Prevention and Intervention
This chapter examines the types of responses society has made to violence against women. There are a number of ways to define and characterize prevention and intervention. This report uses one that best identifies the kinds of responses society can take and the research that can inform those responses. First, however, it notes several other classifications.
The public health perspective classifies ''interventions" into primary, secondary, and tertiary prevention. The goal of primary prevention is to decrease the number of new cases of a disorder or illness. The goal of secondary prevention is to lower the prevalence of a disorder or illness in the population. The goal of tertiary prevention is to decrease the amount of disability associated with the disorder or illness. Although these three categories seem conceptually distinct, in practice there is disagreement over their use (Institute of Medicine, 1994). Another classification is Gordon's (1983, 1987) proposal for universal, selected, and indicated preventive measures. Universal preventive measures are desirable for everyone in a population; selected preventive measures are desirable for those in the population with an above average
risk of acquiring a disorder; and indicated preventive measures are desirable for individuals who are identified as being at high risk for the development of a disorder. Because of frequent confusion over the meaning of the public health classifications, the Institute of Medicine (1994) recommended the use of a combination of it and Gordon's: preventive interventions, broken into three categories modeled after Gordon's; treatment intervention, which includes identification and standard treatments; and maintenance intervention, which aims at reducing relapse and recurrence and promoting rehabilitation.
This report adopts the Institute of Medicine's (1994) use of preventive interventions, but considers treatment and maintenance interventions together under the rubric of treatment interventions. Treatment interventions are separated into individual and community-level interventions: individual treatment interventions are those, such as counseling, that are targeted at the individual; community-level interventions represent more system-oriented interventions, such as criminal justice reforms, rape crisis centers, and battered women shelters. Following this classification, the chapter first discusses preventive interventions. Second, it considers treatment interventions, both the services available to women victims of violence and those, including criminal justice interventions, for offenders.
School-Based Preventive Programs
Preventive intervention efforts have largely consisted of school-based programs on conflict mediation, violence prevention in general, dating violence, sexual abuse, and spouse abuse. There are few data available on how widespread these programs are or to whom they are offered. Sexual assault and rape education programs seem to be increasingly common on college campuses; conflict resolution programs have been in-
stituted in thousands of middle and high schools (Webster, 1993). The programs vary in length, in content, and in degree of theoretical underpinning. Evaluations are rare. The few evaluations that have been done of these programs generally test students' knowledge about and attitudes on relationship violence before and after the prevention program, as well as personal experience with dating violence (Jones, 1991; Jaffe et al., 1992; Kantor and Jasinski, 1995).
In Minnesota, the Minnesota Coalition for Battered Women developed a secondary school violence prevention program and trained secondary school teachers in the use of the curriculum. The approximately 200 teachers who were willing to participate in the evaluation were stratified by junior or senior high, and by rural, suburban, or urban location. Teachers were randomly selected from each of the six subgroups, and their students became the sample for the evaluation. Control groups from the same or nearby schools were also tested. Both groups were given preprogram and post-program tests to assess their knowledge about battering, their attitudes, and their knowledge about the resources available for help in addressing relationship violence. Students who were given the 5-day prevention program improved their knowledge scores significantly more than the control group. However, attitudes among both experimental and control groups showed very little change. There was a posttest significant difference between girls' and boys' scores, with the girls' scoring more in the desired direction. The experimental groups also became more knowledgeable about general resources available for help with relationship violence, such as a hospital or mental health center, although they could not name specific local services (Jones, 1991).
Other studies have found attitudinal changes following school-based intervention programs. Students in four secondary schools in London, Ontario, were involved in a dating violence prevention program (Jaffe et al., 1992). The program involved a large group presentation followed by classroom discussion led by trained facilitators. Questionnaires were
administered to 737 students—selected by means of stratified classroom-level sampling—both 1 week prior to the intervention and 1 week after; at two of the schools, a delayed posttest was also given 6 weeks after the program. Overall, the evaluation showed significant positive changes after intervention on knowledge, attitudes, and behavioral intention. A small group of males, however, showed change in an undesired direction.
Significant decreases in attitudes justifying the use of dating violence were found in a study of a prevention program in a Long Island, New York, high school (Avery-Leaf et al., 1995). The intervention consisted of five weekly sessions incorporated into a health class. The experimental group of 196 students were tested before and after the five-class program, and there was a control group of students whose health classes did not include the dating violence prevention program.
