(Campbell, 1992 Halsted, 1992 Reiss & Roth, 1993 U.S. Attorney���s General Task Force on Family Violence, 1984). Intimate relationships have also been rede- fined. Initially, women married to and living with violent husbands drew the bulk of atten- tion (Gordon, 1988 Pleck, 1987). Other catego- riesofpartner violencevictimshave been recog- nized, including separated and divorced spouses, present and former unmarried cohabitants, and dating and same-sex partners. Despite the broadened perspectiveofintimate partner violence, vio- lence perpetrated by and against adoles- cents in dating rela- tionships has yet to emerge as a major fo- cus of attention among reformers or research- ers. Adolescents, or preteen to high school- aged youths, are in a developmental period when courtship behav- ior is first initiated and when the risk of abuse by or against a dating partner first emerges. Given the crucial nature of this period, it is sur- prising that so little is known about the size and distribution of the problem of dating violence during adolescence and the effectiveness of programs designed to address it. In this review, we describe the major na- tional data sources on prevalence of dating violence and summa- rize the studies that also provide a source, albeit more limited, of prevalence estimates. We discuss what is known about gender, geographic distribu- tion, maltreatment in the home, and justifi- cation of the use of vio- lence in dating rela- tionships. We then review results of dating violence prevention program evaluations. We discuss limitations of the existing studies, areas where more research is needed, and highlight potential explanations for the lack of researcher investment inthistopicarea.Finally,wesuggest that additional investment in high-quality basic researchisneeded toinformthe development of sound theory and effective programs to prevent and intervene in dating violence between adolescent partners. MEASURING THE SIZE OF THE PROBLEM Far more is known about the problem of part- ner violence in terms of prevalence, gender dis- tribution, context, and consequences among college students and adults than among adoles- cents. Very few studies have focused specifi- cally on teens, leaving many questions unan- swered and need for further investigation. However, homicide data, data from two ongo- ing national surveys, (and several cross-sec- tional studies of specific populations) do pro- vide some findings about prevalence of dating violence, and how it varies by sex and context. National Victimization Prevalence Estimates Homicide data and self-report survey data are among the few national data sources on dat- ing violence among adolescents. Homicide data show girls to be at much higher risk than boys for the most extreme form of partner violence. According to the Federal Bureau of Investiga- tion���s (1993-1999) Supplementary Homicide Re- ports, about 10% of all 12- to-15-year-old girls, and 22% of all 16- to-19-year old girls, murdered between 1993 and 1999 were killed by an inti- mate partner. By contrast, intimate partners were the perpetrators of only about 1% of homicides among boys in these age groups. The National Crime Victimization Survey (NCVS) is designed to measure criminal victim- ization among all U.S. households and is con- ducted by the Census Bureau for the Depart- ment of Justice, Bureau of Justice Statistics. Conducted since 1973, the NCVS is a nationally representative household survey, conducted every 6 months with approximately 100,000 in- dividuals at least 12 years of age in about 50,000 124 TRAUMA, VIOLENCE, & ABUSE / April 2004 Despite the broadened perspective of intimate partner violence, violence perpetrated by and against adolescents in dating relationships has yet to emerge as a major focus of attention among reformers or researchers. According to the Federal Bureau of Investigation���s (1993-1999) Supplementary Homicide Reports, about 10% of all 12- to- 15-year-old girls, and 22% of all 16- to-19-year old girls, murdered between 1993 and 1999 were killed by an intimate partner. at UNIV OF ILLINOIS URBANA on May 22, 2009 http://tva.sagepub.com Downloaded from

households. It contains questions on violent and property criminal victimization, including specific questions about violence perpetrated by boyfriends, girlfriends, and present and for- mer spouses (Kindermann, Lynch, & Cantor, 1997). It is perhaps the most widely cited source of national estimates for partner violence among adults. Recent published summary reports of NCVS data do not provide overall prevalence esti- mates for boys and girls combined or estimates only for those age 12 to 18 years 19-year-old youths are also included. The published reports do show physical and sexual victimization by intimate partners to be relatively low for boys and girls 12 to 15 years of age but higher for those 16 to 19 years of age (Rennison & Welchans, 2000). Between 1993 and 1998, the av- erage annual percentage of girls aged 12 to 15 years who suffered victimization was 0.3% and 0.1% of boys of the same age. Among those aged 16 to 19 years, 1.7% of girls but only 0.2% of boys reported violent physical or sexual victimization by an intimate partner. Another source of national data is the Youth RiskBehaviorSurveillanceSystem (YRBSS)sur- vey, sponsored by the Centers for Disease Con- trol and Prevention (CDC) and designed to measure health risk behaviors among U.S. high school students. Conducted annually, it is de- signed to assess a range of health-risk behaviors in a representative sample of American stu- dents grades 9 through 11. Prevalence estimates produced by the YRBSS are several times that reported by the NCVS, with less disparity by sex. In 2001, 9.8% of girls and 9.1% of boys re- ported experiencing physical violence over the previous 12 months at the hands of a dating partner (Grunbaum et al., 2002). Clearly, the victimization estimates of the NCVS and YRBSS differ, and this may be be- cause of the nature of the two data sources. Some obvious differences are that NCVS in- cludes a broader age range than the YRBSS and includes sexual violence, whereas the YRBSS taps only physical violence. These differences are not likely to be solely responsible for pro- ducing the disparate estimates. Another possi- ble explanation may be other differences be- tween the two samples and the manner in which the surveys are conducted. The NCVS is designed to survey a