Women’s Risk for Revictimization by a New Abusive Partner: For What Should We Be Looking?

Abstract

The purpose of this article was to examine the prevalence of, as well as risk factors for, revictimization by a new partner. Data was collected via face-to-face interviews at Time 1 (about 5 weeks after obtaining a protective order against a violent partner [DVO partner]) and at Time 2 (approximately 12 months later). Of those women who reported having a new partner at Time 2 (n = 412), 35.2% reported abuse. Findings indicate that there is a subset of women who are at greater risk of experiencing abuse by future partners: women with greater cumulative lifetime victimization and those who abuse or are dependent on illicit drugs. Intervening with women when they obtain a protective order is a critical point of intervention to reduce women’s risk for revictimization.

One of the most compelling findings in victimization research is that interpersonal victimization is not random; rather, once a woman has been physically or sexually victimized, she is at increased risk for subsequent victimization (Gidycz, Coble, Latham, & Layman, 1993; Gold, Sinclair, & Balge, 1999; Himelein, 1995; Mayall & Gold, 1995; Sappington, Pharr, Tunstall, & Rickert, 1997; Wyatt, Guthrie, & Notgrass, 1992). Many studies of revictimization focus on sexual victimization in childhood and subsequent sexual revictimization in adulthood (Arata, 2000; Breitenbecher, 2001; Classen, Palesh, & Aggarwal, 2005; Gidycz et al., 1993; Gold et al., 1999; Jankowski, Leitenberg, Henning, & Coffey, 2002; Messman & Long, 1996; Messman-Moore & Long, 2002; Roodman & Clum, 2001; Stermac, Reist, Addison, & Millar, 2002). Childhood physical abuse has also been associated with increased risk of physical adulthood victimization (Weaver, Kilpatrick, Resnick, Best, & Saunders, 1997), and individuals who experience both childhood physical and sexual abuse are at greater risk of adult sexual revictimization compared to individuals who experience only childhood sexual abuse (Cloitre, Tardiff, Marzuk, Leon, & Portera, 1996). In fact, several studies have found that any physical or sexual victimization in childhood is associated with a greater risk of adulthood victimization by a partner for women (Coid et al., 2001; Desai, Arias, Thompson, & Basile, 2002).

Yet there is a significant gap in the research regarding partner violence revictimization. Specifically, few studies have focused on women’s risk for abuse by a new partner after leaving an abusive partner. In retrospective studies, sizeable minorities of women with recent histories of partner violence have been found to have prior histories of partner violence, ranging from 34.7% to 41% (Bogat, Levendosky, Theran, von Eye, & Davidson, 2003; Kemp, Green, Hovanitz, & Rawlings, 1995; Krishnan, Hilbert, & Pase, 2001). Moreover, limited research in this area has found some evidence that a history of partner violence increases women’s risk for violence by another partner (Krishnan et al., 2001). In a national probability sample, Woffordt, Mihalic, and Menard (1994) found that among those women who had experienced violence by a marital or cohabiting partner at Time 1 and who were involved with a new partner 3 years later (Time 2), 49% reported violence by a new partner. However, because only a small percentage of the overall sample reported violence at Time 1 and being involved in a relationship with a new partner at Time 2 (n = 26), the finding must be treated with caution. In another study of women who received advocacy services from a domestic violence shelter, 64% had been involved with new partners in the 6 months before the 3-year follow-up, and 19% of these women reported being assaulted by their new partners (Bybee & Sullivan, 2005).

