Application of Attachment Theory to Male Violence Toward ... · Application of Attachment Theory to Male Violence Toward Female Intimates John E. Kesner,!'" Teresa Julian,^ Patrick

Joumal of Family Violence, VoL 12, No. 2, 1997

Application of Attachment Theory to MaleViolence Toward Female Intimates

John E. Kesner,!'" Teresa Julian,^ Patrick C. McKenry^

Ninety one men were interviewed regarding their intimate relationships andchildhood experiences. A secondary analysis of an extant data set wasconducted to explore the applicability of Bowlby's attachment theory inexplaining the etiology of male intimate violence. Various attachment-relatedvariables were hypothesized to be related to male intimate violence. Results ofa hierarchical regression analysis indicated that the attachment cluster ofvariables served as unique predictors of male intimate violence. The male'sperceived relationship support and recollection of maternal relationship werethe two significant predictors of male intimate violence within the attachmentcluster. Race and education were also found to be significant predictors ofmale violence toward a female intimate.

KEY WORDS: attachment theory; male violence; domestic violence; hierarchical regressionanalysis.

INTRODUCTION

Violence perpetrated by male intimates may be the most common ori-gin of injuries to women (Hadiey, 1992; Stark and Flitcraft, 1985), A surveyby the National Institute of Justice reports that an average of a half millionviolent acts are committed against women by intimates each year (U.S. De-

'Department of Early Childhood Education, College of Education, Georgia State University,Atlanta, Georgia 30303.

^Department of Nursing, Otterbein College, Westerville, Ohio 43081.^Department of Family Relations and Human Development and Department of Black Studies,College of Human Ecology, The Ohio State University, Columbus, Ohio 43210.

'*Tt> whom correspondence should be addressed.

211

0885-7482/97/O60(W)211$12J0/O C 1997 Plenum Publishing Corporation

212 Kesner et at.

partment of Justice, 1994), Furthermore, 75% of all homicides of womenare perpetrated by male intimate partners (Campbell, 1992).

Despite much media attention, systematic research conceming malehostility and aggression toward women is limited (Margolin et al., 1988;Steinmetz, 1987). Most studies of domestic violence have been atheoretical,focusing on conceptually unrelated individual factors such as alcohol, stagein the life cycle, and social class—often in isolation from one another (How-ell and Pugliesi, 1988, Gelles and Comell, 1990).

It was thus the purpose of this study to apply concepts from attach-ment to determine their utility in explaining to male violence toward femaleintimates.

ATTACHMENT THEORY

Attachment theory (Bowlby, 1969) is an ethological theory of inter-personal relationships that emphasizes the evolutionary significance of in-timate relationships, especially those in early childhood. Bowlby believedthat individuals possess an inbom need for close attachments to significantothers in order to survive (Bowlby, 1969). The "attachment behavioral sys-tem" is made up of behaviors that elicit care and nurturing from primarycaregiver. Satisfaction of attachment needs by the primary caregiver enablesthe individual to form a secure attachment with the caregiver. Whenstressed, the attachment behavioral system achieves a state of heightenedactivation motivating the individual to seek out the attachment figure forsupport, nurturance and protection. Based on the history of early attach-ment relationships an "internal working model" is formed which serves asthe foundation for future relationships (Bowlby, 1969). The internal work-ing model guides and motivates the individual's relationship behaviors bypredicting the availability of the attachment figure to provide care and nur-turance. Thus, individuals with an insecure model(i.e., one that is based ona history of the attachment figure not meeting attachment needs) will ex-pect future attachment figures to similarly not meet these needs. Individualswho have a history of secure relationships will develop an intemal workingmodel of relationships that anticipates the attachment figures will meet at-tachment needs.

Attachment theory has been used extensively to explain parent-childrelationships (Ainsworth et ai, 1978; Bretherton, 1987; Main et al., 1985;Lyons-Ruth et al., 1990). There is abundant evidence supporting the roleof attachment security in infiuencing social competence in children. How-ever, attachment theorists and researchers believe early childhood relation-

Attachment Theoty 213

ships play a major role in interpersonal relationships throughout thelifespan (Bowlby, 1988),

Adult Attachment

Adult attachment researchers contend that romantic love is an attach-ment process (Hazan and Shaver, 1987). During adult relationships,Ainsworth (1989) argues that attachment issues are still salient and whilenot replacing the initial attachment figure, the intimate other tends to be-come the primary attachment figure for the individual. According to at-tachment theory, individuals in adult intimate relationships are still lookingfor the same sense of security and support from their partners that wasimportant in their secure attachment relationships during early childhood.Hazan and Shaver (1987) have identified adult "attachment styles" thatcharacterize how an individual thinks about intimate relationships. Thesestyles represent characteristic ways of thinking about intimate relationshipsbased on past experiences. Additionally, they guide the interpretation ofincoming relationship information in accordance with the expectations builtinto the model (Bretherton, 1987). These attachment styles have been vali-dated against retrospective accounts of early childhood attachment experi-ences (Shaver and Brennan, 1992).

