Dating Violence in Two High School Samples: Discriminating Variables

The Journal of Primary Prevention, Vol. 18, No. 4, 1998

Dating Violence in Two High School Samples:Discriminating Variables

Annmarie Cano,1 Sarah Avery-Leaf,2 Michele Cascardi,3and K. Daniel O'Leary1,4

Given that dating violence is a serious problem for adolescents, research isneeded to inform dating violence prevention programs which often targetcorrelates of violence. The current two studies examined the multivariatecorrelates of dating violence in two high schools. The Riggs and O'Leary (1989)model of courtship aggression was used to select a number of variables thatare most amenable to change through prevention and intervention strategies.Multivariate analyses, although showing similar discriminating variables formales and females, yielded slightly different patterns of predictors of male andfemale dating violence. The results have implications for the development ofhigh school-based dating violence prevention programs and for the furtherexamination of sex differences in dating violence.

KEY WORDS: dating violence; adolescents; high school students; sex differences; discriminantfunction analysis.

Dating violence among high school students occurs with remarkablefrequency. Surveys indicate that from 9% to 57% of adolescents have hadat least one experience with physical aggression in an dating relationship(Avery-Leaf, Cascardi, & O'Leary, 1994; Henton, Gate, Koval, Lloyd, &Christopher, 1983; Jones, 1987; Kaufman Kantor & Jasinski, 1995; O'Keefe,

1Department of Psychology, The University at Stony Brook, Stony Brook, New York.2At the time of this study, the author was affiliated with the Department of Psychology, TheUniversity of Stony Brook. She is now at the VA Medical Center, Togus, Maine.

3At the time of this study, the author was affiliated with the Department of Psychology, TheUniversity of Stony Brook. She is now at Arbor, Inc., Media, Pennsylvania.

4Address correspondence to K. Daniel O'Leary, Department of Psychology, The Universityat Stony Brook, Stony Brook, New York 11794-2500. Electronic mail may be sent [email protected].

431

© 1998 Human Sciences Press, Inc.

432 Cano, Avery-Leaf, Cascardi, and O'Leary

Brockopp, & Chew, 1986; Roscoe & Kelsey, 1986; Schwartz, O'Leary, &Kendziora, 1996). Further, O'Leary, Barling, Arias, Rosenbaum, Malone,and Tyree (1989) demonstrated that physical aggression during courtshippersisted into the early years of marriage for approximately half of newlymarried couples. In a cross-sectional analysis, Roscoe and Benaske (1985)reported that one half of battered women seeking emergency shelter fromabusive partners indicated physical abuse by a former dating partner. It isclear that relationship violence begins in mid-adolescence and may continueinto marital relationships if undetected and left untreated. As part of alarger effort to develop a curriculum to prevent dating violence among highschool students, two studies were conducted to identify discriminating cor-relates of dating violence that were grounded within theory and were alsoappropriate targets for intervention.

Since one of our primary objectives was to delineate the multivariatecorrelates of dating violence that had the greatest potential for change ina prevention and early intervention curriculum for male and female highschool students, our review of the literature targeted those correlates ofdating violence that may be most amenable to change. This approachprompted us to focus on specific attitudes and conflict based behaviors pre-viously examined as predictors of college and high school dating violence.Specifically, attitudes justifying the use of dating violence, and behaviorsincluding prior experience with aggression against a dating partner, verbalaggression towards a partner, partner's physical and verbal aggression, andjealousy were selected from the literature to be included in the currentstudy (e.g., Bird, Stith, & Schladale, 1991; Bookwala, Frieze, Smith, &Ryan, 1992; Henton et al., 1983; Riggs, 1990; Riggs & O'Leary, in press;Schwartz, O'Leary, & Kendziora, 1996; Smith & Williams, 1992; Tontodo-nato & Crew, 1992).

We will examine the multivariate relationship between these variablesand dating violence within the framework of the Riggs and O'Leary (1989)background-situational model of courtship aggression which is one of thefirst, and most comprehensive models of dating violence. Grounding thismodel in social-learning and conflict theories, Riggs and O'Leary (1989)proposed that background factors (e.g., inter-parental aggression, childabuse, prior aggressive behavior, acceptance of interpersonal violence) mayplace an individual at risk for aggressive behavior. Situational factors thatincrease conflict within the relationship (e.g., drinking, jealousy, verbal ag-gression) may then increase the probability of specific aggressive incidentsamong "at risk" individuals. In other words, background factors increasethe tendency of an individual to behave aggressively and the level of conflictwithin a relationship increases the probability of aggression in a specificsituation.

High School Dating Violence 433

Riggs and O'Leary (in press) tested this model in a sample of 345college undergraduates (232 females and 113 males) using structural equa-tion modeling. Overall, results indicated that for both men and women, amore accepting attitude towards dating aggression and past aggressive be-havior were important background factors in the prediction of dating ag-gression.