The research literature suggests there are a number of risk factors for revictimization. For example, cumulative life trauma may be a good predictor of risk for revictimization, likely through the impact on current trauma-related symptoms (Nishith, Mechanic, & Resick, 2000). In other words, individuals who experience more lifetime trauma may have increased psychological vulnerability to revictimization (Classen et al., 2005; Logan, Walker, Jordan, & Campbell, 2006; Nishith et al., 2000). Whether more trauma-related symptoms are related to an increased risk of entering a relationship with another abusive partner after termination of an abusive relationship has received limited attention. Second, numerous studies have found an association of sexual revictimization with substance use or abuse (Classen et al., 2005). Substance use as a coping mechanism to deal with past trauma is hypothesized to increase women’s risk for subsequent victimization (Logan, Walker, Cole, & Leukefeld, 2002; Wills & Filer, 1996). Yet not many studies have examined the effect of prior partner violence and victim substance use on subsequent partner violence. From a random digit dialing sample, among a subset of women who reported having heterosexual relationships at Wave 1, illicit drug use was significantly associated with women’s violent victimization by a subsequent partner 12 months later (Testa, Livingston, & Leonard, 2003). Third, stressors other than those directly due to partner violence may also have a positive association with psychological distress, which may in turn affect women’s appraisals of threatening situations or reduce internal resources (Hobfoll, Freedy, Green, & Solomon, 1996). Research indicates that individuals who encounter a series of stressful situations are less able to adapt positively to subsequent stressful situations because their internal and external resources are depleted (Baumeister, Heatherton, & Tice, 1999; Hobfoll et al., 1996). Fourth, to our knowledge, negative social interactions, or social obstruction, which is defined as “behaviors which impede, block, or delay normal progress through life, and by which people take away from a person those resources needed to proceed on a chosen course of action” (Gurley, 1990, p. 1), has not been examined as a risk factor for women’s vulnerability to revictimization. Negative social reactions may negatively impact victims’ mental health and add to their social isolation (Goodkind, Gillum, Bybee, & Sullivan, 2003; Gurley, 1990).

Less research has been conducted on identifying protective factors for revictimization. One protective factor that has been examined is social support. Higher levels of social support may protect women from revictimization (Bybee & Sullivan, 2002; Carlson, McNutt, Choi, & Rose, 2002; Tan, Basta, Sullivan, & Davidson, 1995). Higher social support has been found to be associated with lower rates of revictimization by an abusive partner during a 2-year follow-up period among a domestic violence shelter sample (Sullivan & Bybee, 1999).

Based on the lack of research describing the prevalence, as well as risk and protective factors, for women with multiple abusive partners, this study has three primary goals. First, this study examined the prevalence of lifetime victimization at Time 1 among a large sample of women with protective orders against a violent male partner. Second, this study examined the prevalence and description of reports of abuse by a new partner at 12-month follow-up (Time 2). Third, this study examined risk and protective factors for women experiencing victimization by a new abusive partner at Time 2.

METHOD

Participants

Study participants were recruited from court when they obtained protective orders, also known as a domestic violence order (DVO) against a male intimate partner (DVO partner) between February 2001 and November 2003 (n = 756). To be eligible for the study, participants had to meet the following criteria: (a) be female, (b) be 18 years and older or 17 and emancipated, and (c) have obtained a protective order within 6 months of entering the study. The average length of time between issuance of the DVO and entry into the study was 40 days. Approximately 12 months after women participated in the first interview, follow-up (Time 2) interviews were conducted. The follow-up rate for the study was 94% (n = 709).

At Time 2, 585 (82.6%) women reported that they had been involved with a partner since Time 1. Specifically, 33.1% (n = 234) had been involved with the DVO partner and 60.0% (n = 425) with a partner other than the DVO partner at some point since Time 1. Some of the women were involved with the DVO partner and other partners since Time 1, 10.4% (n = 74). Because we were interested in examining the association between recent partner victimization and subsequent involvement with a new abusive partner, women who reported at Time 2 that they had been involved with the DVO partner for the entire period since Time 1 (n = 76) were excluded from the analysis. In addition, one woman was excluded from the analysis for missing data on victimization experiences during the follow-up period. Of the remaining sample (n = 632), 65.2% (n = 412) of the women had been involved with a partner other than the DVO partner, with 47.2% (n = 298) reporting current involvement with that partner at Time 2.

Measures

Demographic and socioeconomic questions were taken from the Risk Behavior Assessment (RBA; Coyle, 1993) and were included in both the Time 1 and Time 2 interviews: (a) age, (b) race, (c) education level, (d) past year income; and from pilot study work (Logan, Walker, Cole, Ratliff, & Leukefeld, 2003): (e) current employment status, and (f) number of children.

