xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001711501
007 cr mnu|||uuuuu
008 060530s2005 flua sbm s000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001320
Differences in frequency and severity of violence for intimate terrorism across genders
h [electronic resource] :
b a test of Johnson's theory /
by Shelly Wagers.
[Tampa, Fla.] :
University of South Florida,
Thesis (M.A.)--University of South Florida, 2005.
Includes bibliographical references.
Text (Electronic thesis) in PDF format.
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
Title from PDF of title page.
Document formatted into pages; contains 85 pages.
ABSTRACT: This study sought to further build on previous empirical findings regarding Johnsons theory that the gender symmetry debate can at least be partially resolved by acknowledging that two distinct subgroups of physical violence exist within intimate partner violence: Intimate Terrorism (IT) and Situational Couple Violence (SCV). According to Johnsons predictions these separate groups can be distinguished by the use of non-violent control tactics. This study focused on testing the ability of non-violent control tactics to predict the frequency and severity of violence within the sub-group intimate terrorism. It further explored Johnsons assertion that intimate terrorism is gender asymmetric with females experiencing a greater amount of victimization. Previous studies demonstrated moderate support that two subgroups do exist within intimate partner violence and that intimate terrorism may be asymmetrical.However, only one of the previous studies included a male sample that was not reflective of the general population. This study will test the gender asymmetry of intimate terrorism by using both a male and female sub-sample form the National Violence Against Women Survey. This studies sample consisted of males and females reporting at least on incident of physical violence by either their current spouse or cohabitating heterosexual partner. The statistical analysis showed moderate support that there are two subgroups within intimate partner violence that can be distinguished by the use of non-violent control tactics. It also demonstrated that for the subgroup intimate terrorism there are some differences across gender when examining severity and frequency of violence. However, only a small amount of the variance in intimate terrorism can be explained by non-violent control tactics.
Adviser: Christine Sellers, PhD.
Power and control.
t USF Electronic Theses and Dissertations.
Differences in Frequency and Severity of Violence For Intimate Terrorism Across Genders: A Test of Johnson's Theory by Shelly Wagers A thesis submitted in partial fulfillment of the requirements for the degree of Master of Arts Department of Criminology College of Arts and Sciences University of South Florida Major Professor: Christine Sellers, Ph.D. John Cochran Ph.D. Wilson Palacios Ph.D. Date of Approval: November 14, 2005 Keywords: domestic, intimate partner, power and control, symmetr y, non-violent control Copyright 2005 Shelly Wagers
Dedication To my best friend and partner Frieda Widera whose time and energy was given to me freely in support of this work and whose lifeÂ’s passion is also reflected among these pages. This work is also de dicated to my family and friends who have always believed in me, no matter what I chose to pursue. I coul d not have completed this project without their unending love and support. I also want to especially acknowledge my grandmother, Rosemarie who always wanted to hear what I had to say. I wish you were here to share in the completion of this project.
Acknowledgements I would like to thank my thesis committ ee chair Dr. Christine Sellers and my two committee members Dr. John Cochcran and Dr. W ilson Palacios for all of their time and guidance. I believe a key to any studentÂ’s succe ss is the ability of their instructors to be mentors who always provide direction, but allo w the student to pursue their own theories. These were the qualities provided to me by my committee. Thank you Dr. Sellers for always being diligent with my writing, as we ll as guiding the theore tical foundations of this project. Thank you Dr. Cochran for cha llenging me to work through the process and ensure this project was methodologically sound. Thank you Dr. Palacios for always being encouraging and providing your truly uni que insight. This combination provided me the tools to truly excel and learn. I also want to thank my employer Remington College who supported my educational goal and allowed the flexibility for me to complete this degree while continuing to teac h. Last but not least I want to acknowledge Frieda Widera for all of her insight into this phenomenon and for editing the preliminary pages of this thesis.
i Table of Contents List of Tables ................................................................................................................ .....ii List of Figures................................................................................................................ ....iii Abstract....................................................................................................................... .......iv Chapter One ................................................................................................................... .....1 Introduction..............................................................................................................1 Chapter Two.................................................................................................................... .....9 Understanding the Gendered Nature of Intimate Partner Violence.........................9 History of the Intimate Partne r Violence Movement and Research............9 Understanding the Causes of the Gender Symmetry Debate.....................14 A proposed Answer to Reconcile the Gender Symmetry Debate..............22 Current Empirical Findings Regarding JohnsonÂ’s Theory........................28 The Purpose of the Current Study..............................................................31 Chapter Three.................................................................................................................. ...34 Methods..................................................................................................................34 Sample........................................................................................................34 Procedures..................................................................................................36 Measures....................................................................................................37 Analytic Strategy.......................................................................................43 Chapter Four................................................................................................................... ...45 Results ....................................................................................................................45 Frequency and Severity of Violence Among Female Victims..................49 Frequency and Severity of Violence Among Male Victims......................53 Comparison Between Male and Female Samples......................................56 Summary....................................................................................................58 Chapter Five................................................................................................................... ....60 Discussion..............................................................................................................60 References .................................................................................................................... 71
ii List of Tables Table 1 Descriptive Statistics for Male and Female Sample..................................47 Table 2 Correlation Matrix for the Male and Female Sample................................48 Table 3 Female Negative Binomial Regression Models: Intimate Terrorism Predicting Frequency of Violence............................................50 Table 4 Female Logistic Regres sion Models: Intimate Terrorism Predicting Severity of Violence.................................................................52 Table 5 Male Negative Binomial Regr ession Models: Intimate Terrorism Predicting Frequency of Violence.............................................................54 Table 6 Males Logistic Regression M odels: Intimate Terrorism Predicting Severity of Violence..................................................................................55 Table 7 Significance Test for Ge nder Differences in Regression Coefficients for Frequency and Severity of Violence................................58
iii List of Figures Figure 1. The Cycle of Violence...............................................................................12 Figure 2. Power and Control Wheel..........................................................................21
iv Difference in Frequency and Severity of Intimate Terrorism Across Genders: A Test of JohnsonÂ’s Theory Shelly Wagers ABSTRACT This study sought to further build on previous empirical findingÂ’s regarding JohnsonÂ’s theory that the gender symmetry deba te can at least be pa rtially resolved by acknowledging that two distinct subgroups of physical viol ence exist within intimate partner violence: Intimate Terrorism (IT) and Situational Couple Violence (SCV). According to JohnsonÂ’s predictions these sepa rate groups can be distinguished by the use of non-violent control tactics. This study focused on testing the ability of non-violent control tactics to predict the frequency a nd severity of violen ce within the sub-group intimate terrorism. It further explored J ohnsonÂ’s assertion that intimate terrorism is gender asymmetric with females experien cing a greater amount of victimization. Previous studies demonstrated moderate support that two subgroups do exist within intimate partner violence and that intimate terrorism may be asymmetrical. However, only one of the previous studies included a male sample that was not reflective of the general population. This study will test the gender asymme try of intimate terrorism by using both a male and female sub-sample form the National Violence Against Women Survey. This studies sample consisted of ma les and females reporting at least on incident of physical violence by either their current spouse or cohabi tating heterosexual partner. The statistical analysis showed moderate support that there are two subgroups within intimate partner violence that can be dis tinguished by the use of non-violent control
v tactics. It also demonstrated that for th e subgroup intimate terrorism there are some differences across gender when examining seve rity and frequency of violence. However, only a small amount of the variance in intimate terrorism can be explained by non-violent control tactics.
1 Chapter One Introduction Over the past thirty years intimate part ner violence has emerged as one of the worldÂ’s most pressing problems. The Unite d Nations has estimated that between 20% and 50% of all women worldwide have experi enced some form of physical violence at the hands of their intimate partner or othe r family members (Kimmel, 2002). According to the U.S. Department of Justice, more than one million cases of intimate partner violence are reported to police each year (see Goldberg, 1999). The U. S. Department of Health and Human Services stated that understanding and preventing intimate partner violence has become a national public health i ssue and listed injury and violence as one of the ten major national health issues (G oldberg, 1999). For the past two decades, efforts to reduce the prevalence and inciden ce rates of intimate partner violence have followed the findings from various empirical studies. For example, new laws and police procedures were established, refuges (Dom estic Violence Shelters) for victims were created, and therapy groups for perpetrators were started, all of which had the same goal or objective of reducing the incidence rates for intimate partner violence. However, the incidence rates along with the domestic homicid e rates are still high, not only in this country but throughout the world. A person may wonder how this is possible when so much has been learned and accomplished over th e past thirty years to prevent intimate partner violence. The answer may be in part because, in the process of working towards
2 understanding and reducing intimate partner vi olence, a great debate among groups has erupted regarding the nature of intimate partner violence, especially regarding the gender of its perpetrators and victims (Kimmel, 2002; Pleck, Pleck, Grossman, & Bart, 1978; DeKeseredy & Schwartz, 1998). Intimate partner violence as an issue was first brought to th e attention of the public by feminist activists in the early 1970Â’ s. Although numerous st udies report that the preponderance of intimate partner viol ence is perpetrated by men against women, a growing number of researchers and political activists claim that women and men are equally victimized (Archer, 2000). As a result, activists for Â“menÂ’s rightsÂ” have suggested that policy efforts re garding this issue have been misplaced because of their failure to include male victims (Kimmel, 2002) These groups argue that intimate partner violence is gender symmetric, which is a di rect contradiction to the argument of the feminists, who state that wome n are disproportionately victimized (gender asymmetry). Feminist activists believe th at although these Â“menÂ’s ri ghtsÂ” groups help to draw attention to the often ignored problem of male victimizati on and female perpetration of violence, their efforts often undermine initiativ es that assist female victims (Kimmel, 2002). In multiple scholarly publications across many disciplines, the empirical findings consistently demonstrate high incidence rates of intimate partner violence and conclude that this is a major issue that needs to be studied and addressed (Dobash & Dobash, 1992; Gelles, 2000; Straus 1993, 1999; Saltzman, 2004; Tjaden & Thonnes, 2000). However, there are also great discrepancies in the lite rature regarding how each gender is affected, and there are no clear agreements about its magnitude for either sex (Johnson, 1995;
3 Johnson & Ferraro, 2000; Kimmel, 2002; Kurz 1989). Empirical studies that are grounded in feminist theory consistently indica te that males are much more likely to be perpetrators and females are disproportionately victims of intimate partner violence compared to men (gender asymmetry). On the other hand, research grounded in family conflict theory has cons istently shown an equal perpetrati on and victimization of intimate partner violence by males and females (Arche r, 2000; Johnson, 1995). This discrepancy between gender symmetry and gender asymme try has led to significant confusion among the general public and policy makers. It has also become an increasingly controversial issue among scholars and at times it even ove rshadows discussions about the prevention of intimate partner violence (Saltzman, 2004). Over the past decade several reasonable e xplanations and possible solutions to the debate have been proposed. For example, a key to measuring any phenomenon is a standard definition. Empirical studies on inti mate partner violence va ry greatly in their definition, which causes differences in how it is being measured (Saltzman, 2004). To resolve this discrepancy, in 1994, the Cent ers for Disease Control created a uniform definition for intimate partner violence (Saltz man, et al. 1999). A second explanation for the differences in research findings is that various types of methodological approaches are used. For example, the feminist and fa mily conflict theorists tend to sample from different types of populations, and the theore tical framework of their surveys vary. Feminist researchers repeatedly use small sa mples from places such as domestic violence shelters or hospital emergency rooms, and generally employ qualitative interviews to obtain detailed information on the context and mo tivation of the violent act. On the other hand, family conflict theorists typically use large random samples of the general
4 population and often employ larg e scale surveys which simp ly count the number of violent acts without accounti ng for context or motivation of the violent act (Johnson, 1995). Yet, even with these plausible explanati ons, the debate still continues today. Due to current complexities, challenges, and c ontinually high prevalen ce rates it is becoming increasingly imperative to stop arguing about gender symmetry or gender asymmetry and begin to propose possible solutions to the deba te. The process of resolving this debate has been compared to solving a puzzle (Dobash & Dobash, 2004). Recently, Johnson took several of the Â“puzzle pieces,Â” such as varying definitions, sampling techniques, and the use of differing methodologies, and proposed a possible solution to the puzzle. He wove these Â“puzzleÂ” pieces toge ther as two different pictures or explan ations rather than one. In other words, he propos ed that within intimate partne r violence there are actually two distinct types of violence occurring. He further theorized that these types of violence are clearly two different phenomena, and the disc repancies in the resear ch are a result of measuring them as a single phenomenon. According to Johnson, feminist research ers have been tapping into a phenomenon he refers to as Â“intimate terrorism,Â” and the family conflict theorists have been measuring the phenomenon he calls Â“situational couple vi olence.Â” The key to distinguishing these two types of intimate partner violence is th e context and motivation behind the violent act. In intimate terrorism, the perpetrator us es physical violence as a motive to maintain a Â“control contextÂ” over the vi ctim and the relationship in general. In this case the physical violence used is only one type of c ontrol method exerted by the perpetrator. In situational couple violence, th e physically violent act is not motivated by a context of
5 control but is a reaction to a current conflict. Johnson states that the gender symmetry versus gender asymmetry debate can be answ ered by this theory. He proposes that intimate terrorism is gender asymmetric, w ith females experiencing disproportionate victimization compared to males, and s ituational couple violence displays gender symmetry. Based on this theoretical fram ework Johnson provided several hypotheses that could be tested empirically. For ex ample, he proposed that the frequency and severity of physical violence would be greate r with intimate terrorism when compared to situational couple violence. The initial studies condu cted regarding this theory were focused on first establishing the major tenet that intimate partne r violence can be divided into two distinct groups based on the perpetratorÂ’s Â“control mo tiveÂ”. To conceptual ize and operationalize the Â“control motiveÂ” in intimate terrorism, Johns on referred to over 30 years of social and feminist research (Johnson, 1995). He specifically uses their definitions of Â“batteringÂ” or Â“battererÂ” to conceptualize his definition of in timate terrorism. Femi nists define battering as a pattern of coercive beha vior that serves to gain pow er and control over another individual. Johnson states that his conceptualization of an in timate terrorist is consistent with the feminist concept of a Â“battererÂ” (Johnson, in press). In order to operationalize a measurement for the Â“control motive,Â” he references the work of Pence and Paymer (1993) and their development of the Power and Control Wheel. The Power and Control Wheel has become the most commonly accepte d and widely used model for Â“battererÂ” treatment programs, and its concepts are cons istently used by advocates to discuss the dynamics of intimate partner violence. Th is model identifies eight areas used by a Â“battererÂ” to control an in timate partner. Then it demonstrates the use of physical
6 violence as the circle that su rrounds or holds all these areas together. Johnson was able to show that individuals who used physical violence on their intimate partner could be divided into two groups based on having ei ther a high or low Â“non-violent control motive.Â” Once he divided the two groups by their high or low Â“control motive,Â” he compared the frequency and severity of violence between the groups. He found that those with high non-violent control had a hi gher mean frequency of physical violence against their intimate partner compared to those with low non-vi olent control (Johnson, 1999; Johnson & Leone, 2005). Although the few studies conducted thus fa r show support for the major tenet of JohnsonÂ’s theory that there may be two distin ct groups within intimate partner violence, much still needs to be done. Previous st udies have compared the mean frequency and mean severity for the two groups of intimat e partner violence differentiated by being either Â“highÂ” or Â“lowÂ” non-violent control. However, this technique can only indicate whether or not the non-violent control variable can distinguish between intimate terrorism and situational coupl e violence in their frequency and severity of violence. Instead, by using a predictive statistical mode l rather than a simple comparison between groups, it is possible to build upon and strengthen th e previous findings in two ways. First, a predictive model allows the researcher to control for other variables that may also affect the frequency and severity of intimat e partner violence. Second, it can tell the researcher how much of the variance in fr equency and severity of violence can be explained by non-violent control tactics. The present study proposes to utilize a predictive rather than a comparison model that will be able to demonstrate how much of the variance in frequency and severity of vi olence is explained by the non-violent control
7 variable. At the same time, it will contro l for the following variables: age, race, employment status, educational level, and le ngth of time together. Controlling for these variables is an important addition to Johns onÂ’s previous studies because each one is identified in various empirical studies as having an effect on incidence rates of intimate partner violence. One weakness in tests of JohnsonÂ’s theory stems from JohnsonÂ’s assertion that intimate terrorism is gender asymmetric, with females disproportionately experiencing victimization. JohnsonÂ’s previous studies fail ed to truly test this hypothesis because he failed to include male participants in his samp le. In fact, males were only included in one study conducted by Graham-Kevan & Archer (2003), and the male population used included traditional age college students and in mates, which is not truly reflective of the general population (Archer, 2000). In order to test J ohnsonÂ’s notion that intimate terrorism is gender asymmetric, both female and male samples reflective of the general population must be included in the study. Th en it is possible to make an objective comparison of the male and female sample for each hypothesis tested and from there draw some conclusions regarding gender asymmetry or symmetry. This study will focus on building upon JohnsonÂ’s previous empirical findings regarding the non-violent control variableÂ’s ability to disti nguish intimate terrorism and situational couple violence by accounting for how much of the variance in frequency and severity of violence can be explained by use of non-violent control ta ctics. Then, it will address JohnsonÂ’s assertion that intimate terrorism is gender asymmetric by comparing the results of the analysis for female and ma le victims of intimate partner violence. The next chapter will provide review of the lite rature and a detailed de scription of JohnsonÂ’s
8 complete theory. Chapter three will review the methods employed to test each hypothesis and how the sample was selected. Chapter four will provide the result s from the analysis conducted. Chapter five will discuss the imp lications of the study for our understanding of the gendered nature of intimate partner vi olence, identify the studyÂ’s limitations, and provide recommendations for future research.
