j interpers violence-2003-graham-kevan-1247-70.pdf

8/19/2019 J Interpers Violence-2003-Graham-Kevan-1247-70.pdf

http://slidepdf.com/reader/full/j-interpers-violence-2003-graham-kevan-1247-70pdf 1/24

10.1177/0886260503256656ARTICLEJOURNALOFINTERPERSONAL VIOLENCE / November 2003Graham-Kevan, Archer / INTIMATE TERRORISM ANDCOUPLE VIOLENCE

Intimate Terrorism and Common Couple Violence

A Test of Johnson’s Predictionsin Four British Samples

NICOLA GRAHAM-KEVANJOHN ARCHER

University of Central Lancashire

This study sought to both replicate and considerably extend the findings of Johnson(1999) that there are two distinct subgroups of physical aggression within relation-ships: intimate terrorismand common couple violence.The presentsample consisted of women residing at Women’s Aid shelters and their partners ( N = 86), male and

female students ( N = 208), men attendingmale treatmentprograms for domestic vio-lence and their partners ( N = 8), and male prisoners and their partners ( N = 192).

Respondents completed measureson physical aggression, injuries sustained, escala-tion of physical aggression, and controlling behaviors. Cluster analysis wasemployed to categorize relationships as either intimate terrorismor common coupleviolence. Frequency analysis showed broad support for Johnson’s findings.

Keywords: relationships; physical aggression; injuries; controlling behaviors;intimate terrorism and common couple violence; sampling

Findings from the Conflict Tactics Scale (CTS) (Straus, 1979) have beencontroversial from the outset. The CTS was first used in the National FamilyViolence Surveys (NFVS) in 1975 and 1985 and, to the surprise of many,indicatedsex symmetry in theuseof physical aggression by menandwomenagainst their partners. These findings appeared to directly contradict theresearch and experience of feminist researchers who told of severe maleaggression toward women within the home. Since then, a large and ever-increasing body of literature has replicated the NFVS findings. Reactions tosuch findingshave been diverse. A complete acceptance of sex symmetry ledto the extrapolation to battered husbands (Steinmetz, 1978) and the call for

1247

Authors’Note: To request reprints, contact Nicola Graham-Kevan, Department of Psychol-ogy, University of Central Lancashire, UK; e-mail: [email protected].

JOURNAL OF INTERPERSONAL VIOLENCE, Vol. 18 No. 11, November 2003 1247-1270DOI: 10.1177/0886260503256656© 2003 Sage Publications

at University of York on March 11, 2016 jiv.sagepub.comDownloaded from

http://jiv.sagepub.com/






more resourcesto be directed to male victims of partner abuse. Many rejectedthese findings, citingboth methodological reasons and the failure of incidentdata to represent the consequences of partner aggression as catalogued inpolice files, court and hospital records, and shelter samples (e.g., Dobash,Dobash, Cavanagh, & Lewis, 1998; Dobash, Dobash, Wilson, & Daly, 1992;Paglow, 1992; Saunders, 1988). Straus has gone to some lengths to addressthese criticisms both theoretically (see, e.g., Straus, 1990) and empirically(seeStraus, 1999;Straus, Hamby, Boney-McCoy, & Sugarman,1996). How-ever the controversy remains.

Archer (2000) published a meta-analytic review of the literature in whichdata were available for both partners. In this, he found that where act-basedmeasures—usually the CTS (Straus, 1979)—were used, there was little dif-ference between males and females in their use of physical aggression in

community and dating samples. However, when samples derived fromwomen in shelters andmenidentified as using high levels of physical aggres-sion here examined, there wasa large effect size in themaledirection. What isparticularly relevant to the present study is that the type of sample was foundto be a significant influence on the magnitude and direction of the effectsizes. Some samples contain only individuals who have been involved inrelationship aggression: Membership is contingent on such experiences,shelter samples being an example. Other sample populations are not com-prised of individuals knownto have been involved in relationship aggression:for example, students. Samples not selected for relationship aggressionappeared more sexually symmetric than samples that were selected for theoccurrence of relationshipaggression. What is apparent from this is that evenwhen using the same instrument, different sampling strategies can lead toconflicting findings.

Johnson (1995; Johnson & Ferraro, 2000) attempted to reconcile the dis-crepant findings of feminist and family violence researchers. He argued thatthere were two distinct types of relationship in which physical aggressionplayed a part. The first was termed “common couple violence” (CCV), inwhich aggression is not“connected to a general pattern of control. It arises inthe context of a specific argument in which one or both of the partners lashout physically” (Johnson & Ferraro, 2000, p. 949). CCV, Johnson proposed,was accessible through large-scale surveys such as those used by family vio-lence researchers. The second type of relationship was termed “patriarchalterrorism” (Johnson, 1995), but this was subsequently changed to the moregender-neutral term of “intimate terrorism” (IT). It was “one of violence asmerely one tactic in a more general pattern of control. The violence is moti-vated by a wish to exert general control over one’s partner” (Johnson &Ferraro, 2000, p. 949). IT was believed not to be accessible through large

1248 JOURNAL OF INTERPERSONAL VIOLENCE / November 2003








population surveys but through selected samples such as women’s shelters,policeand court records, andER admissions. Feminist researchers more typ-ically access these types of samples. He proposed, using research evidence,that the two types of relationship violence could be distinguished from oneanother and concerned nearly nonoverlapping phenomena. The dimensionson which thetwo could be distinguished included theuse of violence withinadyadic relationship context, controlling behaviors, motivation to use vio-lence against one’s spouse, and the likelihood of violence escalating.

Johnson (1999) presented analyses of data collected by Frieze in the1970s. Using cluster analysis, Johnson categorized relationships involvingphysical aggression as CCV (55% male, 45% female), IT (97% male, 3%female), violent resistance (VR) (4% male, 96% female), or mutual violentcontrol (MVC) (50% male, 50%female). These types of relationshipaggres-

sion were then identified as belonging to eithera general survey sample(90%CCV)or a shelter sample (74% IT/VR). Johnson then compared IT andCCVsamples on measures of escalation of violence, severity of male violence (asindexed by injuries sustained by femalepartners), mutualityof violence, andfrequency of violence. He found that relationships labeled IT were morelikely to have involved escalated levels of aggression, involved more injuri-ous aggression, and to be typified by disproportionate levels of aggressionbetween partners than relationships classified as CCV. Johnson didnot, how-ever, find that victims of IT were any less likely to aggress themselves thanwere partners in CCV relationships.