A number of multivariate studies also provide support for the pre-dictive power of background and situational factors identified in the Riggsand O'Leary model (1989). Specifically, Bird, Stith, and Schladale (1991)found that background factors associated with verbal aggression discrimi-nated nonviolent from violent college students. Situational factors such asverbal aggression by the self and partner, and higher levels of physical ag-gression by one's partner, in combination, also relate to the use of datingviolence (Follette & Alexander, 1992). Although these results indicate sup-port for the Riggs and O'Leary model, it is still not clear whether the find-ings apply to high school populations. Further, few studies included bothbackground and situational variables to allow for the multivariate contri-butions of each factor. The inclusion of both background (e.g., past datingviolence) and situational (e.g., partner's physical aggression) variables inthe present study will shed light on these questions.

While the Riggs and O'Leary (1989) model offers promise as a meansfor understanding the etiology of dating aggression, it acknowledges, butfails to explicate, the role of nonphysical expressions of psychological co-ercion and control such as jealous behaviors. As indicated by several re-searchers, jealousy is the most commonly cited precursor of violence amonghigh school and college students (Makepeace, 1981; Roscoe & Kelsey,1986; Roscoe & Callahan, 1985). However, the relationship of nonaggres-sive, coercive behaviors to physical aggression in dating relationships hasbeen largely unexplored (see Stets & Pirog-Good, 1990 for exception), de-spite the fact that their importance has been underscored by several notablefamily violence scholars (e.g., Yllo, 1988). Although research suggests theimportance of control and coercion for both genders, there are data tosuggest that female aggression may be less motivated by a desire to controlanother and more motivated by emotional arousal tied to jealousy. Morespecifically, Bookwala et al. (1992) found that jealousy was associated withthe use of physical aggression for females but not for males. The differentialrole of control and jealousy in explaining gender differences in the expres-sion of dating violence remains unclear, especially when terms such as jeal-ousy are not clearly defined by investigators (e.g., Bookwala et al., 1992).The present study will address these limitations by including operationallydefined jealousy and controlling behaviors as situational factors in the mul-tivariate analysis of dating violence in a high school sample.

434 Cano, Avery-Leaf, Cascardi, and O'Leary

The present study was designed to target potentially modifiable back-ground and situational factors previously associated with college dating vio-lence. Specifically, the multivariate relationships of these attitudes andbehaviors to dating violence among high school students will be examined.The design offers several advantages over previous studies. First, a highschool sample is used. The examination of high school students will aid inthe generalization and relevance of this study's findings toward the devel-opment of high school dating violence prevention programs. Second, vari-able selection was guided by practical and theoretical concerns. Byexamining the most proximal background and situational contributors todating violence, it is anticipated that this strategy will lead to results withdirect implications for further theory development and intervention efforts.Remaining within the Riggs and O'Leary (1989) framework, the followingbackground factors were examined: attitudes justifying the use of datingviolence and past dating aggression. In addition, the following situationalfactors were assessed: physical and verbal aggression by one's dating part-ner, one's own use of verbal aggression against a dating partner, and useof jealous and coercive tactics. These background and situational variableswill be examined in a multivariate fashion to discriminate dating violentand nonviolent high school boys and girls. Conducting analyses separatelyfor males and females allows for comparison between our results and thoseof previous researchers who have reported separate results for males andfemales (e.g., Bookwala et al., 1992; Rigg & O'Leary, in press). Third, amore rigorous approach to measuring control, coercion, and jealous datingbehaviors is attempted and gender differences in the expression of thesebehaviors is explored. That is, more operationalized and explicitly definedvariables will be used to assess controlling and jealousy behaviors.

STUDY 1

Method

Participants

Data from the present study were collected as part of a larger studyto evaluate the efficacy of a dating violence prevention program to alterattitudes associated with dating violence (Avery-Leaf, Cascardi, & O'Leary,1994). All students who were taking health class during the Fall, 1993, andSpring, 1994, completed a series of self-report measures prior to the com-mencement of the prevention program (pre-program evaluation). Only stu-dents who had dating experience (i.e., currently dating or dated in the past)

High School Dating Violence 435

were included in the current study (152 boys and 128 girls). The samplerepresented a racially diverse group of students: 44% Hispanic, 29% Cau-casian, and 19% African American. The mean age of the students was ap-proximately 16.64 (sd = .81) years. Fifty-four percent of the studentparticipants were male and 46% were female.

Measures

Past Dating Aggression Questionnaire (PDAQ). Four questions wereused to assess dating violence in previous dating relationships. Each re-spondent was asked whether he/she and/or his/her partner slapped, pushed,or hit in a past dating relationship. Internal consistency was acceptable ina prior study (Cronbach's alpha = .78; Riggs, 1990).