While all of these programs may change knowledge or attitudes about physical and sexual violence between intimates, no longitudinal studies exist to document whether they have any short- or long-term impact on the commission of dating violence, date rape, or intimate partner violence later in life. A review of evaluations of a broad array of prevention programs aimed at adolescents—including pregnancy prevention, drug abuse prevention, delinquency prevention—found that curricula that only provide information about risks and use scare tactics have little or no positive impact and may even result in more of the undesired behavior (Dryfoos, 1991; National Research Council, 1993). Intensive programs that include social-skill training and follow-up booster sessions may hold more promise, particularly if classroom efforts are part of a more comprehensive, community-wide strategy (Dryfoos, 1991; Webster, 1993).
Public education campaigns, such as those mounted against smoking and drunk driving, are a universal preventive
intervention that have been part of other successful community prevention projects (Institute of Medicine, 1994). The Advertising Council, in conjunction with the Family Violence Prevention Fund, in June 1994 began a public education campaign against intimate partner violence. The campaign consists of television, radio, and print public service announcements "designed to increase public awareness of battering and to motivate individuals to take action to reduce and prevent abuse" (Family Violence Prevention Fund, 1995). The advertising campaign is being evaluated. A preadvertising survey that measured attitudes toward battering was conducted; there will be several postadvertising surveys that will look at advertisement recognition and changes in attitudes about battering, willingness to intervene in battering, and knowledge of community resources (Lieberman Research Inc., 1995).
Separate from public service announcements and other advertising, television programming has the potential to convey antiviolence messages. The recent National Television Violence Study (Mediascope, 1996) suggests that television could be used to send more prosocial messages about violence by showing the negative consequences of violent behavior and nonviolent alternatives to solving problems and by emphasizing antiviolent themes. There has been no research on the effects of such television programming.
To the extent that the threat of criminal justice sanctions deters people from engaging in violent behavior, they can be thought of as preventive interventions. The theory of deterrence is well established in the field of criminal justice (for reviews, see Zimring and Hawkins, 1971; Geerken and Gove, 1975; Gibbs, 1975; Cook, 1977; Blumstein et al., 1978; Tittle, 1980; Paternoster, 1987; Klepper and Nagin, 1989). The theory suggests that increasing the certainty of sanctions increases their deterrent effect (Reiss and Roth, 1993). From this per-
spective, mandatory arrest for intimate partner violence, increasing rates of prosecution for rape and intimate partner violence, and stricter enforcement of protection orders could be considered preventive interventions. (For a more complete discussion of these types of interventions, see the section on Criminal Justice Interventions, below.)
Other Issues in Rape Prevention
The literature on rape prevention includes strategies for rape avoidance and rape resistance, which are considered by some—particularly in the criminal justice field—to be prevention through reduction of opportunity. Rape avoidance entails strategies to be used by women to minimize their risk of sexual assault. These strategies include avoiding dangerous situations, not going out alone at night, keeping doors and windows closed and locked, and other precautions to be taken by women. Although these avoidance techniques may reduce a woman's risk of being sexually assaulted by a stranger, it is not clear they would reduce acquaintance attacks (Koss and Harvey, 1991). These strategies are also criticized as restricting women's activities and as potentially placing the blame on women who are sexually assaulted for not taking adequate precautions (Brodyaga et al., 1975, as cited in Koss and Harvey, 1991). The extent to which a woman chooses to use any particular avoidance strategy may depend on the importance she attaches to the perceived costs and benefits of the strategy (Furby et al., 1991). An emphasis on rape avoidance may actually increase the fear of rape (Koss and Harvey, 1991). Furthermore, avoidance strategies may do little to lower the overall rate of sexual assault; they may simply displace the assault from one potential victim to another.
Rape resistance strategies involve recommendations to women on what to do should they be attacked. Storaska (1975) popularized among law enforcement agencies the theory that women should remain passive in the face of an
attack to avoid angering the attacker and increasing her risk of serious injury or death. The majority of the research on resistance strategies, however, suggests just the opposite. Women who actively resist attack are more likely to thwart rape completion without increasing their risk of serious injury (Javorek, 1979; Bart, 1981; Quinsey and Upfold, 1985; Levine-MacCombie and Koss, 1986; Siegel et al., 1989; Ullman and Knight, 1991, 1992). The success of resistance strategies also appears to be linked to situational factors, such as the proximity of others to the attack site, and offender traits (Koss and Harvey, 1991).