9 Chapter Two Understanding the Gendered Nature of Intimate Partner Violence History of The Intimate Partne r Violence Movement and Research One of the most emotionally and politically charged topics in the social sciences today is the issue of physical and sexual a buse of women by their intimate partner (Yllo, 1988). Historically, this phenomenon was called "domestic violence" and it was not considered a social issue until about thirty y ears ago. However, today the term "domestic violence" is often interchanged with the phr ase "intimate partner violence" (IPV), and it has become common to read newspaper articles and see television programs discussing it. The issue of intimate partner violence was first brought to the public's awareness in the early 1970Â’s as a result of the women' s movement. Initially, the problem was not studied by researchers (Dutton & Gondolf, 2000). Instead, this phenomenon was first identified by feminist activists at the grass-roots level, w ho were speaking out about the violence women were experiencing at the ha nds of their husbands. Their initial focus was on how to keep victims of intimate part ner violence safe. Their work started by establishing underground refuges that evolved into 24-hour Â“domestic violenceÂ” centers (Dobash & Dobash, 1988). It was not unt il the mid-1970Â’s that scholars began conducting empirical research to scientifically study intimate partner violence. Then in the late 1970Â’s into the 1980Â’s, an explosion of research articles began to appear in
10 scientific journals. These articles not onl y studied the prevalence of intimate partner violence, but also tried to explain the causes (Bograd, 1988) As the intimate partner violence research has grown, so have di visions between the scholars who study this phenomenon. The current concep tions of intimate partner vi olence have developed as a result of the convergence of two traditions : the advocacy movement and the social and behavioral research on intimate partner vi olence (Gordon, 2000). In order to better understand intimate partner violence one must first appreciate the history of the "domestic violence" movement, and then exam ine the empirical studies related to it. The domestic violence movement began in 1971 when a small group of women in England were working to put into practice the principles of the women's movement (Dobash & Dobash, 1988). They decided to set up community meetings and an advice center for women. As these women began to talk, they discovered many were experiencing brutal and habitual attacks by their husband or co-habitants. Soon, the locations for these meetings became 24-hour safe refuges for the women and the concept of a "domestic violence shelter" began (Dobash & Dobash, 1988). Shortly after this quiet beginning, the social problem of intimate partne r violence came to the attention of the British public and European scholars. In Europe, Dobash and Dobash began scientifically studying intimate partner violence from the feminist perspective (Dobash & Dobash, 1979). Then in the mid to late 1970Â’s, advocates and sc holars in the United States also began to investigate intimat e partner violence. Lenore Walker began conducting interviews with "battered women" across America and, based on her findings, published a book titled The Battered Woman (Walker, 1979). Walker's book became a significant source of knowledge fo r advocates in the intimate partner violence field and
11 established the theory of the cycle of violence (Walker, 1979). The cycle of violence (see Figure 1) consists of three di stinct phases that continually move in an unbroken circle. The amount of time between phases or to comp lete the cycle is not exact and varies among individuals. Phase one is character ized by tension building and can include arguing, blaming, and anger. It can last for a short or l ong period of time before phase two begins. Phase two is often called the explosion phase and it is when the physical violence or verbal threats of violence occur. This phase is usually over quickly and is episodic. Then phase three begins, which is sometimes referred to as either the honeymoon phase or the calm stage because it is characterized by the offender showing remorse and apologizing for the violence. Th e concepts Walker developed and the cycle of violence are still used toda y by trained clinicians, social workers, and counselors as a basis for understanding the Â“dynami csÂ” of intimate partner violence. At the same time feminist advocates we re promoting societal recognition and criminalization of intimate partner violence, researchers began extending intimate partner violence into the criminological and family studies literature (Dobash & Dobash, 1988; McNeely & Jones, 1980; Yllo, 1988). Alt hough both researchers and advocates were working towards accomplishing the same goa l of ending violence against women, they did not all approach this phe nomenon from a feminist perspective (Yllo, 1988). In fact, within the intimate partner vi olence literature one can find great divisions of thought, a variety of theories, and countle ss numbers of empirical studies. For example, some social scientists (Kaufman & Zeigler, 1993; Dutton, 1980; 1988; 1995) have applied the concepts of social learning theory to explai n intimate partner violen ce. In the beginning, social learning seemed to offer some prom ise for the explanati on of intimate partner
12 violence. It was able to make predicti ons regarding the likelihood that a child who witnessed parental violence would later beco me an abusive spouse. Later, however, these predictions were only partially confirmed and yielded mixed results (Dutton, 1980, 1988, 1995; OÂ’Leary, 1988; Kalmuss, 1984). For example, Kaufman and ZiglerÂ’s (1993) study showed that only 18% of children w ho witnessed parental violence exhibited spousal aggression as adults. Despite mixed results, Ganley (1981) was able to develop a treatment model based on social learning theo ry for court-mandated perpetrators that is still used today. Due to these various findi ngs, many researchers still believe that even though social learning cannot explain all viol ence, it is still an important factor in understanding intimate partner violence (Dobash & Dobash, 2004; Straus, 1991). Figure 1: The Cycle of Violence Cycle of Violence PHASE 1 Increased tension, anger, blaming and arguing. PHASE 3 Calm Stage (this stage may decrease over time). Perpetrator may deny violence, blame drinking, apologize, and promise it will never happen again. PHASE 2 Battering, hitting, slapping, kicking, choking, use of objects or weapons. Sexual abuse. Verbal threats and abuse. From Â“Dynamics of Domestic ViolenceÂ” by F. A. Wide ra, 2002, Instruct ors Manuel p..43 Copyright 2002 by the Florida Regional Community Policing Institute. Adapted with permission from the author.
13 Another prominent area of research in in timate partner violence is to study the personality characteristics and psychopathol ogy of perpetrators. This explanation of intimate partner violence suggests that vi olent individuals may have a personality disorder, violence is not a normal occurrence, and the perpetrators are "sick" (Pagelow, 1984). Researchers who focus in this area ha ve found evidence to support the theory that perpetrators of intimate partne r violence may have distinct ty pes of personality disorders such as narcissistic/a ntisocial, avoidant/dependent, a nd severe pathology (Gondolf, 1997; Hamberger & Hasting, 1986; Saunders, 1992). Thes e empirical studies also suggest that violent men have a higher leve l of depression, lower self-est eem, and a greater need for power compared to men who do not engage in intimate partner violence (Dutton & Strachan, 1987; Julian & Mcknery, 1993; Vivian & Malone, 1997). Based on these findings Dutton and other researchers theorize that these characteristics, such as borderline personality organization, may interact with learned behavior, resulting in anger and violence (Dutton & St arzomski, 1993; Gondolf, 1990). Although the social learning and psyc hopathology research findings have contributed greatly to our unde rstanding of intimat e partner violence, the majority of empirical studies and their results can be clas sified into one of tw o larger perspectives: the feminist or the family conflict model. Approximately thirty years ago, a major disagreement among scholars from these two perspectives regarding the nature of intimate partner violence began. This dis pute is referred to as the gender symmetry versus gender asymmetry debate (Johnson, 1995; Kimmel, 2002) and ca n be traced back to the late 1970s.
14 Understanding the Causes of the Gender Symmetry Debate The debate began when Suzanne Steinmet z published a paper titled Â“The Battered Husband SyndromeÂ”. Based on data collected from the National Family Violence Survey (NFVS), Steinmetz (1977-1978) repo rted that women were as violent as men. She went on to propose that there was a problem of "husband battering" equivalent to the prevalence and seriousness of wife battering (Steinmetz, 1977-1978). At the same time, Straus and Gelles (1979) published their findings from the NFVS which supported SteinmetzÂ’s claim that intimate partner violence was gender symmetrical. This conclusion directly contradicted what the feminists had found in their scholarly work, which was that females were disproportionately victims of intimate partner violence compared to males (gender asymmetry) (Doba sh & Dobash, 1992). Feminists feared that SteinmetzÂ’s study could advers ely impact the "domestic violence" advocacy movement and could put women's lives in danger (M cNeely & Jones, 1980). They accused Steinmetz of using bad data and claimed that her study did not accurately measure intimate partner violence (Pleck, Pleck, Gr ossman, & Bart, 1977-1978). This debate regarding gender symmetry versus gender as ymmetry continued through the 1990Â’s. It also prompted an explosion of empirical studi es in the intimate partner violence literature that continues today. Unfortunately, th roughout most of the 1980Â’s and the 1990Â’s the studies focused more on each side trying to support their perspective rather than objectively understanding and measuring intimate partner violence. For example, McNeely & Jones (1987) asserted that men were just as vi ctimized as women; Saunders (1988) retorted in response that most womenÂ’ s use of violence was in self-defense.