Modified Conflict Tactics Scale (MCTS). A modified version of theConflict Tactics Scale (Straus, 1979) developed by Neidig (1986) is an 18-item instrument that measures the frequency of both verbal and physicalaggression against a romantic partner. After reading a brief introduction,respondents are instructed to recall the frequency of a number of behaviorswithin the context of an argument. Examples of verbal aggression itemsare: "Have you done or said something to spite your boyfriend/girlfriend?"and "Have you insulted or sworn at your boyfriend/girlfriend?" Physicalaggression items include "Have you thrown something at your boy-friend/girlfriend?" and "Have you hit, kicked, or bit your boyfriend/girl-friend?" Respondents indicate how often within the past year each itemoccurred on a five point scale from "never" to "more than 11 times." Theitems are arranged in presumed order of severity to increase respondent'saccuracy of reporting. The CTS is reported to have a low refusal and an-tagonism rate (Straus, 1979) and to have a stable factor structure (Barling,O'Leary, Jouriles, Vivian, MacEwen, 1987; Caulfield & Riggs; 1992).

Dominance/Jealousy Scales (D/J). An abbreviated version of the Psy-chological Maltreatment of Women Inventory was used (PMWI; Tolman,1989). A more recent examination of the PMWI with 1,600 college studentsidentified six factors of this measure (Kaisan & Painter, 1992). Two of thesefactors, dominance and jealousy, were used in the present study. The D/Jscale was formed by retaining the 11 items that loaded the highest on thedominance factor and the 7 items that loaded highest on the jealousy factor.Specific items include "I monitored my partner's time and made him orher account for his or her whereabouts" and "I tried to keep my partnerfrom doing things to help him or herself." Respondents indicate the fre-quency (never to more than 20 times) with which they and their partnerhave engaged in the dominance and jealous behaviors in the year prior to

436 Cano, Avery-Leaf, Cascardi, and O'Leary

assessment. Internal consistency for the dominance/jealousy self scale washigh (alpha = .72) as was the dominance/jealousy partner scale (al-pha = .76).

Acceptance of Interpersonal Violence Questionnaire (AIVQ). Adaptedfrom Riggs (1990), the AIVQ consists of six items to assess the respondent'sacceptance of violence as a means of conflict resolution. Respondents ratethe justification or acceptance of male-to-female and female-to-male push-ing, slapping, and punching on a five point scale from "never" to " almostalways." In the current sample, internal consistency was good for female-to-male and male-to-female aggression (alpha = .74 and alpha = .74, re-spectively).

Results

Preliminary investigation of the data revealed that in order to analyzecomplete cases, 13.5% of the male and 12.43% of the female data (a totalof 37 students) would have to be excluded, reducing the power of the study.Analysis of complete cases is not advised when missing data results in ex-cluding more than 5% of cases (J. W. Graham, personal communication,January 25, 1997). Given this amount of missing data, we applied the EM(Expectation-Maximization) algorithm, an iterative estimation procedure,that has been frequently and successfully used as a method of conductingstatistical analysis with missing data (Graham & Donaldson, 1993; Graham,Hofer, Donaldson, MacKinnon, & Schafer, 1996). The program we used,EMCOV (Graham & Hofer, 1993) consists of two major steps. The firststep (E-step), estimates missing data by using all available data in a re-gression equation predicting the missing data points. The next step (M-step) in the program calculates covariance matrices and regression weightswhich are then plugged back into the E-step. These steps are repeateduntil the convergence criterion are met. That is, the steps are repeateduntil "the changes in the estimates of the covariance matrix are sufficientlysmall and judged to be unimportant" (p. 123; Graham & Donaldson, 1993).The EMCOV procedure produces unbiased parameter estimates and stand-ard errors of missing data as compared to mean substitution and listwisedeletion strategies (Graham et al., 1996). Mean substitution tends to un-derestimate parameter estimates due to restriction of range while listwisedeletion tends to yield highly variable estimates.

High School Dating Violence 437

Discriminant Function Analysis

Two separate discriminant function analyses were conducted to de-termine the ability of the predictor variables to correctly classify males andfemales as aggressive or not aggressive. Participants were classified as ag-gressive if they had engaged in at least one aggressive act toward theirdating partner within a year. The discriminant function for males correctlyclassified 71.12% of males as aggressive and accounted for 49% of the vari-ance in dating violence as indicated by the equation 1 — Wilk's Lambda,a measure of effect size similar to R2 in regression analyses. As shown inTable I, variables that significantly contributed to the classification werehistory of dating aggression by the self and partner, physical aggression bythe partner, and verbal aggression by the self and partner. The linear com-bination of predictor variables was significantly associated with the discrimi-nant function (canonical r = .70), x2(9) = 98.80, p < .00001.

