationalization of constructs were guided by
theory, patriarchy is a difficult construct to define. In this study, patriarchy was
operationalized as sexist attitudes toward women, and data were collected at an individual
level. A logical presupposition is that men's endorsement of sexist attitudes in their
relationships with women is attributable to patriarchal attitudes inherent in the larger
social structure and that men are a product of the patriarchal systems in which they reside.
However, it would be presumptuous to infer that this study provides direct support for the
significance of sociopolitical explanations. In fact, because feminist understandings of
male violence against women are rooted in system-level variables, not individual-level
variables, this study is a very limited test of the feminist hypothesis of battering. Future
studies should include multiple measures operationalizing patriarchal attitudes at an
individual level and/or preferably indices of systemic patriarchy (e.g., women’s economic
I freedom and political freedom). Additionally, if researchers continue to use individual
level variables to test the feminist hypothesis because of the difficult nature of
operationalizing patriarchy at a systems-level, operationalization of patriarchy should be
guided by comprehensive theories of gender role and gender role conflict (0’ Neil, Good,
& Holmes, 1995). Moreover, research is needed to validate the assumption that
individual-level measures of patriarchy and sexism are reflective patriarchal systems.

This study’s conceptualizations of battering from an attachment theory
perspective are also limited. Specifically, such conceptualizations are often based on the
premise that one's attachment style is stable from infancy through adulthood. However,

findings regarding stability of attachment style are inconsistent and relatively new (Hazan

146

& Shaver, 1994). As such, until there is more evidence of the stability of attachment
from infancy through adulthood, interpretations that adult insecure attachment predicting
male violence toward female partners is linked to early relationships with caregivers
should be made with caution. It may be that the insecure attachments of batterers
developed as part of relationships in their formative adolescent years, or that insecure
attachment, irrespective of its developmental trajectory, is a risk factor for increased
anger and violence in relationship to intimate partners.

Directions for Future Research

Additional research exploring the etiology of violence for distinct batterer
subtypes is needed. Exploring behavioral, cognitive, and emotional characteristics
specific to batterer types is also pertinent. Findings from such studies may ultimately
inform clinicians about appropriate interventions. For example, such findings might
suggest different types of treatments for batterers who are generally violent versus those
whose violence is partner-specific. Moreover, research findings that existing programs
for perpetrators have low recruitment and high attrition rates, and may be unsuccessful in
reducing recidivism for battering, suggest the need for additional research that can
generate insights into advancing intervention with this population (Tolman & Edleson,
1995)

Researchers and practitioners alike (see Murphy & Dienemann, 1999) have
advocated for research that randomly assigns different batterers to different treatment
approaches (e. g., cognitive-behavioral, psychodynamic), with effectiveness of the
different treatment approaches compared across batterer subgroups. Implementation of
typology-based intervention programs without such additional research could prove to be
less than cost-effective. Additionally, because studies exploring the effectiveness of
batterer intervention programs are inconsistent (see National Institute of Justice, 1998),
additional outcome studies are needed prior to concluding that current programs are less

than effective. Researchers conducting future studies exploring the effectiveness of

147

batterer intervention programs might also assess whether batterers evidencing pathology
desist from violence after completion of batterer intervention programs or whether
completion of such programs is insufficient in eliminating violent behaviors. Such
studies could provide information regarding the adequacy of current intervention
programs for batterers with personality disorders. Moreover, research exploring factors
(e.g., reactance, psychological mindedness) that might predict the likelihood that a
batterer will successfully complete and/or participate in an intervention program could
help conserve resources allocated to batterer intervention.

Debates about whether typology findings suggest distinct batterer subtypes or a
homogenous group of batterers progressing through distinct phases of violence indicate
the need for research exploring these alternatives. Longitudinal research exploring
associations between length of intimate relationship and frequency as well as severity of
physical and psychological violence could help answer questions about the validity of
batterer subtypes. If such research demonstrated more severe violence and increased
injury to the partner with the progression of the relationship, the hypothesis that batterer
subtypes reflect a continuum of behaviors would be supported. Also, as suggested by this
study’s findings, it is possible that there are both subtypes and distinct patterns
corresponding to subtypes. Future research attempting to match violence patterns with
proposed batterer subtypes could also prove informative in the ongoing debate of batterer
subtypes versus progressive stages of violence.

Another potentially important area of future research is longitudinal studies
demonstrating the stability of attachment orientations from infancy to adulthood.
Explanations of adult intimate relationship violence are borne from expectations that that
adults may react to unmet attachment needs as they might have responded to unmet
attachment needs in childhood (Hazan & Shaver, 1987). However, given the absence of
support for the stability of attachment from infancy through adulthood, interpretations

that adult insecure attachment predicting male violence toward female partners is linked

148

to early relationships with caregivers should be made with caution. In addition, future
research that more specifically explores the nature of associations among fears of
abandonment or engulfment, violence patterns, batterer subtypes, and attachment
orientations should be conducted.

Additional research suggesting links between specific personality disorders and
attachment orientations among batterers should be conducted. Such studies could
enhance the utility of attachment theory in understanding the developmental trajectories
of batter subtypes. Studies that illuminate distinctions between borderline batterers with
anxious attachment and anxiously attached batterers who are not borderline are also
needed to determine the heterogeneity of these two types of batterers. In addition, given
suggested relationships between borderline personality disorder and anxious attachment
(Patrick et al., 1994; West et al. 1993) as well as antisocial personality disorder and
avoidant attachment (Gacano & Meloy, 1992; Gacano et al., 1992), studies exploring the
mediational affects of personality disorders on relationships between attachment and
violence as well as studies exploring the mediational affects of attachment on
relationships between personality disorders and violence are needed. For example,
findings that personality disorders mediate relationships between attachment and violence
might emphasize the need for interventions that especially focus on the impact of clinical
disturbances on relationship violence.

Finally, some methodological suggestions for future research include the
following. First, to date, only one typology study (Holtzworth-Munroe et al., 2000) was
located that included a control group to ensure that conclusions about clusters of
personality characteristics evidenced by the different types of batterers did not generalize
to nonviolent men. Additional typology studies that include control groups comprised of
nonviolent men are needed. Second, to obtain richer data, future studies should include
multiple measures of constructs included in domestic violence research. For example,

including several measures Operationalizing patriarchal attitudes at an individual level

149

and/or preferably indices of systemic patriarchy (e.g., women’s economic freedom and
political freedom) would enhance the reliability and validity of future studies. Finally, it
is essential that firture studies have batterer samples that include violent men recruited
from the community who have not been arrested, men recruited from batterer intervention
programs, and batterers recruited from prison. Given that samples used for most current
studies only include recruit men from batterer intervention programs, it is necessary to
widen the scope of sources from which batterers are recruited to ensure greater

generalizability.

150

 

APPENDICES

151

 

APPENDIX A
Invitation to Participate in the Study

A graduate student from Michigan State University is conducting a study about men and
their relationships with women. This research has nothing to do with court or this

treatment program. You must have an 8th grade reading level to participate in this study,

and you are only eligible to participate if you do have at least an 8th grade reading level.
Total participation for the study itself will take approximately 45 minutes and will be
incorporated into next week's session. Total session time for next week may take longer
than 45 minutes. Your participation is completely voluntary and involves answering a
series of questions. If you chose to participate, you will be given credit for a regularly
scheduled class. If you chose not to participate or are not eligible to participate, in no
way will you be penalized. If you chose not to participate or are not eligible to
participate, you can resume classes in 2 weeks as scheduled. If you are ineligible to
participate or chose not to participate and do not want to miss a week's class, you can
attend another domestic violence class next week at an alternate time. Once again, to be

eligible to participate, you must have an 8th grade reading level. The types of questions
you will be asked include how you think, feel, and behave in your relationships with your
girlfiiend or wife and how you feel about yourself in general. You will not put your
name on any of the questionnaires and your answers will remain anonymous. However,
if you chose to participate, you may have some face-to-face contact with the graduate
student conducting this study.