Partner violence victimization was measured with questions adapted from a study examining incidents reported on protective order petitions (Harrell, Smith, & Newmark, 1993) and from the Conflict Tactics Scales (CTS and CTS2; Straus, 1995; Straus, Hamby, Boney-McCoy, & Sugarman, 1996), and psychological abuse items from Tolman’s Psychological Maltreatment of Women Inventory (PMWI; Tolman, 1989, 1999). Stalking was measured with a question modified from the National Violence Against Women Survey (Tjaden & Thoennes, 1998). Psychological abuse (Cronbach’s alpha = .882), stalking, physical abuse (Cronbach’s alpha = .842), sexual insistence, and threatened/forced sex (Cronbach’s alpha = .819) during the participants’ relationships with their DVO partners as well as in year before Time 1 were measured in the Time 1 interview. The same items were used to measure partner violence by any partner at the Time 2 interview with the reference period, “since the baseline interview.” Additionally, measures of participants’ lifetime experiences of psychological, stalking, physical violence, and sexual violence by a partner prior to the relationship with the DVO partner were included in the Time 1 interview.

In addition, an index on the severity of physical violence by the DVO partner was computed based on weights in the weighted severity index of the CTS (Straus & Gelles, 1990) that were assigned to the physical violence items. The following weights were assigned for acts that occurred in the past year: 1 for twist arm or hair, push or shove, grab, slap; 2 for cause to have an accident on purpose, kick, bit, punch or hit with something, slam against the wall; 5 for beat up, burn or scald on purpose, choke; 6 for threaten with a knife or gun, and try to run down with a car; and 8 for used a knife or fired a gun on the victim. The possible range in scores was 0 to 49. Severe physical violence included any abuse item that received a weight of 2 or more.

Child abuse history was measured in the Time 1 interview with single item questions developed in a pilot study (Logan et al., 2003) about childhood emotional, physical, and sexual abuse by parents or guardians.

Cumulative lifetime victimization was computed based on participants’ reports at Time 1 of the following types of victimization in their lifetime: child emotional abuse by a parental figure, child physical abuse by a parental figure, child sexual abuse by anyone, stalking by someone other than a parent or partner, physical violence by someone other than a parent or partner, sexual violence by someone other than a parent or partner, psychological abuse by a prior partner, stalking by a prior partner, physical violence by a prior partner, sexual violence by a prior partner, psychological abuse by the DVO partner, stalking by the DVO partner, physical violence by the DVO partner, and sexual violence by the DVO partner. Because the literature has consistently shown an association between child apbuse and adult revictimization (Breslau, Chilcoat, Kessler, & Davis, 1999; Lang, Stein, Kennedy, & Foy, 2004), the sum of the types of child abuse was multiplied by three. All affirmative answers to the other types of violence were given a weight of 1. Because all participants experienced at least psychological abuse by the DVO partner, possible scores ranged from 1 to 20. To check the validity of the index, groups were based on the distribution of scores: low, middle, high. These groups were compared on each of the individual types of victimization, and significant differences were found for all the items, except psychological abuse and physical abuse by the DVO partner.

Posttraumatic stress disorder (PTSD) was measured at Time 1 with items taken from the Diagnostic Interview Schedule (DIS; Robins, Helzer, Croughan, & Ratcliff, 1981). Participants were first asked if they had ever experienced any of the qualifying traumatic events. Questions about symptoms of PTSD from the DIS were asked of women who reported a qualifying traumatic event. In addition, this study adapted the PTSD symptom measures to ask participants if they had ever experienced each of the listed symptoms, and if so, if they had experienced those symptoms in the past 30 days. In order to meet criteria for current PTSD, individuals had to have (a) ever experienced a qualifying traumatic event, (b) reported one or more of the re-experiencing symptoms in the past 30 days, (c) reported three or more of the avoidance symptoms in the past 30 days, and (d) reported two or more of the hyperarousal symptoms in the past 30 days (American Psychiatric Association [APA], 2000).

