|Home | About | Journals | Submit | Contact Us | Français|
This study is among the first attempts to address a frequently articulated, yet unsubstantiated claim that sample inclusion criterion based on women’s physical aggression or victimization will yield different distributions of severity and type of partner violence and injury. Independent samples of African-American women participated in separate studies based on either inclusion criterion of women’s physical aggression or victimization. Between-groups comparisons showed that samples did not differ in physical, sexual, or psychological aggression; physical, sexual, or psychological victimization; inflicted or sustained injury. Therefore, inclusion criterion based on physical aggression or victimization did not yield unique samples of “aggressors” and “victims.”
In the field of research on intimate partner violence (IPV), there is growing interest in women’s use of aggression as well as their experiences of victimization (Cook & Goodman, 2006; McHugh & Frieze, 2006; Straus, 2006; Swan & Snow, 2006). Studies that focus on both women’s aggression and/or their victimization often utilize non-probabilistic sampling designs (e.g., convenience sampling) (see meta-analyses by Archer, 2000, 2002). Within non-probabilistic sampling designs, the primary inclusion criterion typically has been either women’s physical aggression or their physical victimization. Little attention has been given, however, to these different inclusion criteria and the characteristics of the resulting samples. Given differences in inclusion criteria, a fundamental methodological question exists: does employing the inclusion criterion of women’s aggression versus their victimization result in statistically different distributions of demographic characteristics and measures of physical, sexual, and psychological IPV and injury in these two samples?
Answers to the aforementioned question have important research and practical implications. First, regarding research implications, findings would affect the ways in which results of research on IPV are interpreted. Results of research with aggression or victimization as the main inclusion criterion may yield study participant samples that are more similar to each other than previously expected, given that women’s physical aggression and victimization are highly correlated (Anderson, 2002; Caetano, Ramisetty-Mikler, & Field, 2005; Field & Caetano, 2005; Hamberger, 2005; Sullivan, Meese, Swan, Mazure, & Snow, 2005). If so, there may be greater external validity in a given study than previously thought regardless of which inclusion criterion was used. If the samples do not differ significantly, investigators and others might then need to interpret IPV study findings based on a different set of assumptions. For example, from a study in which the inclusion criterion is women’s aggression, investigators might attribute findings to the assumption that women in the sample are more aggressive than women included in study samples on the basis of their victimization. In actuality, the same attributions could be made about both samples because each was drawn from a single population of women involved in relationships in which bidirectional IPV occurs. In other words, it is possible that both inclusion criteria generate similar estimates of type and severity of IPV and injury.
Second, regarding practice implications, findings would provide invaluable information to service providers. Providers may make assumptions about women they serve based on the nature of services offered, i.e., whether the services address women’s use of aggression or experiences of victimization. If significant differences between groups by inclusion criteria do not emerge, it would be important for providers to recognize that women who are accessing services because of their aggression likely also are victimized and vice versa. Within the context of social services, for example, where women are seeking mental health or anger management services to address their IPV aggression, findings of non-significant differences by inclusion criteria might suggest that services would be greatly enhanced by the development and utilization of interventions that acknowledge and address women’s victimization in addition to their aggression (i.e., bidirectional IPV). Ultimately, to prevent recurrence of IPV and to enhance women’s and children’s safety, it would be important to integrate IPV services that address both women’s aggression and victimization.
The measurement of IPV is an important topic and one that has sparked much controversy1. One issue in particular is that studies comparing aggression and victimization in intimate relationships often focus only on physical aggression and physical victimization (see Archer, 2000; White, Smith, Koss, & Figueredo, 2000). A more comprehensive understanding of IPV includes sexual and psychological IPV, in addition to physical IPV, given that physical IPV tends to co-occur with other forms of IPV in relationships (Basile, Arias, Desai, & Thompson, 2004; Coker, Smith, Bethea, King, & McKeown, 2000; Smith, Thornton, DeVellis, Earp, & Coker, 2002; West & Rose, 2000). Investigations that include psychological and sexual IPV, as well as other IPV-related experiences such as sustained and inflicted injury could enhance knowledge about the reality of women’s lived experiences. Furthermore, examining IPV comprehensively would allow the determination of whether there are differences between samples on broader measures of IPV than just physical aggression and victimization. Thus, researchers could understand, with greater specificity, risk and protective factors for IPV. This more comprehensive knowledge, in turn, could better inform the development of prevention and intervention programs aimed at reducing and eliminating IPV as well as its negative sequelae such as posttraumatic stress and substance use (Herman, 1993; Wekerle & Wall, 2002).