If you want to participate and have at least an 8th grade reading level, please attend class
next week. If you are not interested or don't have the appropriate reading level to
participate, you can resume classes in 2 weeks or attend an alternate class next week.

If you have any questions or want additional information, you can contact:

Anne Mauricio

Michigan State University

Department of Counseling, Educational Psychology, and Special Education
Erickson Hall

East Lansing, MI 48864

(517) 355-8508 OR e-mail: maurici1@pilot.msu.edu

152

APPENDIX B

CONSENT FORM
I am a graduate student at Michigan State University, and I am seeking your participation
in a study that will provide information about men and their relationships with women.
The project has absolutely nothing to do with court or this treatment program. Your
participation is completely voluntary. The project involves answering a series of paper-
and-pencil questions, and should take approximately 45 minutes of your time. To be

eligible to participate, you must have an 8dI grade reading level. The types of questions
you will be asked include how you think, feel, and behave in your relationships with your
girlfriend(s) or wife and how you feel about yourself in general. For example, you will
be asked how much you agree or disagree with statements such as “It bothers me when
my partner tells me what to do instead of asking,” or “I wish I could have more respect
for myself.” It is important that you answer all questions as honestly as possible. You
will not put your name on any of the questionnaires and your answers will remain
anonymous. Your privacy will be protected to the maximum extent allowable by law.

If you chose to participate, $15.00 will be paid either directly to you or toward your
balance at this counseling center.

If you want to participate, please stay in the room, and you will be handed a packet of

questions to complete. Questionnaires will be completed directly after we review this

consent form. Your completed questionnaires will be reviewed to ensure that you have
answered all the questions. You will be asked to complete any questions that you may
have missed.

Once again, your participation is voluntary. At any time, you may discontinue your
participation or chose not to answer specific questions without any penalty to you. A
decision not to participate will not negatively affect your current circumstances or limit
any future services you might receive. In case you should have any questions about this
project, you can contact me at:

Anne Mauricio

Michigan State University

East Lansing, MI 48864

(517) 355-8508 or e-mail: maurici1@pilot.msu.edu

If you have any concerns about how this research was conducted, you can call:
Dr. David Wright at (517) 355-2180.

 

If you understand the nature of the project and the nature of your participation and
consent to participate in this project, please print and sign your name below on both
consent forms attached. You will keep one copy for your records and return one copy to
me. To protect your anonymity, all consent forms returned to us will be stored separately
from any questionnaires that you complete.

Name (Print): Signature:

 

 

153

 

 

APPENDIX C

Demographic and Background Questionnaire

 

 

1. Age

2. Ethnic Background: (check one) Asian [ ]
Black [ 1
Hispanic [ ]
Native American [ ]
White I I
Other (specify)

3. What was the last year of school you completed? (Check one)

a.) Less than seventh grade C ]

b.) Junior high school (7th-9th grade) C ]

c.) Some high school (10th-1 1th grade) : ]

d.) High school graduate or GED I ]

e.) Some college (at least one year) I ]

f.) College graduate : I

g.) Graduate School I ]

4. Relationship Status:

Married [ 1

Divorced [ ]

Separated [ I

Single (live with girlfiiend) [ ]

Single (do not live with girlfriend) [ ]

5.

Indicate how many significant romantic relationships have you had in the past
year and indicate how many months each one lasted?

 

 

 

Of the relationships you identified above in #5, indicate the number that you used
any minor violence such as hitting, slapping, or pushing?

 

Of the relationships you identified above in #5, indicate the number that you used
any severe violence such as bunting, choking, or cutting?

 

Have you ever been arrested in your lifetime? (please circle one)
Yes or No

If yes, how many times have you been arrested?

 

154

 

 

 

10. Were you ever arrested for assault in your lifetime? (please circle one)

 

 

Yes or No
l 1. If yes, how many times have you been arrested for assault?
12. How many weeks have you been in group at this treatment center?
13. Have you ever been in counseling or group in your lifetime because of being

violent toward your girlfriend or wife? (please circle one)
Yes or No

13a. If yes, how many counseling sessions or groups did you attend all
together?

14. Are you in group now because:
[ ] The court/probation sent you
[ ] you chose to come on your own (i.e., you have not been arrested for
domestic assault and told by your probation officer, lawyer, prosecutor, or
judge to come to group
I ] your wife or girlfriend said she'd leave you if you didn't go to counseling
[ ] other reason (please specify)

 

 

15. Are you still in a relationship or have contact with and the woman who you were
fighting with at the time of your arrest for the assault for which you are now in
group? (please circle one)

Yes or No

15a. If no, how many months has it been since you had contact with
her?

 

15b. If no, why? ] She filed a restraining order

] The judge said I couldn't have contact her

] Contact with her is against the terms of my
probation

] I wanted to end the relationship

] She wanted to end the relationship

] Other (please specify):

16. When there has been violence between you and your girlfriend(s) or wife in the
past year, how often did you hit, slap, or push first?

Never

Once

A few of the times

Half of the times

More than half of the time
All the time

r—iF—IHI—Hr—HI—H
HHHHHH

155

17.

18.

b.)

d.)

f.)

g.)

If you have always hit, slap, or push first when there has been violence between
you and your girlfriend(s)or wife in the past year, how often does she hit, slap, or

push you back?

Never

Once

A few times

Half of the times

More than half of the time
All the time

Other than your wife or girlfiiend(s), how often have you been violent or been in a
physical fight in the past year with: (Check all that apply)

Brother or Sister:

Mother or Father:

Other relatives

Friends

Strangers

Police Officers

Others (please specify)

I—r-Hr—Hr—m-Hr—n
h—Jh—Jl—r—M—IHH

Once
2 to five times
6 - 10 times

More than 10
Never

Once
2 to five times
6 - 10 times

More than 10
Never

Once

2 to five times
6 - 10 times
More than 10

Never

Once

2 to five times
6 - 10 times
More than 10

Never

Once
2 to five times
6 - 10 times

More than 10
Never

Once
2 to five times
6 - 10 times

More than 10
Never

PEP—IMF— f—‘PqF-qf-‘HF—fi l—Hf-HF'WF—H—‘I

HENRI—I

f—HF—HF-H—I ”HF—‘F-fl—‘I

HHHHH Hwfihfiflfl HHHHH

biblh—lh—lh—J

HHHHH HHHHH

 

156

19.

20.

21.

22.

When there has been violence between you and your girlfriend(s) or wife in the
past year, how oflen does yelling and/or name calling lead to hitting, slapping, or
pushing by you?

Never [
Once [
A few of the times [
Half of the times [
More than half of the time [
All the time [

HHHHHH

When there has been violence between you and your girlfriend(s) or wife in the
past year, how often does yelling and/or name calling lead to more severe
violence such as burning, punching, or cutting by you?

Never [ ]
Once [ ]
A few of the times [ ]
Half of the times [ ]
More than half of the time [ ]
All the time [ ]

When there has been violence between you and your girlfriend(s) or wife in the
past year, how often does hitting, slapping, or pushing lead to more severe
violence such as burning, punching, or cutting by you?

 

Never [ I
Once t I
A few of the times I ]
Half of the times I ]
More than half of the time I ]
All the time : I

How often have you been physically violent toward your girlfriend or wife NOT
because you were fighting with her, but for some other reason?