The data set necessary for distinguishing between IT and CCV mustinclude notonly rates of physical aggression forboth self andpartner butalsoinformation on the use of different types of controlling behaviors. The cur-rent data, unlike the earlier data set used in Johnson’s (1999) analysis, wascollected specifically to discriminate between IT and CCV and hence is ide-ally suited for the following analysis, which both replicates, and whereappropriate extends, the analysisperformedby Johnson. Physical aggressionis measured by the CTS, an instrument that has been used with a diverserange of samples (e.g., Giles-Sims, 1983; Laner, 1985) and represents animprovement over the single-item measure used by Johnson (1999). The useof the CTS allows minor and severe physical aggression to be investigatedindividually. Although thisdistinction is not empirically based, it does corre-spond to a legal distinction between simple or common assault and aggra-vated or grievous assault (Straus, 1990). The plastic interval data derivedfrom it lends itself to more sensitive analysis than the nominal physicalaggression measure that Johnson had available to him. Severity of physicalaggression is assessed by two items that relate to the frequency of injuriessustained in all conflicts during the last year, rather than Johnson’s item for

Graham-Kevan, Archer / INTIMATE TERRORISM AND COUPLE VIOLENCE 1249








severity of physical aggression, which relates only to the “most violent” epi-sode. The most violent episode is likely to be atypical because it is identifiedas standing outfromthemore usual levelof injuriousaggression. Controllingbehaviors were measured using a scale devised by the author using literaturefrom the Domestic Violence Intervention Project (DVIP) (Pence & Paymar,1993), with all behaviors being classified using DVIP protocol.

The useof self-reportsof aggression in conjunction with partner-reportedrates may lead to bias. Studies have found that male batterers and their part-ners agree on the frequency of aggression that women perpetrate but differsignificantlyin reports of female victimization,with womenreportinghigherrates than men (Barnett, Lee, & Thelen, 1997; Claes & Rosenthal, 1990;Dobash et al., 1998; Okun, 1986). Whether this wasdue to underreporting bymenor inflationby women cannotbe ascertained,although mendescribedas

batterers have been found to show signs of socially desirable responding,unlike their partners (Dutton & Hemphill,1992). For communityand studentsamples, research has found thatself-reports are consistently lower thanpart-ner reports for both sexes (Archer, 1999; Follingstad, Wright, Lloyd, &Sebastian, 1991; Makepeace, 1986; Moffit, Caspi, Kneger, Magdol, &Margolin, 1997). It is thereforeprudent when investigating sexdifferences toinvestigate possible self-/partner report bias within participant populations.

The present study used a population from England. There are no pub-lished studies that have obtained self- and partner data on the use of physicalaggression within a British shelter sample—indeed, there are very few fromelsewhere. Archer (2000) only found two small-scale studies (Giles-Sims,1983; Okun, 1986) of shelter populations in which CTS-type measures wereavailable. Data from these samples is therefore important to allow informeddiscussion of relationship aggression. To the authors’ knowledge, there areno publishedsources that haveused quantitative methods to assessthe perpe-tration and victimization of controlling behaviors and injuries in shelterpopulations.

The analysis was a replication of that performed by Johnson (1999), withadditionalanalysis performed where appropriate (e.g., univariate analysis of sex by relationship type on rates of aggression). To test the following predic-tions, each respondent and therespondent’s partner needed to be classified asusing either (a) no violence (NV), (b) noncontrolling aggression (NCA), or(c) controlling aggression (CA). The relationship could then be classified aseither (a) nonviolent, in which neither spouse uses violence; (b) individualsinvolved in CCV (NCA used by one or both spouses); (c) individualsinvolved in IT (individuals using CA and their partners using either NV orNCA); (d) VR (individuals who use NCA only but whose spouse uses CA);or (e)MVC (both spouses useCA). Classificationwas based thefrequency of









use of controlling behaviors and whether any acts of physical aggression hadbeen used. The initial analysis identified the above relationship characteris-tics, and subsequent analysis involved only relationships in which violenceplayed a part (therefore all “nonviolent” relationships were omitted).

Johnson (1999) made the following predictions that are testable using theanalytic procedures he outlined:

1. Partner violence occurs in both high and low control contexts.2. IT is primarily male, and in a heterosexual context, it follows that VR will be

primarily female. CCV is sex symmetric.3. IT will result in higher levels (more frequent) of physical aggression than

CCV.4. IT is more likely to escalate than is CCV.5. IT will result in more injuries than CCV.

6. Targets of IT are less likely to be violent than are targets of CCV.7. IT appears almostexclusivelyin shelter samples andCCV almostexclusivelyin survey samples.

8. As a resultof thepatterns predicted, domesticviolence appears to be sexuallysymmetric in the survey samples and exclusively male in shelter samples.

METHOD

Participants

Participants were drawn from a female shelter sample consisting of 43female participantswho were all current residents in Women’s Aid domesticviolence shelters, a mixed-sex student group ( n = 104), 4 men who wereattending a male treatment program (MDVTP) for domestic violence, and amale prisoner sample consisting of 97 inmates from prisons in the north of England (for fuller description of samples, see Archer & Graham-Kevan,2003). All participants reported on their own and their partners’ behaviors.

The sample groups were originally selected to identify IT and CCV. Vic-tims of IT were expected to be detected in the shelter sample, and it washoped that perpetrators could be detected in the male prison sample. Only afraction of theprison group had been cautioned, charged and/or convictedof an assault to a partner: Therefore, the male prisoner sample was included asanother population not selected for the occurrence of aggression. Due to theinability to access perpetrators of “domestic violence” through prisons,MDVTPs were contacted and questionnaires were distributed to their pro-gram users. The response, however, was poor in the MDVTP sample andthereforeno conclusions canbe made regarding this group. A student samplewas accessed, as previous research using students has found sexual symme-









try in reports of partner aggression and therefore this cohort was thoughtlikely to manifest CCV. The age of the participants ranged from 16 to 65years with a mean age of 33 years ( SD = 11.31). The partners of the partici-pants had an age range of 16 to 69 years, with a mean of 33 years ( SD =10.75). The lengthof relationships rangedfrom 1 to 504months, with a meanof 104 months ( SD = 109). A series of one-way ANOVA revealed no signifi-cant difference between the sample groups for the above characteristics.

Johnson (1999) used an “artificially constructed” (p. 7) sample in whicheach respondent supplies information on both his or her own and his or herpartner’s behaviors. Thereports of both members of thedyad areentered intothe data set as separate cases. This technique was thus employed in the pres-ent study, producing the following sample sizes: shelter women and theirpartners ( n = 86), men in MDVTP and their partners ( n = 8), male prisoners

and their partners ( n = 194); and students ( n = 206). Itshould be kept in mind,therefore, that allshelter data are from reports given by women andallprisonand MDVTP data from men. The student sample was mixed-sex: Hence,male and female reports could be either self-reported or partner-reportedbehaviors.