15 This scholarly debate is an issue beca use it involves the two major groups of sociologists/criminologists who study intimat e partner violence, and whose empirical findings have serious implications for pol icy and intervention (Dobash & Dobash, 2004; Johnson & Ferraro, 2000). For example, femini st advocates of intimate partner violence have used their findings to change the le gal system by criminalizing intimate partner violence and instituting mandatory arrest policies (Dasgupta, 2002). On the other hand, men's rights groups, such as the Men's Defense Association, use the findings from family conflict studies to defend their belief that "w idespread bias exists against men" and based on this discrimination request funding for wo men's domestic violence centers to be stopped (Saunders, 2002). The most interesting and perhaps most perplexing part of this debate is that both sides have marshaled la rge amounts of data from empirical studies consisting of large-scale surveys to support their differing perspectives (Archer, 2000; Kimmel, 2002). The feminists rely on the National Crime Victimization (NCV) studies, which are gathered from a variety of sources, as well as clinical studies to defend their argument of gender asymmetry (Kimmel, 2002). The NCV st udies consist of th e National Violence Against Women Survey (NVAW) and the Natio nal Crime Victimization Survey (Archer, 2000). Both of these surveys are conduc ted by government agencies, consist of randomized samplings of households, and unifo rmly find dramatic gender asymmetry for incidents of intimate partner violence (f or a summary see DeKeseredy, 2000; Gelles, 2000; Straus, 1999). These victim surveys provid e important statistics that describe the prevalence of intimate partner violence in general, and specifically that women are disproportionately victimized compared to men. For example, we know that female
16 victims are more likely than male victims to be killed by an intimate partner, suffer more severe injuries, are victims of violence more often, seek more emergency room assistance, and seek injuncti ons for protection more often than males (Dobash & Dobash, 2004; Nazroo, 1995; Osthoff, 2002). Howeve r, victim surveys do not measure or examine whether gender roles and patriarchy ar e responsible for the asymmetry. Instead, most of the insight that has been gained on gender roles and patr iarchy come from indepth studies that have been conducted on c linical samples (Kurz, 1989). For example, many of the clinical studies s how that battering occurs wh en husbands are trying to get their wives to comply with their wishes; over the course of time, batterers increasingly use intimidation and isolation to control their wives. Ba tterers believe their use of violence is justified by thei r wivesÂ’ behavior, and due to limited economic means women are more likely to either stay or return to an abusive partner (Dobash & Dobash, 1979; Pagelow, 1981; Strube, 1988; Brush, 1990; Langen & Innes, 1986, Morse, 1995). On the other side of the debate, the fam ily violence perspective researchers have found gender symmetry within rates of intimate partner violence. The most prominent of these researchers are Straus and Gelles, and they support their argument with two landmark studies conducted in the 1970Â’s and a follow up in 1985 (Straus, Gelles, & Steinmetz, 1980; Straus & Gelles, 1990), along with more than a hundred other empirical studies. A majority of these studies ha ve found evidence to suggest that females perpetrate a Â“violentÂ” act toward their male partner at the same rate or frequency that males perpetrate a Â“violentÂ” act towards their female pa rtner (Archer 2000; Fiebert & Gonzalez, 1997). Straus and GellesÂ’ landm ark studies employed a large scale survey design gathering data from over 8,000 families, and in both, they measured the rate of
17 IPV by using the Conflict Tactics Scale (CTS). The results of these studies are in direct contradiction to the findings from the femini st perspective and rais e troubling questions (Kimmel, 2002). The feminists argue that a lthough the CTS is the most widely used and accepted scale for studying IPV when employing a large scale survey, it is also flawed (Archer, 1999; Dobash & Dobash, 2004; Ost hoof, 2002; Worcester, 2002). When Straus developed the CTS (1979), it scor ed high for reliability and va lidity, but by design it does not measure the context or motivation for the vi olent act. According to feminist scholars, since the CTS simply Â“countsÂ” acts of viol ence in absence of context and motivation, it cannot reliably measure intimate partner viol ence. For example, while evidence from feminist research often suggests that the majority of womenÂ’s vi olence is a result of selfdefense, the CTS will show only that these wo men are violent. According to feminist research these women would in appropriately be considered violent by the CTS when in fact they are also a victim (Dobash, Dobash, Wilson, Daly, 1992; Osthoff, 2002; Saunders, 2002). It is obvious that there is an abundant amount of contradictory information in the intimate partner violence literature which supports both sides of the gender symmetry versus gender asymmetry debate. So the ques tion still remains: How is it possible that even with the rigorous scient ific research methods employed by both sides, the gender symmetry versus gender asymmetry debate st ill persists (Saunders, 2002)? A recent article (Dobash & Dobash, 2004) stated that understanding the intimate partner violence literature and reconciling the disparities is like solving a puzzle. In order to find the solution to the puzzle, researchers must start by Â“focusing on concept formation, definitions, forms of measurement, context, consequences, and approaches to claims
18 making to better understand how researcher s have arrived at such apparently contradictory findings and claimsÂ” (Dobash & Dobash, 2004, p.324). The first step in Â“solving this puzzleÂ” is to understand how each perspective has approached, defined, and conceptualized intimate partner violence. The feminist perspective, which grew out of the domestic violence advocacy movement, examines intimate partner violence on a broad social level. It focuses on the concept of patriarchy and the societal inst itutions that help to maintain patriarchy (Dobash & Dobash, 1979). The feminist scholar s believe that intimate partner violence can be explained by answering the general question "Why do men beat their wives?" instead of asking, "Why did this individual be at his wife?" (Bogra d, 1988). The feminists define intimate partner violence as a pa ttern that can only be understood by examining the social context, which includes the structur e of relationships in a patriarchal society and the imbalance of power and control (Jasinski, 2001). The fe minists also believe it is important to examine contribu ting factors to intimate part ner violence, such as the socialization practices of teaching gender sp ecific rules and the historically male dominated social structure (Smith, 1990; Yllo & Straus, 1990). According to some feminist researchers, these contributing fact ors and patriarchy ar e maintained through traditional marriage (Martin, 1976; Pagelow, 1984). As a result, women occupy a subordinate position in the societ al structure, and violence has become the most overt and active method used to maintain social co ntrol or men's power over women (Bograd, 1988). The family conflict perspectiv e grew out of the family conflict scholarÂ’s work and is generally traced back to the efforts of Straus (1971) and Gelles (1974). Their primary
19 interest was to study a variety of family conflict issues and ho w these conflicts are resolved (Johnson, 1995). In the early 1970Â’s, Straus and Gelles began working together with the primary theoretical focus of ex amining commonalities among forms of family violence and the surprising frequency of vi olence (Yllo, 1988; Straus, 1979). This perspective is considered to be more general than the feminist perspective. It advocates that intimate partner violence is a common o ccurrence that happens within the family by both spouses, rather than an issue of violen ce against women (Jasinski, 2001). Family conflict theorists do not discount the feminist notion of patr iarchy but they believe this focus is too narrow and that violence aff ects all family relationships (Gordon, 2000; Straus, 1999). Family conflict th eorists believe that the origin of violence is the nature of the family structure rather than patriarchy (Straus & Gelles, 1990). Straus argues for example, that violence is legitimized within families by the use of corporal punishment, and it is an accepted resolution to family conflicts (Jasinski, 2001). Since the feminist and family conflict theorists differ in their basic theoretical perspectives on intimate partner violence, their definitions for the purpose of measurement also differ. This is an im portant puzzle piece to understand because Â“the most basic issue in measuring any phenomenon is how we define itÂ” (Desai & Saltzman, 2001). The feminist scholars de fine violence broadl y as any act that is harmful to the victim; alternatively, the family conflict theo rists define violence narrowly and focus on only physical acts that could cause harm (Ge lles, 2000). For example, when feminist scholars define violence as any harmful act, they will consider not only physical harm, but also emotional consequences of physical violence, such as depression, and measure the severity of the harm. On the other hand, family conf lict researcher s like Straus
20 (1979) define violence as any act that had the intention of harming the other person. However, family conflict theorists only incl ude and simply count phys ical acts that could cause physical injury and they do not measur e the severity of the injury or include measurements for emotional consequences. As a result of their opposing definitions, the feminist perspective and family conflict perspective also diffe r on their conceptualization of intimate partner violence. Family conflict researchers conceptualize inti mate partner violence as individual acts of physical violence that occur with in a family when a conflict gets Â“out of handÂ” (Gordon, 2000; Johnson, 1995; Straus & Smith, 1990). Each incident of violence occurs in isolation, is a result of the im mediate conflict, and is not co nnected to a need to control another person. This type of violence usuall y leads to Â“minorÂ” forms of violence, rarely escalating into severe or lif e threatening forms of violen ce (Straus & Smith, 1990). On the other hand, most feminist sc holars conceptualize intimate pa rtner violence as an array of behaviors that include physical acts, psyc hological abuse, verbal attacks, and sexual violence that are not episodic, but actually part of a pattern of behavior (Gordon, 2000). The most common conceptualiza tion of intimate partner violence used by the feminist researchers was explained by Pence and Pa ymer (1993) as "a pattern of physical, psychological, and sexual abuse; coercion, and violence with the intent to dominate and controlÂ” (Pence & Paymer, 1993). They further stated that "violence is used to control people's behavior.... the intention of the batterer is to gain control over their partnerÂ’s actions, thoughts, and feelings" (Pence & Paymer, 1999, 3, pp.1-2). Their research was significant because it established the concep t that intimate partner violence involved Â“batteringÂ”, which was an ongoing pattern of violence that incorpor ates the use of both
21 emotional and physical abuse motivated by th e need to control another person. Pence and Paymer (1993) presented their definition in a visual picture which has become known as the Â“Duluth ModelÂ” or the Â“Power and Control Wheel.Â” Figure 2: Power and Control Wheel From Â“ A Guide for Conducting Domestic Violence AssessmentsÂ” by Domestic Abuse Intervention Project, 2002, appendix C-1. Copywright 2002 by Domestic Abuse Intervention Project. Reprinted with permission of the editors.
22 The Power and Control Wheel (see Figure 2) consists of eight nonviolent control tactics that are like spokes on a wheel, which are held together by phys ical violence. The eight areas of the Wheel are: economic abus e, emotional abuse, coercion and threats, intimidation, using male privilege (patri archy), using children, minimizing, denying and blaming, and isolation (Pence & Paymer, 1993). A Â“battererÂ” (commonly used term for an offender of intimate partner violence) is motivated by his desire to have power over another, and uses a variety of techniques from each of the eigh t areas to maintain control. The wheel conceptualizes physical violence as the overriding control fa ctor used to hold the other eight areas together. The Duluth model is important because it provides a clear understanding of the Â“dynamics Â” of intimate partner viol ence and it is the most commonly used model today when assisting vi ctims, treating batterers, and educating the public. A Proposed Answer to Reconcile the Gender Symmetry Debate These different approaches, definitions, a nd conceptualizations are what drives how each perspective operationalizes and measur es intimate partner violence, which then determines the type of research methods em ployed for a particular empirical study. The vastly different sampling methods used also helps to explain why the findings among these perspectives are so cont radictory. In the intimate part ner violence literature, one will find the following sources of data: clinical, official report data, and social surveys (Gelles, 2000). Since the fe minists conceptualize intimate partner violence broadly and believe it is a result of a so cietal and cultural system of male dominance and patriarchy, they employ methodologies that encapsulate a broad range of psychological and physical harm that is used to control women (Stark & Flitcraft, 1996). To obtain this type of data,
23 the feminists tend to conduct empirical studies that use clinical or agency samples primarily obtained from Â“battered womenÂ’ s sheltersÂ” and emergency room patients (Dobash & Dobash, 1979; Giles-Sims, 1983; Pa gelow, 1981). The clinical and agency samples allow for the researcher s to conduct in-depth interviews and gather detailed data that include measuring the c ontext and motivation of the violent incident. For example, several researchers have found that women who use physical force against their intimate partner are actually battered women stri king out to stop attacks or escape attacks (Dasgupta, 1999; Dobash & Dobash, 1979, 1992; Hamberger, 1997; Saunders, 1988). Although these data are an important and necessary piece for understanding intimate partner violence and assessing the impact of intervention programs, it is generally qualitative in nature and cannot be generali zed to the population (Kimmel, 2002). On the other hand, the family conflict theoristsÂ’ primary methodological concerns are with generating reliable measures of the incidents of violent acts; they ar e less concerned with the context or motivation in which these acts occur (Dobash & Dobash, 2004). As result, family conflict theorists overwhelmingly us e large-scale social surveys of random samples. Johnson has argued that these differences in sampling methods and the differences in how violence is being meas ured by each perspective can explain why feminist and family conflict research findi ngs are so contradictory. Johnson proposed that the gender symmetry versus gender asym metry debate is a result of measuring two distinct types of intimate part ner violence as if they are the same phenomenon. When Johnson compared the feminist (Johnson cal ls Â“shelterÂ”) and family conflict (Johnson calls Â“surveyÂ”) empirical studies he found seve ral key issues that c ould explain the cause
24 of the symmetry debate (Johnson, 1995). The fi rst issue is that bot h types of sampling methods employed by the two perspectives are biased in their own way, causing them to produce two distinct sets of evidence that only contain one of two types of intimate partner violence (Johnson, 1995). He proposed that Â“surveyÂ” samples and Â“shelterÂ” samples reach different segments of th e population, which deal with nearly nonoverlapping phenomena (Johnson, 1995). In effect, neither methodology is misrepresenting the Â“trueÂ” nature of intimate partner violence but is actually measuring different types of intimate partner viol ence (Johnson, 1995). Next, Johnson supported his theory that two distinct types of viol ence are present among intimate partners by establishing the striking differences found betw een the feminist studies and the family conflict theorist studies. For example, the fe minist research consis tently showed a higher per couple frequency of physical violence and greater escalation of physical violence, as compared to the family conflict research (Johnson, 1995). Based on these observations in the literature Johnson went on to define a nd conceptualize two di stinct categories of intimate partner violence: intimate terrorism and situational couple violence. Johnson argues that intimate terrorism is the type of violence that feminist researchers are tapping into and situational coupl e violence is the type that family conflict theorists are measuring. The defining feat ure that separates intimate terrorism from situational couple violence is the perpetratorÂ’s motivati on behind the violence (Johnson, 2001). In intimate terrorism an individualÂ’s use of violence is embedded in a general context or motivation to control their in timate partner, not only temporarily but throughout the entire relationship (Johnson, 2001) On the other hand, situational couple violence does not involve an attempt by either partner to gain gene ral control over the
25 relationship. Instead it erupts situationally when tensions or emotions of a particular conflict lead to someone reacting with a phys ically violent act (Johnson, in press). The distinction is that although both typologies can involve control, situational couple violence is not embedded in an overall motive to control the relations hip, but to win the current conflict (Johnson, 1999). Johnson advises that the key to distinguishing intimate terrorism from situational coupl e violence in empirical studies is to test if there is a general motive to control the victim embedde d within the relationship. This is done by moving the focus from the nature of one viol ent encounter to search for patterns of nonviolent controlling behaviors in th e relationship as a whole. In doing so it is important to understand that the difference between intimate terrorism and situational couple violence is not in the nature of the violent act; the tr ue distinction lies in the degree of control present (Johnson & Ferraro, 2000) Since the key feature that distinguishes intimate terrori sm from situational couple violence is the Â“controlÂ” motive of the pe rpetrator it is important to understand how Johnson conceptualizes and opera tionalizes the control context. Since intimate terrorism is the type of violence that feminists have been studying for thirty years, Johnson uses their findings to define, conceptualize, and ope rationalize the control mo tive. In order to develop a variable that can measure the non-vi olent control motive Johnson refers to the Â“Duluth ModelÂ” and the Power and Control Wheel. JohnsonÂ’s definition of intimate terrorism is reflective of how Pence a nd Paymer defined Â“battering,Â” and he conceptualizes the nonviolent co ntrol motive as a reflection of the eight areas that make up the spokes in the Power and Control Wheel (Johnson, in press). Johnson then operationalizes the nonviolent control motive by using a three-step process. He first
26 identifies questions contained in the surv ey tool that measure each of the eight nonviolent control areas. Then he uses these que stions to develop a s cale ranging from a low to high control motive. From this scale J ohnson identifies the place on the scale that is a cut point to distinguish a Â“highÂ” from a Â“lowÂ” non-violent control motive. Those individuals with a Â“highÂ” control motive are put into the intimate terrorism group and the Â“lowÂ” control motive individuals are put in to the situational couple violence group. Johnson argues that intimate terrorism is what most individuals think of when they hear the term Â“domestic violence,Â” it is gender asymmetr ical, and causes the majority of negative outcomes identified in the feminist research, however it is not the most common type of intimate partner violen ce. In fact, the most common type of intimate partner violence does not involve any at tempt on the part of either party to gain a general control over the relati onship or victim. The most common type of violence is situationally-provoked when te nsions and emotions rise during a conflict between intimate partners. This is what Johnson calls situational couple viol ence. He argues that this type of violence is more gender symmetr ical, occurs less fre quently, and generally does not escalate in severity of physi cal violence over time (Johnson, 1999). In situational couple violence th e physical violence may be mi nor and singular, such as when an argument at some point escalates to the level of a push, gra b, or slap. In these cases the motive for the violence varies from demonstrating extreme anger or frustration to intending to cause serious injury (Johnson, 1999) It is also possibl e that the violence occurred because the in dividual wanted to control that sp ecific argument or situation but the control motive is not part of a ge neral pattern of coercive control.