Findings from two national probability studies are inconsistent regarding the impact of a woman’s racial group on use of aggression and experiences of victimization in intimate relationships (Field & Caetano, 2003; Rennison & Planty, 2003). Field and Caetano found that African American and Latina women were two times more likely to be aggressive than White women and three times more likely to experience victimization than White women after controlling for demographic characteristics and substance use. On the other hand, Rennison and Planty reported that race was no longer a significant predictor of IPV after controlling for the demographics of victim’s gender and annual household income. Given the exploratory nature of the current study, we chose to focus on one racial group: African Americans. Although the research has been inconsistent, there is some evidence that African American women may be at higher risk for IPV compared to other racial groups (Caetano, Field, Ramisetty-Mikler, & McGrath, 2005; Field & Caetano, 2003; Tjaden & Thoennes, 2000).
Extant literature on IPV among African American women suggests that there may be few or no differences between women who are included in studies based on their aggression compare to those included based on their victimization (Caetano, Ramisetty-Mikler et al., 2005). Researchers have offered numerous explanations for this phenomenon. First, regarding women’s use of aggression, research by West and Rose (2000) speculated that since African American women have relatively equal status as men in relationships, women may believe that they have the right to be aggressive if men have been aggressive toward them. Second, Swan and Snow (2006) speculated that a number of factors may increase the likelihood that African American women will use aggression in their relationships. These factors include women’s responsibility to keep the family together, a lack of access to legal and social services, and economic difficulties. Finally, a critical analysis of the literature highlights four themes that might offer insight into the causes of IPV among African Americans; the historical context of enslavement, marginalized socioeconomic status, external barriers such as lack of culturally competent services, and internal barriers such as racial loyalty which may prevent women from reporting IPV (Bent-Goodley, 2001). Taken together, the above research not only provides insight into the unique potential causal factors of IPV among African Americans, but it also illustrates several factors that might explain why African American women tend to endorse levels of their aggression similar to their levels of victimization.
A search of the social science and medical literature produced only one study, Abel (2001), that might provide insight into whether inclusion criterion based on women’s aggression or their victimization generates different distributions of measures of IPV and injury. Abel recruited two samples of women, one sample from a batterer intervention program and another sample from a domestic violence shelter. Results showed that the two samples did not differ significantly on three of the six indicators of victimization, namely, being slapped, witnessing slapping in the home, and witnessing forced sex in the home. The shelter-recruited women reported more instances of being threatened, witnessing threats in the home, and being forced to have sex than the batterer intervention program-recruited women. Because Abel did not report the period for which victimization was assessed, it is unclear if some of the reports reference victimization experiences of earlier adulthood or childhood, e.g., witnessing slapping in the home.
Abel’s (2001) study is relevant to the current investigation in that she recruited women based on their aggression and victimization and examined sexual and psychological victimization in addition to physical victimization. Also salient is that her study showed that batterer intervention program-recruited and shelter-recruited women had comparable levels of victimization on three of the six measures of victimization. In studies of this type, there is an assumption of differences in patterns of aggression and victimization between groups based on the nature of the program or service facility in which the women are involved. The results of Abel’s study suggest that this assumption may be incorrect. These contributions notwithstanding, Abel’s study leaves several questions unanswered. It is unclear whether differences between samples would exist (1) if a community sample rather than a service-utilizing sample was investigated, (2) if the effects of race had been controlled for in the study, (3) if women’s aggression had been measured, (4) if key variables were measured more reliably (e.g., use of validated scales rather than individual items), and (5) if women’s experiences of IPV were measured currently.
To add to the limited research on women’s use of aggression as well as their victimization, the primary aim of this study was to determine if the use of different inclusion criteria was associated with differences in self-reported aggression, victimization, and injury in intimate relationships among African American women. Specifically, does employing the inclusion criterion of women’s aggression or their victimization generate samples whose distributions of physical, sexual, and psychological IPV and injury are statistically different? Given the exploratory nature of the study, no hypotheses are advanced.