Depression questions were adapted from the Mini International Neuropsychiatry Interview (MINI; Sheehan et al., 1997) and were included in the Time 1 instrument. The original concepts and ordering of the depression questions in the MINI were retained. Because the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV-TR) states that a period of at least 2 weeks of depressed mood or loss of pleasure or interest is the first criterion for a major depressive episode (APA, 2000), the first two questions of the MINI were separated (i.e., [a] feel depressed, sad, low, or down and [b] feel less interested in things or less able to enjoy things you used to enjoy) and were used to determine if the subsequent questions relating to additional depressive symptoms were asked. Additionally, response options for questions were adapted from the original Likert scale ratings into a dichotomous yes/no variable.

Substance abuse or dependence measures were adapted from the Addiction Severity Index (ASI; McLellan, Luborsky, O’Brien, & Woody, 1980) to measure alcohol abuse or dependence and drug abuse or dependence at Time 1. Illicit drugs included marijuana, cocaine, hallucinogens, club drugs, and illicit use of sedatives, opiates, and amphetamines. Participants were asked to report if they had ever used each class of substance, and if so, how many months in the past 12 months. Measures for abuse or dependence were based on the criteria in the DSM-IV-TR (APA, 2000), and were modified from the MINI (Sheehan et al., 1997) using the time frame of the 12 months before Time 1. A few questions were added to better assess interference of substance use with occupational functioning and interpersonal relationships as well as efforts to abstain from substance use consistent with DSM-IV-TR criteria. Participants were classified as meeting DSM-IV-TR criteria for substance abuse or dependence based on their self-reported responses to the questions.

The Daily Hassles Scale is used to measure stress as it is experienced in daily life (Kanner, Coyne, Schaefer, & Lazarus, 1981). “Hassles” are defined as “the irritating, frustrating, distressing demands that to some degree characterize everyday transactions with the environment” (p. 3). Thirty-three items were included (e.g., children, time alone, neighborhood), and participants were asked to rate how much of a hassle each item was. Two items related to partners—sex and intimacy—were omitted from the calculation of the hassles score in order to measure participants’ daily stress that was not directly related to partners. Response options ranged from 0 = None to 3 = A lot. Scores for the hassles scale were computed by summing the responses, ranging from 0 to 93. The scale had a Cronbach’s alpha of .896.

The Social Support and Social Obstruction Scale consists of 22 items composed of two distinct components: 10 items measure social support and 12 items measure social obstruction (Gurley, 1990). The instructions asked participants to think about people other than the DVO partner when answering the items. An example of an item in the social support scale is “I know someone who helps me think through what to do about problems.” An example of an item in the social obstruction scale is, “Other people just don’t hear what I say.” The response options for all the items were the following: 0 = Definitely false; 1 = Somewhat false; 2 = Somewhat true; and 3 = Definitely true. The range of scores was 0 to 30 for the social support scale and 0 to 36 for the social obstruction scale, and both scales demonstrated internal consistency reliability, with Cronbach’s alphas of .885 and .845, respectively.

Procedures

In four jurisdictions, female study interviewers approached women who had received a DVO against a male intimate partner to provide them with information about the research project. A total of 756 women completed the Time 1 interview. Female interviewers collected data from participants in face-to-face interviews, which took an average of 3.5 hours at Time 1. Interviews began after women gave informed consent. Participants were compensated for their time. Each participant was provided with a verbal educational protocol about safety and was given a referral resource pamphlet containing pertinent physical health, mental health, and safety planning resources tailored for the community in which she lived.

Approximately 12 months after women participated in the first interview, follow-up (Time 2) interviews were conducted. The same procedures that were used to collect data from participants at Time 1 were used at Time 2. Interviews at Time 2 were shorter on average than Time 1 interviews (M = 2.25 hours) because questions about lifetime victimization were not included.

Data Analysis

Bivariate analyses were conducted to describe the sample by the revictimization group (i.e., [a] no victimization by a new intimate partner at Time 2 and [b] victimization by a new intimate partner at Time 2), and to inform the selection of variables to be included in the multivariate analysis. Logistic regression was used to examine risk and protective factors associated with revictimization. Decisions about the variables to include in the multivariate analysis were also guided by the literature on factors associated with greater risk for revictimization. Predictor variables included the following: (a) sociodemographic variables at Time 1 (age, area, education, income); (b) cumulative lifetime victimization at Time 1; (c) number of months participant was involved with partner other than DVO partner since Time 1; (d) social support at Time 1 (scores on positive social support and social obstruction scales); (e) daily stress at Time 1 (scores on the Daily Hassles Scale); and (f) mental health issues at Time 1 (depression, PTSD, and substance—alcohol or illicit drug—abuse or dependence). To reduce the likelihood of Type I error the p-value was adjusted to .01 (Lipsey, 1990).