27 Johnson argues that the separate incide nts of physical violence in situational couple violence may look exactly like intimate terrorism when the overall control motive is not examined or measured (Johnson, 2000). This is why it is important to begin to incorporate measures of non-viol ent control tactics in intimate partner research. He also stresses the importance of examining the enti re contextual relations hip instead of just counting or measuring an indivi dual incident of violence. He believes this is the only way to begin to identify and separate the tw o types of violence that he has defined and conceptualized. Based on his literature revi ew, Johnson developed the following testable predictions to test his pr oposed ideas (Johnson, 1999, p.9-10). 1. Intimate partner violence occurs in high and low control contexts. 2. In heterosexual relationships intimate terrorism is primarily committed by males and situational couple viol ence is sex symmetric. 3. Intimate terrorism will result in more frequent acts of physical violence compared to situational couple violence. 4. The severity of violence in intimate terrori sm is more likely to escalate over time compared to situational couple violence, therefore resulting in more severe injuries. 5. Victims of intimate terrorism are less likely to return acts of violence as compared to victims of situa tional couple violence. 6. Intimate terrorism is found almost excl usively in Â“shelter Â” populations and situational couple violence is found almost exclusively in Â“surveyÂ” samples. 7. As a result of the predicted patterns intim ate partner violence appears to be gender symmetric in Â“surveyÂ” samples and gender asymmetric in Â“shelterÂ” samples.
28 Current Empirical Findings Regarding JohnsonÂ’s Theory When Johnson first presented his argu ment in the mid 1990Â’s, although it was reasonable, he did not presen t any direct evidence to s upport his prediction. In 1999, however, Johnson published a paper that outlined his predictions, specified the requirements needed in a data set to test hi s predictions, and presented the first empirical evidence from a research study to support his pr edictions. First, he stressed the need for the sample to have the potential to contain e ither perpetrators or victims of both intimate terrorism and situational couple violence (J ohnson, 1999). Then he stated the importance of having measures of not only physical violen ce but also the non-vi olent control tactics needed to search for Â“patterns of genera l power and controlÂ” (Johnson, 1999). Johnson was able to identify an existing data set, co llected in the early 1970Â’s, that provided him measures of non-violent control tactics a nd contained a convenience sample from both the Â“shelterÂ” and Â“surveyÂ” populations. Alt hough this studyÂ’s design was not perfect it did provide support for JohnsonÂ’s theory that two distinct types of intimate partner violence may exist (for further detail see Johnson, 1999). More importantly this study showed that the two types of intimate part ner violence can be distinguished based on a Â“highÂ” or Â“lowÂ” control motive embedded in the relationship. However, the most interesting finding was that his prediction that intimate terrorism occurred only in Â“shelterÂ” samples and situational couple violen ce occurred only in Â“surveyÂ” samples was not supported. He actually found that both typologies existed w ithin both types of populations if a large enough sa mple was taken (Johnson, 1995). The second empirical study conducted to test JohnsonÂ’s predictions was done by Graham-Kevan and Archer (2003). In an effort to include both a Â“surveyÂ” and a Â“shelterÂ”
29 sample they used male and female college students (survey), females from shelters, and males from a prison population. This study did find evidence to support a relationship between aggression and control, which could be e xplained by the existence of sub-groups within violent intimate partner relations hips (Graham-Kevan & Archer, 2003). For example, individuals identified as Â“highÂ” cont rollers were far more likely to use physical violence compared to Â“lowÂ” controllers. They also found a greater frequency and severity of violence with intimate terror ism compared to situ ational couple violence (Graham-Kevan & Archer, 2003). This suppor ts both the feminist philosophies and JohnsonÂ’s current theory that conceptualize ag gression in intimate terrorism as a coercion tactic, which takes place in a general pattern of power and control. However, it found only weak evidence to support th e prediction that intimate terrorism is primarily male (gender asymmetric) which could have been a result of the sampling strategy employed (Graham-Kevan & Archer, 2003). Both of these studies found evidence that contradicted the original prediction, which stated that the populations wher e intimate terrorism could be found are nonoverlapping with the populations where situat ional couple violence may be found. In fact both studies had some evidence to suggest that these two types of intimate partner violence actually overlapped to some degree. This meant that both types of intimate partner violence could be found among both the Â“surveyÂ” and Â“shelterÂ” populations if a large enough sample were collected. Ba sed on these findinings Johnson conducted a third study with a sub-sample (female data only) from the National Violence Against Women Survey (NVAWS). This is a nationa l random sample that collected data from 8,000 males and 8,000 females across the Unite d States (Johnson, 2005). The NVAWS
30 provided a stronger sampling technique to ar gue generalization, and helped to improve upon a weakness in the two previous empirical studies. However, the study using NVAWS data di ffered in its methodological approach to testing JohnsonÂ’s theory. Previously the focus was on obtaining evidence to support the notion that two type s of intimate partner violence exis ted and could be separated from each other based on a non-violent control motive. Now Johnson began to build upon the previous findings by first assuming that two distinct types of intimate partner violence exist and can be separated by a Â“highÂ” or Â“l owÂ” control motive. Then he creates a nonviolent control motive measurem ent tool to separate Â“hig hÂ” controllers from Â“lowÂ” controllers. From this point, Johnson devel ops several hypotheses to test the other predictions regarding intimate terrorism. He found the following: victims of intimate terrorism experience more frequent and more se vere acts of violence compared to victims of situational couple violence, intimate terrorism is less likely to stop, victims of intimate terrorism experience more damage to thei r physical and psychologi cal health, intimate terrorism is more likely to interfere with a victimÂ’s daily activi ties, and victims of intimate terrorism are more likely to leav e and seek help (Johnson & Leone, 2005). These findings were important because Johnson showed that the consequences for victims of intimate terrorism are different (more frequent and severe) from those consequences for victims of situational couple vi olence. He also continued to assert that these findings supported the notion that intimat e terrorism is gender asymmetric (Johnson & Leone, 2005).
31 The Purpose of the Current Study Although each of the three empirical st udies reviewed above supports the major tenet of JohnsonÂ’s theory that two distinct types of intimate partne r violence exist, the question still remains: Has Johnson found e nough evidence to demonstrate that intimate terrorism is gender asymmetric? A major point of controversy in the literature is that family conflict theorists find gender symme try when measuring only the number of violent acts, but even Straus has argued that the injurious consequences of intimate partner violence are asymmetr ical (Straus, 1999). Johns on argues that the non-violent control motive can distinguish two distinct types of intimat e partner violence, which can be identified by certain traits other than inju rious outcomes. He specifically refers to the frequency and severity of physical violence as being measurably different among intimate terrorism and situation couple violence. According to Johnson intimate terrorism is characterized by a Â“highÂ” cont rol motive, which results in frequent and severe physical violence, and situational couple violence is characterized by a Â“lowÂ” control motive and does not experience frequent or severe physical violence. He has successfully differentiated two distinct groups of intimate partner violence by using the non-violent control motive variab le, and demonstrated that the intimate terrorism group experienced a greater frequency and more se vere violence. However, the analytic strategy he employed was a mean comparison of the frequency and severity of physical violence between the intimate terrorism group and situational couple violence group. Although his results established the non-violent control variab leÂ’s ability to distinguish intimate terrorism from situational couple vi olence, they did not demonstrate how much of the variance in frequency and severity of violence the non-violent control variable can
32 explain. In order to cont inue building on JohnsonÂ’s work it is important to use a predictive rather than comparative statistical model that can establish how much of the variance in frequency and severity of viol ence can be explained by the non-violent control variable, while control ling for other variables. One weakness in JohnsonÂ’s work involves his assertion that intimate terrorism is gender asymmetric, despite the fact that he fail ed to include a male sample in his previous work. In order to support the hypothesis that intimate terrorism is asymmetrical (with females being disproportionately victimi zed) while situational couple violence is symmetrical, both a male and female sample mu st be included in the study. The samples collected need to be reflective of the ge neral population, and th e survey instrument, collection methods, hypotheses tested, and statistical analyses employed must be consistent for both the male and female sample s. The purpose of this study is to build upon the previous empirical studies of JohnsonÂ’ s theory by using a different statistical technique to assess how much of the variance in frequency and severity of violence is explained by the non-violent control variable. It will then address the weakness in JohnsonÂ’s assertion that intimate terrorism is gender asymmetric by comparing the results from the female sample to the male sa mple. This study will test the following hypotheses: H1: Female victims of intimate partne r violence are more likely than male victims of intimate partner violence to repor t high non-violent cont rol tactics in their relationships.
33 H2: Among female victims, high non-vi olent control accounts for at least a moderate proportion of the variance in freque ncy, and severity of violence, controlling for other variables. H3: Among male victims, high non-violen t control accounts for little to no proportion of the variance in frequency and se verity of violence, controlling for other variables. H4: High non-violent control is more like ly to lead to greater frequency and severity of violence in female victims than in male victims of intimate partner violence.
34 Chapter Three Methods Sample The data used for the present study are dr awn from a sub-sample of respondents in the National Violence Against Women Su rvey (NVAWS), a cross-sectional national random-digit dialed sample of telephone househ olds in the United States. The purpose of the NVAWS was to further the understanding of violence against women by providing a context in which to place womenÂ’s experiences regarding victimizations of violence. Telephone interviews were conducted from November 1995 to May 1996 by highly trained and experienced interviewers (Tjade n & Thoennes, 2000). The original female sample consisted of 8,000 participants, w ith an average age of 44 years old; 82% identified themselves as white. At the tim e of the survey, 69% reported being employed at least part-time and 20% stated they we re either a homemaker or unemployed. The original male sample consisted of 8,000 partic ipants, with an averag e age of 45 years old; 84% identified themselves as white. At th e time of the surve y, 83% reported being employed at least part-time and 3% stated th ey were either a homemaker or unemployed. Approximately 61% of the male sample and 59% of the female sample reported either being a college graduate or they took some college. The completion rate (once the interview began) was 97% for females and 98% for males.
35 Since most of the published work on intimate partner violence deals primarily with married or heterosexual couples, the decision was made to focus on these populations for the present study. However, it is important to note that some studies survey couples and divide them into separate male and female samples, but this was not the case for the NVAWS. This survey was admi nistered to individuals, so their partners were not represented in the opposite sex samp le. Instead a separate and independent survey was conducted for the male and female sample. The final sample for this study included only those respondents who were he terosexual and married or cohabitating and who reported experiencing at least one incide nt of physical violen ce (physical violence will be defined below) by their current spouse. The final female sample used for this study consisted of 325 particip ants, 33% reported being cu rrently married, with an average age of 38 years old; and 80% identified themselves as white. At the time of the survey, 67% reported being employed at least pa rt-time and 25% stated they were either a full-time homemaker or unemployed. The fina l male sample used for this study consisted of 167 participants, 63% reporte d being currently married, with an average age of 39 years old; and 82% identified themselves as being white. At the time of the survey, 86% reported being employed part-time and 5% st ated they were either unemployed or a homemaker. Approximately 63% of the male sample and 57% of the female sample reported either being a college graduate or that they took some college. The average length of time that the respondents reporte d being with their current spouse or cohabitating heterosexual partner was 12 y ears for the males and 15 years for the females.