Women were recruited from an urban community in New England. Recruitment materials were posted in four urban-area primary care clinics, emergency departments, and local businesses such as grocery stores, laundromats and shops, as well as selected state offices such as the Department of Employment. Recruitment materials did not communicate that the focus of the study was IPV but rather, materials invited women to call if they were in a relationship with a boyfriend or husband. Eligibility was determined by a phone-screen based on the following criteria: (a) female gender; (b) the use of at least one physically aggressive behavior against a male partner within the past six months as measured by selected screening questions from the Conflict Tactics Scale-2 (Straus, Hamby, Boney-McCoy, & Sugarman, 1996); (c) age 18 to 64; (d) residency in the greater urban area; (e) current involvement in a heterosexual relationship of at least six months duration; (f) contact with partner at least twice a week2; and (g) annual household income of less than $50,0003. Data were collected from October 2002 to August 2004. The sample consisted of 412 participants: 150 African American women, 150 Latina women, and 112 non-Latina White women. To address any possible differences by race on the IPV measures, a racially homogenous sample was created. The current study is limited to the 150 women who self-identified as African American. For detailed demographic information see Table 1.
Women were recruited from the same urban community in New England as Study 1. Recruitment materials were identical in content for both studies. Eligibility was determined with the same method and used the same inclusion criteria as Study 1 with the following two exceptions: (a) the main inclusion criterion was that women had to have experienced at least one act of physical victimization by a male partner within the past six months as measured by selected screening questions from the Conflict Tactics Scale-2 (Straus, Hamby, & Warren, 2003), and (b) women could not have spent more than two full weeks apart from their partners during that time. Data were collected from August 2004 to March 2006; there was no overlap in data collection with Study 1. The sample consisted of 212 participants of which 142 self-identified as African American. To maintain comparability with Study 1, only African American women were included.
To ensure that Study 1 and Study 2 samples were independent, data from 14 women who participated in both studies were deleted from the Study 2 sample. Therefore, 128 African-American women who participated only in Study 2 comprise its final sample. For detailed demographic information see Table 1.
The two studies employed similar interview and data-collection procedures. Both employed a cross-sectional, self-report study design and were approved by the first author’s institution’s Human Investigation Committee. For both studies, an eligible participant met with a trained female interviewer at a community-based agency. Interviews were conducted in the same rooms for both studies. In the aggression-inclusion study, women met with a female interviewer of the same race, whereas in the victimization-inclusion study, the race of the interviewer was not matched with the race of the participant.4 An interviewer administered a two-hour, computer-assisted protocol to the participant (NOVA Research Company, 2003). After completion of the interview, the participant was debriefed, remunerated $50, and provided with a list of community resources such as those for domestic violence, employment, food, and benefits assistance, mental health therapy, and substance abuse treatment.
Both studies gathered similar demographic information and utilized identical measures to assess the extent of IPV experiences and injury. The response options for some measures differed between studies. The following section describes the processes used to (re)code variables compared between samples.
The physical abuse, psychological abuse, and injury subscales from The Conflict Tactics Scale-2 (CTS-2) (Straus et al., 1996; Straus et al., 2003)5 were used in both studies to assess women’s experiences of aggression and victimization during the past six months. The CTS-2 is a 78-item self-report measure which assesses physical, sexual, psychological aggression and victimization as well as inflicted and sustained injury and negotiation tactics. The CTS-2 has been used in hundreds of studies on IPV and is shown to have acceptable reliability and validity (Straus et al., 2003). Participants in the victimization-inclusion study were asked to select from the standard CTS response options (never, once, twice, 3 – 5 times, 6 – 10 times, 11 – 20 times, and more than 20 times) to report the frequency with which they and their partners used different aggressive tactics against each other during the past six months. The response options for the aggression-inclusion study were slightly modified from the original CTS response options such that the last two options (i.e., 11 – 20 times and more than 20 times) were collapsed into one (i.e., more than 10 times). The CTS response scales for the victimization-inclusion study were then recoded to match the abbreviated response scales for the aggression-inclusion study. According to the procedures identified by Straus, Hamby, and Warren (2003), the response categories presented as a range were recoded so that the midpoint of each range was the variable’s value (i.e., 3 - 5 = 4; 6 - 10 = 8); more than 10 times was conservatively recoded to a value of 11.