Stress and the Attachment Behavioral System

From an attachment perspective, the effects of stress can be twofold.In childhood, stressful situations tend to activate the attachment behaviors(attachment behavioral system) (Bowlby, 1973), causing the individual toseek out the attachment figure for support and comfort. Excessive stressin childhood, without amelioration by a sensitive caregiver, can negativelyinfiuence the security of the developing attachment relationship, adverselyaffecting one way of dealing with stress. Past research clearly indicates thatstress is highly related to relationship violence (e.g., MacEwen and Barling,1988; Neidig et al, 1986). However, stress levels may trigger attachmentissues which lead to violence. In adult relationships, stress can be alleviatedby various coping mechanisms (e.g., including obtaining support from a se-cure attachment relationship). However, if coping resources are limited dueto adverse childhood experiences, and if the source of stress is the attach-ment relationship, an intense confiict can result. Thus, individuals and fami-lies who find themselves in situations where they are unable to successfullycope with the demands of stress find that their frustrations are likely to

214 Kesner et al.

increase, and this frustration may result in violent behavior (Dollard et al,1939).

Individuals operating with an insecure attachment model may perceiveambiguous behaviors by the intimate other as rejecting and unsupportive.Thus, life stress may be compounded when one is predisposed to expectrejection fi-om the intimate partner, and yet the intimate partner may bethe only source of support an individual has to alleviate stress.

Aggression and the Attachment Behavioral System

Aggression has been linked to the attachment behavioral system duringearly childhood. Bowlby (1988) states that angry behavior is a way thatchildren may communicate to their primary attachment figure that theirattachment behavioral system has been activated and that attachment needsare not being met. This generally occurs in response to a separation wherethe child perceives that the attachment figure is unavailable. Bowlby termedthis "functional anger." That is, it communicates to the attachment figuredispleasure at being separated, and if responded to appropriately by theattachment figure, will facilitate a secure relationship.

A consistent history of insecure attachment experiences may distortthis functional anger into violence by the male. Males who batter theirwives may be utilizing physical violence during times that they perceive anattachment need not being met by their partners. Thus violence by themale may be a dysfunctional coping mechanism for dealing with perceivedunresponsiveness or rejection by the intimate partner.

HYPOTHESES

Separation from the attachment figure in adult relationships becomesa matter of perceived availability of relationship support from the partnerrather than physical proximity. That is, psychological availability of the at-tachment figure replaces the need for physical closeness that is importantin childhood. Thus, the male's perception of the amount of relationship sup-port provided to him by his partner may impact the degree of male per-petuated violence toward his female partner.

An individual with an insecure attachment history may not have theusual coping mechanisms that a secure individual would have (i.e., the at-tachment figure). Therefore, stress may not be able to be alleviated and,thus an individual with an insecure attachment model may be experiencinga high level of negative life events stress.

Attachment Theory 215

Attachment theory contends that the sense of self develops concur-rently with the intemal model of attachment relationships. Consequently,individuals who receive insensitive and inappropriate caregiving duringearly childhood may develop the sense that they are unworthy of love andnurturing (Bretherton 1987, 1992). Additionally, Collins and Read (1990)have found lower levels of self-esteem in adults with insecure attachmentstyles. Thus, the male's level of self-esteem may be an indicator of attachmentsecurity and another important variable in understanding interpersonal vio-lence.

During early childhood, the attachment behavioral system is activatedin an environment of insensitive caregiving. Chronic insensitivity by thecaregiver can result in persistent activation of the attachment behavioralsystem. Bowlby posited that the attachment behavioral system, because ofits inherent survival value, dominates other behavioral systems (Bowlby,1969), Thus, activation of the attachment behavioral system suppressesother behavioral systems, and chronic activation may inhibit the develop-ment of autonomy and individuation (Bretherton, 1987, 1992). If the inti-mate other is the primary attachment figure, the male's perception of howautonomous he is in the intimate relationship may be an indicator of thesecurity of the relationship and may be predictive of male violence towardthe intimate.

Attachment theory contends that the success of initial attachment re-lationship exerts a unique influence on subsequent relationships. Given thefact that the majority of the caregiving in westem society is performed bythe mother this relationship is believed to have special significance. There-fore, recollections of the mother may be an indicator of attachment historyand security.

The purpose of this study was to examine the relationship betweenthe attachment related factors oi perceived relationship support, negative lifeevents stress, self esteem, sense of autonomy, and perceived quality of earlychildhood relationship and male violence toward a female intimate. Thevariables in this study were selected from an extant data set, and were notoriginally chosen to measure attachment. However, they were theoreticallyrelated to attachment concepts to make them useful in this preliminaryexamination of the relationship between attachment and male violence.