For females, the discriminant function correctly classified 73.54% of fe-males as aggressive. Table I presents the correlations of the predictor variableswith the discriminant function. Significant contributors to this function werehistory of dating aggression by the self and partner, physical aggression by thepartner, one's own use of verbal aggression, and dominance/jealousy tacticsused by self and partner. This combination of predictor variables accountedfor 40% of the variance in female dating violence and was significantly asso-ciated with the discriminant function (canonical r = .63), x2(9) = 62.13,p < .00001. Results are presented in Table I. Although similar discriminatingvariables emerged for boys and girls (self-reported and partner's history ofdating violence, partner's physical aggression, and self-reported verbal aggres-sion), there were some differences. For instance, partner's verbal aggressionalso contributed to male violence while self-reported and partner dominanceand jealousy tactics also predicted female dating violence.

STUDY 2

Method

Participants

The data collected for the present study were obtained for a largerstudy evaluating the efficacy of a dating violence prevention program to alterattitudes associated with dating violence (Avery-Leaf, Cascardi, O'Leary, &Cano, in press). All students taking health classes in a Long Island, NewYork high school in the Fall semester of 1994 completed a battery of ques-

438 Cano, Avery-Leaf, Cascardi, and O'Leary

Table I. Discriminant Function Analysis: Study 1 — Correlationof Predictor Variables with Discriminant Function

Predictor Variable

Background variablesPDAQ - selfPDAQ - partnerAIVQ male-to-femaleAIVQ female-to-male

Situational variablesPhysical MCTS - partnerVerbal MCTS - selfVerbal MCTS - partnerD/J - selfD/J - partner

MalesN = 152

.80a

.49a

.35

.24

.68"

.48"

.45"

.25

.30

FemalesN = 128

.71a

.45"

.09

.04

.64"

.40a

.36

.57"

.48"

Key: PDAQ = Past Dating Aggression Questionnaire; MCTS =Modified Conflict Tactics Scale; D/J = Dominance/JealousyQuestionnaire; AIVQ = Acceptance of Interpersonal ViolenceQuestionnaire.a Indicates a significant discriminating variable.

tionnaires as part of the larger study prior to the implementation of theprevention program. The school at which this study was conducted was adifferent school than that in which Study 1 was conducted. As in the previousstudy, only subjects who had personal experience in dating relationships wereincluded in the present study (100 females and 117 males). The racial/ethniccomposition was predominantly Caucasian (83%); 9% were Hispanic/Latino,3% were African American, and 2% were Asian American. The mean ageof the students was 16.43 years (sd = .88 years).

Measures

As stated earlier, all of the measures used in Study 1 are includedin Study 2.

Results

As in Study 1, the EMCOV procedure was used to correct for missingdata, as 17.2% male and 12.9% female data (a total of 33 students) wouldhave to be excluded if complete case analysis were conducted. Again, thiswould result in a substantial loss of power.

High School Dating Violence 439

Table II. Discriminant Function Analysis: Study 2 — Correlationof Predictor Variables with Discriminant Function

Predictor Variable

Background variablesPDAQ - selfPDAQ - partnerAIVQ male-to-femaleAIVQ female-to-male

Situational variablesPhysical MCTS - partnerVerbal MCTS - selfVerbal MCTS - partnerD/J - selfD/J - partner

MalesN = 117

.36

.63a

.28

.25

.52a

.57a

.54a

.45"

.34

FemalesN = 100

.42"

.29

.23

.12

.67"

.47"

.33

.42°,25

Key: PDAQ = Past Dating Aggression Questionnaire; MCTS =Modified Conflict Tactics Scale; D/J = Dominance/JealousyQuestionnaire; AIVQ = Acceptance of Interpersonal ViolenceQuestionnaire.a Indicates a significant discriminating variable.

Discriminant Function Analysis

Discriminant function analyses were conducted separately for malesand females, to determine the best set of independent variables to correctlyclassify aggressive respondents. For males, the discriminant function cor-rectly classified 92.4% of males as aggressive. The variables that signifi-cantly contributed to this function were dating violence by a past datingpartner, frequency of aggression by the partner, verbal aggression by selfand partner, and self-reported dominance and jealousy tactics. These re-sults are presented in Table II. There was a strong association betweengroups and predictors (canonical r = .77), x2(9) = 99.60, p < .0001. Thediscriminant function accounted for 59% of the between-group variabilityas indicated by 1 — Wilk's Lambda, a measure of effect size.