36 Procedures The NVAWS collected data independently from the male and female participants and separate data sets were created (Tjadon & Thonnes, 2000). For purposes of this study the two data sets continue d to be maintained separate ly. The procedure used to create the male and female sub-samples for th is study and all of the statistical processes were conducted independently on both the male and female data sets. This allowed for each hypothesis to be tested individually on the male and female samples, providing an objective comparison of the resu lts. The NVAWS used the same questionnaire for both male and female participants, collecting info rmation regarding the following six areas: 1) their general fear of violence and the ways in which they managed their fears, 2) emotional abuse they had experienced by mar ital and cohabitating pa rtners, 3) physical assault they had experienced as children by a dult caretakers, 4) physical assault they had experienced as adults by any type of perpetra tor, 5) forcible rape or stalking they had experienced by any type of perpetrator, and 6) incidents of threatened violence they had experienced by any type of perpetrator. Th e NVAWS data were then further categorized into fifteen different sections the following of which were used for this study: physical victimization; power, control, and emotional abuse; and charact eristics of current spouse or partner. The female and male sub-samples used for this study were created by using a twostep process. First, an initial sub-sample was created for each data set by using an item from the questionnaire that aske d respondents to identify thei r current marital status. A filter was then used to delete all cases that did not respond as either currently married or currently living as a couple at least part-time with a member of the opposite sex. The
37 second step used the physical victimization section to identify t hose participants who reported at least one act of phys ical violence committed against them as an adult. This section contained responses to a twelve-item yes or no vers ion of the Conflict Tactics Scale (Straus, 1979). The original Conflict Ta ctics Scale is the most widely used and commonly accepted scale used in the intimate partner violence literature, although the instrument has been subjected to criticis m (Archer, 2000; Johnson, 1995; Straus 1990). The twelve-item physical victim ization section asked particip ants the following: after you became an adult did any other adult, male or female everÂ… throw something at you that could hurt youÂ… push, grab, or shove youÂ… pull your hairÂ… slap or hit youÂ… kick or bite youÂ… choke or attempt to drown youÂ… hit you with some objectÂ… beat you upÂ… threaten you with a gunÂ… threaten you with a knife or other weapon besides a gunÂ… use a gun on youÂ… use a knife or other weapon besides a gun? A ny participant who responded Â“yesÂ” to at least one of these items was then asked to identify their relationship to the perpetrator who committed the violent act against them. A filter was then used to retain only those individuals who responded that the perpetrator was either their current spouse or cohabitating heterosexual partner. This two-step process created a male and female data set which contained only thos e individuals who reported at least on one incident of violence by eith er a current spouse or cohab itating heterosexual partner. Measures Frequency of Violence. Frequency of violence was measured by using an item that asked participants how many different time s their partner had done at least one of the twelve Conflict Tactic Scales items to them. The responses ranged from 1-97 for both the females (M=5.85) and the males (M= 4.42). Unfortunately, the NVAWS did not
38 measure the frequency for each of the individu al twelve items used to measure physical violence. Therefore, this variable cannot be us ed to calculate a frequency for each type of physical violence measured. It simply gets a co unt of the number of times any or all of these physically violent acts occurred and presents them as one final number. Severity of Violence. Severity of violence was measured by using the following seven physical violence items: did any adult ma le or female everÂ…choke or attempt to drown youÂ… hit you with some objectÂ… beat you upÂ… threaten you with a gunÂ… threaten you with a knife or other we apon besides a gunÂ… use a gun on youÂ… use a knife or other weapon on you besides a gun? The re spondentÂ’s answers to each of these items were measured as Â“yesÂ” or Â“noÂ”. These seven questions were selected based on the original Conflict Tactic s ScaleÂ’s division of severe viol ence versus non-severe violence (Straus, 1979). In order to cr eate a severe violence scale th e items were recoded into 1 = yes and 0= no. A principal components anal ysis was performed on the seven items for both the male and female data sets. The female eigenvalue was 3.11, with 44.9% of the variance explained by one component and th e male eigenvalue was 3.056 with 43.66% of the variance explained by one factor. The CronbachÂ’s alpha for the female sample was .783 and the male sample was .779. This scale was then dichotomized as follows: 0-1 severe types of violence reported were categorized as non-severe vi olence and 2-7 types of severe violence reported were categorized as se vere violence. A dichotomized variable was used instead of the scale because the questionnaire item as ks only about the number of different types of severe violence, rather than the number of times severe violence occurred. It is possible that someone choked twelve times may have been subjected to greater Â“severityÂ”
39 than someone who was choked once and threaten ed with a knife once. If the variable remained undichotomized, then the latter would be classified as higher severity than the former. A cut point of two or more types of severe violence reported was chosen because Johnson argues that although intimate terror ism will usually involve more severe violence, it is possible for an incident of se vere violence to occur in situational couple violence. However, multiple incidents of se vere violence are more likely to occur in intimate terrorism. Since it was not possible to measure the frequency of each type of severe violence the cut point of two or more types was used to represent the repetitive nature of intimate terrorism as opposed to an isolated incident in situational couple violence. Non-violent Control Tactics. As stated previously, the non-violent control variable, according to Johnson, is the key to distinguishing intimate terrorism from situational couple violence. The NVAWS include d a total of thirteen items that represent operationalizations of the categ ories contained in Pence a nd PaymerÂ’s (1993) Power and Control Wheel. From these thirteen items, a to tal of seven were sele cted to be included in the present study because they were also used in the Canadian Violence Against Women Survey (Johnson, 1996), closely re sembled items in the Psychological Maltreatment of Women Survey (Tolman, 1989) and they were previously used to measure JohnsonÂ’s theory of intimate te rrorism (Johnson & Leone, 2005). The seven items used included the following: thinki ng about your current husband (wife)/partner would you say s/he... is jealous or possessive?...t ries to limit your contact with family or friends?...insists on knowing who you are with at all times?... calls you names or put downs in front of others?... makes you feel inadequate?... shouts or swears at you?...
40 prevents you from knowing or having access to the family income even when you ask? The response options to each item were Â“noÂ” or Â“yesÂ”. A principal components analysis was c onducted separately on both the original NVAWS female and male samples and the sub-sa mples used for this survey to determine if the items represented more than one constr uct. The results of the principal components analysis did not indicate a si gnificantly different result for this studyÂ’s smaller subsample as compared to the larger NVAWS sa mple. The results for this studyÂ’s female sample had an eigenvalue 2.58 with 36.8% of the variance explained by one factor and this studyÂ’s male sample had an eigenva lue of 2.35 with 33.56% of the variance explained by one factor. This suggested th at a reasonable scale could be constructed from these seven items for both males and females. The score for the non-violent control tactics variable included the number of contro l tactics that the respondent reported his/her current spouse or cohabitating heterosexual partner used against them, with a potential range of 0-7. The CronbachÂ’s alpha for male participants was .65 and for female participants it was .70. The re liability test was also conduc ted on the larger original NVAWS sample and compared to this sub-sa mple. The CronbachÂ’s alpha for the larger sample did not differ at a level of statistical significance from this studyÂ’s sub-sample. In order to operationalize a distinction between intimate terrorism and situational couple violence based on JohnsonÂ’s previous work, it is necessary to transform the nonviolent control tactic scale in to a dichotomized variable (Johnson & Leone, 2005). Since the principal components analysis and reliabi lity test for this studyÂ’s male and female samples yielded results almost identical to JohnsonÂ’s previous study, the decision was made to follow his cut point for dichotomiza tion of 0-2=low non-viol ent control and 3-7=
41 high non-violent control. Johnson had used a cluster-K analysis to determine an appropriate cut point for this variable (f or further discussion, see Johnson & Leone, 2005). Using this dichotomization of the nonviolent control tactics scale, those spouses or partners who used three or more of the se ven control tactics (high non-violent control) were categorized as intimate terrorism. Thos e spouses or partners using two or fewer of the control tactics (low non-vi olent control) were categor ized as situational couple violence. Control Variables. Since the male and female data sets were not merged it was not necessary to control for gender. Howe ver, the following variables were used as controls for all hypotheses tested: relationship type, age, level of education, race, employment status, and length of time together. These variable s were chosen as controls because research has shown that they can im pact the incidence rates for intimate partner violence. For example, individuals who are of traditional college ag e, have less formal education, or members of a minority group te nd to report a higher in cidence of intimate partner violence compared to older adults, individuals with more education, or those reporting to be white (Archer, 2000). Also, studies show conflicting findings regarding the effect of length of time together. So me studies find that in the beginning of a relationship the violence may be frequent, but at a certain point in time the rates of physical violence may decrease (Archer, 2000). Researchers theorize that physical violence may not be necessary after a certa in point to maintain control or once the individuals are married (Worcester, 2002). For purposes of this study the following control variables were recoded from their original measure in the NVAWS: relationship type, employment, and level of education.
42 The variable relationship type was m easured by a question in the NVAWS that asked participants to identify their current relationship status. Only those cases that reported being either currently married or c ohabitating with a heterosexual partner were retained for this study. This variable wa s then coded as follows: married=1 and cohabitating=0. In order to control for th e effect of employment the original question was recoded into the following two dichot omized measures: Â“unemployedÂ” or Â“other incomeÂ”. The first measured unemployment against being employed and the second measured other types of income against being employed. Th e Â“unemploymentÂ” variable was coded and labeled as follows: all indi viduals who indicated being either unemployed or a homemaker were coded as unemployed=1 and a ll other responses=0. The variable Â“other incomeÂ” was coded and labeled as follows: a ll individuals who indi cated being retired, military, student, or other were coded as Â“other incomeÂ”=1 and all other responses=0. In order to control for th e effect of educational level the original question was again recoded into several dichotomized vari ables. On the NVAWS, responses to the question regarding level of education achieved were originally coded into the following categories: up to eighth grade completed, a bove eighth grade but le ss than a high school diploma, received a high school diploma, completed some college, 4yr college graduate, beyond four year degree. For purposes of this study level of educa tion was recoded into the following three dichotomized variables: high school graduate=1, some college=1, and college graduate=1, with each having all other responses=0. Each of these variables were compared against individuals who did not graduate from high school.
43 The NVAWS asked individuals to identify their racial background as one of the following: white, black or Afri can-American, Asian or Pacific Islander, American Indian or Alaskan Native, or Mixed Race. For this study race was labeled and coded as follows: white=1 and all the other categories were co mbined and labeled as others=0. It is important to note that individuals who iden tify as Hispanic were measured using a separate question that was not included in this study. Ther efore, the Hispanic population was not included in this study. The variable length of time together was measured based on the number of years the respondent reported living w ith their spouse or heterosexua l partner. Those stating the length of time together was less than one year were coded as zero. Analytic Strategy. In order to build on the previous studie s testing JohnsonÂ’s theory, it was important to use predictive statistical models to test if the amount of non-violen t control tactics used (high controller) could predic t intimate terrorism. Johns on identified severity and frequency of violence as key factors that occur in intimate terrorism versus situational couple violence (Johnson, 1995; Johnson, 1999; Johnson & Leone, 2005). Therefore, two different types of regression models were used to test if the use of high non-violent control tactics could predict both frequency and severity of violence for both males and females. For the frequency variable, negativ e binomial regression was used instead of Poisson regression or OLS regression for two reasons. First, freque ncy of violence was measured as a natural count of a rare incident which can cause a variable to be skewed. The Kolmogorov-Smirnov test results for both th e female and male data set determined that this dependent variable had a statistically significant departure from normality and
44 was severely skewed to the right (Long, 1997). As a result, the standard errors in an OLS regression model could be biase d, causing a false positive on a significance test (Gardner et al., 1995). The second reas on was that the alpha test on the Poisson model determined that over dispersion was present in the fre quency of violence variable. This made the negative binomial regression model a bette r fit than the Poisson model (Long 1997). Alternatively, since the severi ty of violence variable was dichotomized, logistic regression was used instead of OLS (Gardner et al., 1995). In order to address JohnsonÂ’s assertion that intimate terrorism is gender asymmetric, significance tests of the difference between the males and females for the regressi on coefficients for fr equency and severity of violence were used (Brame et al ., 1998; Paternoster et al., 1998).