Specifically, it was hypothesized that attachment related variableswould be correlated with and serve as unique predictors of male intimateviolence; that is, they would account for unique variance in a model pre-dicting male violence against a female intimate beyond that which wouldbe accounted for by significant demographic variables.

216 Kesner et al.

METHODS

Subjects

A total of 94 men were interviewed; 41 were defined as physically vio-lent toward their spouse, and fifty were defined as nonviolent. Three sub-jects were eliminated from the final data analysis because of incompleteinterviews.

Some of the violent males were recmited from a population of menparticipating in a group program for male batterers at a community mentalhealth center in a large midwestern city. Most of the men participating inthis mental health program were court referrals; others were therapeuticreferrals. Study inclusion criteria included (a) first-time participation in theprogram, and (b) involvement in an intimate relationship with a female.Agency personnel assisted in the recmitment of subjects.

Other participants in the violent group and all of the nonviolent sub-jects were solicited through newspaper advertisements for participants ina study of men's health and family roles. Those men whose Confiict/TkcticsScale scores indicated any physical violence toward their female intimatein the past year were assigned to the violent group. Additional inclusioncriteria for the comparison group of nonviolent men were: (a) involvementin an intimate relationship for at least the past year (i.e., married, cohabi-tating or just recently divorced); and (b) annual incomes of less than thirtythousand dollars per year and educational levels of less than fourteen years,comparable to the men participating in the group program for male bat-terers at the mental health center.

After all participants were screened, an appointment was scheduledfor a one hour, face-to-face interview. Subjects were interviewed by trainedinterviewers. The participants were informed of their voluntary status, thatthey could withdraw at anytime, and that their responses were confidential.Consent forms were reviewed and signed, and subjects were reimbursed$25 for their participation in the study. Data collection for both groups ofmales took place at a university medical research center. A combinationof interview and self-report measures were used.

Sample Characteristics

As shown in Tkble I, the mean ages of the nonviolent and violentgroups were comparable (violent = 31.8, and nonviolent = 33). Approxi-mately percent of the violent males were Caucasian, and 38% were Afri-can-American, whereas Caucasians represented 88% and African-American

Attachment Theory 217

represented 12% of the nonviolent group. Both groups were similar interms of marital status. However, a higher percentage of the violent groupreported being separated from their spouses, whereas members of the non-violent group were more likely to report being married. Violent subjectsreported incomes of less than $5000 more frequently than nonviolent males.The majority of the incomes of both nonviolent and violent men fell withinthe $10,000 to $25,000 range. Although both groups had roughly similareducational levels, the participants that were categorized in the violentgroup had almost 1 year less education (x = 12,71 years) than the non-violent group (x = 13.38 years). Most participants reported good to excel-lent health; only four violent males and four nonviolent males perceivedtheir health as less than good.

Instrumentation

The interview consisted of various demographic background questionsas well as standardized instmmentation. Demographic data included re-spondent's age, race, marital/cohabitation status, socioeconomic status, his-tory of parental abuse, and health status. Poverty status was defined by an

Thhie I, Selected Sociodemographic Characteristics of Study Participants

Violent Males(« = 41)

Nonviolent Males(« = 50)

AgeX = 31.8 {SD = 7.63)

RaceWhite = 62% (n = 26)Black = 38% {n = 16)

Marital statusMarried = 31.0% (n = 13)Separated = 21.4% (n = 19)Divorced = 11.9% (n = 5)Single = 35.7% (/» = 15)

tncomeLess than $5,000 = 21.4% (n = 9)$5,000 to $10,000 = 11.9% (n = 5)$10,000 to $24,999 = 50.0% (« = 1)$25,000 to $30,000 = 14.3% {n = 6)

EducationX = 12.71 {SD = 2.2)

Health statusExcellent = 40.5%Good = 40.0%Fair = 9.5%Poor = 0.0%

X = 33.0 {SD = 7.7)

White = 88% (n = 44)Black = 12% {n = 6)

Married = 46% (« = 23)Separated = 6% (n = 3)Divorced = 10% (n = 10)Single = 28% {n = 14)

Less than $5,000 = 6% (n = 3)$5,000 to $10,000 = 20% (n = 10)$10,000 to $24,999 = 50% {n = 25)$25,000 to $30,000 = 24% {n =12)

X = 13.4 {SD = 1.7)

Excellent = 56.0% (n = 28)Good = 36.0% {n = 18)Fair = 8.0% (n = 4)Poor = 0.0% {n = 0)

218 Kesner et at.

annual family income of $10,000 or less; the Federal poverty line for afamily of three was approximately $9900 in 1989 (U.S. Bureau of the Cen-sus, 1991). Males whose incomes were greater than $10,000 were catego-rized as not being within the poverty range. Educational level was definedby the number of completed years of education. The race of the subjectwas defined either as Caucasian or African-American; no other racialgroups were represented.