Measures

The use of controlling behaviors was measured using a 24-item behav-ioral scale, Controlling Behaviors Scale (CBS). The CBS was developed bythe author, using literature from the Domestic Abuse Intervention Project(DAIP) (Pence & Paymar, 1993). The DAIP literature cites examplesof con-trolling behaviors consistently reported (by both victims and perpetrators) asbeing used by violent men against their partners. The CBS uses behavioralcategories and does not include any items involving actual physical aggres-sion. The CBS can be scored to derive an overall controlling behaviors total,or five subscores, each of which is a particular type of control tactic. Thesubscales are Using Economic Abuse, Using Coercion and Threats, UsingIntimidation, Using Emotional Abuse, and Using Isolation (see Appendix Afor items and alpha coefficients).

Physical Aggression

Frequency of acts of physical aggression . A modified version of Straus’sCTS (1979) was used to measure the occurrence and frequency of physical









aggression. Although this version of the CTS contains fewer acts of physicalaggression than the CTS-2 (Straus et al., 1996), it is the most widely usedphysical aggression measure (Archer, 2000) and has been found to be sen-sitive enough to discriminate between different samples (Archer, 2000;Graham-Kevan & Archer, 2003). Only the 17-item Aggression subscale wasused and of that, only the last 8 items—those involving acts of physicalaggression—were analyzed. The items and alpha coefficients are shown inAppendix B. These 8 items yield the following variables: Items k, l, m wereclassified by Straus as minor acts of physical aggression and Items n, o, p, q,and r classified as acts of severe physical aggression. When scoring theques-tions, the frequencies foreach item were added together to obtaina total CTSscore. The frequencies for Items 1 to 3 were added together to obtain a minorCTS score. The frequencies for Items 4 to 8 were added together to obtain a

severe CTS score. These scores were calculated for both the respondent andthe respondent’s partner. This gave the following values: minor CTS self score, minor CTS partner score, severe CTS self score, and severe CTS part-ner score.

Level of escalation . Level of escalation was assessed by an item at the endof thephysicalaggression andinjury items.One item asked,“During thetimeyou and your partner have been/were together, has the use of physical forceincreased, stayed the same, or decreased?” Responses were coded as 1(increased ), 2 ( stayed the same ), or 3 ( decreased ) for both self and partner.

Severity of violence . Severity of violence was assessed by 2 items takenfrom Morse (1995): Each had a 5-point response format ranging from 0(never ) to 4 ( always ). The items were introduced by the following sentence:“Regarding the past year with your partner, or the last year you were withyour partner, please answer the following questions: 1. How many timeswere you (your spouse/partner) physically injured (e.g., knocked down,bruised, scratched, cut, choked, bones broken, eyes or teeth injured?); 2. Inhowmany of these fights in which you(your spouse/partner)were physicallyinjured did you (he/she) go to the doctor, clinic, or hospital for medical treat-ment?”Two variables were coded:actualphysical injury andsevere physicalinjury. Thesecorrespond to the followingcategories usedby Johnson (1999):(a)no physical injury; (b)actualphysical injury; (c) severe injury, no trauma;and (d) severe injury with trauma.









RESULTS

Cluster Analysis of Controlling Behaviors Profiles

A K-means cluster analysis was conducted using SPSS Version 7 to codeindividuals as either high or low on the five types of controlling behaviors:economic, threats, intimidation, emotional abuse, and isolation. Self- andpartner reports were treated as separate cases. A two-cluster solution wasselected, using Euclidean distance as a measure of dissimilarity. The clustermembership was saved as a variable. The meaning of the two clusters isapparent by contrasting the value for the cluster centers for both clusters oneach of the five types of controlling behaviors (see Table 1). In accordancewith Johnson (1999), the two clusters have been termed “high” and “low”control. High controllers useeconomiccontrol more than 3.5 times more fre-quently than low controllers, threats more than 5 times more, intimidationnearly 6 times more, emotional nearly 5 times more, and isolation nearly 5times more.

Two Forms of Physical Aggression:Controlling and Noncontrolling Physical Aggression

To investigate the first prediction, that partner physical aggression occursin both high- and low-control clusters, the frequencies of controlling andnoncontrolling physical aggression were calculated. For this purpose, physi-calaggression wastreated as a discrete variable, with those whohadused anyact of aggression being classed as physical aggressive. The frequencies inTable 2 show that there are individuals who have used physical aggressionagainst their partners in both the high- and low-control clusters. The propor-tions are similar to those found by Johnson (1999). Overall, there were 239


TABLE 1: Control Tactics by Cluster

Economic EmotionalControl Threats Intimidation Abuse Isolat ion

High control ( n = 69)Mean ( SD) 2.41 (1.11) 1.67 (0.90) 2.42 (0.91) 2.81 (0.95) 3.16 (0.93)

Low control ( n = 405)Mean ( SD) 0.65 (0.66) 0.73 (0.99) 0.40 (0.44) 0.58 (0.61) 0.74 (0.70)

NOTE:Reportsonbothmenand womenfromonepartner( N = 492). Typesof controlare derivedfrom subscales of the Controlling Behaviors Scale.








physically aggressive individuals in the present sample; of these, 27% werehigh controllers (Johnson reported 32%).

The partner reports were likewise coded. Therefore, each party in a rela-

tionship was coded as being NV, using NCA, or using CA. If neither partyused any physical aggression, the relationship was called NV. Dyads inwhich only NCA was used (by one or both partners) were labeled CCV.Dyads in which the respondent used NCA and their partner used CA werelabeled VR. Dyads in which the respondent used CA and his or her partnerused no physical aggression or NCA were labeled IT. Dyads in which boththe respondent and their partner used CA were called MVC.

Thefrequencyof each type of relationshipwas found to be as follows:NV49%( n = 253), IT11% ( n =53),VR6%( n =30),MVC3%( n = 16), and CCV28% ( n = 140).

Sex and Physical Aggression

The second prediction relates to sexual symmetry/asymmetry. It was pre-dicted that IT would be perpetrated primarily by men and VR by women andthat CCV would be sexually symmetric. Table 3 presents the frequencies of each type of relationship by thesex of the individuals. IT is, as predicted, pri-marily male (87%). Although this supports the second prediction, it is not asnear to being absolute as Johnson (1999) reported (96%). VR is clearlyfemale (90%) in the present sample but is still slightly lower than thatreported by Johnson (96%). CCV is almost sexually symmetric (45% malesand 55% females), which supports the prediction.

Characteristicsof IT and CCV for both males and females . To investigatethe remaining predictions, the subsequent analyses centered on perpetrationof IT and CCV only. The third prediction relates to frequency of physicalaggression within the two relationship categories.


TABLE 2: Physical Aggression by Control Type

Physically Aggressive Not Physically Aggressive N

High control 95% (69) 5% (4) 100% (73)Low control 41% (170) 59% (249) 100% (419)

N 253 239 492

NOTE: Data on both self- and partner reports ( N = 492). Participants were classified as high orlow control based on cluster membership. Participants who indicated that they had used anyphysical aggression during the last 12 months were classified as physically aggressive.