This report does not consider rape avoidance or rape resistance to be preventive interventions, the goal of which should be reduction in rates of perpetration. Some researchers also consider rape prevention to mean minimizing the psychological impact of sexual assault; this report considers that topic under interventions for victims, not as prevention.
Interventions For Victims
There is no universal system of services available to victims of battering or sexual assault; they vary from community to community. Interventions may occur in the criminal justice system, the health care system, the social service system, the mental health system, or some combination of systems. As noted above, the discussion of interventions is divided into those whose main focus is on the individual and those whose focus is institutional or community based; individual-level interventions seek to ameliorate the consequences of individual victimization; community-level interventions seek to change systems' responses to victims.
Individual counseling and peer support groups are probably the services most used by battered women. A survey of 250 victims of battering in New York City who called 911 for
help found that 43 percent of the callers wanted counseling services and 42 percent said they wanted someone to talk to about their feelings (Taylor, 1995). However, few data exist on how many battered women actually seek counseling services: of those seeking counseling services in the New York City study, Taylor (1995) found approximately two-thirds actually received them. Although specific therapy elements have been recommended for use with battered women (e.g., Walker, 1994), the panel found no evaluation studies of individual counseling or support groups for battered women.
Counseling services are also available for couples in which the woman has been battered or otherwise victimized, but there remains much debate in the field over the merits and advisability of couples counseling (Dobash and Dobash, 1992; Edleson and Tolman, 1992; Gondolf, 1993). Many practitioners and researchers argue that couples counseling is never appropriate when violence is present because it endangers women. Other counseling providers argue that couples counseling that is specifically designed to address the use of aggression may be beneficial for couples in discordant or mildly violent relationships (Pan et al., 1994; O'Leary et al., 1995). Because couples counseling is generally viewed as an intervention for the perpetrator, evaluations of it are addressed below in the section on interventions for batterers.
Mental health interventions with rape victims have received more study than those with battered women. Treatment approaches designed to address the postrape psychological consequences have been developed, and in some instances evaluations were undertaken to assess their effectiveness. However, in a review of the rape treatment outcome studies, Foa et al. (1993) concluded that few studies used an approach that would permit drawing conclusions about effectiveness. In most cases, there was no control group so it was not possible to determine whether improvements were a function of the passage of time or the intervention. The early studies that randomly assigned victims to different conditions produced mixed results. Veronen and Kilpatrick (1982) devel-
oped a brief, focused intervention implemented immediately postrape and designed to be prophylactic; they found that active treatment was no more therapeutic than assessment only. In contrast, a comparison of different six-session therapy approaches found that all three types of intervention produced improvement while victims on a naturally occurring waiting list did not improve (Resick et al., 1988).
Two more recently reported studies examined the effectiveness of specific treatments for victims suffering from posttraumatic stress disorder (PTSD). Foa and colleagues (1991) compared stress inoculation training—a combination of cognitive-behavioral and relaxation techniques to teach clients to control their fear, prolonged exposure—reliving the rape scene in the imagination in order to confront fear, and supportive counseling. They found that all treatments produced improvements at posttreatment, but at a 3-month follow-up, exposure appeared to be the most effective for PTSD symptoms. A cognitive processing approach designed to address maladaptive beliefs, as well as rape-related fears, reduced symptoms compared to a waiting-list control group (Resick and Schnicke, 1992). These studies support the conclusion that treatment for rape victims can be helpful and that specific types of treatment may be more effective for certain symptoms.
Crisis-Oriented Services: Shelters, Rape Crisis Centers, and Advocacy
A recent survey (Plichta, 1995) found 1,800 programs, of which approximately 1,200 were shelters, targeted at battered women in the United States. The programs offer a variety of services including hotlines, temporary shelter services, group and individual counseling, legal advocacy, social service referral and advocacy, services for children of abused women, transitional housing, child care, and job training. Public edu-
cation and changing social norms with respect to battering are also an integral part of "the shelter movement." In addition to shelters, individual social workers, psychologists, and clinics that offer therapy and counseling services undoubtedly provide services to some victims of battering.