Perceived Relationship Support

Male perceived relationship support was measured by the supportsubscale of the Autonomy/Relatedness Inventory (ARI) (Schaefer and Edg-erton, 1982). This subscale was computed by summing four items on theARI that assessed the degree to which the intimate partner provides emo-tional support to the respondent. Items asked how much the partner "Isthere when I need her," or "Does what she can to make things easier forme," Five possible responses ranged from "not at all like her" to "verymuch like her." These items are thought to assess issues salient to attach-ment theory. Whether an individual feels that he can turn to the attachmentfigure when stressed is critical for secure attachment formation. Reliabilityfor relationship support was found to be .91. Vahdity was moderate tostrong as evidenced by correlations with the Spanier Dyadic AdjustmentScale (Schaefer and Edgerton, 1982). The Cronbach alpha reliability forthe entire scale in this study was .72.

Self-Esteem

Self-esteem was measured by the Rosenberg (1979) Self-Esteem Scale.This measure is a 10-item scale with a 5-point response range from(l)"strongly disagree" to (5) "strongly agree." Items tap issues of esteemsuch as personal qualities, abilities, and general evaluation of the self. Thismeasure has been found to have high intemal consistency. In this study, theCronbach alpha coefficient was .91. The development of the sense of selfis linked to the development of attachment models, thus, the level of self-esteem may indicate the security of the attachment model of an individual.

Sense of Autonomy

Sense of autonomy within the interpersonal relationship was measuredby four items of the autonomy subscale of the Autonomy/Relatedness In-

Attachment Theory 219

ventory (Schaefer and Edgerton, 1982). Short statements about the auton-omy of the interpersonal relationship are asked covering topics such as per-sonal freedom, disagreements, and decision making. Attachment theorycontends that an individual strives to achieve a balance between autonomyand satisfying attachment needs. Failure to alleviate attachment needs willadversely affect the individuals abilify to achieve autonomy. Five possibleresponses ranged from "not at all like her" to "very much like her." TheCronbach alpha reliability for this scale was ,69.

Perceived Quality of Early Chiidhood Relationship

Perceived quality of early childhood relationship was measured by ask-ing the respondent his recollection of his mother when growing up andhow often his mother expressed love and caring. This study variable wasmeasured on a Likert-type scale ranging from "Never" to "Rarely." Anindividual's recollection of the maternal relationship is a good indicator ofattachment history. This perception is a critical component of many attach-ment history measures (George, Kaplan, and Main, 1987; Pottharst andKessler, 1991), thus demonstrating face validity.

Perceived Life Stressors

Perceived life stressors were measured with the Life Experiences Sur-vey (Sarason et al, 1978); this instrument assesses perceived life stressorswithin the past year. This 57-item self-report measure allows for respon-dents to indicate events that they have experienced during the past yearand rate whether they perceived these events as positive or negative stres-sors. Items covered topics related to work, home life (including intimaterelationships), finances, and personal health. Summing the ratings provideda negative change score, a positive change score, and a total change score.This measure was only administered once, so test-retest reliability is notpossible to determine. However, Sarason et al report test-retest correla-tions for negative change were ,19 and ,53 (p < .001), positive change .56and .88 (p < ,001), and total change score .63 and ,64 (p < .001).

Aggression/Male Perpetrated Physical Violence Toward the Female Intimate

Aggression/male perpetrated physical violence toward the female inti-mate was measured by the Conflict-Tkctics Scale (CTS) (Straus, 1979). The(CTS) is a 19-item self-report measure in which the individual reports on

220 Kesner et at.

the tactics used in conflicts with the intimate partner during the past year,and then if these tactics have ever been used. The CTS has 19 descriptionsof conflict tactics that range from less to most severe with the individualanswering how many times in the past year each tactic has been used. These19 items are broken down into three subscales: (a) reasoning; (b) verbalaggression; and (c) violence. The respondent answers in relation to himselfand his intimate partner, using a 7-point scale ranging from 0 times tomore than 20 times. For these analyses, only the violence subscale was usedwhich includes only actual acts of physical violence against an intimate,Cronbach's alpha was ,82 for the entire scale and .83 for the violencesubscale. The nature of this measure yields a substantial number of subjectsindicating no violence. This can create a highly skewed sample. Thus, alog procedure (Moore and McCabe, 1989) was utilized to normalize thedistribution.