Never [ ]
Once [ ]
Twice [ ]
3-5 times [ ]
6-10 times [ ]
11-20 times [ ]
more than 20 times [ ]

What was that reason(s)?

 

 

157

 

23.

24.

25.

As a result of any of the violence between you and your girlfriend(s) or wife in
the past year, how often did she suffer the following injuries?

Mild scratch, small bruise, or swelling:

Never

Once

2 to 5 times

6 to 10 times

More than 10 times

r—HF-Ir—Ir-m—I
h—JHH—JHh—J

Moderate Injury such as a fracture, minor burn, cut, or large bruise:

Never

Once

2 to 5 times

6 to 10 times

More than 10 times

HHHHH
HHHHH

Severe Injury such as major wounds, severe bleeding or burns, knocked out:

Never

Once

2 to 5 times

6 to 10 times

More than 10 times

F—H—ir—I—H—H
I—=Jl-—rlh=rlh—rlh——l

As a result of any of the violence between you and your girlfriend(s) or wife in
the past year, how often did she miss work?

6 to 10 times
More than 10 times

Does not work : ]
Never : ]
Once I ]
2 to 5 times I ]
= l
. ]

 

As a result of any of the violence between you and your girlfriend(s) or wife in
the past year, how often did she miss her regular activities such as errands,
shopping, or family and social activities?

Never

Once

2 to 5 times

6 to 10 times

More than 10 times

F-HP-Hr—IF—rr—i
h—Jl—lhflh—Ih—l

158

As a result of any of the violence between you and your girlfriend(s) or wife in the
past year, how often did she need go to the hospital?

Never [ ]
Once [ ]
2 to 5 times [ ]
6 to 10 times [ ]
More than 10 times [ ]

As a result of any of the violence between you and your girlfi'iend(s) or wife in
the past year, how often did she need any medical attention?

Never

Once

2 to 5 times

6 to 10 times

More than 10 times

r—ir—HF-H—u-H
HHHHH

 

159

APPENDIX D
Basic Reading Inventory

Directions: Please read the following passage. Then, answer the corresponding
questions. It is very important that you do not leave any questions blank.

Our Environment

Besides using plants and animals for food, people use the hides of animals for
shoes, the wood from trees to build houses, the fiber from the cotton plant to make skirts
and shirts, and the wool from sheep to make suits and coats. Even the synthetic fibers
that people use are made from matter found in the environment.

People and the environment are interdependent, but that is not the whole story.
Modern people can do much more; they can use science and technology to change their
environment. Because of their advanced brains, people can investigate and use their .1
precious environment. "

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Questions L
1. What is this passage about? i
2. What do modern people use to change the environment?
3. Why are people able to investigate their environment?
4. What are synthetic fibers made from?
5. What are some things in the environment that people use?
6. What does the passage say people use to make skirts and shirts?
7. What reasons could you give for why our environment is precious?
8. What are some of the ways in which people have changed their environment?

 

 

9. Do you think that it is a good thing for people to change their environment? Why?

 

 

10. What does synthetic mean?

 

 

160

APPENDIX E

Personality Diagnostic Questionnaire - Revised
Borderline Personality Disorder Scale

Directions: The purpose of this questionnaire is for you to describe the kind of person
that you are. When answering the questions, think about how you have tended to feel,
think, and act over the past several years. Please Circle T for true or F for False.

1a. My feelings towards another person can often change drastically. T F
1b. People that I have worshipped have often ended up disappointing T F
3. lnaem a moody person. T F
4a. I rarely get so angry that I lose control. T F
4b. I have gotten into more real, physical fights than most people. T F
5a. I have never threatened to kill myself. T F
5b. 1 often hurt myself accidentally. T F
6a. My friends, my goals, or my beliefs are always changing. T F
6b. 1 have had more than my share of identity crises. T F
7. I feel empty or bored much of the time. T F
8. If I know a relationship is going to end, I'll let it end peacefully. T F
2./2 I have done things on impulse (such as those below) that can get T F

me into trouble.

Check all those that apply to you:

 

 

 

 

 

a. Spending more money than I have

b. Having sex with people I hardly know.
c. Drinking too much.

(1. Taking drugs.

e. Eating binges.

f.

Shoplifting.

 

161

 

APPENDIX F

Personality Diagnostic Questionnaire — Revised
Antisocial Personality Disorder Scale

Directions: The purpose of this questionnaire is for you to describe the kind of person
that you are. When answering the questions, think about how you have tended to feel,
think, and act over the past several years. Please Circle T for true or F for False.

l. I have no trouble keeping jobs or staying in school. T F

'11

2. I have been in trouble with the law several times T
(or would have been if I had been caught).

3b. I have beaten my wife, or kids when I felt that they deserved it.
I have often just not paid my bills.

Some people consider me a drifter.

 

I drive pretty well even if I have been drinking.

~1"l*'l"'l"i~1
'fl'fi‘fi'fi'flfi

4

5

6. Lying comes easily to me and I often do it.
7

8

Other people have felt that I have not taken proper care of my
children.

9. I am the kind of person who stays in a relationship (with a T F
spouse, lover, etc.) for a long time.

10. I don't care if other people get hurt so long as I get what I want. T F

B/3. When I was a kid (before age 15), I was somewhat of a T F
juvenile delinquent, doing some of the things below:
Check all those that apply to you:

I skipped school alot.

I ran away from home overnight more than once.
I used to start fights with other kids.

I used a weapon in fights more than once.
I forced someone to have sex with me.

I was physically cruel to people.

I was physically cruel to animals.

I broke windows or destroyed property.

I set fires.

I lied alot.

I cheated other people.

I robbed or mugged people.

 

 

 

 

 

 

 

 

 

 

 

=r=°rz=rqo rm .0-9 .01»

 

162

 

APPENDIX G
Experiences in Close Relationships

Directions: The following statements concern how you feel in romantic relationships.
I am interested in how you generally experience relationships, not just in what is
happening in a current relationship. Respond to each statement by indicating how much
you agree or disagree with it. Write the number in the space provided, using the
following rating scale:

 

 

 

 

Disagree Neutral/Mixed Agree
Strongly Strongly
1 2 3 4 5 6 7
1. I prefer not to show a partner how I feel deep down.

2. I worry about being abandoned.

3. I am very comfortable being close to romantic partners.

4. I worry alot about my relationships.

5. Just when my partner starts to get close to me I find

myself pulling away.

I worry that romantic partners won't care about me as

much as I care about them.

I get uncomfortable when a romantic partner wants to be

very close.

I worry a fair amount about losing my partner.

I don't feel comfortable opening up to romantic

partners.