RESULTS

Abuse by New Partner at Time 2

Two groups were formed based on reports of any victimization by a new partner. At Time 2, 150 (23.7%) women reported any victimization by a new partner, and 483 (76.3%) reported no victimization experiences by a new partner, including 71 women who did not have a new partner during this period but had the opportunity to have a new partner because they were not involved with the DVO partner during the entire follow-up period. Of those women who reported having a partner (n = 412) other than the DVO partner during the follow-up period, 35.2% reported any abuse: 34.0% reported psychological abuse, 6.8% reported stalking, 10.7% reported any physical violence, 7.0% reported severe physical violence, and 1.2% reported threatened/forced sex by the new partner.

Sociodemographic Factors at Time 1

About half of the participants (50.5%) were rural residents. The mean age of participants was 31.7 (SD = 9.3), the majority of participants were White (83.1%), 13.4% were Black, and the remaining 3.5% were other races or biracial. The majority of participants (55.5%) were unemployed at Time 1, with 34.8% being employed full-time and 9.7% being employed part-time. Less than one-third of participants (29.3%) had less than a high school diploma or GED. about one-third (33.0%) had a high school diploma or GED, and over one-third (37.7%) had at least some college education. The mean annual income was low: $10,499. Participants had an average of 1.5 (SD = 1.25) children under age 18. There were no sociodemographic differences by group.

Victimization Experiences

Results of the bivariate analysis on the relationship between new partner victimization experiences at Time 2 and child abuse, acquaintance- or stranger-perpetrated violence, prior partner victimization, and victimization experiences in the relationship with the DVO partner are presented in Table 1. Significantly more women who experienced victimization by a new partner reported emotional and sexual abuse by a parent or guardian compared to women who did not report victimization by a new partner at Time 2. No significant differences were found in victimization by someone other than a parent or partner (i.e., acquaintance, other relative, or stranger).

TABLE 1.

Prior Victimization Experiences by New Partner Victimization Group

	Victimization by Partner Other Than DVO Partner Since Time 1		t or χ2 Statistic
	No (n = 482)	Yes (n = 150)
Childhood abuse by a parent or guardian
Emotional abuse	34.1%	48.7%	10.351*
Physical abuse	24.5%	25.3%	.039
Sexual abuse	9.4%	18.0%	8.291*
Lifetime victimization perpetrated by someone other than a parent or partner
Stalking	8.3%	12.7%	2.579
Physical violence	16.2%	22.7%	3.299
Sexual violence	29.9%	34.7%	1.227
Prior partner victimization
Psychological abuse by a partner prior to DVO partner	40.5%	54.0%	8.530*
Stalked by a partner prior to DVO partner	14.7%	26.0%	10.107*
Any physical abuse by a partner prior to DVO partner	36.3%	52.0%	11.736*
Threatened/forced sex by a partner prior to DVO partner	13.5%	23.3%	8.330*
Victimization by DVO partner ever in relationship
Psychological abuse	100%	100%
Stalked	53.1%	67.3%	9.413*
Any physical violence	95.6%	98.7%	2.982
Threatened/forced sex	25.1%	30.7%	1.821
Weighted severity index ever in relationship (0-49)	16.8	19.5	−2.548
Cumulative lifetime victimization index (0-20)	6.4	8.0	18.149**

^*p < .01.

^**p < .001.

More women reporting victimization by a new partner at Time 2 reported a history of partner victimization before the relationship with the DVO partner. Specifically, more women who had experienced victimization by a new partner reported a history of psychological abuse, stalking, any physical violence, and threatened/forced sex by a partner prior to the DVO partner. Abuse by the DVO partner during the relationship was examined by group (see Table 1). All women in the sample reported psychological abuse by the DVO partner, and high percentages of participants in both groups reported any physical violence. The only significant difference by group was found in stalking. More women who had been victimized by a new partner at Time 2 had been stalked by the DVO partner compared to women who did not report victimization by a new partner at Time 2. Scores on the lifetime cumulative victimization index were significantly higher for those women who reported victimization by a new partner at Time 2 than women who did not report victimization by a new partner.

