Click for next page ( 84


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 83
5 Identifying and Treating Offenders ver the past four decades, efforts to determine what motivates ) men to commit crimes of violence against women have not led to ~ definitive answers. While past research focused on sexual assault, recent research examines battery and homicide against intimate partners. New evidence from longitudinal studies connects developmental prob- lems, such as early conduct disorder, to later violence against women (Moffitt et al., 2001~. In addition, researchers are formulating theories to categorize rapists, batterers, and other types of individuals who victimize women and others and exploring what treatments may help prevent of- fenders from repeating their behaviors. CHARACTERISTICS OF OFFENDERS Motivations of Men Who Rape Explanations for why men rape range from individual determinants- such as physiology and neurophysiology, personality traits, attitudes, al- cohol abuse, and psychopathology to sex and power motives, social learning theories, dyadic contexts, institutional influences, societal influ- ences, and multifactorial models (for a review see National Research Council, 1996~. The two most common explanations are that rape is sexu- ally motivated or motivated by a desire for power. However, it has gener- ally been accepted that multiple causal factors are involved in sexually aggressive behavior (National Research Council, 1996~. Most researchers and clinical investigators agree that sexual offend- 83

OCR for page 83
84 RESEARCH ON VIOLENCE AGAINST WOMEN ers are a heterogeneous group with differing motivations. The develop- ment of taxonomies based on these differences may enhance the efficacy of clinical decisions and provide guidance for the study of etiology, re- cidivism, and the life-span adaptation of sexual offenders. Knight and Prentky (1990) developed one such taxonomy of rapists by identifying specific categories of offenders common to available typologies and de- termining type-defining variables. Their model identifies nine types of rapists falling into one of four descriptive summary categories based on motivation: opportunistic, pervasively angry, sexual, and vindictive. For opportunistic rapists, sexual assault was found to be an impul- sive, usually unplanned, predatory act, occasioned more by contextual and immediately antecedent factors than by protracted or stylized sexual fantasy. These offenders are likely to have a life history of poor impulse control. Pervasively angry rapists are described as motivated primarily by nonsexualized, undifferentiated anger. They show gratuitous aggres- sion, directed equally toward men and women. Sexually motivated rap- ists are characterized as driven by either protracted sexual or sadistic fan- tasies or preoccupations. Sadistic types may be overt or muted, depending on the level of aggression or physically damaging behavior they exhibit during the rape. For nonsadistic types, rape was found to be a result of sexual arousal, distorted views of women and sex, and feelings of inad- equacy regarding the perpetrator's sexuality and masculine self-image. Finally, Knight and Prentky's (1990) taxonomy describes vindictive rap- ists as intending for the sexual assault to result in physical harm, degrada- tion, and humiliation of the victim. Their anger and aggression are cen- tered primarily on women. Motivations of Men Who Batter In most studies, wife batterers are treated as a homogeneous group. Recent research, however, indicates that men who exhibit violence toward their spouses are heterogeneous along theoretically important dimen- sions. Thus, developing and comparing subtypes of violent men and un- derstanding how each subtype differs from nonviolent men may help in identifying the various underlying processes that contribute to violent behavior. Holtzworth-Munroe and Meehan (2002) describe a typology that in- tegrates several theories of aggression into a developmental model of dif- fering types of violence by husbands (see also Holtzworth-Munroe and Stuart, 1994~. That model highlights the importance of correlates of male violence, including both historical correlates violence in the family of origin and association with delinquent peers, gleaned from the broader literature on delinquency and proximal correlates individual charac-