10. I often wish that my partners feelings for me were
as strong as my own feelings for her.

11. I want to get close to my partner, but I keep
pulling back.

12. I often want to merge completely with romantic partners,
and this sometimes scares me.

13. I am nervous when partners get too close to me.

14. I worry about being alone.

15. I feel comfortable sharing my private thoughts and
feelings with my partner.

16. My desire to be very close sometimes scares people
away.

17. I try to avoid getting too close to my partner.

18. I need a lot of reassurance that I am loved by my partner.

19. I find it relatively easy to get close to my partner.

20. Sometimes I feel that I force my partners to show more
feeling, more commitment.

 

 

 

 

>0.°°>’.°‘

 

 

 

 

 

 

 

 

 

 

 

 

163

 

 

Disagree Neutral/Mixed Agree

Strongly Strongly
1 2 3 4 5 6 7
21. I find it difficult to allow myself to depend on
romantic partners.
22. I do not often worry about being abandoned.
23. I prefer not to be too close to romantic partners.
24. If I can't get my partner to show interest in me, I get
upset or angry.
25. I tell my partner just about everything.
26. I find that my partner(s) don't want to get as close as
I would like.
27. I usually discuss my problems and concerns with my
partner.
28. When I am not involved in a relationship, I feel somewhat
anxious and insecure.
29. I feel comfortable depending on romantic partners.
30. I get fi'ustrated when my partner is not around as much
as I would like.
31. I don't mind asking romantic partners for comfort,
advice, or help.
32. I get frustrated if romantic partners are not available
when I need them.
33. It helps to turn to my romantic partner in times of need.
34. When romantic partners disapprove of me, I feel really
bad about myself.
35. I turn to my partner for many things, including
comfort and reassurance.
36. I resent it when my partner spends time away from me.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

164

 

 

APPENDIX H
Attitudes Toward Women Scale

Directions: The statements listed below describe attitudes toward the roles of women
in society which different people have. There are no right or wrong answers, only
opinions. You are asked to express your feelings about each statement by circling a letter
to indicate your opinion.

A = Agree Strongly,

B = Mildly Agree

C = Disagree Mildly

D = Disagree Strongly.

1 . Swearing and obscenity are more repulsive in the A B C D
speech of a woman than a man.

2. Under modern economic conditions with women A B C D
being active outside the home, men should share in
household tasks such as washing dishes and doing laundry.

3. It is insulting to women to have the "obey" clause A B C D
remain in the marriage service.

4. A woman should be as free as a man to propose marriage. A B C D

5. Women should worry less about their rights and more A B C D

about becoming good wives and mothers.

6. Women should assume their rightful place in business A B C D
and all the professions along with men.

7. A woman should not expect to go to exactly the same A B C D
places or have quite the same freedom of action as a man.

8. It is ridiculous for a women to run a locomotive and a A B C D
man to darn socks.

9. The intellectual leadership of a community should be A B C D
largely in the hands of men.

10. Women should be given equal opportunity with men for A B C D
apprenticeship in the various trades.

11. Women earning as much as their dates should bear A B C D
equally the expense when they go out together.

12. Sons in a family should be given more encouragement A B C D
to go to college than daughters.

13. In general, the father should have greater authority than A B C D
the mother in the bringing up of children.

165

14.

15.

Economic and social freedom is worth far more to
women than acceptance of the ideal of femininity
which has been set up by men.

There are many jobs in which men should be given
preference over women in being hired or promoted.

166

ABCD

ABCD

APPENDIX I

FOR THE NEXT FIVE QUESTIONS PLEASE RESPOND USING THE FOLLOWING
SCALE:

1 = Agree Strongly,

2 = Mildly Agree

3 = Mildly Disagree

4 = Disagree Strongly.

1. The only way to get my needs met in my relationships 1 2 3 4
with my girlfiiend(s) or wife is to threaten to hit, slap, or
push her.

2. Sometimes its necessary to threaten to hit, slap, or
push my girlfriend(s) or wife to get the relationship, 1 2 3 4
or household, to run smoothly.

3. The worst thing in a relationship is if a woman is afraid I 2 3 4
of a man.

4. Threatening to hit, slap, or push my girlfiiend(s) 1 2 3 4
or wife helps the relationship run smoothly.

5. I don't get the respect that I deserve from my 1 2 3 4

girlfiiend(s) or wife unless I threaten to hit, slap,
or push her.

167

 

APPENDIX J
Conflict Tactics Scale

Directions: It is normal for couples to have conflicts. There are a lot of different
things that you can do when you have conflict or disagreement with your partner. We
would like you to try and remember what went on during the past year when you had
disagreements. Please circle a number for each of the things listed below to show how
often you did it IN YEAR BEFORE YOUR ARREST FOR WHICH YOU ARE NOW
IN GROUP.

 

0 = Never

1 = Once

2 = Twice

3 = 3-5 times

4 = 6 - 10 times
5 = 11-20 times
6= more than 20

1. Discussed an issue calmly: 0 1 2 3 4 5 6

2. Got information to back up

my side of things: 0 l 2 3 4 5 6
3. Brought in or tried to bring

in someone to help settle

things: 0 1 2 3 4 5 6

4. Insulted or swore at her: 0 1 2 3 4 5 6

5. Sulked or refused to talk 0 l 2 3 4 5 6
about an issue:

6. Stomped out of the room 0 1 2 3 4 5 6
or house or yard:

7. Did or said something to 0 1 2 3 4 5 6
spite her:

8. Cried: 0 1 2 3 4 5 6

9. Threatened to hit or throw 0 l 2 3 4 5 6
something at her:

10. Threw, smashed, hit, or 0 l 2 3 4 5 6
kicked something:

1 1. Threw something at her: 0 1 2 3 4 5 6

12. Pushed, grabbed, or shoved 0 1 2 3 4 5 6
her:
13. Slapped her: 0 1 2 3 4 5 6

168

 

14.

15.

16.
17.
18.

19.
20.

21.

Kicked, bit, or hit her with
a fist:

Hit or tried to hit her with
something:

Beat her up:
Choked her:

Threatened her with a
knife or gun:

Used a knife or fired a gun:

Forced your partner to have

sex with you or someone else:

Have you been violent toward your wife or girlfriend(s) in the past year in anyway

not mentioned above?

Yes[] No[]

If yes, please explain:

 

 

How often have you been violent in this way?

Once

Twice

3-5 times

6-10 times

1 1-20 times

more than 20 times

r—H-Hi—In—u-qr-q
HH—fl—Jl—H—JH

 

APPENDIX K
Psychological Maltreatment against Women Scale

Directions: This questionnaire asks about actions you may have taken in your
relationship with your partner. Answer each item as carefully as you can by placing a
number beside each one as follows:

1 = Never

2 = Rarely

3 = Occasionally

4 = Frequently

5 = Very Frequently

IN THE SIX MONTHS BEFORE MY ARREST FOR WHICH I AM NOW IN
GROUP:

 

 

 

 

 

 

 

l. I called my partner names.

2. I swore at my partner.

3. I yelled and screamed at my partner.

4. I treated my partner like an inferior.

5. I monitored my partner's time and made
her account for her whereabouts.

6. I used my partner's money or made important financial
decisions without talking to her about it.

7. I was jealous or suspicious of her friends.

8. I accused my partner of having an affair

 

with another man.

9. I interfered in my partner's relationships

 

with other family members.

10. My partner tried to keep me from doing

 

things to help myself.

 

1 1. l restricted my partner's use of the telephone.

12. I told my partner her feelings were irrational or crazy.

 

13. I blamed my partner for my problems.

 

14. I tried to make my partner feel crazy.

 

170

APPENDIX L
Social Desirability Scale

Directions: For each of the following statements, please indicate your feelings by
placing a number in the blank before the statement. Use the following list of choices.

”$2922!”

10.

1 = Strongly Agree

2 = Agree

3 = Mildly Agree

4 = Neither Agree nor Disagree
5 = Mildly Disagree

6 = Disagree

7 = Strongly Disagree

No matter who I am talking to, I am always a good listener.

 

I have sometimes taken unfair advantage of another person.

 

I am always courteous, even to people who are disagreeable.

 

I sometimes try to get even, rather than forgive and forget.

 

I am quick to admit making a mistake.

 

I sometimes feel resentfirl when I don't get my own way.

 

I am always willing to admit making a mistake.

 

There have been occasions when I have taken advantage of

 

someone.

I would never think of letting someone else be punished for

 

my wrongdoing.

At times, I have wished that something bad would happen to

 

someone I disliked.