The discriminant function also correctly classified 86.9% of femalesas aggressive. Partner's dating violence, self-reported verbal aggression,self-reported use of violence in past dating relationships, and self-reporteddominance and jealousy tactics were significant contributors to this func-tion. Table II presents the correlations of these variables with the discrimi-nant function. The function significantly accounted for 65% of the variancein dating violence; there was also a strong association between groups andpredictors, (canonical r = .80), x2(9) = 96.94, p < .00001. For both boys

440 Cano, Avery-Leaf, Cascardi, and O'Leary

and girls, partner's physical aggression, self-reported verbal aggression, andself-reported dominance and jealousy tactics significantly discriminated vio-lent from nonviolent participants. However, differences also emerged: self-reported history of dating violence also contributed to female datingviolence while partner's history of dating violence and verbal aggressionalso contributed to male violence.

DISCUSSION

The current studies used the Riggs and O'Leary (1989) model of dat-ing violence as a framework for investigating the association of modifiableattitudinal and relationship-specific behaviors with high school students' useof dating violence in an effort to inform prevention programs. As expected,both of the present studies demonstrated that both background and con-textual factors multivariately discriminated between dating violent and non-violent boys and girls. We concentrate our discussion on results that werefound consistently across the two studies. Specifically, for males and fe-males in both studies, situational factors (partner's dating violence and self-reported verbal aggression) consistently and significantly predicted datingviolence, taking into account all other background and situational measures.This was expected, as it confirmed past research findings (Bookwala et al.,1992; Follette & Alexander, 1992; Riggs et al., 1990). These results suggestthat prevention programs focused on these discriminating correlates willbe equally effective with both boys and girls.

However, sex differences also emerged in the discriminant functionanalyses. Primarily self-reported background and situational variables in-cluding history of dating violence, verbal aggression, and jealousy behaviorsdiscriminated between violent and non-violent females. However, discrimi-nating variables for boys consisted primarily of partner-attributed behaviorsincluding partner's physical aggression, verbal aggression, and history ofdating violence. These findings support previous research demonstratingsex differences in the correlates of dating violence (Bookwala et al., 1992;Follette & Alexander, 1992; Riggs et al., 1990); however, more research isneeded before definitive conclusions are made given differences in studymethodology and assessment instruments across studies.

One reason for these sex differences in the present study may be thatfemales are more knowledgeable or "tuned in" to their own behaviors intheir relationships, thereby endorsing a greater number of self-related,rather than partner-attributed behaviors. Males, on the other hand, maybe more likely to minimize or deny their use of aggressive behaviors, whichis reflected in the association between dating violence and their reports of

High School Dating Violence 441

their partners' negative behaviors in past and current relationships ratherthan their own behaviors. Alternatively, as Follette and Alexander (1992)suggest, findings such as these may imply a pattern of conflict escalation.Perhaps females attempt to resolve conflicts verbally but do not judgethemselves effective unless they use physically and verbally aggressivestrategies. In response to this physical violence (along with the accompa-nying verbal aggression and jealousy tactics) the male partner may then bephysically violent to the female. These results call for studies examiningthe context of specific arguments that lead to male and female dating vio-lence, as this type of research may yield more detailed information aboutthe interactional nature of dating violence. In other words, what kind ofinteractional sequences within the couple are most likely to lead to vio-lence? How does each partner influence the other?

Interestingly, we found that none of the attitudinal measures involvingthe use of dating violence were significant discriminating variables. Re-search examining the relationship between attitudes and dating violencehas been inconsistent (Bookwala et al., 1992; Burke, Stets, & Pirog-Good,1989; Gate, Henton, Koval, Christopher, & Lloyd, 1982; Deal & Wampler,1986; Foo & Margolin, 1995; Henton et al., 1983; Stets & Pirog-Good,1990). One possible explanation for this inconsistency is that attitude meas-ures have not been consistently used across studies. Perhaps the validationof measures across a variety of samples could improve this situation. Wehave found that using measures of attitudes that give information aboutspecific situations rather than acontextual measures allows for a greaterrange of responses, thereby reducing floor effects that could account foran absence of correlations (Slep, Avery-Leaf, Cascardi, & O'Leary, 1996).Because the relationship between attitudes and dating violence has beendifficult to understand, we suggest further use and refinement of contextualmeasures such as those used by Foo and Margolin (1995). These measuresseem better suited to tap attitudes toward behaviors that are not sociallydesirable.

Another explanation for inconsistencies across studies concerning therole of attitudes may be the potentially different within-school cultures. La-voie, Hebert, and Dufort (1995) found attitudes to be differentially corre-lated with dating violence and suggested that these differences were dueto inter-school differences in terms of culture and norms. Although Study1 was conducted in a more racially diverse school than Study 2, we did notfind different patterns of correlates from school to school. In fact, we foundmore similarities between schools than differences, suggesting that the ro-bust and consistent associations we found in the present study are foundacross different school cultures and subgroups. Perhaps inconsistencies maybe attributed, in part, to regional differences.