45 Chapter 4 Results The two purposes of this study were to build upon previous empirical findings regarding JohnsonÂ’s theory and to test his assertion that intimate terrorism is primarily experienced by females. Previous research used a comparative model to establish a correlation between intimate terrorism (non-vi olent control) and frequency and severity of violence. In order to build upon previous findings, this study used two predictive models to test the non-violent control variableÂ’s ability to distinguish intimate terrorism from situational couple violence by account ing for the amount of the variance in frequency and severity of violence that can be explained by the use of non-violent control tactics. The question addressed by these mode ls are as follows: are female victims of intimate terrorism (those reporting high non-viol ent control) more likely to experience frequent and severe acts of physical violence compared to male victims of intimate terrorism? Negative binomial regression was utilized to examine the frequency of physical violence and logistic regression was used to examine the severity of physically violent acts. The question of gender symmetry was tested by using significance tests of the difference between males and females for the regression coefficients for frequency and severity of violence. Preliminary analyses produced descriptiv e statistics and bivariate correlations for each of the items used in the regression mode ls. The descriptive statistics in Table 1
46 revealed that males and females differed fr om each other on only a few variables. For example, women were significantly more lik ely than men to report relationships of greater duration and being unemployed; furt her, women were signi ficantly less likely than men to be married and to be college gra duates. Of greater importance to the present study was that males were significantly more likely than females to report severe violence; moreover, there was no significant gender difference in the reported mean frequency of violence and no si gnificant gender difference in reports of partnerÂ’s use of non-violent control tactics. Therefore, among married and cohabitating heterosexual male and female victims of intimate partner violence, intimate terrorism, as measured by non-violent control tactics, and the frequenc y of violence both a ppear to be gender symmetrical. Additionally, viol ence against men appears to be more severe than violence against women in these data. However, univari ate statistics can reveal only a small part of the entire picture of the gendered na ture of intimate partner violence. The bivariate correlation analysis, pres ented in Table 2, revealed interesting differences and similarities between the fema le and male samples. Among males, for example, only age was significantly correlate d with high non-violent control, with younger male victims reporting partners using in timate terrorism tactics. By contrast, among females, high non-violent control tactics were significan tly correlated with greater frequency of violence, being a high school gr aduate, and cohabitating rather than being married. It would appear, th en, that while males and females did not differ in the proportions reporting rela tionships characterized by high nonviolent control, their risk factors for involvement in such relationships were quite different. Men and women were more similar, however, on their correlates of severe violence. For both the male and
47 female samples severe violence was associated with greater frequency of violence and a shorter duration of the relationship. However, age had an inverse relationship to severe violence only among men, and lack of a colle ge degree was associated with severe violence only among women. Interestingly, cohabitating males were more likely to experience severe violence, while married fema les were more likely to experience severe violence. Table 1: Descriptive Statistics for Male and Female Sample Variable Item Males (n=167) Female (n=325) Z Score Partners use of Non-Violen t Control Tactics 31% 27% .909 Mean Frequency of Violence 4.2 5.8 -1.57 Severity of Violence 38% 20% 4.09* Number of Years Together 12 15 -2.84* Married 63% 33% 4.54* Age 39 38 .640 High School Graduate 30% 34% -.888 Some College Completed 32% 34% -.444 College Graduate 31% 23% 1.91* Unemployed / Homemaker 5% 25% -6.66* Non-employed Income from Other Source 9% 8% .385 White 82% 80% -.541 Positive z scores indicate that scores for ma les were higher than scores for females. *p<.05
48 Table 2 Correlation Matrix for the Male and Female Sample Females High NV Control Severe Vio Freq Vio Yrs. Together Age Unemploy Other Income HS Grad Some College College Grad Married Cohab. Race High NV Control 1 .098 .137* .036 -.032 .056 .085 .113* -.048 -.096 -.183* -.073 Severe Vio .079 1 .246* -.114* -.085 .066 .026 .031 .041 -.167* .121* .018 Freq Vio -.017 .217* 1 -.063 -.050 .084 .017 .137* -.076 -.089 .086 .084 Yrs. Together -.079 -.232* -.078 1 .824* .100 .134* .001 -.060 -.031 -.469* .083 Age -.236* -.210* -.029 .788* 1 .044 .188* -.055 -.052 .076 -.336* .098 UnEmploy .122 -.118 -.049 -.117 -.105 1 -.165* .213* -.138* -.192* -.108 .005 Other Income -.028 .057 -.088 .164* .250* -.070 1 -.035 .013 -.075 -.053 .023 HS Grad .066 .001 -.083 -.152* -.152 .037 -.022 1 -.510* -.387 -.051 .021 Some College -.056 .034 -.013 .095 .041 -.032 .056 -.446* 1 -.390* .056 -.066 College Grad -.076 -.100 .060 .059 .130 -.090 -.076 -.440* -.458* 1 .044 .021 Married Cohab .149 -.394* -.142 .421* .240* -.121 -.066 -.074 -.017 .080 1 -.010 Race -.006 -.152 .024 .107 .124 .033 -.015 -.099 -.052 .111 .068 1 Males (* significant at the .05 level)
49 Frequency and Severity of Violence Among Female Victims In the univariate analysis the high non-violent control variable did not distinguish female victims from male victims of intimat e terrorism, thus providing no support for the first hypothesis. However, the bivariate anal ysis did reveal gende r differences in the correlations between risk factor s and high non-violent control re lationships. In particular, high non-violent control was correlated with gr eater frequency of vi olence among female victims, but not among male victims. Multiv ariate analysis can fu rther elucidate these relationships among the variables by controlli ng for other variables known to affect the frequency and severity of intimate partner violence. Johnson theorizes that high use of non-violen t control tactics is a key factor in distinguishing intimate terrorism from s ituational couple violence (Johnson, 2005). He also proposed that intimate terrorism is gender asymmetric with females disproportionately being victimized. J ohnson also argues that another important difference between intimate terrorism and situ ational couple violence is the amount of frequency and severity of violence experi enced by the victims (Johnson, 2005). Based on JohnsonÂ’s conceptualization of intimate terrorism physical vi olence is repetitive and over time increases in severity. Th erefore, in intimate terrorism there should be a significant correlation between high non-viol ent control tactics and the frequency and severity of violence for females. The second hypothesis predicts that the us e of non-violent control tactics accounts for at least a moderate amount of variation in the frequency of violence against female victims. In order to examine how much of the variation in frequency of violence the nonviolent control variable can explain, both bi variate and multivariate negative binomial
50 regression models were used. The results for the female sample are displayed in Table 3. The bivariate model demonstrat ed a statistically significan t relationship between high non-violent control and greater frequency of violence for the female sample; however, only 1% of the variance was explained. The multivariate model, controlling for other known correlates of intimate partner violen ce, showed that high non-violent control remained a statistically significant predicto r in the female sample, but the full model explained only 5% of the variance in frequency of violence. In the multivariate model, frequency of violence was also associated w ith being married versus cohabitating with a partner, shorter duration of the relationship, being non-white, completing some college, and being a college graduate. Table 3: Female Negative Binomial Re gression Models: Intimate Terrorism Predicting Frequency of Violence Bivariate Multivariate b SE % Change b SE % Change High Non-violent Control .671 .154 95.6 .824* .160 128 Relationship (Married=1) .569* .169 76.6 Yrs. Together -.023* .010 -2.3 Un-employ (=1) .114 .177 Other Income (=1) -.002 .302 HS Graduate (=1) -.118 .283 Some College (=1) -.592* .286 -44.7 College Graduate (=1) -.838* .305 -56.8 Age .015 .011 Race .584* .180 79.4 Constant 1.529 .082 .889 .082 Chi Square 19.99* 80.26* -2 Log-Likelihood 869.954 823.731 Pseudo R Square .01 .05 *p<.05
51 The second hypothesis also predicted that the use of non-violent control tactics accounts for at least a moderate amount of th e variation in severity of violence against female victims. In order to examine how much of the variation in seve rity of violence the non-violent control variable can explain, agai n both bivariate and multivariate logistic regression models were used. These result s are displayed in Ta ble 4. The bivariate model did not demonstrate a statistically significant relationship between high nonviolent control variable and severe violence for the female sample, and again explained only 1% of the variance in severe violence. The multivariate model, controlling for other known variables of intimate partner violence, showed that high non-violent control was a statistically significant predictor of severe violence and the full model explained 18% of the variance. Moreover, adding the control variables revealed a suppressor effect. Specifically, being married had a negative a ssociation with non-vi olent control and a positive association with severe violence at the bivariate level, which rendered the bivariate relationship between high non-violent control and se vere violence to be nonsignificant. When being married was controlled in the multivariate model, the true significant positive relationship between high non-violent control and severe violence appeared. In the multivariate model, severe violence was also associated with being unemployed/homemaker versus being employed in either a full or part-time job. In summary, the second hypothesis received onl y weak partial support; high non-violent control tactics in combination of other known correlates of intimate partner violence, predict almost none of the va riance in frequency of violen ce and only a modest amount of the variance in severity of violence among fe male victims of intimate partner violence.
52 Table 4: Female Logistic Regression Models: Intimate Terrorism Predicting Severity of Violence Bivariate Multivariate b WaldExp(b) b Wald Exp(b) High Non-violent Control .521 (.304) 2.939 1.683 .742* (.298) 6.211 2.099 Relationship (Married=1) .995* (.319) 8.961 2.60 Yrs. Together -.041 (.023) 3.012 .960 Un-employ (employed=1) .619* (.319) 3.764 1.86 Other Income (employed=1) .354 (.477) .550 1.42 HS Graduate (=1) -.654 (.431) 2.30 .520 Some College (=1) -.502 (.442) 1.29 .605 College Graduate (=1) -1.80 (.583) 9.50 .166 Age .020 (.022) .854 .020 Race .414 (.350) 1.40 1.513 Constant -1.573 (.176) 79.91 .207 -1.79 (.799) 4.99 .168 Chi Square 2.853 45.641* -2 Log-Likelihood 305.477 350.818 Nagelkerke R Square .014 .18 *p<.05 Standard errors in parentheses The Wald and Exp. (b) were not calculated for un-employ males because the sample size for this variable was too small.
53 Frequency and Severity of Violence Among Males Since Johnson theorized that situational couple violence is gender symmetric and intimate terrorism was gender asymmetric, w ith females disproporti onately experiencing victimization, he proposed that although males, report being victims of intimate partner violence they primarily are experiencing si tuational couple violence. Therefore, for males the effect of high non-vi olent control tactics should no t be significan tly correlated with either the frequency orse verity of violence. The thir d hypothesis predicts that the use of non-violent control tac tics accounts for little to none of the variation in the frequency of violence and severity of violence against male victims. In order to examine how much of the frequency of violence the non-violent cont rol variable can explain for males both bivariate and multivariate negative binomial regression models were used. The results for the male sample are displa yed in Table 5. The bivariate model as expected, did not demonstrate a statistically significant associati on between the high nonviolent control and frequency of violence fo r the male sample, and it was not able to explain any of the variance in that variable The multivariate model, controlling for other known correlates of intimate partner violence demonstrated statis tical significance for the model, but there was no statistically si gnificant relationship be tween high non-violent control and frequency of viol ence and only 4% of the vari ance in frequency of violence was explained. However, in the multivar iate model, frequency of violence was associated with cohabitating with a partner and being employed instead of being either unemployed or obtaining income from another source other than employment.
54 Table 5: Male Negative Binomial Regression Models: Intimate Terrorism Predicting Frequency of Violence. Bivariate Multivariate b SE % Change b SE % Change High Non-violent Control -.072 .176 .209 .191 Relationship (Married=1) -.597* .185 -44.9 Yrs. Together -.008 .014 Un-employ (=1) -.851* .414 -57.3 Other Income (=1) -.942* .314 -61.0 HS Graduate (=1) -.611 .324 Some College (=1) -.190 .326 College Graduate (=1) -.201 .335 Age .008 .013 Race .125 .206 Constant 1.469 .097 1.78 .513 Chi Square .17 30.19* -2 Log-Likelihood 415.626 396.696 Pseudo R Square .00 .04 *p<.05
55 T able 6: Males Logistic Regression Mo dels: Intimate Terrorism Predicting Severity of Violence Bivariate Multivariate b WaldExp(b) b Wald Exp(b) High Non-violent Control .347 (.346) 1.001 1.414 .736 (.449) 2.680 2.087 Relationship (Married=1) -1.88* (.457) 16.99 .152 Yrs. Together -.004 (.032) .019 .996 Un-employ (=1) -21.77 000 000 Other Income (=1) .151 (.662) .052 1.163 HS Graduate (=1) -1.48 (.823) 3.25 .227 Some College (=1) -1.08 (.816) 1.75 .340 College Graduate (=1) -1.45 (.831) 3.06 .234 Age -.017 (.030) .330 .983 Race -.546 (.472) 1.34 .579 Constant -.588 (.197) .556 2.91 (1.27) 5.30 18.4 Chi Squared .996 43.88* -2 Log-Likelihood 214.586 168.816 Nagelkerke R Squared .01 .33 *p<.05 Standard error in parentheses (The Wald and Exp. (b) could not be calcu lated because the number of unemployed males was too small to produce a valid result.)
56 The prediction that among male victims the high non-violent control variable would account for little to none of the variati on for severe violence was also examined by using both a bivariate and multivariate logistic regression model. The results for males regarding severity of violence are displaye d in Table 6. The bivariate model did not demonstrate statistical significance for the relationship between non-violent control and severe violence, and only 1% of the varian ce in severe violence was explained. When other known correlates of intimate partner viol ence were controlled in the multivariate model, however, the model became statistica lly significant and explained 33% of the variance. Nevertheless, high nonviolent control still failed to predict severe violence for male victims. However, severe violence was as sociated with cohabitating with a partner. These results demonstrate support for the thir d hypothesis, that the use of non-violent control tactics is not a useful predictor of the frequency and severity of violence among male victims of intimate partner violence. Comparison Between Male and Female Samples The preceding analyses have demonstrated that while female and male victims of intimate partner violence in heterosexual, married or cohabitating relationships do not differ in levels of non-violent control tactic s, frequency of violence, or severity of violence, as JohnsonÂ’s theory would predict, the influence of high non-violent control on frequency and severity of violence does a ppear to vary by gender. These findings suggest the possibility that intimate partner violence is not gender symmetrical even within a large random sample of survey re spondents using measures from the Conflict Tactics Scale. However, in order to determine if these results support JohnsonÂ’s assertion that intimate terrorism is gender as ymmetric it is important to explore if the
57 difference between the male and female samp les are statistically significant for both frequency and severity of violence. The fourth hypothesis predicts that high nonviolent control is more likely to lead to greater frequency and severity of violence in female victims than in male victims of intimate partner violence. This empirical ques tion is explored by us ing a significance test for the difference between the male and fema le samples on the regression coefficients that were significant in either the male or female multivariate models for frequency and severity of violence (Brame et al, 1998; Pate rnoster et al, 1998). The results in Table 7 show a statistically significant difference be tween males and females for the effect of non-violent control on the freque ncy of violence. This demo nstrates that the non-violent control variable is statistically more likely to produce higher frequency of violence for female victims of intimate partner violence than it is for male victims. Conversely, there was no significant gender difference in the effect of non-violent control on the severity of violence. However, relationship type also demonstrated differential effects by gender. Being married was significantly more lik ely to produce both higher frequency and severity of violence for female victims than for male victims. Additionally, being unemployed and earning other income were significantly more likely to increase the frequency of violence among females. Th ere were no significant gender differences, however, in the effects of higher education on frequency and severity of violence. Based on the results of these analyses the fourth hypothesis in this study, which predicted that high non-violent control is mo re likely to lead to greater frequency and severity of violence for female victims than ma le victims receives at least partial support. Although high non-violent control tactics may be present about equally the relationships
58 of both male and female victims of intimate pa rtner violence, the effect of those control tactics on the nature of the violence does vary by gender. High non-violent control plays a significantly greater role in the victimizati on of women than in the victimization of men in intimate relationships, e nhancing the frequency if not the severity of violence among women but not among men. Table 7: Significance Test for Gender Differences in Regression Coefficients for Frequency and Severity of Violence. Frequency of Violence (Z) Severity of Violence (Z) High Non-violent Control -2.49* -.011 Relationship (married vs. cohabitating) -4.64* -5.16* Years Together 1.06 .956 Unemployed -2.15* **Not calculated Other Income -2.16* -.249 Some College .928 -.623 College Graduate 1.42 1.02 ( Positive z scores indicate that coefficients for males were higher than coefficients for females.) *p<.05 ** Could not be calculated because the numbe r of unemployed males in the sample was too small to produce a valid standard error. Summary To summarize the results of the study, the initial analysis revealed that female and male victims of intimate partner violence in heterosexual and married or cohabitating relationships do not differ in the extent to which their partners ma ke use of non-violent control tactics. In other words, victims of intimate terrorism are not more likely to be females than males, when no other variable s are accounted for. However, non-violent control tactics vary by gender in their influe nce on frequency of vi olence. Specifically, high non-violent control tactics are significantly likely to lead to higher frequency and severity of violence for female victims but not for male victims. Nevertheless, while high
59 non-violent control accounts for none of the variance in frequency and severity of violence for male victims, its influence on fr equency and severity of violence for female victims is also negligible.