To create more reliable measures than the single item measures employed by Abel (2001), multiple-item scales were created. CTS-2 subscale scores were created by summing the items on each scale. The scores for physical aggression (aggression-inclusion: Cronbach’s α = .87; victimization-inclusion: Cronbach’s α = .89) and victimization (aggression-inclusion: Cronbach’s α = .93; victimization-inclusion: Cronbach’s α = .90) each were created by summing the 12 items on those scales. The scores for women’s use of psychological aggression (aggression-inclusion: Cronbach’s α = .75; victimization-inclusion: Cronbach’s α = .84) and psychological victimization (aggression-inclusion: Cronbach’s α = .80; victimization-inclusion: Cronbach’s α = .82) each were created by summing the eight items on those scales. The scores for women’s inflicted injuries (aggression-inclusion: Cronbach’s α = .68; victimization-inclusion: Cronbach’s α = .69) and women’s sustained injuries (aggression-inclusion: Cronbach’s α = .70; victimization-inclusion: Cronbach’s α = .73) each were created by summing the six items on those scales.
For the purposes of post-hoc analyses, the CTS-2 subscales for physical aggression, psychological aggression, and injury were broken down further according to severity (Straus et al., 2003). Of the twelve physical aggression items, seven are categorized as severe and five as minor. Of the eight psychological aggression items, four are categorized as severe and four as minor. Of the six injury items, four are categorized as severe and two as minor.
To gain additional and more specific information about unwanted sexual experiences, the Sexual Experiences Survey (SES) was used to assess unwanted sexual contact, attempted rape, sexual coercion, and completed rape (Koss, Gidycz, & Wisniewski, 1987). The SES was modified for the purposes of both the aggression-inclusion and victimization-inclusion studies to assess women’s sexual aggression in addition to their sexual victimization during the past six months. Given that the SES has been used largely with college populations and requires a fairly high reading level, the measure was revised to facilitate comprehension among study participants. For example, a standard SES item reads, “have you had a man attempt sexual intercourse (get on top of you, attempt to insert his penis) when you didn’t want to by threatening or using some degree of force (twisting your arm, holding you down, etc.) but intercourse did not occur?” Rather, in both studies, a brief introduction to the section of the measure that queries attempted forced sexual intercourse explained “the next set of questions asks about when your partner tried to insert his penis but the sex did not happen.” This explanation was followed by, for example, “has your partner tried to make you have sex by using force like twisting your arm or holding you down, or by threatening to use force?”
The SES is a widely used measure that has demonstrated good reliability and validity (Koss et al., 1987). The response options and coding scheme were the same as those used for the CTS-2. For the use of sexual aggression scale, Cronbach’s α = .84 in the aggression-inclusion study, and Cronbach’s α = .73 in the victimization-inclusion study. For the sexual victimization scale Cronbach’s α = .91 in the aggression-inclusion study and Cronbach’s α = .89 for the victimization-inclusion study. No post-hoc analyses were conducted to compare samples by SES severity given that minor versus severe distinctions parallel to the CTS-2 are not made on the SES.
Independent samples t-tests were performed to examine between groups differences on measures of physical, sexual, and psychological aggression and victimization, inflicted and sustained injury, and measures of demographics that were continuous in nature (i.e., age and income). Chi-square tests of independence were used to determine if there were group differences on the categorical variables (i.e., employment, education, relationship duration, and cohabitation status). Because multiple comparisons were made, Bonferroni adjustments were employed to reduce Type I error.
Detailed demographic descriptions of each sample are noted in Table 1. No significant differences between the aggression-inclusion and the victimization-inclusion samples were observed using the adjusted Bonferroni significance criteria.
Results from the independent-samples t-tests comparing the physical, sexual, and psychological aggression and victimization scores as well as the inflicted and sustained injury scores for aggression-inclusion and victimization-inclusion samples are presented in Table 2. There were no significant differences between the samples in their scores for (a) physical, sexual, or psychological aggression; (b) physical, sexual, or psychological victimization; and (c) inflicted or sustained injury using the adjusted Bonferroni significance criteria.
Post-hoc analyses were conducted to determine if acts labeled as minor versus severe according to the CTS-2 coding scheme differed between the two samples (Straus et al., 2003). Independent-samples t-tests were conducted to compare physical and psychological aggression and victimization severity scores as well as the injury severity scores between the two samples. There were no significant differences on any of these measures (see Table 3)6.