OCR for page 83
IDENTIFYING AND TREATING OFFENDERS 85 teristics of violent behaviors. According to this model, batterers can be classified into three categories according to severity/frequency of marital violence, generality of violence (marital only or extrafamilial), and the batterer's psychopathology or personality disorders: Family-only batterers are predicted to engage in the least marital violence, the lowest levels of psychological and sexual abuse, and the least violence outside the home. Holtzworth-Munroe and Stuart (1994) iden- tify little or no psychopathology in this group. Dysphoric/borderline batterers are predicted to engage in moder- ate to severe wife abuse, but little abuse outside of the family. Violent/antisocial batterers are predicted to engage in high levels of marital violence and the highest levels of extrafamilial violence crimi- nal behavior, arrests, and substance abuse. Holtzworth-Munroe and Stuart (1994) theorize that family-only batterers engage in aggressive acts as a result of a combination of stress (personal and/or marital) and low-level risk factors, including childhood exposure to marital violence and a lack of relationship skills. After such incidents, however, the men's low levels of psychopathology and related problems (low impulsivity, low attachment dysfunction), combined with their lack of hostility toward women and lack of positive attitudes toward violence, lead to remorse and prevent an escalation of the aggression. In contrast, Holtzworth-Munroe and Stuart hypothesize that dyspho- ric/borderline batterers come from backgrounds of parental abuse and rejection and therefore have difficulty forming a stable, trusting attach- ment with an intimate partner. Thus, these individuals tend to be highly jealous and dependent on, yet fearful, of losing their wives. Moreover, they are likely to be impulsive, lack marital skills, and have attitudes that are hostile toward women and supportive of violence. This group re- sembles batterers studied by Dutton (1995), who suggests that these men's early traumatic experiences lead to borderline personality organization, anger, and insecure attachment, which, when the men become frustrated, result in violence against their adult attachment figure the wife. Finally, generally violent/antisocial batterers are predicted to re- semble other antisocial, aggressive groups. They are expected to have ex- perienced high levels of violence in their family of origin and association with deviant peers. They are impulsive, lack skills (marital and nonmarital), have hostile attitudes toward women, and view violence as acceptable. For this category of batterers, marital violence is conceptual- ized as part of their general use of aggression and engagement in antiso- cial behavior. Later research by Holtzworth-Munroe et al. (2000) supports this ty-

OCR for page 83
86 RESEARCH ON VIOLENCE AGAINST WOMEN pology and adds a fourth subgroup low-level antisocial which falls between the family-only and generally violent/antisocial groups. This group exhibits moderate levels of antisocial behavior, general violence, and marital violence, similar to the levels predicted for the family-only group. Recent data suggest that such typologies make it easier to understand and predict the outcomes of treatment. Tones and Gondolf (in press) re- port that although reassault following treatment is not predicted by batterer personality profiles, it is predicted by a history of arrest at intake and drunkenness during the follow-up period. Gondolf and White (2001) identify a wide variety of personality profiles among repeat reassaulters and note subclinical or low levels of personality dysfunction among the majority of these individuals. A clinical implication of batterer typologies such as that described by Holtzworth-Munroe and Meehan (2002) is that treatment outcomes might be improved by matching interventions to types of batterers. For example, in a comparison of two treatment models, Saunders (1996) found that of- fenders with dependent personalities had significantly lower rates of recidivism than those found in other types of offenders in process- psychodynamic treatment groups, and that offenders with antisocial per- sonalities had lower rates of recidivism in structured, feminist-cognitive- behavioral groups. Furthermore, it may be important to consider the ap- plicability of interventions developed in other fields, such as criminal justice. For example, the potential usefulness of intensive rehabilitation supervision close monitoring of offenders in the community, along with rehabilitation focused on criminogenic needs is receiving attention in work with antisocial groups (Gendreau et al., 1994~. Holtzworth-Munroe and Meehan (2002) recommend that prospective, longitudinal studies be conducted to identify the developmental pathways that result in different types of violent husbands. Such studies would al- low researchers to examine the constructs assumed to predict the use of violence among samples of adolescents or children, and to then observe the relationship between those variables and the emergence of relation- ship violence as study participants enter intimate relationships. For ex- ample, the Dunedin Multidisciplinary Health and Development Study identified developmental pathways to violent behavior, including part- ner violence (see Moffitt et al., 2001~. Data from this study suggest that antisocial behavior is predictable across time. A developmental history of conduct problems was found to be the strongest predictor of perpetration of intimate-partner violence for both males and females (Moffitt and Caspi, 1998~. This was true even when such a history was compared with previously established risk factors, such as low family socioeconomic sta- tus, conflicted early family relations, and weak childhood cognitive func-