60 Chapter 5 Discussion Although most researchers agr ee that intimate partner violence is a national public health issue with far reaching consequences, there is still little to no agreement on the gendered nature of intimate partner viol ence. This gender symmetry versus gender asymmetry debate is important to examine and resolve because it has led to a significant amount of confusion among the general public and policy makers and it has become an increasingly controversial i ssue among scholars that at times overshadows discussions regarding the prevention of in timate partner violence. Curre ntly, there is a push in the literature to gain further unde rstanding into the issues that brought about the gender symmetry debate and to better understand th e gendered nature of intimate partner violence. In the late 1990Â’s Johnson proposed an d moderately tested a theory that within intimate partner violence there are actually two different and distinct phenomena, which he called intimate terrorism a nd situational couple violence. He proposes that the key to distinguishing intimate terrorism and situa tional couple violence is to examine the context and motivation behind the physically vi olence acts that occur in intimate partner violence. According to Johnson (2001), in intimate terrorism the motivation is to maintain a control context over the victim and the relationship in general. He argues that physical violence is only one type of control, and intimate terr orists will also have a high use of non-violent control tac tics. He further asserts th at within intimate partner
61 violence, females disproportionately experience intimate terrorism while situational couple violence is gender symmetric. The purpose of this study was to furthe r explore JohnsonÂ’s two main premises that there are two different types of intimate partner violence, which can be distinguished by the motivation behind the physical violence the frequency and the severity of the violence, and that within intimate partner vi olence females disproportionately experience intimate terrorism. The motivation behind the violence was explained by examining the gendered nature of the use of non-violent cont rol tactics by the victim Â’s partner. This study found that when none of the other known correlates of intimate partner violence are controlled, the difference in a partnerÂ’s use of high non-vi olent control tactics among male and female victims of intimate partner violence is not statistically significant. According to this studyÂ’s preliminary results, the heterosexual partners of female victims are not using a greater number of non-viol ent control tactics than the heterosexual partners of the male victims. Therefore, the prediction that female victims of intimate partner violence are more likely than male victims to experience high non-violent control is not supported. This result could lead to an artificial conclusion that the motivation behind the physical violence does not vary ac ross gender. However, these preliminary results did not allow for other known correlates of intimate partner violence to be controlled, nor could they elucid ate what type of effect nonviolent control tactics have on frequency and severity of violence for male and female victims of intimate partner violence. This study found that when other known correlates of intimate partner violence are controlled, high non-violent control does have a different effect on frequency and
62 severity of violence among male and female victims of intimate partner violence. More specifically, for female victims but not for male victims high non-violent control is a statistically significant predictor for both fre quency and severity of violence. In other words, although for both male and female vic tims of intimate partner violence there does not appear to be a gender difference in use of high non-violent contro l tactics there is a gender difference on the effect of non-violent control tactics fo r frequency and severity of violence experienced. Based on this studyÂ’s results, female victims of intimate partner violence with partners who use many non-vi olent control tactics are more likely to experience frequent and severe acts of physical violence. These results give the appear ance that there is suppor t for one of JohnsonÂ’s main premises that female victims of intimate partner violence experience more intimate terrorism than male victims of intimate part ner violence as measured by frequent and severe acts of violence, but when the gender di fferences are tested onl y the effect of nonviolent control on frequency of violence is significant. Therefore, although high nonviolent control has a significan t effect on severity of violen ce for female victims but not male victims, the difference between the gende rs is not strong e nough to conclude that non-violent control has a stronge r effect on severity of violence for females than for males. However, this same conclusion is not true for frequency of violence. The effects of non-violent control tactics do vary by ge nder in their influence on frequency of violence. Therefore, there is some s upport for JohnsonÂ’s prediction that intimate terrorism is gender asymmetrical, with females subjected to intimate terrorism experiencing physical violence more freque ntly than males subjected to intimate terrorism.
63 Since the effect of non-violent control t actics demonstrated a difference between male and female victims of intimate partner violence, with regard to the frequency but not the severity of violence, it is not possi ble to conclude definitively that intimate terrorism is gender asymmetrical. It is impor tant to explore further why the results for frequency and severity were not the same. In other words why is there a gendered difference for frequency of violence and not se verity of violence? One explanation may lie in the way severity of violence was meas ured for this study. Frequency of violence was measured by using a count variable that demonstrated the number of times a victims experienced a violent act. The severity of violence variable was put into a scale based on how many types of severe violence a person re ported experiencing. The manner in which severe violence was measured is a weakne ss in this study and it is important to understand how the way it was measured may a ffect the results. A person who reported experiencing only one type of severe violen ce was not given the opportunity to state the number of times they experienced the violence. Therefore, they would not be classified as experiencing severe violence. However, a person who reported two types of severe violence was counted as experiencing severe violence in this study. This means, for example, that a person who has been choked fi ve times but did not report any other type of severe violence would not have been cap tured as severe but a person who was choked once and beaten up once would have been c ounted. Based on this operationalization of the variable, many individuals may have b een omitted who actually experienced two incidents or more of severe violence. Since the measure for severe violence was weak and all of this studyÂ’s findings regarding seve re violence are weak, it is important not to draw any definitive conclusions regarding the gendered nature of severe violence or the
64 effect of the non-violent control variable on severe violence. It is important that future studies more fully measure severity and try account for not only the different types of severe violence but also how many times each type occurs In addition to the findings on the rela tionship among gender, intimate terrorism, frequency, and severity, several other findings from the analysis are noteworthy. Based on this study it appears that relationship type is a signifi cant variable when examining frequency and severity of violence among bot h female and male victims of intimate partner violence. Perhaps the most interes ting finding is that being married has an opposite effects for male and female victims. Among females, being married predicts greater frequency and severity of violence. In fact, for females being married increases the likelihood of experiencing frequent vi olence by 76.6%. Married females are also more likely to experience severe violence co mpared to those who cohabitate with their partner. It has been argued in the feminist research that perhaps the marriage license is considered a hitting license (Yllo, 1988). This study would show support for this notion for female victims but not for male victims. In fact, for males the results were the opposite. Male victims who reported cohabitati ng with their heterosexual partners rather than being married were more likely to expe rience both frequent and severe violence. These findings suggest that it is important to explore further why the type of relationship has an opposite effect on freque ncy and severity of violence for male and female victims of intimate partner violence. Other significant findings were the effects of educa tion level and employment status on frequency and severity of violence among the male and female samples. For females, obtaining a higher education, especia lly being a college gra duate, leads to lower
65 frequency of violence but among males educatio nal level obtained had little to no impact on frequency of violence. However, educationa l level did not significa ntly affect severity of violence for either males or females. Similarly, although among females being employed versus unemployed had little to no impact on frequency of violence, among males being employed significantly reduced the frequency of violence. Yet when severity of violence is examined the results show that unemployed females are significantly more likely to experience se vere violence and among males employment status is not signifi cantly correlated with severe violence. The findings are interesting because both employment and educational level have been discussed in intimate partner literature in different capacities. It has been argued, that in general, intimate partner violence is a crime that reaches across all socioeconomic and educational levels, but at the same time the majority of victims in shelters tend to be less educated and make less money (Archer, 2000; Walker, 1979; Yllo & Straus, 1990). It could also be argued that achieving edu cation and being employed are also forms of attaining independence, giving people a sense of power and control over their lives, and for females to achieve a college degree a nd maintain employment would be a direct contradiction to the ideas of patriarchy. The amount of job opportunities available and the types of salaries obtainable could be directly related to educational level and it is much more challenging to isolate an indivi dual who goes to college or is employed (Pence & Paymer, 1993). However, it is par ticularly interesting that for males only employment status and not educational level are significantly correla ted to frequency of violence. This could be an indication that the type of intimate partner violence men experience is different form what females e xperience. In other words, our knowledge of
66 the dynamics of intimate partner violence are primarily derived from studies examining female victimization and characteristics of male perpetrators. It is possible that the effect of educational level on intimate partner violence for male victims, or as Johnson proposes intimate terrorism, are different than its eff ect for females. It is possible that this variation among males and females for educa tional level is an indication that the measures used to determine the use of power and control or non-viol ent control tactics in the Â“Duluth ModelÂ” are actually gender biase d. Meaning, it is possibl e to theorize that previous research has provide d a solid understanding of fema le victimization in intimate partner relationships but not fo r male victimization and it can not be assumed that there is not a gendered difference. One of the main goals of this study was to begin to examine some of the gendered differences of intimate partner violence by te sting JohnsonÂ’s theory regarding intimate terrorism and situational couple violence. This study found support for JohnsonÂ’s theory that two types of intimate partner violence ma y exist and they can be differentiated at least in part by the use of non-vi olent control tactics. It al so supports JohnsonÂ’s assertion that females experience intimate terrorism diffe rently than do males. Finally, it also demonstrates the importance of examining J ohnsonÂ’s theory more cl osely and continuing to test his propositions. The issues within th e intimate partner literature can certainly be addressed by the basic tenets of JohnsonÂ’s theory, but there are still many unanswered questions. Perhaps one of the most important is the notion of female violence. JohnsonÂ’s definition of non-violent control is grounded in feminist theory that has studied maleÂ’s use of violence for thirty years. Most res earchers would agree that this is a good starting point to measure intimate terrorism, but the following question still remains: Do females
67 use the same types of control tactics? In other words, do these findings suggest that men are not victims of intimate terrorism or are th e measures used actually gender-biased and not tapping into the tactics used by female partners? A key area that needs to be continually a ddressed in future research regarding not only JohnsonÂ’s theory but any studies of intim ate partner violence is the examination of the gender differences, not only in prevalence rates or consequences, but in how male and female victims and offenders differ in their contextual experience regarding intimate partner violence. Although, this study was ab le to make some comparisons between male and female samples to begin to address th e gender symmetry debate, a key limitation to this study is that the male and female data se ts were never merged. Maintaining separate male and female data sets allowed fo r a comparison between males and females regarding key variables such as, non-violent control tactics, frequency of violence and severity of violence but it prevented the abil ity to truly control fo r gendersÂ’ effect on the variables of interest. It is possible that just like relations hip type presented a suppressor effect on the non-violent control variable, ge nder may have a unique effect on the ability of the non-violent control vari able to explain the frequenc y and/or the severity of violence. It cannot be concluded that ma les are not suffering from this phenomenon or determined to what degree they experience it until we better unde rstand womenÂ’s use of violence or power and control (O stoff, 2002; Worchster, 2002). In order to better measure intimate part ner violence, determine how similar or different the prevalence rates are among male s and females, or to explore gendered differences it is becoming increasingly im portant to examine the context in which the physically violent act occurs. A weakness to this study and many other studies of
68 intimate partner violence is the way in which violence is being measured fails to account for the context of the violence. Most of ten in large scale victimization surveyÂ’s individuals are simply asked to count or repo rt whether or not a specific act of physical violence occurred. This was how the NVA WS measured both physical and severe violence; therefore, just like most other empirical studies wh ich use large scale surveys, this study also failed to account for the contex t in which the physical violence occurred. With this type of measurement it is not po ssible to know if the respondent is reporting that their partner engaged in a physically viol ent act due to being th e aggressor or if the partner was actually acting in self-defense. In order to fully understand and capture correctly if the person reporting victimization is in fact the victim and not the aggressor, it is necessary to also examine why physic al violence was used. For example, a respondent may have answered Â“yesÂ” that thei r partner threatened th em with a knife. However, the question did not account for the co ntext in which that knife was used. It could have been used aggressively to threaten the person reporting vict imization or in self defense because the person reporting victimi zation was actually choking their partner, causing their partner to grab a kitchen knife and use it to pr event from continuing to be choked. Without accounting for the context, it is not possible to asses or control for the possibility that some respondents reporting be ing victimized by a phys ically violent act were actually the aggressors. Johnson states that a possible third category of intimate partner violence may be what he calls a violen t resister, which is a person who is being victimized by an intimate terrorist but us es physical violence in self defense (Johnson, 2000). With out the context of the violent act being accounted for it is not possible to test for this group or identify how it may affect the results.
69 In conclusion, although this study sought to test JohnsonÂ’s theory, and begin to resolve pieces of the intimate partner violence puzzle, much work still needs to be done. JohnsonÂ’s theory provides a basic framework th at begins to not only resolve the gender symmetry debate but perhaps offers a better explanation for this complex phenomenon called intimate partner violence and the conf licting findings which surround it. It is important to first recognize that the findings from the three previous empirical studies which tested JohnsonÂ’s theory, along with this study demonstrate enough evidence to support further exploration and testing of JohnsonÂ’s propositions In order to continue moving forward in our knowledge of intimate pa rtner violence it is necessary to use the basis of JohnsonÂ’s theory to asses the gender differences within intimate partner violence and begin to acknowledge that perhaps males and females are not experiencing the same phenomenon. Moreover researchers and advocat es need to reach an understanding that although both intimate terrorism and situational couple violen ce are important issues to resolve, they posses their own dynamics, which require different techniques for prevention and intervention. Reaching these goals and objectives means future research are faced with two primary challenges. First, large scale surveys used to measure intimate partner violence need to include some type of measure for non-violent control tactics. At this time, although it is not certain if the measures used by Johns on for non-violent control tactics are gender biased thirty years of empirical studies have estab lished that they are reliable for at least females. Until further research ca n be done to explore female perpetrators of intimate partner violence or the male victimÂ’s experience, this measure at least allows further exploration into JohnsonÂ’s theory re garding female victimization. The second
70 challenge is to recognize that although mu ch has been learned about intimate partner violence over the last thirty years, a great deal is still unknown. It is necessary to recognize that this phenomenon is very comp lex and requires much more sophisticated research tools and methods than have been used thus far. In order to address the issues of context and motivation behind physical vi olence, the gender differences, and the complexities of intimate partner violence, it is important to not continually rely on only large scale surveyÂ’s but begi n to also use qualitative studies. Although, qualitative methods are difficult and time consuming, it is necessary to include them with quantitative studies to help improve our understanding of the context in which the violence occurs. Such mixed methods appro ach will also allow for further exploration into the gender differences in how intimat e partner violence is experienced, how the genders differ in their use and motivation fo r use of non-violent control, or to even discover if the genders use different types of non-violent control tac tics. This greater understanding will help to resolve the ge nder symmetry debate and hopefully bring together all those individuals who are worki ng towards ending intimate partner violence.