Mental Health Indicators at Time 1

The groups were examined for significant differences in mental health indicators at Time I. About two-fifths of the sample reported experiencing symptoms that met the study criteria for PTSD, and about half reported symptoms meeting study criteria for depression with no differences by group (see Table 2). More than one-third (38.0%) of the women who reported victimization by a new partner at Time 2 met criteria for alcohol abuse or dependence at Time 1, which was twice the percentage when compared to women who reported no new victimization (19.1%). Significantly more women in the revictimized group reported symptoms that met the study criteria for illicit drug abuse or dependence (39.3% versus 16.0%).

TABLE 2.

Mental Health Problems, Stress, Major Life Events, and Social Support at Time 1 by Partner Victimization Group

	Victimization by Partner Other Than DVO Partner Since Time 1		F or χ2 Statistic
	No (n = 482)	Yes (n= 150)
PTSD	38.9%	42.7%	.685
Depression	50.4%	54.7%	.828
Alcohol abuse or dependence	19.1%	38.0%	22.711**
Illicit drug abuse or dependence	16.0%	39.3%	36.959**
Score on Daily Hassles	19.3	22.9	7.830*
Mean score on Social Support Scale	24.7	23.7	3.968
Mean score on Social Obstruction Scale	11.4	13.3	8.470*

^*p < .01.

^**p < .001.

Stress, Social Support, and Social Obstruction at Time 1

Measures of participants’ stresses, social support, and social obstruction are presented in Table 2. Women reporting revictimization by a new partner at Time 2 had significantly higher mean scores on the Daily Hassles Scale compared with those women reporting no victimization by a new partner. Women in the sample reported high levels of social support, with mean scores near the top of the maximum value (30) for the scale, and no differences by group. Scores on the Social Obstruction Scale were lower, and women in the revictimized group reported significantly higher scores than women in the not revictimized group.

Multivariate Analysis

Table 3 presents the results of a logistic regression analysis predicting the odds of women experiencing victimization by a new partner in the follow-up period, time between Time 1 and Time 2 (M = 11.7 months). The results indicate the overall model was statistically reliable (χ² = 114.113, df = 13, p < .001) and the Nagelkerke R² = .254. Results indicate that cumulative lifetime victimization, length of involvement with the new partner, and meeting criteria for drug abuse/dependence were significantly positively associated with partner violence by a new partner. Specifically, women with more cumulative lifetime victimization experiences and women with drug abuse or dependence were significantly more likely to experience victimization by a new partner between Time 1 and Time 2. Further, longer involvement with a new partner increased the likelihood of experiencing victimization by the new partner. None of the indicators of daily stress, social support, or social obstruction were significantly associated with the victimization by a new partner. Moreover, meeting study criteria for PTSD and depression were not significantly associated with victimization by a new partner.

TABLE 3.

Logistic Regression Predicting Partner Violence by a New Partner During the Follow-Up Period

	β	Odds Ratio (CI)	Wald
Age	−.015	.985 (.961, 1.011)	1.305
Area	.128	1.137 (.723, 1.788)	.309
Highest level of education	−.013	.987 (.879, 1.108)	.051
Annual income at Time 1	−.003	.997 (.935, 1.063)	.009
Cumulative lifetime victimization index at Time 1	.078	1.081 (1.023, 1.142)	7.671*
Number of months involved with new partner since Time 1	.168	1.183 (1.129, 1.240)	49.649**
Social Support score at Time 1	−.016	.985 (.943, 1.028)	.495
Social Obstruction score at Time 1	.003	1.003 (.966, 1.042)	.028
Daily Hassles score at Time 1	.016	1.016 (.999, 1.034)	3.462
Depression in the 2 weeks before Time 1	.041	1.042 (.608, 1.548)	.027
Met study criteria for PTSD in the 30 days before Time 1	−.030	.970 (.638, 1.700)	.016
Met study criteria for alcohol abuse/dependence in the year before Time 1	.641	1.899 (1.151, 3.134)	6.305
Met study criteria for illicit drug abuse/dependence in the year before Time 1	.721	2.056(1.221, 3.460)	7.361*

^*p < .01.