442 Cano, Avery-Leaf, Cascardi, and O'Leary

Strengths and Weaknesses of the Current Studies

The present study has a number of strengths. First, we used bothempirical studies and the Riggs and O'Leary (1989) model of dating vio-lence as a basis from which to study dating violence in a high school sample.As we discussed earlier, few models exist explaining dating violence andmuch of the current research in this topic has been conducted with collegesamples (e.g, Riggs et al., 1990). It was not clear whether results obtainedin college student samples were generalizable to those results obtained inyounger populations. The findings from the two studies conducted hereinsuggest that similar factors are associated with high school and college dat-ing violence. Therefore, prevention efforts in high school that are focusedon the discriminating correlates of dating violence may be extremely usefulin decreasing and perhaps eliminating violent behavior in relationships incollege and later adulthood.

Second, we were also able to integrate theory and research concerningcontrol and jealousy in our examination of dating violence. Operationaliz-ing and defining these concepts as distinct behaviors made it possible toexplore the association of these variables with dating violence, thus inform-ing possible interventions. Moreover, we were able to show consistenciesof relationships between two high schools, thus increasing the inter-schoolreliability of our results. This is an important strength, as high school stu-dents are understudied, particularly with regard to dating and dating vio-lence. As school-based interventions seem the best approach to theprevention of a pervasive, long-term problem with partner aggression, re-search with this age group is sorely needed.

Although these strengths are important and increase the validity ofconclusions made from this study, there is a weakness that needs to beaddressed. We experienced a substantial loss of data in the two studies.Although we were able to utilize a well-validated missing data approach(EMCOV; Graham & Hofer, 1993), one cannot deny the fact that havingcomplete data on all participants is much more desirable. Thus, an impor-tant next step is to replicate this study with a more complete set of data,perhaps collected in the presence of classroom monitors. Although a morecostly procedure of data collection, this will ensure less missing data.

Implications for Prevention Programs

The high rates of dating violence reported in a number of studies(e.g., Avery-Leaf et al., 1994; Henton et al., 1983; Kaufman Kantor & Jas-inski, 1995; Schwartz, O'Leary, & Kendziora, 1996) underscore the need

High School Dating Violence 443

for primary prevention and early intervention programs. Several programsto curtail the use of dating violence have been developed and implemented(Cascardi & Avery-Leaf, 1994; Jaffe et al., 1992; Kaufman Kantor & Jas-inski, 1995; Werk, Brown, & Shields, 1995). These programs typically targetbackground (e.g., attitudes) and situational behaviors associated with datingviolence. Emphasis on these factors is supported largely by the growingbody of research focused on correlates of dating violence for college stu-dents (Follette & Alexander, 1992; Pedersen & Thomas, 1992; Riggs, 1993;Riggs, O'Leary, & Breslin, 1990; Stets & Pirog-Good, 1990).

However, many of the studies examining the correlates of dating vio-lence do so without looking at the practical implications for violence pre-vention programs. In the present study, we found that past experience withdating violence was associated with current violence, suggesting that pro-grams be presented to children before they begin to date. Perhaps prevent-ing the onset of dating violence may decrease the likelihood that thisbehavior would occur during adolescence and young adulthood.

Our findings that a history of dating violence and verbal and physicalaggression by the partner as well as self-reported verbal aggression are allhighly associated with one's own use of dating violence suggest that at-tempts should be made to educate students of the long-term consequencesof violence. For instance, one can become engaged in a pattern of repeatedaggression which may result in negative psychological impact and physicalinjury (e.g, Roscoe & Benaske, 1985). Students should be encouraged tothink about past experiences with dating violence and how they could haveresolved the conflict differently. For example, what could they have doneto prevent themselves from becoming violent? What could their partnerhave done? With the help of a thoughtful teacher, students may be ableto brainstorm ways in which they could have prevented violence and learnnew ways of preventing it in the future. One strategy (Time Out) is forboth partners to mutually agree to stop an escalating verbal argument,separate for a mutually agreed upon time (e.g., 1 hour), and then to calmlydiscuss the problem again.

Dominance and jealousy behaviors are highly associated with currentreports of violence. Jealousy behaviors, which are often attempts to main-tain valued relationships (Salovey & Rodin, 1988), can be targeted byteaching students alternative means of maintaining their relationships suchas open communication about one's goals and desires for the relationship.Teaching assertive communication strategies may aid in reducing the useof indirect strategies such as making the partner account for his/her time.This is especially relevant for females since the results herein indicated astrong relationship between female dominance and jealousy tactics and dat-ing violence. It is clear that prevention programs need to be multi-faceted

444 Cano, Avery-Leaf, Cascardi, and O'Leary

if they are to be successful. Therefore, combining components targeting avariety of background and situational factors is necessary. As additionalresearch is conducted, more variables will need to be included in preventionprogram curricula.