RESULTS

Univariate Analyses

Initial univariate analysis indicated that demographic variables, male'seducational level and race, were significantly correlated (negative) with malephysical violence against an intimate partner. Because of the dichotomousnature of the sample (either African American or Caucasian) a single di-chotomous variable for race was created; subjects were coded either as 1 ifthey were Caucasian, or 0 if they were African American, T-tests were per-formed out on these and other demographic variables (i.e., male's education,male's race, male's income, male's marital status, and male's physical health)to determine differences between the violent and nonviolent groups. Racewas the only one of these demographic variables that showed significant dif-ferences between violent and non-violent males (t = 3,46, p < .01),

Additionally, self-esteem, perceived relationship support, perceived re-lationship autonomy, and the male's recollection of his mother, were signifi-cantly related (negative) with male physical violence. Male negative life eventsstress was positively related to male violence as indicated in Tkble II,

Multiple Regression Analysis

A hierarchical regression analysis was used to address study hypothesesbecause the intent of the study was to demonstrate a unique effect of at-tachment factors on male violence beyond that variance explained by demo-

Attachment Theory 221

•c

p II

O II ° II p

ON ON .1), ON

I II 3 II ° II

o II S II 33 II

p II p II ^ II

O\ « ON » ON • ON

o n

I

N II j S

l' I

11-I.." oo -a

II ,-i II

ON ON

; II : : II

m II c5

Ili111.," o

p II

^ II

S II

e

* o\

i"

li V V V

222 Kesner et al.

graphic variables. Hierarchical regression determines if the addition of aset of variables significantly increases the i?^ accounted for beyond a pre-vious set of variables (Cohen and Cohen, 1983). The independent variableswere analyzed by structuring them in two clusters, demographic and at-tachment.

The first cluster, included traditional demographic variables (i.e., raceand educational level). Education and race were included because of theirsignificance in relation to violence in the univariate analyses. The modelthat included only demographic variables accounted for 13% (F = 6.45, p< .01) of the variance in predicting male violence. Both male's educationand race were significant predictors of male violence (see Ikble III).

The second cluster included the first cluster as well as the attachmentrelated variables. The analysis indicated that the addition of attachmentvariables (i.e., perceived quality of early childhood mother-son relationship,self esteem, relationship support, sense of autonomy, and perceived life stres-sors) significantly increased the R^ of the model by 27% (F = 7.50, p <,01). The model that included demographic and attachment related vari-ables accounted for 40% of the variance in predicting male violence (F =7,88, p < .001). The best predictor in this model was male perceived re-lationship support (F = 8.09, p < .01) (see Tkble IV). Other predictors inorder of significance were race (F = 6.21, p < .05); recollection of mother(F = 5.15, p < .05); and education(F = 2.96, p < .05).

Results of a stepwise regression using all independent variables yieldedfour significant variables, and also indicated that male perceived relation-ship support was the best predictor in the model. In order of entry intothe regression equation, the variables were as follows: male perceived re-lationship support (F = 24.06,;? < .001); male life events stress (F = 7.39,p < .01); male's recollection of mother (F = 5,34, p < .05); and male'srace (F = 6.40, p < .01).

Because stress may interact with childhood and adult variables, result-ing in variations in attachment with level of stress, interaction terms were

Ibhle tu. Multiple

Variable

Male's educationMale's race

F^ = .13F = 6.45 p < .01.

Regression Analysis of Male Violence Tbward FemaleSelected Demographic \^riables

StandardizedRegressionCoefficient

-.10-.63

StandardError

.05

.22

F

4.498.42

tntimates and

P <

.05

.01

Attachment Theory 223

Ihble IV Multiple Regression Analysis of Male Physical Violence Tbward a Femaletntimate and Selected Independent Variables

Variable

Male's educationMale's raceMale's recollection of motherMale's preceived relationship

support

^= .40F = 7.88 p < .001.

StandardizedRegressionCoefficient

-.07-.49-.16

-.07

StandardError

.04

.20

.07

.02

F

2.966.215.15

8.09

P <

.05

.05

.05

.01

created with two child attachment related variables and three adult attach-ment related variables. None of the interaction terms proved to be signifi-cantly related to male violence against a female intimate.

DISCUSSION

Results of these analyses indicate the potential value of an attachmenttheoretical perspective in predicting male physical violence against a femaleintimate. The model developed through the data analysis supports a majortenet of Bowlby's (1969) attachment theory (i.e., early parent-child rela-tionships may infiuence intimate relationship behavior throughout an indi-vidual's life span). Results of this study indicated that the cluster ofattachment-related variables was significantly related to male violence ininterpersonal relationships. Specifically, violence in males was related to:(a) a perceived deficiency in love and caring from their mother while grow-ing up; (b) lower self-esteem; (c) perception of less relationship support;(d) perceptions of low relationship autonomy and (e) number of recentlife stressors.

The strongest predictor of male violence in the regression model wasmale's perceived relationship support from his female intimate. Althoughnot specifically designed to measure attachment, the adult's abilify to feelsupport from the intimate other may be a direct reflection of attachmentsecurify. The fact that the male's recollection of his relationship with hismother was also a significant predictor of male aggression further supportsthe salience of attachment issues in understanding male violence. Althoughthis variable was not specifically designed to assess childhood attachment

224 Kesner et al.

history, it assesses the male's current understanding of his relationship withhis mother. The perception of childhood attachment relationships is a criti-cal component of adult attachment status (George et al, 1987).