The current study extends the work of Abel (2001) by comparing community samples of African American women to determine if women’s reports of aggression, victimization, and injury differ based on the inclusion criterion of women’s physical aggression or their victimization. There is a preconceived notion that relatively non-overlapping groups of aggressors and victims exist; however, results of the current study show that among these samples of women recruited from an urban community, this is not the case. The current study illustrated that the two samples of participants were more similar to each other than previously expected, in that the two groups of women had similar demographic characteristics and profiles of IPV. These findings suggest that investigators may need to consider different assumptions when interpreting IPV study results. For example, when women’s aggression is used as the inclusion criterion of a study, one might believe that the research findings reflect women who are more aggressive as opposed to women who are more victimized and therefore, should be generalized to women who are aggressive only. In fact, the findings may generalize to both aggressors and victims because the samples were drawn from a single population of women involved in relationships in which bidirectional IPV occurs. Similarly, service providers, including mental health, medical, and criminal justice professionals, need to be trained that a victim also may be an aggressor and vice versa, for the purposes of providing their clients with appropriate and comprehensive interventions and support services.
This study has limitations worthy of note. First, data on women’s aggression and their victimization were self-reported. It is possible that the reports of women’s partners might differ from the women’s on measures of aggression or victimization; however, research has shown that women’s reports are reliable (Caetano, Schafer, Field, & Nelson, 2002; Magdol, Moffitt, Caspi, & Silva, 1998). Second, both the aggression-inclusion and victimization-inclusion samples were homogenous in terms of socio-economic status, race, and geographic area; they were low-income, African American women living in an urban environment. Although this demographic homogeneity was intentional, it limits generalizability. This study should be replicated with women of different socio-economic status, from other racial and ethnic groups, and those who live in other geographic areas. Third, given that recruitment and participation was voluntary in both studies, it is likely that the most severely abused and coercively controlled women were not represented in either sample. Last, given that non-significant differences were determined using the adjusted Bonferroni significance criteria, and the p < .05 for comparisons of sexual aggression and victimization, a question still exists about whether or not sexual aggression and victimization are different between samples. Future research should investigate this issue further.
These limitations notwithstanding, this study is a unique contribution to IPV research because it focused solely on community-based samples of women, rather than service-utilizing or treatment-seeking samples that are more frequently represented in the literature. A strength of this study also is the participation of racial minority women, a population that until recently has been relatively under-studied as a focal group. A final strength is that a wide range of measures of IPV-related behaviors were included such as sexual and psychological measures as well as injury; this measurement approach permits a broader examination of relationship violence, which, in turn, may better inform the development of community-based IPV prevention programs and interventions. Building from these strengths while addressing the limitations are avenues for researchers to more rigorously test the effects, if any, inclusion criteria have on additional measures of women’s wellbeing.
The research described here was supported, in part, by grants from the National Institute on Drug Abuse (R03 DA17668, K23 DA019561) National Institute of Justice (2001-WT-BX-0502) and the University of South Carolina Research Foundation.
Tami P. Sullivan, Ph.D. is an Assistant Professor in the Division of Prevention and Community Research and Director, Family Violence Research and Programs, Department of Psychiatry, Yale University School of Medicine. Dr. Sullivan’s program of research focuses on understanding (a) precursors, correlates, and outcomes of women’s victimization and their use of aggression in intimate relationships, and (b) the co-occurrence of IPV, posttraumatic stress, and substance use with specific attention to daily processes and intensive longitudinal data. She is particularly interested in risk and protective factor research that informs the development of interventions to be implemented in community settings.
Jennifer Titus, M.P.H. is a Project Director at the Institute for Juvenile Research in the University of Illinois at Chicago’s Department of Psychiatry. She holds a Master’s degree in Public Health from Yale University and a Bachelor of Arts in Psychology and Sociology from Washington and Lee University. Her current research interests include the social and economic determinants of health, psychiatric epidemiology, and the design and evaluation of interventions to promote child and adolescent health.
Laura J. Holt, Ph.D., is an Assistant Professor of Psychology at Trinity College in Hartford, CT. Her program of research focuses on the identification of risk and protective factors for problem behaviors in adolescents and adults, such as academic failure and high school dropout, substance abuse and dependence, and interpersonal violence. She also examines the extent to which community- and school-based universal and selective preventive interventions can impact adolescents’ academic and psychosocial trajectories.