OCR for page 83
IDENTIFYING AND TREATING OFFENDERS 87 tioning and educational difficulties. Such longitudinal cohort studies can eliminate possible biases of retrospective studies, such as faulty memory. The committee believes that future research should focus on learning more about perpetrators of violence generally and use this information as the context for specific studies of domestic violence. In the past, re- search on domestic violence has not taken theories of general violence into account, focusing instead on theories of patriarchy and power rela- tions without considering possible explanatory variables from other dis- ciplines or from the longitudinal literature on the development of violent behavior (National Research Council and Institute of Medicine, 1998~. Research should take into account the possibility that battering is caused by a more complex set of hierarchical influences, such as weak social controls, situational arousal, and psychopathology, that cause violence in general (Fagan, 1996~. The batterer typology described by Holtzworth- Munroe and Meehan (2002) shows how research on violence in general can inform research on male batterers by using a more inclusive set of variables. In addition, social ecological studies (as discussed in Chapter 3) might inform future research on batterers by linking batterer typologies with research on the social conditions that may shape various battering behaviors. Links Between Women's Victimization and Offending Research has shown that women who engage in violence are often victims of violence themselves. According to an analysis by the Bureau of Justice Statistics (Harrow, 1999), 45 percent of women in state prisons who had committed a violent offense reported that they had been physically or sexually abused at some time during their lives. And 37 percent of female offenders (probationers and federal and state prison and jail inmates) re- ported they had been physically or sexually abused before age 18 a sig- nificantly higher percentage than is reported for the general population. In some cases, violence may be the only immediate response to dan- ger for a threatened woman. However, Moffitt et al. (2001) report a large number of supporting analyses suggesting that women's violence against partners cannot be completely explained as defensive. In the Dunedin sample, for example, women who had engaged in partner violence were 4.4 times more likely than male offenders to have committed a violent crime against someone other than their partner. Both theoretical and empirical models need to be expanded to enable a better understanding of such female-perpetrated violence, including any commonalities this violence may have with that perpetrated by men, as well as the violent victimization experienced over the life course by women who engage in violent behavior. Research designs should include

OCR for page 83
88 RESEARCH ON VIOLENCE AGAINST WOMEN an examination of the environments, as well as the cultural context, of violent women. More research is also needed on the increased rates of incarceration of women, changes in the percentages of women in prisons and jails for drug-related and other nonviolent offenses and for violent offenses, and the implications of these changes (Kruttschnitt and Gartner, 2003). INTERVENTIONS The fields of domestic violence and sexual assault have developed separately, with distinct service and research traditions. One reason for this separation is the tendency of programs for sexual offenders to target a greater variety of offender types (rapists and child molesters within and outside the family, juvenile and adult offenders). Despite this separation, the two fields share some similar treatment orientations, innovations, and controversies. As noted throughout this report, researchers and practitio- ners might benefit from placing their work in the broader context of vio- lent criminals in general and discussing domestic violence and sexual as- sault as subtypes of general criminal offending. Types of Treatment Programs Programs for both domestic violence offenders and sex offenders tend to integrate several approaches through a number of phases that first ex- pand offenders' definitions of abuse and hold them responsible for it, and then teach them alternative reactions and behaviors (Gondolf, 1997; Free- man-Longo et al., 1995~. The most common format for delivering services is men's groups (Marshall, 1999), followed by individual counseling (Pirog-Good and Stets-Kealey, 1985~. Saunders and Hamill (2002) report that cognitive-behavioral ap- proaches are especially prevalent in sex offender treatment programs, which emphasize highly specific techniques such as relapse prevention (Burton and Smith-Darden, 2001; Marshall and Serran, 2000~. Such pro- grams are also more likely than domestic violence treatment programs to explore and attempt to resolve past traumas, with more than 70 percent of the former programs focusing on childhood victimization (Burton and Smith-Darden, 2001; Freeman-Longo et al., 1995~. Most interventions can be classified along several dimensions based on their underlying assump- tions. Table 5-1 summarizes key features of treatment programs for do- mestic violence and sexual assault offenders.