^**p < .001.

DISCUSSION

Research suggests that one of the primary considerations for women when deciding whether to terminate a violent relationship is ending the violence (Choice & Lamke, 1997; Logan, Walker, Jordan, & Campbell, 2004). One option available to women to try to reduce or stop the violence is through the use of protective orders (Logan, Shannon, Walker, & Faragher, 2006). Some research has examined women’s risk for future partner violence after seeking a protective order (Carlson, Harris, & Holden, 1999; Kelitz, Hannaford, & Efkeman, 1997; Harrell & Smith, 1996; Holt, Kernic, Wolf, & Rivara, 2003; Kaei, 1994; Klein, 1996; Logan, Shannon, Walker, & Cole, in press; McFarlane et al., 2004; Mears, Carlson, Holden, & Harris, 2001; Ptacek, 1999). Most of the studies found a reduction in violence after women received the protective order, showing that criminal justice system interventions that hold perpetrators accountable for their actions often reduce the violence (Carlson et al., 1999; Holt et al., 2003; Kaei, 1994; Kelitz et al., 1997; McFarlane et al., 2004; Ptacek, 1999).

What has been less researched is women’s risk for revictimization by new partners following termination of the relationship with the person against whom she obtained a protective order. In fact, the time of separation from a violent partner is risky for women, not only because of the risk of continued violence by the partner, but also because ending the violence with one partner does not preclude revictimization by a future partner (Woffordt et al., 1994). The current study findings suggest that a significant minority of women who terminate their relationships with an abusive partner become involved with a subsequent abusive partner. Within this relatively short follow-up period, on average 11.7 months, about one-third (35.2%) of the women who had a new partner reported abuse by that new partner. Study findings emphasize the need for interventions with partner violence victims to address socioeconomic, psychological, and systemic factors that may contribute to increased risk for abuse by any partner. Because of the immediacy of the threat and danger of revictimization by a current or recent abusive partner, interventions may give only cursory, if any, attention to discussion of future intimate relationships and how women can decrease their risk of revictimization by subsequent partners.

Furthermore, this study found three factors that were associated with increased likelihood of being involved in a new abusive relationship. Specifically, women with greater cumulative lifetime victimization, those who abused illicit drugs, and those who had been in a relationship with a new partner for longer were more likely to report abuse by that new partner. Individuals who experience more lifetime trauma may have increased psychological vulnerability to revictimization (Nishith et al., 2000). However, even though one of the hypothesized mechanisms that increases women’s vulnerability for revictimization is psychological distress, in particular trauma-related symptoms (e.g., PTSD, anxiety, dissociative symptoms), the findings from this sample of female protective order recipients did not support this hypothesis. It is possible that trauma-related symptoms may have less to do with partner revictimization than other psychological factors. In a review of the empirical literature on sexual revictimization, Classen et al. (2005) found that revictimized individuals, meaning those who experienced childhood physical and/or sexual abuse as well as adulthood sexual violence, experienced more problems in their interpersonal relationships. Thus, it is possible that women with more extensive victimization histories, not just childhood victimization, experience interpersonal difficulties, which then increase their risk of revictimization. Further research on interpersonal and psychological functioning of women leaving abusive relationships is needed to better understand risk factors for abuse by subsequent partners.