According to attachment theory, individuals who received insensitiveand inappropriate caregiving from their primary attachment figure in child-hood are more likely to have expectations for similar treatment in adultattachment relationships. That is, insecure attachments in childhood createmodels about intimate relationships that may persist into adulthood. Theaggression that served a functional purpose in childhood by communicatingattachment needs may become distorted into violence by the male whoperceives the same insensitivify in his attachment relationship in adulthoodeven though it may not exist (Mayseless, 1991).

Life events stress also appeared to have a significant effect on maleviolence. Although not significant in the multiple regression analysis, lifeevent stress was the second variable to enter a stepwise regression analysis.In terms of attachment per se, the lack of interaction effect between at-tachment and life events stress on the dependent variable suggests that lifestress did not confound the attachment system resulting in greater violence.

The findings of this study did not indicate that the female intimateswere ineffective primary attachment figures. It may be that men with in-secure attachment histories may have unrealistic expectations which theirpartners can never fulfill. In these analyses male's perception of relation-ship support is a key predictor of male intimate violence. Such findingswould imply that female intimates may be in a vulnerable position withmen who have insecure attachments. Consciously or unconsciously, theadult male may be predisposed to believe that the new primary attachmentfigure will not be available to assist him in times of stress and thus reactwith extreme aggressive behaviors.

In addition, the findings of this study indicated that race and educationwere important variables in understanding male violence toward female in-timates. Past findings relative to the relationship between African-Americanmen, decreased educational opportunities, and the propensify for violencehave been inconsistent (Allen and Straus, 1980; Stark and McEvoy, 1970).Void (1986) notes that violence among African-Americans is better ex-plained by structural factors such as resource inequalify. It is important tonote that in the stepwise regression analysis, race was the last variable toenter the model. Although a significant predictor, it only added 4% to thevariance accounted for in this model. However, in this sample, Caucasiansand African-Americans had similar earning power and educational attain-ment. Also, there were no racial differences in perceived life stressors. Inaddition, African-Americans in this study were significantly more likely toview their relationship with their female partner negatively. According to

Attachment Theory 225

Staples (1988), there is tension in the Black male and Black female rela-tionship because of the lack of opportunities available for Black males tosucceed in the traditional good provider role. Also, the greater inde-pendence of Black women makes it even more difficult for men to engagein traditional role behaviors, what Franklin (1984) terms "incompatible roleenactments," Ball and Robbins (1986) found, in fact, that being marriedwas a stressor for Black men and was actually related to lower levels ofwell-being.

CONCLUSIONS

In terms of applying attachment theory to understanding male intimateviolence, again caution should be exercised in drawing conclusions fromthis data. First, this research is based on the selection of study participantswho corresponded to specific demographic characteristics (i.e., charac-teristics similar to the men who were mandated by the courts to participatein a communify agency's batterers program).

Second a noncausal model was used, it is not possible to state thedirection of the relationship between male violence and the variables ofinterest. For example, violent episodes may have influenced perceptions ofthe intimate relationship. However, it was not the purpose of this study toestablish causalify, but merely to determine the relationship of specifictheoretically related variables to male violence toward female intimates.

Also, this was a secondary analysis of an extant data set. The variablesused were not originally designed to measure attachment and although theyare conceptually similar to attachment issues there generalizabilify is lim-ited. However, these data do provide some tentative support for an attach-ment perspective . In so doing, this study helps to fill a void in the literaturewhich has traditionally been bereft of theory.

Despite these limitations, these findings suggest some tentative impli-cations for prevention and treatment programs. The findings in this studysuggests the infiuence of early relationships in forming a foundation forfuture relationships. Thus, programs and policies that attempt to reducethe incidence of adult male violence toward female intimates might profitfrom primary interventions that begins in childhood by providing educa-tional opportunities to parents that enable them to provide the secure andnurturing caregiving that is critical to the child's optimal development. In-dividuals who provide a secure base for their children may help to preventdysfunctional anger.

Also, the study's findings suggest the utilify of individual and familytherapy. Violence-prone individuals still may be dealing with attachment

226 Kesner et al.

issues that remain from their own childhoods. Individual and family therapymay be the means of providing couples with appropriate interpersonal ex-pectations, insight into their own behavior, individual behavioral responsi-bilify, and new behaviors associated with appropriate attachment-relatedbehaviors and nonviolence.

The findings of this study also suggest the importance of the two demo-graphic variables included in the analysis. That is, the significance of edu-cation and race underscores the importance of macro level indicators thatwould increase the resource base and self esteem of lower socioeconomicstatus African-American males. Along with macro system changes, it is criti-cal that culturally relevant methods of both studying and interviewing Af-rican-American and lower SES families be developed and implemented.