The present study examined correlates of dating violence for highschool boys and girls and focused on how such correlates can give us insighttoward developing prevention programs that are effective for both boys andgirls. It is hoped that more researchers studying dating violence will turntheir attention toward high school students and children of younger ages.Although it is more difficult to conduct research in public high schoolsthan in colleges and universities, the information obtained is valuable andworthwhile. These programs will teach students skills that can be used fora lifetime, from dating through marriage, as well as applied to other inter-personal relationships.

ACKNOWLEDGMENTS

This study was supported by a National Institute of Mental HealthTraining Grant MH19107 to the fourth author. We would like to thankJohn Graham for his consultation on data analysis and the anonymous re-viewers for their comments on earlier drafts of this manuscript. Portionsof this manuscript were presented at the 4th International Family ViolenceResearch Conference, Durham, N.H., 1995, and the Association for theAdvancement of Behavior Therapy Conference, Washington, D.C., 1995.

REFERENCES

Avery-Leaf, S., Cascardi, M., & O'Leary, K.D. (1994). Efficacy of a dating violence preventionprogram. Poster presented at the 102nd annual meeting of the American PsychologicalAssociation, Los Angeles, CA.

Avery-Leaf, S., Cascardi, M., O'Leary, K.D., & Cano, A. (in press). The efficacy of a datingviolence prevention program on attitudes justifying aggression. The Journal of AdolescentHealth.

Barling, J., O'Leary, K.D., Jouriles, E.N., Vivian, D., & MacEwen, K. (1987). Factor similarityof the conflict tactic scales across samples, spouses, and sites: Issues and implications.Journal of Family Violence, 2, 37-54.

Bird, G.W., Stith, S.M., & Schladale, J. (1991). Psychological resources, coping strategies, andnegotiation styles as discriminators of violence in dating relationships. Family Relations,40, 45-50.

Bookwala, J., Frieze, I.H., Smith, C., & Ryan, K. (1992). Predictors of dating violence: Amultivariate analysis. Violence and Victims, 7, 297-311.

Burke, P.J., Stets, J.E., & Pirog-Good, M.A. (1989). Gender identity, self-esteem, and physicaland sexual abuse in dating relationships. In M. A. Pirog-Good & J.E. Stets (Eds.) Violencein dating relationships: Emerging social issues. New York: Praeger Publishers.

High School Dating Violence 445

Cascardi, M., & Avery-Leaf, S. (1992). Dating Violence Prevention Curriculum. Unpublishedmanuscript. The University at Stony Brook, Stony Brook, New York.

Cascardi, M., Avery-Leaf, S., & O'Leary, K.D. (1994). Building a gender-sensitive model toexplain male and female use of dating violence: Preliminary results. Poster presented atthe 102nd annual meeting of the American Psychological Association, Los Angeles, CA.

Gate, R.M, Henton, J.M., Koval, J., Christopher, F.S., & Lloyd, S. (1982). Premarital abuse:A social psychological perspective. Journal of Family Issues, 3, 79-90.

Caulfield, M.B., & Riggs, D.S. (1992). The assessment of dating aggression: Empiricalevaluation of the conflict tactics scale. Journal of Interpersonal Violence, 7, 549-558.

Deal, I.E., & Wampler, K.S. (1986). Dating violence: The primacy of previous experience.Journal of Social and Personal Relationships, 3, 457-471.

Follette, V. M., & Alexander, P.C. (1992). Dating violence: Current and historical correlates.Behavioral Assessment, 14, 39-52.

Foo, L., & Margolin, G. (1995). A multivariate investigation of dating aggression. Journal ofFamily Violence, 10, 351-377.

Graham, J. W., & Donaldson, S.I. (1993). Evaluating interventions with differential attrition:The importance of nonresponse mechanisms and use of followup data. Journal of AppliedPsychology, 78, 119-128.

Graham, J.W., & Hofer, S.M. (1993). EMCOV Reference Manual. Institute for PreventionResearch, University of Southern California.

Graham, J.W., Hofer, S.M., Donaldson, S.I., MacKinnon, D.P., & Schafer, J.S. (1996). Analysiswith missing data in prevention research. In K. Bryant, M. Windle, & S. West (Eds.),New Methodological Approaches to Alcohol Prevention Research. Washington, D.C.:American Psychological Association.

Henton, J.M., Gate, R.M., Koval, J., Lloyd, S., & Christopher, S. (1983). Romance andviolence in dating relationships. Journal of Family Issues, 4, 467-482.

Jones, R.J. (1987). Program evaluation in crime and delinquency. In E.K. Morris & J.Braukmann (Eds.) Behavioral approaches to crime and delinquency: A handbook ofapplication, research, and concepts. New York: Plenum Press.