Perpetration of minor acts of physical aggression . A Sex × RelationshipType ANOVA was conducted on the perpetration of minor acts of physicalaggression. A significant main effect was found for relationship type, F (1,189) = 56.79, p < .0005, withIT ( m = 8.5) perpetrating significantly moreactsof minor physical aggression than CCV ( m = 3.2) individuals. There was asignificant maineffect foundfor sex, F (1, 189) = 6.67, p < .05,with men ( m =6.0) perpetrating significantly more acts of minor physical aggression thanwomen ( m = 4.6). There was a significant interaction between relationshiptype and sex, F (1, 189) = 8.13, p < .01. Simple effects analysis found that ITmales ( m = 8.9) used significantly more minor physical aggression thanIT females ( m = 5.6), t = 2.25, df = 51, p < .03.

Perpetration of severe acts of physical aggression . A Sex × RelationshipType ANOVA was conducted on the perpetration of severe acts of physicalaggression. A significant main effect was found for type of relationship, F (1,189) = 49.48, p < .0005, with IT ( m = 10.2) perpetrating significantly moreacts of severe physical aggression than CCV ( m = 2.0) individuals. A signifi-cant main effect was found for sex, F (1, 189)= 6.10, p < .05, with men ( m =6.3) perpetrating significantly more acts of severe physical aggression thanwomen ( m = 4.1). There was a significant interaction between relationshiptype and sex, F (1, 189) = 7.84, p < .01. Simple effects analysis found that ITmen ( m = 10.8)perpetrated significantly moresevere actsof physical aggres-sion than CCV men ( m = 1.9), t = 8.74, df = 57.7, p < .01.

To control for the possibility that sex differences were driven by samplebias (through self-reports and partner reports), a further analysis was con-ducted. By using only true self-reports and treating self-reports and partner


TABLE 3: Individual Aggressive Behavior in a Dyadic Context Classified Accordingto Relationship Category

Husbands (%) Wives (%) N

Total physical aggression 50 50 239Mutual violent control 50 50 16Intimate terrorism 87 13 53Violent resistance 10 90 30Common couple violence 45 55 140

NOTE: Physically aggressive individualsonly as reportedby either husbandsor wives ( N = 239).Mutual violent control= bothpartners using controlling aggression; intimate terrorism= partici-pantuses controllingaggression; violent resistance= participant uses noncontrolling aggressionand partner uses controlling aggression; common couple violence = one or both partners usesnoncontrolling aggression.








reports as a within-subjects factor, it is possible to test whether males andfemales differin theextent to which they reportself-and partner behaviors. Asignificant interaction would indicate that they do. Neither Sex by (Report)ANOVA for minor physical aggression, F (1, 66) = 2.25, p > .05, nor forsevere physical aggression, F (1, 66) = 0.22, p > .05, proved significant.

Partner perpetration of minor acts of physical aggression . A Sex × Rela-tionship Type ANOVA of partner’s use of minor aggression revealed therewas no significant main effects for relationship type, F (1, 189) = 3.59, p >.05, or for sex, F (1, 189) = 0.69, p > .05, and no significant interactionbetween relationship type and sex, F (1, 189) = 0.40, p > .05.

Partner perpetration of severe acts of physical aggression . A Sex × Rela-tionship Type ANOVA of partner’s use of severe physical aggressionrevealed no significant main effect of relationship type, F (1, 189) =3.58, p >.05, orof sex, F (1, 189)= 2.91, p > .05, andno significant interactionbetweenrelationship type and sex, F (1, 189) = 0.32, p > .05.

The implication of the above analysis is that there are significant differ-ences in theratesof physical aggression between IT andCCVindividualsbutnot in theirpartners’behaviors. ITsdo perpetrate more acts of both minor andsevere acts of physical aggression than CCV individuals, and whereas CCVrespondents showed sexual symmetry in the perpetration of physical aggres-sion, ITs did not. IT men perpetrated significantly more acts of minor physi-cal aggression than IT women, and IT men perpetrated significantly moreacts of severe physical aggression than CCV men.

Escalation of physical aggression . The fourth prediction stated that ITwould be more likely to escalate than CCV. To investigate the relationshipbetween escalation and sample group, frequencies were calculated. Table 4shows that ITs were more likely to escalate their use of physical aggressionthan CCV perpetrators, χ 2 = 32.54, df = 2, p < .0005.

Injuries inflicted on partners . Severity of physical aggression wasindexed by examining injuries to partners. The frequencies for injuries topartners by relationship type are presented in Table 5. The two groups dis-played discordant rates for injuries not requiring medical attention, χ 2 =70.14, df = 4, p < .0005, and injuries that required medical attention, χ 2 =28.61, df = 4, p < .0005. In both cases, victims of IT were significantly more

likely to sustain these than were victims of CCV.









Reciprocity of physical aggression . To investigate the reciprocity of phys-ical aggression frequencies of uni- and bidirectional 1 physical aggression bysexof perpetrator werecalculated (seeTable6). Chi-squareanalysis revealedthat targets of IT were significantly less likelyto usephysical aggression thanwere targets of CCV, χ 2 = 5.663, df = 1, p < .05.

Relative rates of perpetrator and partner’s use of physical aggression . Toinvestigateperpetrator and victim use of physical aggression further, the rela-tive couple frequency of use of physical aggression, calculated by subtract -ing the self-reported physical aggression score from the reported partnerphysical aggression for each dyad were calculated (see Table 7). An unre-


TABLE 4: Level of Escalation by Type of Relationship Category

De-escalated No Change Escalated N

Intimate terrorism 11% (4) 11% (4) 78% (28) 100% (36)Common couple violence 48% (33) 32% (22) 20% (14) 100% (69)

N 37 26 42 105

NOTE: Perpetrators of intimateterrorism and common couple violenceonly ( N = 105).The itemon escalationwas notincludedfor allparticipants.Participants wereasked if their aggression hadreduced (de-escalated), stayed the same (nochange), or increased (escalated) over the course of their relationship. Intimate terrorism = participant uses controlling aggression; commoncoupleviolence = one or both partners uses noncontrolling aggression.

TABLE 5: Severity of Violence by Relationship Category Measured by Injuries toPartner

No Injury Not No Injury Injury Minor Requiring Requiring RequiringPhysical Medical Medical Medical Injury Attention N Attention Attention N

Intimate terrorism 25% 76% 100% 57% 43% 100%(13) (40) (53) (30) (23) (53)

Common couple violence 64% 36% 100% 87% 13% 100%(90) (50) (140) (122) (18) (140)

N 103 90 193 152 41 193

NOTE: Perpetrators of intimate terrorismand common couple violence only( N = 193). Thefirstthree columns of percentages relate to the item measuring minor injury. The second three col-umns of data relate to injury requiring medical attention. No physical injury = never injured;

injury not requiring medical attention = injured without the need for medical treatment; injuryrequiring medical attention = if partner saw doctor for injuries. Intimate terrorism = participantuses controllingaggression;commoncouple violence= oneor bothpartners usesnoncontrollingaggression.








lated t test revealed that IT ( m = 11.8) used significantly more physicalaggression relative to their partners than CCV did ( m = 0.9), t = –9.5, df = 59,

p < .0005.