Little data exist on how many clients are served by the various programs or who those clients are. It does appear that the services are inadequate to meet the needs of all victims of battering who seek them. For example, in New York City in March 1995, about 300 women and children a week were denied emergency shelter due to lack of space (O'Sullivan et al., 1995).
There is some information on the women who do use shelter services. They tend to be from low socioeconomic groups, possibly because they have fewer resources available to them than women from higher socioeconomic groups. For example, in one study (O'Sullivan et al., 1995) 76 percent of the sample were on public assistance; another 1 percent had no income whatsoever. In another large sample of women using services in Texas (1,482 battered women in shelters and 650 battered women using nonresident shelter-based programs), Gondolf and Fisher (1988) found that a substantial portion lived in poverty. Over one-half of their sample had no personal income, and three-fourths of the women's husbands made less than $15,000 per year. The women who sought only nonresident services tended to be from higher socioeconomic groups than those who sought resident shelter services. In a sample from a shelter in a medium-sized Midwestern city, 81 percent of the women were receiving some type of government assistance, and 60 percent lived below the poverty line (Sullivan et al., 1994). It is often suggested that women with more economic resources may be able to pay for temporary shelter, for example at a hotel, and that they obtain other services through private means, such as individual counseling.
Shelter service seekers also have low educational attainment. In the Texas sample, about one-half of the women had
not completed high school, and only about one-fifth had some post-high-school education (Gondolf and Fisher, 1988). Similarly, 45 percent of the New York sample had not finished high school, 31 percent had a high school diploma, and 23 percent had attended some college.
The racial characteristics of shelter users seem to depend on the location of the shelter. In Texas, 57 percent of the women were white, 15 percent were African American, and 29 percent were Hispanic. In New York City, 52 percent of shelter users were African American, 39 percent were Latina, and 9 percent were white or other (O'Sullivan et al., 1995). In the Midwestern sample, 45 percent of the participants were white, 43 percent African American, 8 percent Latina, and 1 percent Asian American (Sullivan et al., 1994). In preliminary data on shelters in the deep South, Donnelly and Cook (1995) found residents to be primarily white: only 2 of the 16 shelters they had surveyed (to date) targeted women of color. In recruiting a sample of abused women for a study in Newark, New Jersey, Joseph (1995) found more African American abused women in shelters for the homeless than in shelters for victims of battering. Since such characteristics as socioeconomic status, race, ethnicity, and educational level of all the women in a region who may need shelter services are not known, it cannot be determined if the shelter populations are representative of the battered women in an area.
Race and ethnic origin may be important factors in assessing the needs of women who use shelter services and in understanding barriers that may exist to obtaining services. For example, in comparison with white and African American women, Hispanic women in Texas shelters had been married the longest, had lower education, employment, and job status, and tended to report the longest duration of abuse, thereby experiencing more socioeconomic barriers to ending victimization (Gondolf and Fisher, 1988). There are few descriptions in the literature of programs that are successfully serving minority communities (Norton and Manson, 1997).
There has been little evaluation of the services offered to
battered women. Basic data on how many women and their children receive services, and the type of services received, are not available in any systematic form. Some state coalitions collect information from services within their state, but there is no uniform definition of what constitutes services. The Centers for Disease Control and Prevention of the U.S. Department of Health and Human Services is in the process of fielding an inventory of service availability and clientele served in order to develop some of this information.
Most studies of shelters and the outcome of shelter stays are descriptive. The panel's literature search found no experimentally controlled evaluations comparing outcomes for women using shelters with outcomes for women not using shelters. Some studies compare women's experiences preshelter stay to postshelter (Cannon and Sparks, 1989), compare groups of shelter residents based on services provided either during or after the shelter stay (Gondolf and Fisher, 1988; Cox and Stoltenberg, 1991; Sullivan et al., 1992, 1994), or compare shelter residents to women seeking other services (Berk et al., 1986). Recruiting samples of battered women to study can be quite difficult. Most of the studies have taken samples from women who seek help at a shelter or other service (for example, the courts). Using samples drawn from services makes it impossible to have a control group who has not sought services. Some researchers advertise for subjects through magazines, newspapers, or other media outlets. Although this method of recruiting may bring forth both women who have used shelter services and who have not, the sample is subject to self-selection bias. Of course, neither recruitment method produces a random sample.