Kaisan, M., & Painter, S.L (1992). Frequency and severity of psychological abuse in a datingpopulation. Journal of Interpersonal Violence, 7, 350-364.

Kaufman Kantor, G., & Jasinski, J.L. (1995). Prevention of teen dating violence: Evaluationof a multidimensional model. Paper presented at the 4th International Family ViolenceResearch Conference, Durham, N.H.

Lavoie, F., Hebert, M, & Dufort, F. (1995). Predictive variables identifying dating violencevictims and aggressors among grade ten students. Paper presented at the 4th InternationalFamily Violence Research Conference, Durham, N.H.

Makepeace, J. M. (1981). Courtship violence among college students. Family Relations, 30,97-102.

Neidig, P. (1986). Modified Conflict Tactics Scale. Unpublished scale. Behavioral SciencesAssociates, Beaufort, South Carolina.

O'Keefe, N.K., Brockopp, K., & Chew, E. (1986). Teen dating violence. Social Work, 31,465-468.

O'Leary, K.D., Barling, J., Arias, I., Rosenbaum, A., Malone, J., & Tyree, A. (1989).Prevalence and stability of physical aggression between spouses: A longitudinal analysis.Journal of Consulting and Clinical Psychology, 57, 263-268.

Pagelow, M.D. (1980). Woman-battering: Victims and their experiences. Beverly Hills, CA: SagePublications.

Pedersen, P., & Thomas, C.D. (1992). Prevalence and correlates of dating violence in aCanadian university sample. Canadian Journal of Behavioural Science, 24, 490-501.

Riggs, D. S. (1993). Relationship problems and dating aggression: A potential treatment target.Journal of Interpersonal Violence, 8, 18-35.

Riggs, D. S. (1990). Tests of a theoretical model of self-reported courtship aggression.Unpublished doctoral dissertation. The University at Stony Brook. Stony Brook, NewYork.

446 Cano, Avery-Leaf, Cascardi, and O'Leary

Riggs, D.S., & O'Leary, K. D. (in press). Violence between dating partners: Background andsituational correlates of courtship aggression. Journal of Family Violence.

Riggs, D. S., & O'Leary, K. D. (1989). A theoretical model of courtship aggression. In M.A. Pirog-Good & I.E. Stets (Eds.) Violence in dating relationships: Emerging social issues,New York: Praeger Publishers.

Riggs, D. S., O'Leary, K. D., & Breslin, F. C. (1990). Multiple correlates of physical aggressionin dating couples. Journal of Interpersonal Violence, 5, 61-73.

Roscoe, B., & Benaske, N. (1985). Courtship violence experienced by abused wives:Similarities in patterns of abuse. Family Relations, 34, 419-424.

Roscoe, B., & Callahan, J. E. (1985). Adolescents' self-report of violence in families and datingrelations. Adolescence, 20, 545-553.

Roscoe, B., & Kelsey, T. (1986). Dating violence among high school students. Psychology, 23,53-59.

Salovey, P., & Rodin, J. (1988). Coping with envy and jealousy. Journal of Social and ClinicalPsychology, 7, 15-33.

Schwartz, M., & O'Leary, S. G. (1996). Physical aggression in a high school dating population:A study of interpersonal conflict. Unpublished manuscript, State University of New Yorkat Stony Brook, Stony Brook, New York.

Slep, A.S., Avery-Leaf, S., Cascardi, M., & O'Leary, K.D. (1996). Attitudes accepting of datingaggression: A comparison of 3 measures. Paper presented at the 30th annual meeting ofthe Association for the Advancement of Behavior Therapy, New York, New York.

Smith, J.P., & Williams, J.G. (1992). From abusive household to dating violence. Journal ofFamily Violence, 7, 153-165.

Stets, J. E., & Pirog-Good, M.A. (1990). Interpersonal control and courtship aggression.Journal of Social and Personal Relationships, 7, 371-394.

Straus, M. (1979). Measuring intrafamily conflict and violence: The conflict tactics (ct) scales.Journal of Marriage and the Family, 41, 75-88.

Tolman, R.M. (1989). The development of a measure of psychological maltreatment of womenby their male partners. Violence and Victims, 4, 159-177.

Tontodonato, P., & Crew, B.K. (1992). Dating violence, social learning theory, and gender:A multivariate analysis. Violence and Victims, 7, 3-14.

Werk, A., Brown, T.G., & Shields, N. (1995) Adolescent attitudes regarding violence: Aschool-based small group program. Paper presented at the 4th International FamilyViolence Research Conference, Durham, N.H.

Yllo, K. (1988). Political and methodological debates in wife abuse research. In K. Yllo andM. Bograd (Eds.) Feminist perspectives on wife abuse, Newbury Park, Sage Publications.