Sampling Strategies and the Sex Asymmetry Debate

The final analysis centers on the distribution of the four types of relation-ship aggression in relation to sex and sample. The frequencies are presentedin Table 8. The seventh prediction stated that IT would appear almost exclu-sively in shelter samples and CCV almost exclusively in survey samples. Inthis study, 70% of all IT were found in the shelter sample, 13% were found inthe male prisoner sample, 17% were found in the student sample, and per-haps surprisingly, none were found in the male treatment program sample.When one looks at CCV only, 6% of these were found in the shelter sample,60% were found in the male prisoner sample, 29% were found in the student


TABLE 6: Mutuality of Violence by Relationship Category

Perpetrator Only Both N

Intimate terrorism 43% (23) 57% (30) 100% (53)Common couple violence 26% (36) 74% (104) 100% (140)

N 59 134 193

NOTE: Intimateterrorism and common couple violenceonly ( N = 193). Classified as perpetratoronly ifparticipanthadusedany physicalaggressionin the past year andhis orher partner hadnotused any andclassifiedas both ifparticipantandhis orher partner hadboth used physicalaggres-sion in the past year.

TABLE 7: Difference in Frequency of Violence by Relationship Category ( N = 193)

–23 –14 –4 5 15 21 26 to to to to to to to –15 –5 4 14 20 25+ 32 N

Intimate terrorism 2% 0% 13% 28% 21% 11 26% 100%(1) (0) (7) (15) (11) (6) (13) (53)

Common couple violence 0% 9% 78% 12% 1% 0% 0% 100%(0) (12) (109) (17) (2) (0) (0) (140)

N 1 12 116 32 13 6 13 193

NOTE: Score calculated by subtracting participant’s score on the Conflict Tactics Scale (CTS)fromhis or herpartner’s scoreon the CTS.A negative difference indicatesmore physicalaggres-sion perpetratedby the partner of the participant, and positive differences indicate more partici-pant physical aggression compared to his or her partner.








sample, and 6% were found in the male treatment program sample. There-fore, 94% of all CCV relationships were found in nonselected samples. Theseventh prediction is therefore supported, although the term “almost exclu-sively” is overstating the distribution of IT.

Theeighth prediction statedthatas a resultof patterns predictedin thesec-ond and seventh predictions, domestic violence appears to be sex symmetricin survey samples and exclusively male in shelter samples.

Using Johnson’s categories, Table 8 shows that relationship aggressionappears to be predominately sex symmetric in the student, male prison, andmale treatment samples and more, although not completely, sex asymmetricin the shelter sample. Johnson (1999) reported 99% of men and 80% of women in the shelter sample as perpetrating some physical aggressionagainst their partner. This study found similar frequencies for men(98%) butlower values for women (60%), the difference being statistically significant(see above). Support for prediction 8 is provided by the type of physicalaggression utilized in the samples. In all but the shelter sample, it is (usingJohnson’s categories) sex symmetric. Theprison samplehas80%CCVand afurther 8% MVC, the student sample has 76% CCV and 2% MVC, and themale treatmentprogram samplehas 80%CCV and20% MVC. This is in con-trast to theshelter sample, in which 86%of malesperpetrate IT butnone of itsfemales do so. The vast majority of shelter women’s physical aggression isVR (81%); no males in the shelter sample are reported to use VR. The occur -


TABLE 8: Percentages (and Numbers) of Participants Classified in Each RelationshipCategory by Sampling Strategy and Sex

Students Prisoners Shelter MDVTP

Type of Relationship Men Women Men Women Men Women MenWomen Total

Intimate terrorism 13 4 4 9 68 2 0 0 100(7) (2) (2) (5) (36) (1) (0) (0) (53)

Violent resistance 0 17 10 3 0 70 0 0 100(0) (5) (3) (1) (0) (21) (0) (0) (30)

Common couple 9 19 30 30 3 3 3 3 100violence (13) (27) (42) (42) (4) (4) (4) (4) (140)

Mutual violent control 6 6 31 31 6 6 6 6 100(1) (1) (5) (5) (1) (1) (1) (1) (16)

NOTE: Includes all violent relationships ( N = 239). All percentages relate to the percentage of violent relationships that involve each specific type of violent relationship. Mutual violent con-trol = both partners using controlling aggression; intimate terrorism = participant uses control-lingaggression; violent resistance= participant uses noncontrolling aggressionand partner usescontrolling aggression; common couple violence = one or both partners uses noncontrollingaggression.








rence of CCV in the shelter sample is low (9%) and is indeed considerablylower than Johnson (1999) found (24%). The same levels of MVC werereported in both studies (2%).

DISCUSSION

The main objective of the present study was to investigate the possibleexistence of subgroups within relationships reporting the occurrence of physical aggression. Johnson (1995) suggested that the existence of sub-groups may help to reconcile the apparently conflicting findings of feministandfamily violence perspective research by challengingthe beliefin a mono-lithic model of relationship aggression.

The crux of the argument put forward by Johnson (1995, 1999) was thattwo qualitatively different types of aggression exist. The first type is CA setwithin a general framework of relationship control. The motivation behindthe use of this type of physical aggression is to maintain overall control overone’s relationship. The second type of physical aggression is NCA, which incontrast to CA, represents a reaction to particular stressors. Therefore,according to Johnson, we would expect aggression to occur in both high andlow controllers. Physical aggression was found among both high and lowcontrollers. It was not distributed evenly, however. High controllers were farmore likely to also use physical aggression than were low controllers. Thisrelationship between control and aggression supports both feminist and evo-lutionary perspectives that conceptualize aggression as a coercion tactic(e.g., Walker, 1989; Wilson & Daly, 1993).

Physical aggression and control are interpersonal behaviors and theyshould be studied within the relationship dyad. Therefore, respondents andtheir partners were classified as using no physical aggression, NCA, or CA.Dyads in which only NCA was used (by one or both partners) were labeledCCV. Dyads in which the respondent used NCA and his or her partner usedCA were labeled VR. Dyads in which the respondent used CA and his or herpartner used no physical aggression or NCA were labeled IT. Dyads in whichboth the respondent and his or her partner used CA were called MVC.

IT wasfound to be, as expected, primarilymale. However, the“maleness”of intimate terrorism may well be an artifactof the sampling procedure used.Indeed, if the shelter data is omitted, IT shows sexual symmetry. VR isclearly female, again reflecting the shelter sample: Once this is removed, thefemaleto male ratio changes from 9:1 to 2:1. The influenceof theshelter dataon IT and VR analysis must be borne in mind when attempting to draw popu-lation-wide conclusions. Johnson’s analyses used either a sample containing









“battered women” (Johnson, 1999) or female respondents identified throughthe National Violence Against Women Survey (Johnson, 2000). Bothsources are more likely to contain female victims of severe violence thanmale victims. Therefore, although valuable in assessing the distribution of CCV and IT in femalepopulations, it is problematic to extrapolate to sex dif-ferences in mixed-sex populations.