In addition to the difficulty of a bias-free sample for study recruiting, researchers have not agreed on what outcome to measure. While all shelters probably have as an ultimate goal, a violence-free life for the women, many shelters' immediate goal is women's empowerment and support for them in whatever decisions they make. There currently is no clear way to measure empowerment. Many studies look at living
arrangements after a shelter stay as a proxy for violence, assuming no violence if the woman is no longer living with her abuser. Other studies have asked shelter residents where they planned to live following the shelter stay and used this planned living arrangement as a proxy for the presence or absence of violence. It is obviously easier and less expensive to use planned living arrangement as an outcome measure than to try to follow a large sample of women over time to determine their actual living arrangements or the actual level of violence in their relationships. And it is true that planning to leave may be an important step in ending a violent relationship, whether or not the woman actually leaves after any given shelter stay. However, Berk et al. (1986) argue that the outcome measure should be amount of abuse at some extended follow-up period. They note that planned and actual living arrangements differ and, furthermore, that leaving a batterer does not ensure a violence-free life. For example, Sullivan et al. (1994) found that more than one-fourth of the women in their sample who had ended their relationships with their abusers continued to be physically abused by them. However, using amount of violence or abuse is also problematic as an outcome measure for programs aimed at victims, because it depends on the behavior of offenders who were not in the program. Some argue that it is unfair to judge a program on the basis of measures over which the program can exert no control or influence. A relatively new tactic is to measure stage of change in the women who are participating in the program. An instrument for this measurement is under development (Brown, 1995).
Sullivan and colleagues (Sullivan and Davidson, 1991; Sullivan et al., 1992, 1994) have carried out a longitudinal study of women who have received shelter services, some of whom were randomly assigned to receive follow-up advocacy services. At 6 months postshelter stay, there were no significant differences in the level of violence being experienced between those who had received advocacy services and those who had not. 1 The 10 weeks of advocacy services did result
in certain positive outcomes for the women. Women who had received advocacy services were more likely to have made an effort to obtain other resources they needed. Overall, however, women in both groups experienced significantly less violence and fewer injuries than they had 6 months prior to entering the shelter, suggesting that their shelter stay was beneficial. They also reported significantly lower levels of depression, fear and anxiety, and emotional attachment to their batterers and increased feelings of personal control, an overall higher quality of life, and increased satisfaction with their social supports.
A study by Berk et al. (1986) also found that, for some women, shelters appeared to limit new incidents of violence in the 6 weeks following shelter stays. Specifically, women who were actively taking control of their lives, as evidenced by taking a number of help-seeking actions in addition to a shelter stay, were more likely not to be victimized again after a shelter stay. For other women—those who did not take other actions—the impact of the shelter stay was either neutral or even triggered retaliation by their abusers. The sample in this study consisted of 155 battered women in Santa Barbara County, California, drawn from women who went either to the local shelter or whose cases were referred to the county prosecutor's office. Statistical adjustments were made to compensate for the nonrandom assignment to experimental (shelter stay) and control (no shelter stay) groups.
Using where women planned to live after leaving the shelter (i.e., with the batterer or apart from him) as their outcome measure, Gondolf and Fisher (1988) investigated the prediction of living arrangement using service-related variables associated with economic independence, intervention decisions made, and number of shelter services used. The most influential predictors of planned living arrangements were having the batterer in counseling and the three economic independence variables. The batterers' being in counseling was the single strongest predictor of returning: 53 percent of the women whose batterers were in counseling planned to return
to them, compared with only 19 percent of those whose batterers were not in treatment. Even the effect of having independent income was overshadowed by the effect of treatment for batterers: 16 percent of women with their own income planned to return to their batterers, compared with 38 percent of women with their own income and whose batterers were in counseling.
Using a pre- and post-quasi-experimental quantitative design, supplemented by in-depth interviews, Tutty (1995) assessed social support/isolation, stress/coping, and self-esteem of women who used shelter follow-up programs designed to assist women who had decided to live independently. Social workers from the programs visited women in their homes for 1-2 hours per week to provide counseling and advocacy services. These visits began immediately after the shelter stay and continued for 3 to 6 months. The women in the sample significantly improved their scores for appraisal support (the availability of someone to talk to about one's problems) as measured by the Interpersonal Support Evaluation List after 3 months in the follow-up program. There was also some evidence of an increase in self-esteem by the end of the program. Although the sample size was quite small, this study supports the positive impact of postshelter services found by Sullivan et al. (1992, 1994).