OCR for page 83
IDENTIFYING AND TREATING OFFENDERS TABLE 5-1 Types of Offender Treatment Programs Program Type Description 89 Underlying Assumption Domestic Violence Skills training Group members rehearse Offenders have socially learned positive behaviors to build behavioral deficits and skills. behavioral excesses. Cognitive Attempts to restructure faulty Faulty patterns of thinking lead approaches patterns of thinking and build to negative emotions, which in awareness of core belief turn lead to abusive behavior. systems developed in childhood. Sex role Helps men see the negative Rigid socialization of constricted resocialization effects of constricted male roles male roles results in male Building awareness of control tactics Family systems approaches and the benefits of gender equality (Saunders, 1984~. Helps men take ownership of their intentions to control others (Pence and Paymar, 1994~. Analyze and change communication patterns of couples. dominance. Abusive men use isolation demeaning language, control of finances, and other means to control their victims. Couples unknowingly engage in repeated cycles of interaction that may culminate in abuse (Neidig et al., 1984~. Trauma-based Attempt to resolve childhood Abusive men cannot empathize approaches traumas, in particular the well with others because they traumas of witnessing parental are cut off from their own Sexual Assault violence and being physically abused by parents (Browne, Saunders, and Staecker, 1997~. painful memories of childhood traumas. Comprehensive evaluation and risk assessment Psycho-social-sexual education Analyze abusive behavior to identify warning signs and prevent future offending. Program staff use a structured curriculum to work with the offender and family members to prepare him for treatment and enhance the ultimate impact of other intervention components. Offenders engage in a chain of thoughts, feelings, stimuli, and behavior that leads to an offense. Continued

OCR for page 83
So TABLE 5-1 Continued RESEARCH ON VIOLENCE AGAINST WOMEN Program Type Description Underlying Assumption Process treatment Sex offense-specific individual, group, and family therapies are used to achieve honesty, responsibility, empathy, and remorse. Focused treatment Trains the abuser to gain Offenders have low impulse control over specific acts of control in specific situations. offending and to rehearse the skills necessary to maintain recovery. Behavioral Monitors and controls the Opportunity increases the supervision abuser's everydayliving likelihood of offending. environment to minimize his opportunity to reoffend and teach him internal control. Case management Manages interventions with the abuser along the continuum of care so that accountability and participation in offense-specific treatment are ensured, and the offender's placement in the least-restrictive safe setting is maintained. Medication Some form of medication, Medication can reduce sexual usually antiandrogens, is used urges in the most compulsive in about half the programs in offenders. conjunction with other methods (Burton and Smith-Darden, 2001~. SOURCE: Saunders and Hamill (2002~. Outcome Studies Research on treatment outcomes has only begun to emerge and is in general of poor quality. Past evaluations have been nonexperimental and therefore unable to rule out nontreatment effects, or have used reports of official records, which are likely to greatly underestimate rates of vio- lence. Many studies also have not included sufficient follow-up periods to measure recidivism or other long-term outcomes, nor have such outcomes been uniformly identified, conceptualized, and operationalized (problems

OCR for page 83
IDENTIFYING AND TREATING OFFENDERS 91 with evaluation studies are discussed by Gondolf, 2001, 1997; Marques, 1999; McConaghy, 1999; Holtzworth-Munroe et al., 1995; Tolman and Edleson, 1995; Hamberger and Hastings, 1993; and Rosenfeld, 1992~. Domestic Violence Saunders and Hamill (2002) report that few of the more than 30 stud- ies of program effectiveness they reviewed had rigorous designs that would allow for firm conclusions (see reviews by Babcock et al., in press; Davis and Taylor, 1999; and Tolman and Edleson, 1995~. More-rigorous evaluations tend to use experimental designs and longer follow-up peri- ods (e.g., 1 to 4 years). These studies have suggested that the structure of treatment groups and the length of treatment may influence the effective- ness of treatment (Edleson and Syers, 1991, 1990), that treatment over a longer time span may be more effective than that over a shorter period (Davis et al., 2000), and that the success of certain treatments may depend on the typology of the offenders (Saunders, 1996~. Experimental studies have revealed no differences in outcome between couples' groups and men's groups, although both have been associated with significant reduc- tions in abuse (O'Leary et al., 1999; Brannen and Rubin, 1996~. Nor have differences in outcome been found among four types of treatment involv- ing male naval personnel: cognitive-behavioral men's groups, cognitive- behavioral "quasi-couples groups" (low partner attendance), rigorous monitoring, and stabilization and safety planning (pretreatment screen- ing, safety planning, and referrals for women) (Dunford, 2000~. Meta- analyses using partner and official reports indicated that the effects of treatment were small (Babcock et al., in press; Levesque, 1998~. Sexual Abuse Saunders and Hamill (2002) report the results of several recent meta- analyses of outcomes of sex offender treatment programs. The most rigor- ous such analysis, coordinated by the Association for the Treatment of Sexual Abusers, involved 42 studies of psychological treatment in both community and institutional settings (Hanson et al., 2000~. Of those who received treatment, 12 percent had committed another sexual offense, ver- sus 18 percent in comparison groups; for any type of offense, the rates were 29 and 42 percent, respectively. For the 15 most rigorous studies, the sexual offense recidivism rates were 10 percent for the treated and 17 per- cent for the untreated offenders. Both institutional and community-based treatments were associated with lower rates of recidivism. Many programs supplement individual and group therapy with medications that lower the level of testosterone. These medications pro-