171

REFERENCES

Adams, D. (1988). Counseling men who batter: A profeminist analysis of five
treatment models. In M. Bograd & K. Yllo (Eds), Feminist perspectives on wife abuse
(pp. 176-199). Beverly Hills, CA: Sage.

Adams, D. (1990). Identifying the assaultive husband in court: You be the judge.
Response to the Victimization of Women and Children, 13, 13-16.

Adams, D. , & McCormick, A. (1982). Men unleaming violence: A group
approach. In M. Roy (Eds), The abusive partner: An analysis of domestic battering (pp.
170-197). New York: Van Nostrand Reinhold.

Aiken, L., & West, S.G. (1991). Multiple regression: Testing and interpreting
interactions. Thousand Oaks, CA: Sage Publications, Inc.

 

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

American Psychiatric Association (1987/1994). Diagnostic and statistical manual
of mental disorders (3rd ed. /4th ed.). Washington, DC: Author.

 

Arias, I., & Beach, S. R. H. (1987). Validity of self-reports of marital violence.
Journal of Family Violence, 2, 139-149.

 

Avis, J. M. (1992). Where are all the family therapists? Abuse and violence
within families and family therapy's response. Journal of Marital and Family Therapy,
E8L225-232.

Bachman, R. (1994). Violence against women: A national crime victimization
survey report. Washington, DC: Bureau of Justice Statistics.

 

Ballard, R. (1992). Short forms of the Marlowe-Crowne social desirability scale.
Psychological Reports, 71,1 155-1 160.

 

Ballard, R., Crino, M.D., & Rubenfeld, S. (1988). Social desirability response
bias and the Marlowe Crowne Social Desirability Scale. Psychological Reports, 63, 227-
237.

Bartholomew, K., & Horowitz, L.M. (1991). Attachment styles among young
adults: A test of a four-category model. Journal of Personality and Social Psychology,
_6_1J_226-224.

Bartholomew, K., & Shaver, P. R. (1998). Methods of assessing adult
attachment: Do they converge? In J. A. Simpson & W. S. Rholes (Eds.), Attachment
theory and close relationships (pp. 25-45). New York: Guilford Press.

 

Bograd, M. (1984). Family systems approaches to wife battering: A feminist
critique. American Journal of Orthopsychiatry, 54, 558-568.

 

Bograd, M. (1988). Feminist perspectives on wife abuse: An introduction. In K.
Yllo & M. Bograd (Eds), Feminist perceptive on wife abuse, (pp. 11-26). Beverly Hills,
CA: Sage.

172

Bograd, M. (1990). Why we need gender to understand human violence. Journal
of Interpersonal Violence, 5, 132-135.

 

Bograd, M. (1992). Values in conflict: Challenges to family therapists' thinking.
Journal of Marital and Family Therapy, 18, 245-256.

 

Bookwala, J ., & Zdaniuk, B. (1998). Adult attachment styles and aggressive
behavior within dating relationships. J oumal of Social and Personal Relationships, 15,
175-190.

Bowlby, J. (1944). Forty-four juvenile thieves: Their character and home life.
International Journal of Psychoanalysis, 21, 154-178.

 

Bowlby, J. (1978). Attachment and loss. Vol.1]: Segaration. New York: Basic
Books.

Bowlby, J. (1979). Psychoanalysis as art and science. International Review of
Psychoanlysis, 6, 3-14.

 

 

Bowlby, J. (1980). Attachment and loss. Vol. III: Loss. New York: Basic
Books.

 

Bowlby, J. (1982). Attachment and loss. Vol. 1: Attachment. New York: Basic
Books.

Bowlby, J. (1984). Violence in the family as a disorder of the attachment and
caregiving systems. The American Journal of Psychoanalysis, 44, 9-27.

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

 

Brennan, K. A., Clark, C. L., & Shaver, P. R. (1998). Self-report measurement of
adult attachment: An integrative overview. In J. A. Simpson & W. S. Rholes (Eds),
Attachment theory and close relationships (pp. 46-76). New York: Guilford Press.

 

Brennan, K. A., & Shaver, P. R. (1995). Dimensions of adult attachment, affect
regulation, and romantic relationship functioning. Personality and Social Psychology
Bulletin, 21, 267-283.

 

Brennan, K. A., & Shaver, P. R. (1998). Attachment styles and personality
disorders: Their connections to each other and to parental divorce, parental death, and
perceptions of parental caregiving. Journal of Personality, 66, 83 5-878.

Browne, A. (1993). Violence against women by male partners: Prevalence,
outcomes, and policy implications. American Psychologist, 48, 1077-1087.

 

Bureau of Justice Statistics. (1994). Violence against women: A national crime
victimization report. Washington, DC: US. Department of Justice.

 

Bureau of Justice Statistics. (1995). Violence against women: Estimates from a
redengned survey. Washington, DC: US. Department of Justice.

 

Caesar, P. L., & Hamberger, L. K. (1989). Treating men who batter: Theory,
practice, and programs. New York, NY: Springer-Verlag.

 

173

Campbell, J. C., & Lewandowski, L. A. (1997). Mental and physical health
effects of intimate partner violence on women and children. The Psychiatric Clinics of
North America, 20, 323-374.

 

Carden, A. D. (1994). Wife abuse and the wife abuser: Review and
recommendations. The Counseling Psychologist, 22, 539-582.

Cohen, J. (1969). Statistical power analysis for the behavioral sciences. New
York: Academic Press.

Cohen, J. (1992). A power primer. Psychological Bulletin, 112,155-159.

Coleman, D. H., & Straus, M. A. (1986). Marital power, conflict, and violence in
a nationally representative sample of American couples. Violence and Victims, 1, 141-
157.

Cook, T.D., & Campbell, D.T. (1979). Quasi-exocrimentation: Design and
analysis issues for field settings. Chicago: McNally College Publishing Company.

 

 

 

Cone, J. D., & Foster, S. L. (1993). Dissertations and theses from start to finish.
Washington, DC: American psychological Association.

Crino, M. D., Svobada, M., Rubenfeld, S., & White, MC. (1983). Data on the
Marlowe-Crowne and Edwards Social Desirability Scales. Psychological Reports, 53,
963-968.

Crowne, D. P., & Marlowe, D. (1964). The approval motive. New York: Wiley.

 

Daugherty, C. G., & Dambrot, F. H. (1986). Reliability of the attitudes toward
women scale. Educational and Psychological Measurement, 46, 449-453,

Dobash, R. E., & Dobash, R. (1979). Violence against wives. New York: Free
Press.

Dutton, D. G. (1988). The domestic assault of women: Psychological and
criminal justice perspectives. Boston, MA: Allyn & Bacon.

 

Dutton, D.G. (1994a). Patriarchy and wife assault: The ecological fallacy.
Violence and Victims, 9, 167-182.

 

Dutton, D.G. (1994b). The origin and structure of the abusive personality.
Journal of Personality Disorders, 8, 181-191.

 

Dutton, D. G. (1995). Intimate abusiveness. Clinical Psychology: Science and
Practice 2 207-224.

 

Dutton, D. G. (1997). Client personality disorders affecting wife assault post-
treatment recidivism. Violence and Victims, 12, 37-50.

 

Dutton, D. G., Bodnarchuk, M., Kropp, R., Hart, S. D., & Ogloff, J. P. (1997).
Client personality disorders affecting wife assault post-treatment recidivism. Violence
and Victims, 12, 37-50.

 

174

 

Dutton, D. G., & Browning, J. J. (1988). Power struggles and intimacy anxieties
as causative factors of wife assault. In G.W. Russell (Ed.), Violence in intimate
relationships (pp.l63-176). New York: PMA Publishing Corporation.