71 References Cited Archer, J. (1999). Assessment of the reliabilit y of the conflict tactic scales: A Metaanalytic review. Journal of Interpersonal Violence 14, 1263-1289. Archer, J. (2000). Sex differences in aggression between heteros exual partners: A Meta-analytic review. Psychological Bulletin 126, 651-680. Bograd, M. (1988). Feminist perspectives on wif e abuse: An introduction. In K. Yllo & M. Bograd (Eds.), Feminist Perspectives on Wife Abuse (pp. 11-26). Newbury Park, CA: Sage. Brame, R., Paternoster, R., Ma zerolle, P., & Piquero, A. (199 8). Testing for the equality of maximum-likelihood regression coefficients between two independent equations. Journal of Quantitat ive Criminology, 14, 245-261. Brush, L. D. (1990). Violent acts and injurious outcomes in married couples: Methodological issues in the National Survey of Families and Households. Gender and Society 4, 56-67. Dasgupta, S. D. (1999). Just like men? A critical view of violence by women. In M. F.Shepard & E. L. Pence (Eds.). Coordinating Community Response to Domestic Violence: Lessons from Duluth and Beyond (pp. 195-222). Thousand Oaks, CA: Dasgupta, S. D. (2002). A framework for understanding wom enÂ’s use of non-lethal violence in intimate het erosexual relationships. Violence Against Women, 8, 1364-1389. DeKeseredy, W. S., & Schwartz, M. D. (1998) Measuring the extent of women abuse in intimate heterosexual relationships: A cr itique of the Conflict Tactics Scale. retrieved from VAWnet Web site: http://www.vawnet.org/vn1/library/general/ AR_ctscrit/html. DeKeseredy, W. S. (2000). Current contro versies on defining non-lethal violence against women in intimate heterosexual relationships. Violence Against Women 6, 728-731.
72 Desai, S., & Saltzman, L. E., (2001). Measurement issues for violence against women.In C. M.Renzetti, J. L. Ed leson, & Bergen, R. K. (Eds.), Sourcebook On Violence Against Women (pp. 35-52). Thousand Oaks, CA: Sage. Dobash, R. P. & Dobash, R. E. (1979). Violence Against Wives New York, NY: Free Press. Dobash, R. E. & Dobash, R. P. (1988). Re search as social action: The struggle for battered women. In K. Yllo & M. Bograd (Eds.), Feminist Perspectives on Wife Abuse (pp. 51-74). Newbury Park, CA: Sage Dobash, R. E., & Dobash, R. P. (1992). Women, Violence, and Social Change London and New York: Routledge. Dobash, R. P., Dobash, R. E., Wilson, M. & Daly, M., (1992). The myth of sexual symmetry in marital violence. Social Problems 39, 71-91. Dobash, R. P. & Dobash, R. E. (2004). WomenÂ’s violence to men in intimate relationships Working on a puzzle. British Journal of Criminology 44, 324-349. Domestic Abuse Intervention Project (2002). A Guide for Conducting Domestic Violence Assessments. Duluth, MN: Domestic Abuse Intervention Project. Dutton, D. G. (1980). Some so cial psychological research of relevance to an understanding of domestic violence. In C. T. Griffiths & M. Nance (Eds.) The Female Offender (pp.-81-92). Simon Fraser University Press. Dutton, D. G. (1988). Profiling wife assaulters: Some evidence for a trimodal analysis. Violence and Victims 3(1), 5-30. Dutton, D. G. (1995). Trauma symptoms and PTSD profiles in perpetrators of abuse. Journal of Traumatic Stress 8, 299-315. Dutton, M. A., & Gondolf, E. W. (2000). Wife Battering. In Ammerman & Hersen (Eds.) Case Studies in Family Violence, 2nd ed (323-348). New York, NY. Kluwer Academic / Plenum Publisher. Dutton, D. G., & Strachan, C. E. (1987). Motivational ne eds for power and spousespecific assertiveness in assaultive and non-assaultive men. Violence and Victims 2, 145-156. Dutton, D. G., & Starzomski A. J. (1993). Borderline personality in perpetrations of psychological and physical abuse. Violence and Victims, 8, 327-337. Fiebert, M. (1997). Annotated bi bliography: References exam ining assaults by women on their spouses / partners. Sexuality and Culture 1, 273-286.
73 Ganley, A. (1981). Partici pantÂ’s manual: Court-mandated therapy for men who batterA three day workshop for professionals. Washington D.C.: Center for WomenÂ’s Policy Studies. Gardner, W., Mulvey, E. & Shaw, E. (1995) Regression analyses of counts and rates: Poission, overdispersed poisson, and negative binomial models. Psychological Bulletin, 118, 392-404. Gelles, R. J. (1974). The Violent Home: A Study of Physical Aggression Between Husands and Wives. Newbury Park, CA: Sage. Gelles, R. (2000). Estimating the Incidence a nd prevalence of violence against women. Violence Against Women 6, 7, 784-804. Giles-Sims, J. (1983). Wife Battering New York, NY: Guilford. Goldberg, C. (1999, November 23). Spouse abuse crackdown, surprisingly, nets many women. New York Times p. A16. Gondolf, E. W., (1990). Psychiatric response to fam ily violence: Identifying and confronting neglected danger Lexington, MA: Lexington Books. Gondolf, E. W. (1997). Multisit e evaluation of batterer inte rvention systems: A summery of findings for a 12-month follow-up (Int erim Report). Washi ngton DC: CenterÂ’s for Disease Control and Prevention, U.S. Department of Health and Human Services. Gordon, M. (2000). Definitional issues in violence against women: Surveillance and Research from a violence research perspective. Violence Against Women 6, 7, 747-783. Graham-Kevan, N., & Archer, J. (2003). Intimate Terrori sm and Common Couple Violence: A test of JohnsonÂ’s predic tions in four British Samples. Journal of Interpersonal Violence 18, 11, 1247-1270. Hamberger, L. K., & Hastings, J. (1986). Pe rsonality correlates of men who abuse their partners: A cross-validation study. Journal of Family Violence 1, 323-341. Hamberger, L. K. (1997). Female offenders in domestic violence: A look at actions in their context. Journal of Aggression, Maltreatment and Trauma 1, 117-129. Hamberger, L. K., & Guise, C. E. (2002). MenÂ’s and womenÂ’s use of intimate partner violence in clinical samples. Violence Against Women 8, 1305-1355.
74 Jasinski, J. L. (2001). Theor etical explanations for violenc e against women. In C. M. Renzetti, J. L. Edleson, & Bergen, R. K. (Eds.), Sourcebook on Violence Against Women (pp. 5-21). Thousand Oaks, CA: Sage. Johnson, M. P. (1995). Patriarchal terro rism and common couple violence: Two forms of violence against women. Journal of Marriage and the Family 57, 283-294. Johnson, M.P. (1999). Two types of violence against women in the American family: Identifying patriarchal terrorism and common couple violence. Paper presented at the National Council on Family Re lations annual meetings, Irvine, CA. Johnson, M.P. (2001). Conf lict and Control: Symmetry and asymmetry in Domestic violence. In A. Booth, A.C. Crouter, & M. Clements (Eds.), Couples in Conflict (pp.95-104). Mahwah, NJ: Erlbaum. Johnson, M.P. (in press) G endered Communication and Intimate Partner Violence. In B.J. Dow & J. T. Wood (Eds.), The Sage Handbook of Gender and Communication Thousand Oak, CA: Sage. Johnson, M.P., & Ferraro, K.J. (2000). Research on Domestic Violence in the 1990s: Making Distinctions. Journal of Marriage and the Family 62, 948-963. Johnson, M.P., & Leone, J.M. (2005). The Diffe rential Effects of In timate Terrorism and Situational Couple Violence: Findings from the Nationa l Violence Against Women Survey. Journal of Family Issues 26, 3, 322-349. Julian, T. W., & McKenry, P. C. (1993). Mediators of male violence toward female intimates. Journal of Family Violence 8, 39-56. Kalmuss, D. (1984). The in tergenerational transmission of marital aggression. Journal of Family Violence 8, 39-56. Kaufman, J., & Zigler, E. (1993). The intergenerational transmission of abuse is overstated. In R. J. Gelles & D. R. Loseke (Eds.), Current Controversies on Family Violence Newbury Park, CA: Sage. Kimmel, M. S. (2002). Â“Gender sy mmetryÂ” in domestic violence. Violence Against Women, 8, 1332-1363. Kurz, D., (1989). Social Science perspectives on wife abuse: Current debates and future directions. Gender and Society 3, 489-505. Langen, P. A. & Innes, C. A. (1986). Preventing domesti c violence against women. Washington DC: US Department of Justice.
75 Long, J. (1997). Regression models for categorical and limit ed-dependent variables. Thousand Oaks, CA: Sage Martin, D. (1976). Battered Wives San Francisco, CA: Glide. McNeely, R. L., & Jones, J. M. (1980). Ref uge from violence: Designing shelter services for battered women. Administration in Social Work 4. 71-81. McNeely, R. L., & Jones, J. M. (1987). T he truth about domestic violence: A falsely framed issue. Social Work 32, 485-490. Morse, B. J. (1995). Beyond the Conflict Ta ctics Scale: Assessing gender differences in partner violence. Violence and Victims 10, 251-272. Nazroo, J. (1995). Uncovering gender differences in the use of marital violence: The effect of methodology. Sociology 29, 475-494. OÂ’Leary, K. D. (1988). Physi cal aggression between spouses: A social learning perspective. In V. B. Van Hasselt, R. Morrison, A. S. Bellack, & M. Herson (Eds.), Handbook of Family Violence New York, NY: Plenum. Osthoff, S. (2002). But, Gertr ude, I beg to differ a hit is no t a hit is not a hit. Violence Against Women 8, 1521-1544. Pagelow, M. D. (1981). Fa ctors effecting womenÂ’s decision to leave violent relationships. Journal of Family Issues 2, 391-414. Pagelow, M. D. (1984). Family Violence New York, NY: Praeger. Paternoster, R., Brame, R., Mazerolle, P ., Piquero, A. (1998). Using the Correct statistical test for the equality of regression coefficients. Criminology, 36, 859866. Pence, E., & Paymar, M., (1993). Education groups for men who batter: The Duluth Model New York, NY: Springer. Pleck, E., Pleck, J., Gro ssman, M., & Bart, P. (1978). The battered data syndrome: A comment on SteinmetzÂ’s article. Victimology, 2, 680-684. Saltzman, L. E. (2004). Defi nitional and methodological issues related to transitional research on intimate partner violence. Violence Against Women 7, 812-830. Saltzman, L. E., Fanslow, J. L., McMahon, P. M., & Shelley, G. A. (1999). Intimate partner violence Surveillance: unifo rm definitions and recommended data elements (version 1.0). Atlanta, GA: CDC, National Center for Injury Prevention and Control.
76 Saltzman, L. E., Green, Y. T. Marks, J. S., Thacker, S. B. (2000). Violence against women as a public health issue. American Journal of Preventative Medicine 19, 4, 325-329. Saunders, D. G. (1988) Other Â‘truthsÂ” about domestic vi olence: A reply to McNeely and Robinson-Simpson. Social Work 33. 179-198. Saunders, D. G. (1992). A ty pology of men who batter: Three types derived from cluster analysis. American Journal of Orthopsychiarty 62, 264-275. Saunders, D. G. (2002) Are physical assaults by wives and girlfriends a major social problem? Violence Against Women 8, 1424-1448. Smith, M. D. (1990). Patriarchal ideology and wife beating: A test of feminist hypothesis. Violence and Victims 5, 257-273. Stark, E., & Flitcraft, A. (1996). Women at Risk: Domestic Violence and WomenÂ’s health Thousand Oaks, CA: Sage. Steinmetz, S. K. (1977-78). The battered husband syndrome. Victimology 2, 499-509. Straus, M. A. (1971). Some social antecedents of physica l punishment: A linkage theory interpretation. Journal of Marriage and the Family 33, 658-663. Straus, M. A. (1979). Measuri ng intrafamily conflict and vi olence: The Conflict Tactics (CT) Scales. Journal of Ma rriage and the Family, 41, 75-88. Straus, M. A. (1991). New theory and old canards about family violence research Social Problems 38, 180-197. Straus, M. A. (1993). Physical assaults by wives: A majo r social problem. In R. J. Gelles & D. R. Loseke (Eds.) Current controversies on family violence. Newbury Park, CA: Sage. Straus, M. A. (1999). The controversy of domestic violence by women: A methodological, theoretical, and sociology of science analysis. In X. B. Arriaga & S. Oskamp (Eds.), Violence in Intimate Relationships (pp 17-44). Thousand Oaks, CA: Sage. Straus, M. A., Gelles, R. J. & Steinmetz, S. K. (1980). Behind Closed Doors: Violence in the American Family. Garden City, NY: Anchor Press/Doubleday. Straus, M. A. & R. J. Gelles (Eds.) (1990). Physical Violence in American Families: Risk Factors and Adaptations to ViolenceIn 8,145 Families. New Brunswick, NJ: Transaction.
77 Strube, M. J. (1988). The decision to leave an abusive relationship: empirical evidence theoretical issues. Psychological Bulletin 104, 236-250. Swan, S. C. and Snow, O. L. (2002). A typology of womens use of violence in intimate relationships. Violence Against Women 8, 3, 286-319. Tjaden, P. & T hoennes, N. (2000). Extent Nature and Consequences or Intimate Partner Violence: Findings from the Na tioanl Violence Against Women Survey. Washington, DC: US Department of Justic e, National Institute of Justice. Tolman, R.M. (1989). The devel opment of a measure of psyc hological maltreatment of women by their male partners. Violence and Victims 4, 159-177. Vivian, D., & Malone, J. (1997). Relations hip factors and depressive symptomatology associated with mild and severe husband-to-wife physical aggression. Violence and Victims 12, 3-18. Worcester, N. (2002). WomenÂ’s use of force. Violence Against Women 8, 1390-1415. Walker, L. (1979). The Battered Women. New York, NY: Harper & Row. Widera, F. (2002). Dynamics of Domestic Violence. Saint Petersburg, FL: Florida Regional Community Policing Institute. Yllo, K. (1988). Political and methodological debat es in wife abuse research. In K. Yllo & M. Bograd (Eds.), Feminist Perspectives on Wife Abuse (pp. 28-50). Newbury Park, CA: Sage. Yllo, K. & Straus, M. A. ( 1990). Patriarchy and violence against wives: The impact of structural and normative fa ctors. In M. A. Straus & R. J. Gelles (Eds.), Physical Violence in American Families: Risk Fa ctors and Adaptations to Violence in 8,145 families (pp. 383-399). New Bruns wick, NJ: Transaction.