In future efforts to determine the viabilify of an attachment perspectivein predicting domestic violence, researchers should utilize established at-tachment instruments to more fully determine the applicabilify of this the-ory to family violence. In addition, future research might provide insightsinto the role of attachment in domestic violence by identifying the copingmechanisms of nonviolent males who report poor attachment relationshiphistories. Data also should be gathered regarding the role of significantothers that may provide compensatory attachment relationships, therebyproviding a secure foundation for future relationships. There are other datathat would help establish a linkage between attachment and violence. Arethere biochemical changes or other relationship supports that counterbal-anced self-esteem and autonomy issues when the attachment figure wasinadequate? Are there specific child care issues that occurred at key pointsthat seem to buffer the amount of aggression associated with attachmentneeds? Such future work is needed to elaborate on the preliminary findingsof this study. In doing so, social scientists could provide a much neededtheoretical explanation of the etiology of male violence toward a femaleintimate.

REFERENCES

Ainsworth, M. D. S. (1989). Attachments beyond infancy. Am. Psychologist 44: 706-716.Ainsworth, M. D. S., Blehar, M. C , Waters, E., and Wall, S. (1978). Patterns of Attachment:

A Psychological Study of the Strange Situation. Erlbaum, Hillsdale, NJ.Allen, C , and Straus, M. (1980). Resources, power, and husband-wife violence. In Straus,

M., and Hotaling, G. (eds.). The Social Causes of Husband-Wife Violence, University ofMinnesota Press, Minneapolis.

Ball, R. E., and Robbins, L. (1986). Marital status and life satisfaction among Black Americans/ Marr. Fam. 48: 389-394.

Bandura, A. (1973). Aggression: A Social Leaming Analysis, Prentice-Hall, Englewood Cliffs,NJ.

Attachment Theory 227

Bowlby, J. (1969). Attachment and Loss. Basic Books, New York.Bowlby, J. (1973). Separation, Basic Books, New York.Bowlby, J. (1988). A Secure Base, Basic Books, New York.Bretherton I., and Waters, E. (1985). Growing points of attachment theory and research.

Monogr. Soc. Res. Child DeveL 50 (1-2, Serial No. 209).Bretherton, I. (1992). The origins of attachment theory: John Bowlby and Mary Ainsworth.

DeveL PsychoL 28(5): 759-775.Bretherton, I. (1987). New perspectives on attachment relations: security, communication, and

internal working models. In Osofsky, J. (ed.). Handbook of Infant Development, JohnWiley and Sons, New York.

Buss, A. (1971). Aggression pays. In Singer, J. (ed.). The Control of Aggression and Violence,Academic Press, New York.

Campbell, J. (1992). A review of nursing research on battering. In Carolyn, P. R., andSampselle, M. (ed.). Violence Against Women: Nursing Research, Education, and PracticeIssues, Hemisphere Publishing Corporation, New York.

Check, J. (1988). Hostility toward women: Some theoretical considerations. In Russell, G.(ed.), Violence in Intimate Relationships, PMA Publishing Corp, New York.

Cohen J., and Cohen, P. (1983). Applied Multiple Regression-Correlation Analysis for theBehavioral Sciences, Lawrence Erlbaum Associates, Hillsdale, NJ.

Collins, N. L., and Read, S. J. (1990). Adult attachment, working models, and relationshipquality in dating couples. /. PersonaL Social PsychoL 58: 644-663.

Dollard, J., Doob, L., Miller, N., Mowrer, O., and Sears, R. (1939). Frustration and Aggression,Yale tJniversity Press, New Haven, CT .

Erickson, E. F, Sroufe, L. A., and Egeland, B. (1985). The relationship between quality ofattachment and behavior problems in preschool in a high risk sample. Monogx. Soc. Res.Child DeveL 50: 209.

Fagan, R., Bamett, O., and Patton, J. (1988). Reasons for alcohol use in marital violent men.Am. J Drug Alcohol Abuse 14: 371-392.

Farrington, K. (1986). The application of stress theory to the study of family violence:Principles, problems, and prospects. /. Fam. VioL 1: 131-147.

Franklin, C. W. (1984). Black male-female conflict: Individually caused and culturally nurtured./. Black Studies 15: 139-154.

Gelles, R. (1987). Family Violence, Sage Publications, Newbuiy Park.Gelles, R. J., and Comell, C. P (1990). Intimate Violence in Families, Sage, Newbuiy Park.George, C , Kaplan, N., and Main, M. (1987). The adult attachment interview, LJnpublished

manuscript. University of California, Berkeley.Giles-Sims, J. (1983). Wife Battering: A Systems Theory Approach, Guilford, New York.Hadiey, S. (1992). Working with battered women in the emergency department: A model

program. / Emerg. Nurs. 18: 18-23.Hazan, C , and Shaver, P. (1987). Romantic love conceptualized as an attachment process. J.

PersonaL Social PsychoL 52: 511-524.Hotaling, G., Finkelhor, D., Kirkpatrick, J., and Straus, M. (1989). Family Abuse and Its

Consequences, Sage, Newbury Park, CA.Howell, M. J., and Pugliesi, K. L. (1988). Husbands who harm: Predicting spousal violence

by men. / Fam. VwL 3: 151-168.Julian, T, and McKenry, P (1989). Relationship of testosterone to men's family functioning

at mid-life: A research note. Aggress. Behav. 15: 281-289.Julian, T, and McKeniy, P (1993). Mediators of male violence toward female intimates. /.