 

Dutton, D. G., Ginkel, C. V., & Starzomski, A. (1995). The role of shame and
guilt in the intergenerational transmission of abusiveness. Violence and Victims, 10,
121-131.

Dutton, D. G., Saunders, K., Starzomski, A., & Bartholomew, K. (1994).
Intimacy-anger and insecure attachment as precursors of abuse in intimate relationships.
Journal of Applied Social Psychology, 24, 1367-1386,

Dutton, D. G., & Strachan, C. E. (1987). Motivational needs for power and
spouse-specific assertiveness in assaultive and nonassaultive men. Violence and Victims,
2, 145-156.

Dutton, D. G., & Starzomski, A. J. (1993). Borderline personality in perpetrators
of psychological and physical abuse. Violence and Victimp, 8, 327-337.

 

Dutton, D. G., Starzomski, A. J ., & Ryan, L. (1996). Antecedents of abusive
personality and abusive behavior in wife assaulters. Journal of Family Violence, 11, 113-
132.

Edleson, J. L., Guttman, E., & Sela-Amit, M. (1995). Cognitive styles and
socialized attitudes of men who batter: Where should we intervene. In S. M. Stith & M.
A. Straus, (Eds) Understandingpartner violence (pp. 69-76). Minneapolis, MN:
National Council on Family Relations.

 

Edleson, J. L., & Tolman, R. M. (1992). Intervention for men who batter.
Newbury Park, CA: Sage Publications.

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

 

Eisikovits, Z. G, Edleson, J. L., Guttman, E., & Sela-Amit, M. (1995).
Cognitive styles and socialized attitudes of men who batter: Where should we intervene?
In S. M. Stith & M. A. Straus (Eds), Understanding partner violence: Prevalence,
causes, consequ_ences, and solutions (pp. 69-76). MN: National Council on Family
Relations.

Elbow, M. (1977). Theoretical considerations of violent marriages: Personality
characteristics of wife abusers. Social Casework, 58, 515-526.

 

F inn, J. (1986). The relationship between sex role attitudes and attitudes
supporting marital violence. Sex Roles, 14, 23 5-244.

 

F loumoy, P. S., & Wilson, G. L. (1991). Assessment of MMPI profiles of male
batterers. Violence and Victims, 6, 309-320.

 

Fonagy, P., Leigh, T, Steele, M., Steele, H., Kennedy, R., Mattoon, G., Target,
M., & Gerber, A. (1996). The relation of attachment status, psychiatric classification,
and response to psychotherapy. Journal of Consulting and Clinical Psychology, 64, 22-
31.

175

 

F ossati, A., Maffei, C., Bagnato, M., Donati, D., Donini, M., F iorilli, M., Novella,
L., & Ansolidi, M. (1998). Brief communication: Criterion validity of the personality
diagnostic questionnaire in a mixed psychiatric sample. Journal of Personality Disorders,
_1_2_, 172-178.

 

Frieze, I.H., & Browne, A. (1989) Violence in marriage. In L. Ohlin & M.
Tonry (Eds), Family violence: Crime and justice, a review of research (pp 163-218).
Chicago: University of Chicago Press.

F uendeling, J. (1998). Affect regulation as a stylistic process within adult
attachment. Journal of Personal and Social Relationships, 15, 291-322.

Gacano, C. B., & Meloy, J. R. (1991). A Rorschach investigation of attachment
and anxiety in antisocial personality disorder. Journal of Nervous and Mental Disease,
179, 546-552.

Gacano, C. B., & Meloy, J. R. (1992). The Rorschach and the DSM-III-R
antisocial personality: A tribute to Robert Lidner. Journal of Clinical Psychology, 48,
393-406.

Gacano, C. B., Meloy, J. R., & Berg, J. L. (1992). Object relations, defensive
operations, and affective states in narcissistic, borderline, and antisocial personality
disorder. Journal of Personality Assessment, 59, 32-49.

 

Ganley, A. (1989). Integrating feminist and social-learning analyses. In P.L.
Ceasar & K.L. Hamberger (Eds), Treating men who batter: Theory, practice, and
programs (pp. 196-235). New York: Springer.

Giles-Sims, J. (1983). Wife battering: A systems-theory approach. Beverly
Hills, CA: Sage.

Goldstein, D., & Rosenbaum, A. (1985). An evaluation of the self-esteem of
maritally violent men. Family Relations, 34,425-428.

 

Gondolf, E. W. (1985). Men who batter: An integrated approach for stopping
wife abuse. Holmes Beach, FL: Learning Publications.

 

Gondolf, E. W. (1988). Who are those guys? Toward a behavioral typology of
batterers. Violence and Victims, 3, 187—202.

 

Gottman, J.M., Jacobson, N.S., Rushe, R.H., Wu Shortt, J, Babcock, J ., La
Taillade, J. J ., & Waltz, J. (1995). The relationship between heart rate activity,
emotionally aggressive behavior, and general violence in batterers. Journal of Family
Psychology, 9, 227-248.

 

Greenwald, H. J., & Satow, Y. (1970). A short social desirability scale.
Psychological Reports, 27, 13 1-135.

 

Gunderson, JG. (1984). Borderline personality disorder. Washington, DC:
American Psychiatric Press.

Haase, R.F., & Ellis, M.V. (1987). Multivariate analysis of variance. Journal of
Counseling Psychology, 34, 404-413.

 

176

Haj-Yahia, M. M. (1997). Predicting beliefs about wife beating among engaged
Arab men in Israel. Journal of Interpersonal Violence, 12, 530-545.

Hale, G., Zimostrad, S., Duckworth, J ., & Nicholas, D. (1988). Abusive partners:
MMPI profiles of male batterers. Journal of Mental Health Counseling, 10, 214-224.

Hamberger, L. K., & Hastings, J. (1988). Characteristics of male spouse abusers
consistent with personality disorders. Hospital and Community Psychiatry, 39, 763-770.

Hamberger, L. K., & Hastings, J. (1991). Personality correlates of men who
batter and nonviolent men: Some continuities and discontinuities. Journal of Family
Violence, 6, 131-147.

 

Hamberger, L. K., & Lohr, J. M. (1989). Proximal causes of spouse abuse: A
theoretical analysis for cognitive behavioral interventions. In P.L. Ceasar & K.L.
Hamberger (Eds), Treating men who batter: Theory, practice, and prggrams (pp. 53-76).
New York: Springer.

Hastings, J. E., & Hamberger, L. K. (1988). Personality characteristics of spouse
abusers: A controlled comparison. Violence and Victims, 3, 31-48.

Hazan, C., & Shaver, P. (1987). Romantic love conceptualized as an attachment
process. Journal of Personality and Social Psychology, 52, 511-524.

Hazan, C., & Shaver, P. (1994). Deeper into attachment theory. Psychological
Inquiry, 5, 68-79.

Heise. L. (1998). Violence against women: An integrated, ecological framework.
Violence Against Women, 4, 262-290.

 

Heppner, P., Kivlighan, D. M., & Wampold, B. E. (1992). Research desiggr in
counseling. Pacific, Grove, CA: Brooks/Cole Publishing Co.

 

Holtzworth-Munroe, A. (1992). Social skills deficits in maritally violent men:
Interpreting the data using a social-information processing model. Clinical Psychology
Review, 12, 605-617.

 

Holtzworth-Munroe, A., Meehan, J.C., Herron, K., & Rehman, U. (2000).
Testing the Holtzworth-Munroe and Stuart batterer typology. Journal of Consulting and
Clinical Psychology, 68, 1000-1019.