nationally representative sample of American households, while the YRBSS represents American high school stu- dents. NCVS respondents are interviewed in the presence of other household members, whereas YRBSS respondents complete an anon- ymous paper-and-pencil instrument in relative privacy within a classroom environment. The lack of privacy during the NCVS interviews has been criticized for its potential to suppress ado- lescent���s reporting, who may be reluctant to re- veal victimization in front of parents (Finkelhor & Ormrod, 1999). Similar differences between classroom-administered surveys and house- hold surveys have been noted in relation to teen reports about substance abuse (e.g., Gfroerer, Wright, & Kopstein, 1997), supporting the no- tion that there is potential underreporting of sensitive, personal information among teens participating in household surveys. Finally, the NCVS presents questions within a criminal victimization context. Adolescents in Hickman et al. / DATING VIOLENCE AMONG ADOLESCENTS 125 KEY POINTS OF THE RESEARCH REVIEW ��� Although research on intimate partner violence among adults has dramatically expanded over the past 30 years, comparatively little is under- stoodaboutpartnerviolenceamongadolescents. ��� Although dating violence is clearly an important problem for teens, no consensus has emerged about the prevalence and gender distribution of violence between adolescent dating partners, and two major sources of national data produce widely divergent estimates. ��� Few studies have evaluated programs for adoles- cents designed to prevent dating violence, and the results of the existing studies do not present conclusive evidence about the efficacy of these programs. ��� There are a number of challenges that face re- searchers interested in investigating adolescent dating violence and this may explain the dearth of research in this area. ��� Despite these challenges, greater attention on conducting methodologically rigorous descrip- tive studies of the phenomenon and high quality evaluations ofprogramsdesigned toprevent dat- ing violence among adolescents are greatly needed. at UNIV OF ILLINOIS URBANA on May 22, 2009 http://tva.sagepub.com Downloaded from

the NCVS may underreport criminal victimiza- tion of all sorts because violence by peers may not be viewed as a legitimate crime (Finkelhor & Ormrod, 1999). By contrast, the YRBSS asks students about dating violence in the context of other health risk behaviors, including smoking, sexual behavior, dietary habits, alcohol and drug use, accidental injuries, and physical fights with peers. Single Study Prevalence Estimates Estimates from published single studies of dating violence victimization among adoles- cents are more consistent with those of the YRBSSthan NCVS,howevermany single-study estimates of physical victimization are higher than those produced by the YRBSS. These stud- ies generally employ a format similar to the YRBSS, that is pencil-and-paper questionnaires completed within a classroom environment. Many use the Conflict Tactics Scale (CTS) (Straus, 1979), modified for use in a classroom with adolescents. Studies examining overall prevalence of dating violence among adoles- cents often share similarities in their samples but have produced quite disparate estimates. In the single studies, per- petration estimates range from 26% to 46% for physical violence and 3% to 12% for sex- ual violence. For victim- ization, estimates range from a low of 9% to a high of 23% for physical perpetration and vic- timization, further lim- iting the ability to draw conclusions. Table 1 provides a summary of these studies, their samples, and prevalence estimates. Patterns of Adolescent Dating Violence In addition to examining the overall preva- lence of adolescent dating violence, under- standing how dating violence is distributed is important for the development of theory and promising programming. Unfortunately, stud- ies share few common descriptive variables seeking to capture risk and protective factors. Of the published studies displayed in Table 1, the only variable assessed across all studies was gender. Two studies examined geographic dis- tribution of victimization. Another two studies focusing on perpetration examined justification of the use of violence and observation of vio- lence between parents. Because other variables wereexamined only ina single study,we cannot assess how findings may compare across studies. We first discuss gender. For studies using separate measures of physical and sexual vic- timization, estimates of physical victimization among girls range from 8% to 57% and 6% to 38% among boys. For sexual victimization, esti- mates fallbetween 14%to43%forgirlsand0.3% and 36% for boys. Physical violence perpetra- tion estimates for girls range from 28% to 33% and from 11% to 20% for boys. Estimates of sex- ual violence perpetration among girls range from 2% and 24% and 3% and 37% for boys. See Table 2 for a summary of these studies. RoscoeandCallahan (1985) foundthat 10%of girls and 6% of boys in their midwestern high school sample experienced physical dating vio- lence. In addition, in a midwest high school sample and using a modified CTS instrument, Molidor and colleagues found a much larger percentage of girls and boys reporting ever ex- periencing physical dating violence (36% and 37%, respectively) but less gender disparity in such victimization (Molidor, Tolman, & Kober, 2000).1 Although not assessed by Roscoe and Callahan, Molidor and colleagues did find a considerable gender difference in sexual vio- lence. Of girls, 18% had suffered sexual violence compared to 0.3% of boys. Examining only perpetration, Schwartz and colleagues (Schwartz, O���Leary, & Kendziora, 1997) found a great deal of difference in dating violence in their suburban adolescent sample. On a combined measure of sexual and physical violence, 44% of girls reported perpetrating at least one suchact compared toonly 16% ofboys. Boys were more likely to report committing se- rious acts of violence, such as threats with a gun or knife, beating up a partner whereas girls 126 TRAUMA, VIOLENCE, & ABUSE / April 2004 Although not assessed by Roscoe and Callahan, Molidor and colleagues did find a considerable gender difference in sexual violence. Of girls, 18% had suffered sexual violence compared to 0.3% of boys. at UNIV OF ILLINOIS URBANA on May 22, 2009 http://tva.sagepub.com Downloaded from