CCV, as expected, is almost sexually symmetric. However, as CCV isdefined as one or both partners using NCA, classifying it as sexually sym-metric is circular. When one examines the proportion of males to femalesusing no violence,men arefound to be overrepresented by almost3:1. There-fore, although the majority of the relationships classified as CCV do involvemutual physical aggression, where this not the case women are over-represented as sole perpetrators. Morse (1995), Riggs (1993), and O’Leary,

Barling, Arias, and Rosenbaum (1989) all found that in one-sided assaults,women were more likely to be the sole perpetrator than men. This suggeststhat further distinctionsmayneed to be made in relation to CCV. Thedynam-ics of relationships in which only one person uses physical aggression, evenif it does not appear to be control-orientated, may well differ in importantways from truly bidirectional physical aggression. Future research that usesboth quantitative and qualitative methods is needed to shed light on theserelationships.

Johnson stated that IT was characterized by frequent, escalating violencethat is not reciprocal. This is in contrast to CCV, which was described asbeing less frequent, nonescalating, and more likely to be reciprocal. Thesepatterns were found in the present study, with IT individuals using signifi-cantly more acts of physical aggression than CCV individuals. Sex differ-ences in partners’ use of physical aggression within a dyadic context werealso investigated, the rationale being that that women perpetrating IT mightdiffer from their male counterparts in the level of physical aggression theyreceive from their partners. It would be reasonable to suppose that a manvic-timized by IT aggression would be less afraid and more able to adequatelydefend himself than a woman and hence be more likely to retaliate. Thus, heshould manifest higher levels of physical aggression than either female vic-tims of IT or CCV. Contrary to this expectation, thepartnersof IT women didnot differ from other partners.

Another way to examine sex differences in rates of physical aggression isby calculating a respondent’s physical aggression as a proportion of all (bothself and partner) physical aggression within IT relationships. The presentdata reveals that female ITs show proportions that are lower for minor acts(female = .79 and male = .84) but higher for severe acts (female = .98 andmale = .89) of physical aggression. This leads to theconclusion that although









systematic assaults.Fromthe relative frequencies of perpetrator/targetuse of physical aggression it is clear that the two groups display differing profiles,with IT reporting values more than 10 times higher than CCV individuals. Inthe present sample, 85% of the IT had been aggressive 5 times or more thantheir partners, a figure higher than Johnson’s (71%), whereas 13% of theCCV individualshad. Thevast majority of CCV(78%) showedapparent rec-iprocity, whereas only 13% of IT dyads did. These findings are in line withearlier research on shelter women (Giles-Sims, 1983; Okun, 1986) and gen-eral populations.

In summary, it appears that IT and CCV can be distinguished from oneanother along the dimensions highlighted by Johnson (1995, 1999). The factthat the attributes of IT may be due to its selected status does not detract fromone of thecentral tenetsof Johnson’s proposition, that of qualitatively differ-

ent typesof physicallyaggressive relationships.Thisextends earlier researchthat found evidence of subgroups of physically aggressive relationships butfailed to identify the dimensions on which they could be differentiated(Riggs, 1993; Vivian & Langhinrichsen-Rohling, 1994). In this study, con-trolwascentral to classifying physicallyaggressive relationships. Thosewhoused high levels of control presented a classic domestic violence profile.Controlling behaviors are therefore a crucial element in understanding thedynamics of relationships that may be likely to require intervention at manydifferent levels (e.g., law enforcement, child protection, medical, and judi-cial).Stark (1995) believedthat it wasthe pattern of coercivecontrol that wasmore central to understanding battered women than the actual physicalaggression. This pattern, at itsmostextreme,violatesa person’s human rightsthrough the restriction of liberty. Concentrating on physical aggressionaloneobscures the differences between relationships, leading to inaccurate report-ingof levelsof classic domesticviolence, inappropriate advicebeinggiven topeople in physically aggressive relationships, and a legal system that treatsdomestic assaults in the same way as stranger assaults by examining actswithout context.

The distribution of relationship types was also broadly in line with John-son’s (1999) predictions. The selected sample was predominantly comprisedof male IT and women using VR, a profile consistent with feminist researchfindings. The nonselected samples mainly consisted of CCV. The presentfindings suggest that data drawn using the same instruments and methodolo-gies but from different populations would appear contradictory. There areseveral implications arising from these findings. Generalizations should notbe made from shelter populations to general populations or vice versa, asthese samples are likely to contain qualitatively different types of aggressiverelationship. TheCTS appears to be an adequate measure of physical aggres -









sion for both selected and nonselected samples. In seeking support for a par-ticular model of relationship aggression, both feminist and family violenceresearchers may have inadvertently created an adversarial arena in whichsubsequent research is judged. Such a climate isnotconduciveto open debateand enquiry.

The present studyprovidesbroadsupport for Johnson’s (1999) contentionthat there aretwo main types of physical aggression accessed by feminist andfamily violence researchers. However, further studies are necessary beforegeneralizations can confidentlybe made for populations as a whole. There isalso a need to investigate IT in a general population to explore itssexual com-position and attributes in a nonselected sample. With that qualification, itappears that controlling behaviors are a crucial element in distinguishingbetween physical aggression that is defined by femalevictims as so problem-

atic that it warrants leaving one’s home and fleeing to a Women’s Aid shelterand physical aggression that is not accompanied by frequent use of control-ling behaviors. It would appear that controlling behaviors are a risk markerfor high frequency, injurious, escalating physical aggression. It is thereforeimportant thatboth researchers and practitioners recognize thisdimensionof relationship conflict. Couples that report low levels of controlling behaviorsin conjunction with physical aggression may be involved in a qualitativelydifferent experience compared to high-control couples.

There areseverallimitationsto the findingsof thepresent study. Themea-sures used included a new scale, the CBS. However, the scales available inpublished sources were problematic. Some contained items of physicalaggression (e.g., Foshee, 1996; Shepard & Campbell, 1992), and others con-tained ambiguous items, such as “blamed for problems” (Toleman, 1989).Others contained items that were unsuitable for dating and childless couples(Shepard & Campbell, 1992; Toleman, 1989), or they assessed feelingsrather than behaviors (e.g., Smith, Earp, & DeVellis, 1995), or they were notbroad enough for the present study (e.g., Stets, 1991). The CBS has not beenvalidated by comparing it with existingscales, and this needs to be addressedin future research. The item “threaten to leave/commit suicide” is problem-atic. These behaviors are not interchangeable and have not been shown to berelated. This item should be split into two separate items to avoid confusion.Although the overall internal consistency of the CBS for men and womenwas good, there wasvariation in individual subscales. Economic was low forwomen and especially men, as was threats. Intimidation, emotional, and iso-lation were all high; however, as most subscales had five items or less, thealphas areacceptable.The MDVTP samplehadvery low response rates, lim-iting the interpretation of this data set.