Holtzworth-Munroe, A., & Stuart, G. L. (1994). Typologies of male batterers:
Three subtypes and differences among them. Psychological Bulletin, 116, 476-497.

Hotaling, G. T., & Sugarrnan, D. B. (1986). An analysis of risk markers in
husband to wife violence: The current state of knowledge. Violence and Victims, l, 104-
124.

Huitema, B. E. (1980). Analysis of covariance and alternatives. New York:
John Wiley & Sons, Inc.

Hyler, S. E., Rieder, R. 0., Williams, J. B. W., Spitzer, R. L., Hendler, J ., Lyons,
M. (1988). The Personality Diagnostic Questionnaire: Development and preliminary
results. Journal of Personality Disorders, 2, 229-237.

 

177

Hyler, S. E., Skodol, A. E., Oldham, J. M., Kellman, H. D., & Doidge, N. (1992).
Validity of the Personality Diagnostic Questionnaire-Revised: A replication in an
outpatient sample. Comprehensive Psychiatry,33, 73-77.

Hyler, S. E., Skodol, A. E., Oldham, J. M., Kellman, H. D., Oldham, J. M., &
Rosnick, L. (1990). Validity of the Personality Diagnostic Questionnaire-Revised:
Comparison with two structured interviews. American Journal of Psghiatry, 147, 1043-
1048.

Johns, J. L. (1997). Basic Reading Inventory: Pre-primer through grade twelve
and early literacpassessments Dubuque, Iowa: Kendall/Hunt Publishing Company.

 

Johnson, J. G., & Bomstein, R. F. (1991). Personality Diagnostic Questionnaire-
Revised (PDQ-R) personality disorder scores and negative life events independently
predict changes in Hopkins Symptom Checklist (SCL-90) psychopathology scores.
Journal of Psychopathology and Behavioral Assessment, 13, 61-71.

Kemberg, O. (1975). Borderline conditions and pathological narcissism. New
York: Jason Aronson.

Koss, MP. (1990). The women’s mental health research agenda: Violence
against women. American Psychologist, 45, 374-380.

 

Lansky, M. R. (1993). Family genesis of aggression. Psychiatric Annals, 23,
494-499.

 

Lerman, L. G. (1992). The decontextualization of domestic violence. 1h_e
Journal of Criminal Law and Criminology, 83, 201-208.

Livesley, W. J ., Schroeder, M. L., & Jackson, N. J. (1990). Dependent
personality disorder and attachment problems. Journal of Personality DisordersA, 131-
140.

L00, R., & Thorpe, K. (2000). Confirrnatory factor analyses of the full and short
versions of the Marlowe-Crowne Social Desirability Scale. Journal of Social
Psychology, 140, 628-635.

 

Lopez, F. G., & Brennan, K. (2000). Dynamic processes underlying adult
attachment organization: Toward an attachment theoretical perspective on the healthy and
effective self. Journal of Counseling Psychology, 47, 283-300.

Macchietto, JG. (1992). Aspects of male victimization and female aggression:
Implications for counseling men. Journal of Mental Health Counseling, 14, 375-392.

Main, M. (1990). Cross-cultural studies of attachment organization: Recent
studies, changing methodologies, and the concept of conditional strategies. Human
Development, 33, 31-47.

 

Main, M., & Solomon, J. (1990). Procedures for identifying insecure-
disorganized/disoriented infants in the Ainsworth strange situation. In M. Greenberg, D.
Cichetti, & M. Cummings (Eds), Attachment in the preschool years: Theory, research,
and intervention (pp. 121-160). Chicago, IL: University of Chicago Press.

 

178

 

Mayseless, O. (1991). Adult attachment patterns and courtship violence. Family
Relations, 40, 21-28.

 

McHugh, M. C., & Frieze, I.H. (1997). The measurement of gender-role
attitudes: A review and commentary. Psychology of Women Quarterly, 21, 1-16.

Miller, S. (1994). Expanding the boundaries: Toward a more inclusive and
integrated study of intimate violence. Violence and Victims, 9, 183-194.

 

Millon, T. (1981). Disorders of personality. New York: Wiley.

Millon, T. (1987). Millon Clinical Multiaxial Inventory-II (2"‘1 ed;
Minneapolis: National Computer Systems, Inc.

Murphy, C. M., & Dienemann, J. A. (1999). Informing the research agenda on
domestic abuser intervention through practitioner-researcher dialogues. Journal of
Interpersonal Violence, 14, 1314-1326.

 

Murphy, C. M., Lee-Meyer, S. L., & O' Leary, K. D. (1993). Family of origin
violence and MCMI-II psychOpathology among partner assaultive men. Violence and
Victims, 8, 165-176.

 

 

National Institute of Justice. (2000). Prevalence, incidence, and consequences of
violence against women. Washington, DC: US. Department of Justice.

 

 

National Institute of Justice. (1998). Batterer intervention: Program approaches
and criminal justice strategies. Washington, DC: US. Department of Justice.

 

O' Grady, K. E. (1988). The Marlowe-Crowne and Edward Social Desirability
scales: A psychometric perspective. Multivariate Behavioral Research, 23, 87-101.

0’ Neil, J.M., Good, G. E., & Holmes, S. (1995). Fifteen years of theory and
research on men’s gender role conflict: New paradigms for empirical research. In R.F.
Levant & W.S. Pollack (Eds), A new psychology for men (pp. 164-206). New York,
NY: Basic Books.

 

f

O’ Neil, J. M., & Harway, M. (1997). A multivariate model explaining men’s
violence toward women. Violence Against Women, 3, 182-203.

 

Pagelow, M. D. (1992). Adult victims of domestic violence: Battered women.
Journal of Interpersonal Violence, 7, 87-120.

Patrick, M., Hobson, R. P., Castle, D., Howard, R., & Maughan, B. (1994).
Personality disorder and the mental representation of early social experience.
Development and Psychopathology, 6,375-3 88.

 

Pence, E. (1989). Batterer programs: Shifting from community collusion to
community confrontation. In P.L. Caesar & L. K. Hamberger (Eds), Treating men who
batter: Theory, practice, and programs (pp. 24-50). New York: Springer.

 

Roberts, N., & Noller, P. (1998). The associations between adult attachment and
couple violence: The role of communication patterns and relationship satisfaction. In J.
A. Simpson & W. S. Rholes (Eds), Attachment theory and close relationships (pp. 317-
350). New York: Guilford Press.

 

179

Rosenbaum, A., & Mauiro, R. D., (1989). Eclectic approaches in working with
men who batter. In P.L. Ceasar & K.L. Hamberger (Eds), Treating men who batter:
Theory, practice, and programs (pp. 165-195). New York: Springer.

 

Rosenstein, D. S., & Horowitz, H. A. (1996). Adolescent attachment and
psychopathology. Journal of Consulting and Clinical Psychology, 2,244-253.

Russell, M. (1988). Wife assault theory, research, and treatment: A literature
review. Journal of Family Violence, 3,193-208.

 

Sable, P. (1997). Attachment, detachment, and borderline personality disorder.
Psychotherapy, 34, 171-181.

 

Sack, A., Sperling, M. B., F agen, G., & Foelsch, P. (1996). Attachment style,
history, and behavioral contrasts for a borderline and normal sample. Journal of
Personality Disorders, 10, 88-102.

 

Saunders, D. G. (1992). A typology of men who batter: Three types derived
from cluster analysis. American Journal of Orthopsychiatry, 62, 264-275.

Saunders, D. G. (1993). Husbands who assault: Multiple profiles requiring
multiple responses. In N. Z. Hilton (Ed.), Legal responses to wife assault (pp. 9-34).
Newbury Park, CA: Sage.