OCR for page 83
92 RESEARCH ON VIOLENCE AGAINST WOMEN vice relief to some compulsive sex offenders (Robinson and Valcour, 1995; Fedoroff et al., 1992), but their mandated use is controversial (Miller, 1998; Prentky, 1997~. Prentky (1997) cautions strongly against this approach as an exclusive treatment for sexual aggressors. Areas for further research that may prove fruitful include matching of offender types to types of treatment and matching of treatment to the motivational stage of the offender. Recognition of how types of offenders differ along key theoretical dimensions, such as developmental back- ground, social context, and mental disorder, should be part of the concep- tual development and testing of offender interventions (Fagan, 1996~. improving Quality-of-Outcome Studies Most offender treatment programs have been based on theoretical assumptions, but their outcomes are as yet unclear. The committee rec- ommends that improvements be made to the quality of outcome studies and that all treatment programs be evaluated. Saunders and Hamill (2002) provide suggestions for such improvements. First, evaluations should in- clude a comprehensive set of outcome measures and develop ways of reducing the possibility of faked improvements. Second, they should pro- vide evidence on whether the treatments proposed were actually the ones delivered; this could be done by standardizing treatments through use of a manual, by providing close supervision, or by employing other ap- proaches that can demonstrate treatment integrity. Third, evaluations should also include information on participant selection into treatment and attrition during and after treatment, as both methodological concerns affect the generalizability of study findings. Offenders who enter and com- plete treatment, especially voluntary programs and court-mandated pro- grams that do not have consequences for noncompliance, may differ in important ways from offenders who do not. Finally, program evaluations should use experimental designs involving comparison or control groups whenever possible if firm conclusions are to be drawn. Reducing Program Attrition One of the biggest problems in working with offenders is the high rate of attrition in treatment programs. Saunders and Hamill (2002) note that men who engage in domestic violence and sexual assault have little desire to participate in treatment programs, even when the programs are court-ordered or could lead to an earlier end to incarceration, and only a small percentage enroll in treatment voluntarily. Several methods have been proposed for reducing program attrition. One approach involves a coordinated community-wide response that in-

OCR for page 83
IDENTIFYING AND TREATING OFFENDERS 93 eludes criminal justice sanctions; although most available evidence does not demonstrate that this approach keeps men in treatment, evidence sug- gests that criminal justice mandates may help reduce attrition among younger, less-educated men (Daly and Pelowski, 2000; Saunders and Parker, 1989~. One method that has had some success in improving reten- tion is the use of a marathon orientation group a 12-hour group session designed to provide an overview of upcoming treatment, as well as to teach concrete skills (Tolman and Bhosley, 1991~. Another promising in- novation uses video and discussion to facilitate the development of feel- ings of compassion (Stosny, 1994~. More-traditional approaches that rely on supportive phone calls and handwritten notes from therapists at the outset of treatment and after any missed sessions have also been shown to be related to continuation in treatment (Taft et al., 2001~. Most convicted sex offenders who reside in the community are man- dated to participate in treatment programs as a condition of probation or parole. Increasingly, probation and parole departments are using specially trained personnel to supervise sex offenders (English et al., 1996~. Whether these methods actually decrease attrition is not known. Assessing Dangerousness and Risk of Recidivism A variety of instruments have been developed to assess the likelihood that an offender will engage in additional battering or sexual assault after completion of a treatment program, although the predictive validity of these instruments has not been established (Dutton and Kropp, 2000; Roehl and Guertin, 2000; Websdale, 2000~. Practitioners in programs for domestic violence offenders may be required to warn or protect potential victims if they believe lethal violence is imminent (Hart, 1988; McNeill, 1987~. They may also be asked to make predictions about the recurrence of severe violence in order to provide specialized treatment or recom- mend closer supervision. Campbell's (1986) Danger Assessment instrument for use with bat- tered women is a 15-item checklist that correlates with violence severity and distinguishes between victims who go to emergency rooms and those who do not (Campbell, 1995~. MOSAIC-20 (de Becker, 1997) emphasizes the role of the survivor's intuition, which some evidence suggests is an accurate way to predict severe violence (Weisz et al., 2000~. More research is needed to determine the predictive usefulness of survivor's intuition. Other risk assessment tools focus on the risk of recidivism. The Spou- sal Assault Risk Assessment Guide (Kropp et al., 1999) uses reports from as many sources as possible, including interviews with offenders and vic- tims, official records, and standardized tests. The Domestic Violence In- ventory relies on a questionnaire administered to offenders and contains