The CTS did not contain acts of sexual aggression. Although physicalaggression and sexual aggression have typicallybeen studied independently,this separation has no theoretical basis. Future research needs to investigatesexual aggression in conjunction with physical aggression and controllingbehaviors. The CTS does not distinguish between self-defensive and offen-sive physical aggression. In mutuallyviolent relationships, it is oftendifficultto establishwhether physical aggression wasused offensivelyor defensively.However, studies that have asked men and women why they used physicalaggression have found a variety of reasons, including anger expression andcoercive control (Follingstad et al., 1991; Hamberger, Lohr, Bonge, & Tolin,1997). Indeed, Giordano and Cernkavich (1999) found that women’s angerself-concept was a significant predictor of their perpetration of relationshipaggression and that it held greater predictive power for women than men.

These findings suggest that although reasons for using physical aggressiondiffer among individuals, there is no consistent sex difference.

APPENDIX AThe Controlling Behaviors Scale and Alpha Coefficients

Economic. Each measure of economic control is the mean of five items with a 5-point responseformatrangingfrom 0 ( never ) to4( always ).Theitemswereas follows:

1. Did you/your partner disapprove of the other working or studying?2. If yes, didyou/yourpartner tryand preventor make difficult the other working

or studying?3. Did you/your partner feel it was necessary to have control of the other’s

money (e.g., wage, benefit)?

4. If yes, did you/your partner give theother an allowance or require other to ask for money?5. Did you/your partner have knowledge of the family income? a

For females, Cronbach’s alphafor the4-itemscaleis .66andfor males,it is.48.

Threats. Each measure of the useof threats is the mean of four items with a 5-pointresponse format ranging from 0 ( never ) to 4 ( always ). The items were as follows:

1. Did you/your partner make or carry out threats do something to harm theother?

2. Did you/your partner threaten to leave the other and/or commit suicide? b

3. Did you/your partner threaten to report the other to welfare?4. Didyou/your partner encourage the otherto doillegalthings he/shewouldnot

otherwise have done?










Intimidation . Each measure of the use of intimidation is the mean of five itemswith a 5-point response format ranging from 0 ( never ) to 4 ( always ). The items wereas follows:

1. Did you/your partner use looks, actions,and/or gestures to change the other’sbehavior?

2. If yes, did you/your partner make the other afraid when this was done?3. Did you/your partner smash property when annoyed/angry?4. If yes, was it the other’s property?5. When angry, did you/your partner vent anger on household pets? c


Emotional . Each measure of the use of emotional abuse is the mean of five itemswith a 5-point response format ranging from 0 ( never ) to 4 ( always ). The items wereas follows:

1. Did you/your partner put the other down when they felt the other was getting“too big for their boots”?

2. If yes, did you/your partner putthe otherdown in front of others (friends,fam-ily, children)?

3. Did you/your partner try to humiliate the other in front of others?4. Did you/your partner tell the other that he/she was going crazy?5. Did you/your partner call the other unpleasant names?

Cronbach’s alpha for the 5-item scale is .80 for females and .87 for males.

Isolation. Each measure of the use of isolation is the mean of five items with a 5-point response format ranging from 0 ( never ) to 4 ( always ). The items were as

follows:

1. Did you/your partner restrict the amount of time the other spent with friendsand/or family?

2. If you/your partner went out, didthe other want to know where the other wentand who the other spoke to?

3. Did you/your partner limit the other’s activities outside the relationship?4. Did you/your partner feel suspicious and jealous of the other?5. If yes, was this used asa reasonto monitor and controlthe other’s activities?

Cronbach’s alpha for the 5-item scale is .81 for females and .81 for males.In the overall 24-item CBS, Cronbach’s alpha was .90 for females and .89 for

males.

a. Item 5 was reversed scored.b. Note that this item is included under “Use of Privilege” by Johnson (1999) but is includedunder “Threats” by Pence and Paymar (1993).c. Notethat thisitem is includedunder“Using Children”by Johnson(1999)but is includedunder“Intimidation” by Pence and Paymar (1993).









APPENDIX BConflict Tactics Scale Items

Each item of physical aggression had a 5-point response format ranging from 0(never ) t o 4 ( always ). Respondents were instructedas follows:“Here is a list of thingsyou and your current or most recent spouse/partner might have done when you had aconflict. Taking all disagreements into account, not just the most serious ones, indi-cate how frequently each of you did the following during the conflict:”

k. threw something at the other one,l. pushed, grabbed, or shoved the other one,m. slapped the other one,n. kicked, bit, or hit with a fist,o. hit or tried to hit with something,p. beat up the other one,

q. threatened with a weapon (e.g., a knifea

),r. used a weapon (e.g. a knife).

Cronbach’s alphasfor theminoraggression scale were .96for females and.90 formales. For the severe aggression scale, the Cronbach’s alphas were .94 for femalesand .86 for males.

a. Note that the original format specifies “knife or gun”; however, “gun” was omitted due to themore limited availability of guns in Britain when compared to the United States.

NOTE

1. Bidirectional refers to physical aggression used at any time in the 12-month period, asopposed to in the same episode.

REFERENCES

Archer,J. (1999).An assessmentof thereliabilityof the ConflictTactics Scales: A meta-analyticreview. Journal of Interpersonal Violence , 14 , 1263-1289.

Archer, J. (2000). Sex differences in aggressionbetweenheterosexual partners:A meta-analyticreview. Psychological Bulletin , 126 , 651-680.

Archer,J., & Graham-Kevan, N. (2003).Do beliefs about aggression predict physicalaggressionto partners? Aggressive Behavior , 29(1), 41-54.

Barnett, O. W., Lee, C. Y., & Thelen, R. E. (1997). Gender differences in attributions of self-defense and control in interpartner aggression. Violence Against Women , 3 , 462-481.

Claes, J. A., & Rosenthal, D. M. (1990). Men who batter women: A study in power. Journal of Family Violence , 5 , 215-224.

Dobash, R. P., Dobash, R. E.,Cavanagh,K., & Lewis, R. (1998). Separateand intersecting reali -ties: A comparison of men and women’s accounts of violence against women. Violence

Against Women , 4 , 382-414.









Dobash, R. P., Dobash, R. E., Wilson, M., & Daly, M. (1992). The myth of sexual symmetry inmarital violence. Social Problems , 39 , 71-85.

Dutton, G. D., & Hemphill,K. J. (1992). Patterns of socially desirableresponding amongperpe-trators and victims of wife assault. Violence and Victims , 7 , 29-39.