 

Saunders, D. G. (1996). F eminist-cognitive-behavioral and process-
psychodynarnic treatments for men who batter: Interaction of abuser traits and treatment
models. Violence and Victims, 11, 393-413.

 

Scharfe, E., & Bartholomew, K. (1994). Reliability and stability of adult
attachment patterns. Personal Relationships, 1, 23-43.

 

Scharfe, E., & Bartholomew, K. (1995). Accommodation and attachment
representation in young couples. Journal of Social and Personal Relationships, 12, 389-
401.

Schmidt, J. D., & Sherman, L. W. (1996). Does arrest deter violence? In E. S.
Buzawa & C. G. Buzawa (Eds), Do Arrests and restraining orders work? (pp. 43-53).
Thousand Oaks, CA: Sage Publications, Inc.

Sharpsteen, D.J., & Kirkpatrick, LA. (1997). Romantic jealousy and adult
romantic attachment, Journal of Personality and Social Psychology, 72, 627-640.

Shavelson, R. J. (1988). Statistical reasoning for the behavioral sciences.
Boston: Allyn and Bacon, Inc.

Sheldon, A. E., & West, A. M. (1990). Attachment pathology and low social
skills in avoidant personality disorder: An exploratory study. Canadian Journal of
Psychiatry, 35, 596-599.

 

 

Simpson, J.A. (1990). Influence of attachment styles on romantic relationships.
Journal of Personality and Social Psychology, 59, 971-980.

 

180

 

Simpson, J .A., Rholes, W.S., & Nelligan, J .S. (1992). Support-seeking and
support-giving within couples in an anxiety-provoking situation: The role of attachment
styles. Journal of Personality and Social Psychology, 62, 434-446.

Smith, M. D. (1990). Patriarchal ideology and wife beating: A test of the
feminist hypothesis. Violence and Victims, 5, 257-273.

 

Smith, M. D. (1994). Enhancing the quality of survey data on violence against
women: A feminist approach. Gender and Society, 8, 109-127.

Smith, R. L., & Bradley, D. W. (1980). In defense of the Attitudes Toward
Women Scale: An affirmation of reliability and validity. Psychological Reports, 47, 511-
522.

Sorenson, S. B., & Telles, C. A. (1991). Self-reports of spousal violence in a
Mexican-American and Non-Hispanic white population. Violence and Victims, 6, 3-16.

Spence, J. T., & Hahn, E. D. (1997). The attitudes toward women scale and
attitude change in college students. Psychology of Women Quarterly, 21, 17-34.

Spence, J. T., & Helmrich, R. (1972). The attitudes toward women scale: An
objective instrument to measure attitudes towards the rights and roles of women in
contemporary society. Catalogue of Selected Documents in Psychology, 2, 66-67.

Spence, J. T., Helmrich, R., & Stapp, J. (1973). A short version of the attitude
toward women scale. Bulletin of the Psychonomic Society, 2, 219-220.

Steiger, J. H. (1980). Tests for comparing elements of a correlation matrix.
Psychological Bulletin, 87, 245-251.

 

Strachan, C. E., & Dutton, D. G. (1992). The role of power and gender in anger
responses to sexual jealousy. Journal of Applied Social Psychology, 22,1721-1740.

Straus, M. A. (1979). Measuring intrafamily conflict and violence: The Conflict
Tactics Scales. Journal of Marriage and the Family, 41, 75-88.

Straus, M. A. (1990a). Measuring intrafamily conflict and violence: The Conflict
Tactics Scales. In M. A. Straus & R. J. Gelles (Eds), Physical violence in American
families (pp. 29-48). New Brunswick: Transaction Publishers.

Straus, M. A. (1990b). The Conflict Tactics Scales and its critics: An evaluation
and new data on validity and reliability. In M. A. Straus & R. J. Gelles (Eds), Physical
violence in American families (pp. 49-74). New Brunswick: Transaction Publishers.

Straus, M. A. (1990c). New scoring methods for violence and new norms for the
Conflict Tactics Scales. In M. A. Straus & R. J. Gelles (Eds), Physical violence in
American families (pp. 535-560). New Brunswick: Transaction Publishers.

 

Straus, M. A. (1991). Conceptualization and measurement of battering:
Implications for public policy. In M. Steinman (Ed.), Woman battering: Policy
resmnses (pp. 19-47). Cincinnati: Anderson Publishing Co.

181

 

Straus, M. A., & Gelles, R. J. (1986). Societal change and change in family
violence from 1975 to 1985 as revealed by two national surveys. Journal of Marriage and
the Family, 48, 465-479.

 

Straus, M., Gelles, R., & Steinmetz, S. (1980). Behind closed doors: Violence in
the American family. Garden City, NY: American Press.

 

Sugarrnan, D. B., & Frankel, S. L. (1996). Patriarchal ideology and wife assault:
A meta-analytic review. Journal of Family Violence, 11,13-40.

Swim, J. K., & Cohen, L. L. (1997). Overt, covert, and subtle sexism: A
comparison between the Attitudes Toward Women and the Modern Sexism Scales.
Psychology Women Quarterly, 21, 103—118.

 

Tabachnick, B. F., & F idell, L. S. (1989). Using multivariate statistics. New
York: Harper Collins.

Tolman, R. M. (1989). The development of a measure of psychological
maltreatment of women by their male partners. Violence and Victims, 4, 159-177.

Tolman, R. M. (1999). The validation of the psychological maltreatment of
women inventory. Violence in Victims, 14, 25-37.

 

Tolman, R. M., & Edleson, J. L. (1995). Intervention for men who batter: A
review of research. In S. Stith & M. A. Straus (Eds), Understanding partner violence:
Prevalence, causespconsec uences, and solutions (pp. 262-273). Minneapolis, MN:
National Council on Family Relations.

 

Tolman, R.M., Edleson, J .L., & Fendrich, M. (1996). The applicability of the
theory of planned behavior to abusive men’s cessation of violent behavior. Violence and
Victims, 11, 341-354.

 

Troy, M., & Sroufe, L. A. (1987). Victimization among preschoolers: Role of
attachment relationship history. Journal of the American Academy of Child and
Adolescent Psychiatry, 26, 166-172.

 

Tweed, R. G., & Dutton, D. (1998). A comparison of impulsive and instrumental
subgroups of batterers. Violence and Victims, 13, 217-230.

 

Walker, L. E. A. (1996). Abused women and survivor therapy. Washington DC:
American Psychological Association.

 

West, M., Keller, A., Links, P., & Patrick, J. (1993). Borderline disorder and
attachment pathology. Canadian Journal of Psychiatry, 38 (Supplement 1), 16-21.

Wierzbicki, M, & Gorman, J. L. (1995). Correspondence between students'
scores on the Millon Clinical Multiaxial Inventory-II and Personality Diagnostic
Questionnaire-Revised. Psychological Reports, 77, 1079-1082.

 

Woike, B. A., Osier, T. J., & Candela, K, (1996). Attachment styles and violent
imagery in thematic stories about relationships. Personality and Social Psychology
Bulletin, 10, 1030-1034.

 

182

Yllo, K. (1983). Sexual equality and violence against wives in American states.
Journal of Comparative Family Studies, 14, 67-86.

Yoder, J. D., Rice, R., Adams, J., Priest, R. F ., & Prince, H. T. (1982).
Reliability of the Attitudes Toward Women Scale and the personal attributes
questionnaire. Sex Roles, 8, 651-657.

 

Zeanah, C. H., & Zeanah, PD. (1989). Intergenerational transmission of
maltreatment: Insights from attachment theory and research. Psychiatry, 52, 177-196.

183