OCR for page 83
94 RESEARCH ON VIOLENCE AGAINST WOMEN the following subscales: truthfulness, alcohol use, control, drug use, vio- lence, and coping with stress (Risk and Needs Assessment, 1996~. Research on the prediction of recidivism by sex offenders has led to the creation of at least 24 measures of the offender's personal and legal history, including offender's age, victim's age and gender, and details of the offenses (Doren, 1999~. Of the 24 risk assessment measures identified by Doren, however, only 8 have been validated for assessing the risk of reoffending. The use of statistically validated measures as opposed to clinical intuition is viewed as increasingly important because the former are usually more accurate. Of the measures frequently used in the evalua- tion of sex offenders, a recent study of predictive validity showed that the Violence Risk Appraisal Guide, RRASOR, and Static-99 were best at pre- dicting sexual recidivism (Barbaree et al., 2001~. Research on predicting sex offender recidivism may benefit from the more-general research on risk prediction devices for serious offending (see, e.g., Lattimore et al., 1995~. Culturally Competent Interventions Many newly developed programs are addressing the diverse cultural backgrounds of clients (Lewis, 1999; Marshall et al., 1998~. It is posited that treatment programs might be more successful if the cultural and world views of clients were better supported by treatment models and clinicians' knowledge base (Iones et al., 1999~. Subcultural mistrust of the dominant culture, help-seeking patterns in collectivist cultures, and sub- cultural communication styles might better explain resistance to treatment than intrapsychic mechanisms (Lewis, 1999~. However, the committee could find no assessments of the effectiveness of such programs compared with treatment programs that do not specifically consider cultural factors. Williams and Becker (1994) distinguish among "color-blind" pro- grams that do not take race or ethnicity into account, "culturally focused" programs that pay attention to historical and contemporary experiences of particular cultural groups, and "culturally centered" programs that place a particular culture at the center of treatment and use culturally significant rituals. Some programs give men the choice of same-race or mixed-race groups. In addition to the Afrocentric models being devel- oped by Williams (1994), programs are available for Southeast Asians, Native Americans, Latinos, and other ethnic and racial groups (Carrillo and Tello, 1998; Healey et al., 1998~. Saunders and Hamill (2002) note that evaluations of efforts to coordinate and consult with the minority com- munity should provide useful guidance, and they recommend that quali- tative research on same-race groups be extended to further assess the im- pact of these groups on treatment attrition and outcome.

OCR for page 83
IDENTIFYING AND TREATING OFFENDERS CONCLUSIONS 95 More research is needed on treatment for perpetrators of violence against women. Possible subjects for future research include matching of what is known about different types of offenders to appropriate interven- tions, methods to reduce program attrition, the usefulness of risk assess- ment tools and survivor's intuition in predicting recidivism, and increased attention to training in cultural competence and the use of culturally spe- cific interventions. Research should also address women who commit vio- lent acts, and should further explore interactions between women's expe- riences of violent victimization and their own violent offending. Despite an accumulation of studies evaluating programs for domestic violence offenders, rigorous studies are few, and firm conclusions about the effectiveness of interventions cannot be drawn. The quality of out- come studies needs to be improved for all treatment programs.