Foshee, V. A. (1996). Gender differences in adolescentdating abuse prevalence, types, and inju-ries. Health Education Research , 11 , 275-286.

Follingstad, D. R., Wright, S., Lloyd, S., & Sebastian, J. A. (1991). Sex differences in motiva-tions and effects in dating violence. Family Relations , 40 , 51-57.

Giles-Sims, J. (1983). Wife battering: A systems theory approach . New York: Guilford.Giordano, P. C., & Cernkavich, S. A. (1999). Gender and antisocial behavior. In D. M. Scott,

J. Breiling, & J.D. Maser (Eds.), Handbookof antisocial behavior (pp.496-510). NewYork:John Wiley.

Graham-Kevan, N., & Archer, J. (2003). Physical aggression and control in heterosexual rela-tionships: The effect of sampling. Violence and Victims , 18(2), 181-196.

Hamberger, L. K., Lohr, J. M., Bonge, D., & Tolin, D. F. (1997). An empirical classification of motivations for domestic violence. Violence Against Women , 3 , 401-423.

Johnson, M. P. (1995). Intimate terrorism and common couple violence: Two forms of violenceagainst women. Journal of Marriage and the Family , 57 , 283-294.

Johnson,M. P. (1999, November). Two types of violence against women in theAmerican family: Identifying intimate terrorism and common couple violence . Paper presented at the annualmeeting of the National Council on Family Relations, Irvine, CA.

Johnson, M. P. (2000). The differential effects of patriarchal terrorism and common couple vio-lence: Findings from the National Violence Against Women Survey . Paper presented at theTenth International Conference on Personal Relationships, Brisbane, Australia.

Johnson,M. P., & Ferraro, K. J. (2000).Researchon domesticviolencein the 1990s: Making dis-tinctions. Journal of Marriage and the Family , 62 , 948-963.

Laner, M. R. (1985). Unpleasant, aggressive, and abusive activities in courtship: A comparisonof Mormon and non-Mormon college students. Deviant Behavior , 6 , 145-168.

Makepeace, J. M. (1986). Gender differences in courtship violence and victimisation. Family Relations , 35 , 383-388.

Moffitt, T. E., Caspi, A., Kneger, R. F., Magdol, L., & Margolin, G. (1997). Do partners agree

about abuse in their relationship?A psychometric evaluationof interpartner agreement. Psy-chological Assessment , 9 , 47-56.

Morse, B. J. (1995).Beyondthe ConflictTactics Scale:Assessing sex differences in partner vio-lence. Violence and Victims , 10 , 251-272.

Okun, L. (1986). Woman abuse: Facts replacing myths . Albany: State University of New York Press.

O’Leary, K. D., Barling,J., Arias, I.,& Rosenbaum,A. (1989).Prevalenceand stabilityof physi-cal aggressionbetweenspouses:A longitudinalanalysis. Journal of Consulting and ClinicalPsychology , 57 , 263-268.

Paglow, M. D. (1992). Adult victims of domestic violence: Battered women. Journal of Inter- personal Violence , 7 , 87-120.

Pence, E., & Paymar, M. (1993). Education groups of men who batter: The Duluth model . NewYork: Springer.

Riggs, D. S. (1993).Relationshipproblemsand dating aggression. Journal of Interpersonal Vio-lence , 8 , 18-35.

Saunders, D. (1988). Wife abuse, husband abuse, or mutual combat? In K. Yllö & M. Bogard(Eds.), Feminist perspectives on wife abuse . Newbury Park, CA: Sage.Shepard, M. F., & Campbell, J. A. (1992). The Abusive Behavior Inventory: A measure of psy-

chological and physical abuse. Journal of Interpersonal Violence , 7 (3), 291-305.









Smith, P. H., Earp, J. A., & DeVellis, R. (1995). Measuring battering: Development of theWomen’s Experience With Battering (WEB) Scale. Women’s Health: Research on Gender,

Behavior, and Policy , 1(4), 273-288.Stark, E. (1995). Representing woman battering: From battered woman syndrome to coercive

control. Albany Law Review , 58 , 973-1026.Steinmetz, S. K. (1978). The battered husband syndrome. Victimology: An International Jour-

nal , 2, 499-509.Stets, J. E. (1991). Psychological aggression in dating relationships: The role of interpersonal

control. Journal of Family Violence , 6 , 97-114.Straus, M. A. (1979). Measuring intrafamily conflict and violence: The Conflict Tactics (CT)

Scales. Journal of Marriage and the Family , 41 , 75-88.Straus, M. A. (1990). The Conflict Tactics Scales and its critics: An evaluation and new data on

validityand reliability. In M. A. Straus & R. J. Gelles (Eds.), Physical violence in American families: Risk factors and adaptations to violence in 8,145 families . New Brunswick, NJ:Transaction Publishing.

Straus, M. A. (1999). The controversy over domestic violence by women: A methodological,theoretical,and sociologyof science analysis.In X. B. Arriaga & S. Oskamp (Eds.), Violencein intimate relationships (pp. 17-44). Thousand Oaks, CA: Sage.

Straus, M. A., Hamby, S. L., Boney-McCoy, S., & Sugarman,D. B. (1996). The revised ConflictTactics Scale (CTS-2). Journal of Family Issues , 17 , 283-316.

Toleman, R. M. (1989).The developmentof a measure of psychologicalmaltreatmentof womenby their male partners. Violence and Victims , 4(3), 159-177.

Vivian, D., & Langhinrichsen-Rohling, J. (1994). Are bi-directionally violent couples mutuallyvictimized? A gender-sensitive analysis. Violence and Victims , 9 , 107-124.

Walker, L. E. A. (1989). Psychology and violence against women. American Psychologist , 44 ,695-702.

Wilson, M., & Daly, M. (1993). An evolutionary psychological perspective on male sexualproprietariness and violence against wives. Violence and Victims , 8 , 271-294.

Nicola Graham-Kevan is a lecturer of psychology at the Universityof CentralLancashire.She was educated at the University of Central Lancashire (B.Sc., psychology with law)andis currently completinga doctoralprogram at University of Central Lancashire. Her current research interests are partner aggression, stalking, fear of crime,and evolution-ary psychology.

John Archer is a professor of psychology at the University of Central Lancashire, Pres-ton. He was educated at the Universities of Wales (B.Sc., zoology) and Bristol (Ph.D.,

psychology). His earlier work concerned hormones and behavior in animals and thesocial and developmental psychology of gender. His current interests are human aggres -sion, grief andloss, and evolutionary psychology. He has written more than100 journalarticles andbook chaptersanda numberof editedandauthored books,the most recent of which are The Nature of Grief (Routledge, 1999) and Sex and Gender (coauthored with

Barbara Lloyd) (Cambridge University Press, 2002).


j interpers violence-2003-graham-kevan-1247-70.pdf

Documents