Bonnie S. Fisher is a Professor in the Division of Criminal Justice at the University of Cincinnati. She is currently coediting the Encyclopedia of Victimology and Crime Prevention with Professor Steven Lab and writing two books about sexual victimization in the ivory tower and the social construction of campus crime. Her current research includes examining the scope and nature of alcohol-and drug enabled sexual assault among college women, recurring sexual and violent victimization, methods and measures used to study victimization, and comparative cross-national analysis of violence against women.
Suzanne C. Swan, PhD, is an assistant professor in the Department of Psychology and the Women’s & Gender Studies Program at the University of South Carolina. She received her PhD in social and personality psychology from the University of Illinois. Prior to coming to the University of South Carolina, she was the director of Family Violence Programs at the Yale School of Medicine’s Department of Psychiatry. She conducts research in the area of intimate partner violence, with particular interests in women who use violence in intimate relationships, and preventive interventions with college students to reduce interpersonal violence on college campuses.
David L. Snow, Ph.D. is a Professor of Psychology in Psychiatry, Child Study Center, and Epidemiology & Public Health at Yale University School of Medicine, and is Director of The Consultation Center and Division of Prevention and Community Research in the Department of Psychiatry. His work has focused extensively on the design and evaluation of preventive interventions and on research aimed at identifying key risk and protective factors predictive of psychological symptoms, substance use, and family violence. He also has special interests in the protective and stress-mediating effects of coping and social support and in methodological and ethical issues in prevention research.
1A number of IPV scholars have highlighted the IPV measurement controversy. The controversy is centered on the measurement of IPV by assessing only the frequency of physically aggressive acts. A discussion of this very important issue is beyond the scope of this manuscript; interested readers are referred to Belknap and Melton (2005) Johnson (2006) and Miller and Meloy (2006) for information on this topic.
2This was determined a priori to methodologically control for variability in IPV frequency/severity that may be associated with frequency of contact with her intimate partner.
3This was determined a priori to methodologically control for the differential resources associated with higher income.
4Independent samples t-tests were conducted to determine if, in the victimization-inclusion study, there were differences in aggression, victimization, or injury based on whether or not the interviewer’s race was the same or different from the participant’s. Results showed that for physical, sexual, and psychological aggression and victimization, as well as for inflicted and sustained injury, no significant differences emerged between participants interviewed by a same-race interviewer and those who were not.
5The CTS-2 measure in Straus et al., 1996 is the same exact measure as in Straus et al., 2003; the more recent citation refers to the measure and manual as they are owned by Western Psychological Services.
6Even though, according to the adjusted Bonferonni significance criteria, there were no demographic differences between samples, all analyses also were conducted comparing samples on measures of IPV and injury controlling for employment since p < .05. Still, no significant differences emerged between samples.
Tami P. Sullivan, 389 Whitney Avenue, Department of Psychiatry, Yale University School of Medicine, New Haven, CT 06511, Phone: (203) 789-7645, Fax: (203)562-6355.
Jennifer A. Titus, Department of Psychiatry, University of Illinois at Chicago, Chicago IL, 654 1/2 West Oakdale Ave., Apt. 1E, Chicago, IL 60657, Cell phone: (845) 242-6751, Work phone: (312) 996-9510, Work fax: (312) 413-1036.
Laura J. Holt, Trinity College, Hartford, CT, Department of Psychiatry, Yale University School of Medicine, New Haven, CT, 300 Summit Street Hartford, CT 06106, Phone: (860) 297-4019, Fax: (860) 297-2538.
Suzanne C. Swan, Department of Psychology and Women’s and Gender Studies Program, University of South Carolina, Columbia SC 29208, Phone 803-777-2558, Fax 803-777-9558.
Bonnie S. Fisher, Division of Criminal Justice, University of Cincinnati, Cincinnati OH, 600 Dyer Hall, ML 210389, Cincinnati, OH 45221, Home phone: (513) 871-3544, Fax: (513) 556-3303.
David L. Snow, 389 Whitney Avenue, Department of Psychiatry, Yale University School of Medicine, New Haven, CT 06511, Phone: (203) 789-7645, Fax: (203)562-6355.