Fam. VwL 8: 39-56.Lutkenhaus, P, Grossman, K. E., and Grossman, K (1985). Infant-mother attachment at

twelve months and style of interaction with a stranger at the age of three years. ChildDeveL 56: 1538-1542.

Lyons-Ruth, K., Connell, D. B., Grunebaum, H. U., and Botein, S. (1990). Infants as socialrisk: matemal depression and family support services as mediators of infant developmentand security of attachment. Child DeveL 61: 85-98.

228 Kesner et al.

MacEwen, K., and Barling, J. (1988). Multiple stressors, violence in the family of origin, andmarital aggression: A longitudinal investigation. /. Fam. VioL 3: 73- 86.

Main, M., Kaplan, N., and Cassidy, J. (1985). Security in infancy, childhood, and adulthood:A move to the level of representation. Monogr. Soc. Res. Child DeveL 50: 66-104.

Margolin, G., Sibner, L., and Gleberman, L. (1988). Wife battering. In Van Hassett, V,Morrison, R., Bellack, A., and Herson, M. (eds.). Handbook of Family Violence, PlenumPress, New York.

Mayseless, O. (1991). Adult attachment patterns and courtship violence. Fam. Relat. 40: 21-28.Neidig, P, Friedman, D., and Collins, B. (1986). Attitudinal characteristics of males who have

engaged in spouse abuse. /. Fam. VioL 1: 223-233.Price, S. J., McKeniy, P C. (1988). Divorce: A Major Family Thansition, Sage, Newbuiy Park,

CA.Rosenberg (1965). Society and the Adolescent Self-image, Princeton University Press, Princeton,

NJ.Sarason, I., Johnson, J., and Siegel, J. (1978). Assessing the impact of life changes:

Development of the life experiences survey. / Consult. Clin. PsychoL 64: 932-946.SAS Institute, Inc. (1982). SAS User's Guide, SAS Institute, Inc, Caiy, NC.Schaefer, E. S., and Edgerton, M. (1982). Autonomy and relatedness inventory {ARI),

Unpublished manuscript. School of Public Health, University of North Carolina-ClhapelHill.

Shaver, P., and Brennan, K. (1992). Attachment styles and the "big five" personality traits:Their connection with each other and with romantic relationship outcomes. PersonaLSocial PsychoL BulL 18: 536-545.

Sroufe, L. A., and Waters, E. (1977). Attachment as an organizational construct. Child DeveL49: 1184-1199.

Staples, R. (1988). The Black Am. Family. In Mindel, C. H., Habenstein, R. W, and Wright,R. (eds.). Ethnic Families in America, Elsevier, Amsterdam, The Netherlands.

Star, B. (1980). Patterns of Family Violence. Social Casework 61: 261-263.Stark, E., and Flitcraft, A. (1985). Souse abuse. In Surgeon General Violence and Public Health

Sourcebook, Center for Disease Control, U.S. Public Health Service, Atlanta.Stark, R., and McEvoy, I. (1970). Middle-class violence. PsychoL Today 4: 52-54.Steinmetz, S. K. (1987). Family violence: Past present and future. In Sussman, M. B., and

Steinmetz, S. K. (eds.). Handbook of Marriage and the Family, Plenum Press, New York,pp. 725-765.

Straus, M. A. (1980). Social stress and marital violence in a national sample of American families.Symposium on Forensic Psychology and Psychiatry, New York Academy of Sciences.Xnnals of the New York Academy of Sciences.

Straus, M. A. (1979). Measuring intrafamily conflict and violence: the conflict tactics (CT)scale. / Mam Fam. 41: 75-88.

Straus, M., and Gelles, R. (1986). Societal change and change in family violence from 1975to 1985 as revealed by two national surveys. /. Marr. Fam. 48: 785-793.

U.S. Bureau of the Census (1991). Statistical Abstract of the United States: 1991, U.S.Government Printing Office, A\^shington, DC.

Void, G. B. (1986). Theoretical Criminology, Oxford University, New York.Waters, E., Wippman, J., and Sroufe, L.A. (1979). Attachment, positive affect and competence

in the peer group: TWo studies in construct validation. Child DeveL 50: 821-829.Weiss, (1982). Attachment in adult life. In Parkes, C. M., and Hinde, J. (eds.). The Place of

Attachment in Human Behavior, Basic Books, New York.Weiss, R. S. (1991). The attachment bond in childhood and adulthood. In Parkes, C. M.,

Stevenson-Hinde, J., and Marris, P (eds.), Attachment Across the Life Cycle, Routledge,London, pp. 42-62.

Wodarski, J. (1987). An examination of spouse abuse: Practice issues for the profession. Clin.Social Work J, 172-187.