Women’s abuse or dependence on illicit drugs was also significantly associated with victimization by a new partner. Assortative mating has been hypothesized as an explanation for increased risk of victimization among women who use substances (Logan et al., 2002; Logan, Walker et al., 2006). Briefly stated, women who use substances often socialize with other substance users (e.g., informal drug-using networks, in bars, clubs) and thus have an increased likelihood of becoming involved in an intimate relationship with a man who uses substances (Laumann, Ellingson, Mahay, Paik, & Youm, 2004). Research has consistently associated perpetrator substance use, including alcohol and illegal drugs, with a variety of male-to-female partner violence tactics (Fals-Stewart, Golden, & Schumacher, 2003; Leonard, 2001; Lipsky, Caetano, Field, & Larkin, 2005; Moore & Stuart, 2004; Testa, 2004). Therefore, a possible link between women’s illicit drug abuse or dependence and partner revictimization may be because women who use substances have an increased risk of becoming involved with a partner who is a substance user, and thus, are at increased risk of intimate partner violence (IPV) because of their partner’s substance use. Thus, assessment of perpetration of IPV among men in substance abuse treatment is recommended (Fals-Stewart & Kennedy, 2005), although it has been shown to be the exception rather than the rule (Schumacher, Fals-Stewart, & Leonard, 2003). Integration of psychoeducational materials about the use of power, control, and violence in intimate relationships into substance abuse interventions for men is needed to reduce the risk of future partner violence. In addition, substance abuse treatment programs may consider making referrals to batterer treatment programs (Fals-Stewart & Kennedy, 2005).

The association between illicit drug abuse/dependence and subsequent partner revictimization has implications for substance abuse treatment providers as well as service providers who work with IPV victims. The cycle of victimization leading to victims using substances as a coping strategy, which in turn renders them more vulnerable to revictimization, indicates that treatment providers need to address the multiple and interconnected problems of clients to achieve successful outcomes (Schumm, Hobfoil, & Keogh, 2004). Entry into substance abuse treatment and interventions for IPV victims present a unique opportunity for practitioners to intervene with women on multiple issues that may ultimately reduce women’s risk for future victimization and substance abuse (Logan, Walker, et al., 2002, 2006). For example, interventions in treatment facilities should assess for lifetime victimization histories, including IPV, because of the high comorbidity of depression, PTSD, and substance abuse among revictimized women (Logan et al., 2002), and should incorporate therapy and education about interpersonal functioning and beneficial relationship norms. Collaborative interventions across community agencies (i.e., a domestic violence coordinating council) may facilitate the identification of screening, assessment, and intervention methods that could more efficiently address victims’ complex needs and problems (Mears, 2003) by bringing service providers from multiple agencies together to address systemic issues.

Time with the new partner was also associated with victimization. Increased exposure to a new partner increases the opportunity for abuse. This may suggest that more women in new relationships that were shorter may find themselves at risk for abuse by that partner but the 12-month follow-up was not long enough to capture this risk. When examining the trajectory of abuse, it is often typical for the first few months of the relationship to be free of physical abuse and for other kinds of abusive cues to be misinterpreted (e.g., controlling, monitoring, and stalking behaviors; Logan, Cole, Shannon, & Walker, 2006). This is another reason that risk of revictimization, by the initially violent partner as well as new partners, should be discussed with all women exiting abusive relationships.

Social obstruction was associated with revictimization by a new partner in the bivariate analysis, but not in the multivariate analysis, when controlling for other factors. Scores on the social support measure exhibited a possible ceiling effect. It is possible that the variance for this measure was limited, which dampened any actual effect that social support may have had on women’s risk for revictimization.

Limitations of the study must be discussed. First, the correlational data prevent making causal inferences. Second, the purposive sample limits generalizability, even though the use of a protective order sample has strengths, such as obtaining a sample of women with a sufficiently high threshold of violence for a judge to grant a protective order (Logan, Shannon, & Walker, 2005). Third, the data was self-reported and thus is subject to the response biases that are possible with survey methods, such as recall bias and social desirability.

The current study provides information on the trajectory of IPV victimization by examining victims’ subsequent IPV experiences. Findings offer support for the importance of cumulative lifetime victimization on women’s risk for partner revictimization by future partners. Study findings suggest that intervening with women who have extensive victimization histories requires even more attention to risk for future victimization and women’s safety than with women without extensive victimization histories. Additionally, the strong association between illicit drug abuse or dependence and risk for revictimization by a new partner emphasizes the need for incorporating discussion of interpersonal functioning in substance abuse treatment and the assessment of substance abuse for women seeking help for IPV. Intervening with women around the time of obtaining a protective order, or when they are terminating a relationship with a violent partner, is a critical point of intervention to possibly reduce women’s risk for future revictimization. Finally, research on IPV and risk of revictimization needs to expand the examination not only to revictimization by the initially abusive partner but also by new partners.