Rape crisis centers have been providing services for victims for more than 20 years (Koss and Harvey, 1991). This specialized response evolved when rape victims could not expect that family, friends, or medical and legal systems would either understand the nature of their trauma or respond supportively. The grass roots rape movement took on the tasks of ensuring that victims had access to informed and sympathetic advocate-counselors to assist with the emotional consequences of rape and to deal with the appropriate systems. These centers also considered community education, system reform, and empowerment of women and victims to be central to their mission. As in the shelter movement,
changing the larger social climate has been as important an activity as providing direct services to victims.
Research on rape crisis services is descriptive in nature, cataloguing the kinds of services offered (O'Sullivan, 1976; Gornick et al., 1983; Burt et al., 1984) or other characteristics of the programs (Harvey, 1985). Most such services consist of 24-hour crisis lines offering information and support, advocacy in the form of information about the medical and legal systems, accompaniment to medical and legal appointments or court appearances, and supportive postrape counseling. The counseling has often been offered as a group treatment. In many cases, the services have been delivered by trained volunteers, but increasingly mental health professionals supervise or carry out the counseling.
Although the provision of these services for rape victims has come to be standard fare in many communities, there has been no systematic assessment of their effectiveness in helping reduce victims' distress. And these programs may not lend themselves to conventional methods of evaluation, with individual victims using pre- and postintervention measures of distress or assignment to various treatment conditions. Given that virtually all rape victims experience initial distress and that the purpose of intervention is to alleviate that distress with a supportive response, it is not reasonable to impose such an expectation. However, the characteristics of programs perceived as highly effective by their communities and by experts in the sexual assault field have been described (Harvey, 1985). The elements include maintaining a commitment to both victim services and social change, having a cohesive philosophy regarding program values and action, and developing the capacity to change in response to self-evaluation or shifts in the community or societal climate.
Health and Criminal Justice Services
Interventions in the health sector usually involve the treatment of injuries incurred in a physical or sexual assault.
For intimate partner violence, emphasis has been on identification of battered women and appropriate referral (Council on Scientific Affairs, 1992). In addition, the American Medical Association (1992) has issued guidelines for physicians on identifying and treating victims of intimate partner violence (McAfee, 1995). In spite of increased attention by the medical profession to the issue of intimate partner violence in the past several years, it appears that violence is often overlooked or not documented as a cause of injury (e.g., Abbott et al., 1995). However, as noted in Chapter 2, documentation of abuse in medical records is not without potential costs to the victim, such as denial of health insurance coverage.
Besides treating the injuries of a sexual assault victim, a hospital emergency room is often critical to gathering forensic evidence to be used if legal charges are brought against an assailant. Sexual assault nurse examiner (SANE) programs have been established in some cities to provide comprehensive care to sexual assault victims (Ledray and Arndt, 1994). Although there may be variation, all forensic nursing examinations of sexual assault victims include five essential components: treatment and documentation of injuries; treatment and evaluation of sexually transmitted diseases; evaluation of pregnancy risk and prevention; crisis intervention and arrangements for follow-up; and collection of medical and legal evidence while maintaining the proper chain of evidence (Hazlewood and Burgess, 1995).
Most of the criminal justice system responses relate to sanctions for perpetrators of violence and so are discussed below. However, aspects of the criminal justice system can be thought of as interventions for victims. Victims of sexual assault and intimate partner violence often face barriers that keep them from making use of the criminal justice system. Battered women may be reluctant to involve the criminal justice system out of fear of retaliation and fear of the criminal justice system itself (e.g., Fischer and Rose, 1995). Victims of sexual assault may be reluctant to report their victimization out of fear of the stigma attached to sexual assault,
fear of retaliation, and fear of the ordeal of court (e.g., Kilpatrick et al., 1992). Although there has not been much research, it appears that changes in the criminal justice system could improve women's experiences. Efforts to make the criminal justice system more responsive to victims' concerns are being tried in various communities. Such efforts include reforming laws; training police, prosecutorial, and court personnel; creating special units to investigate sex crimes; and providing victim advocates to assist women through the criminal justice process. These efforts have not been widely studied.
Increasing access to the criminal justice system may benefit women in a number of ways—increasing women's safety, improving women's sense of self-efficacy, and making a statement about the community's intolerance of violence against women. The degree to which arrest or protective orders do or do not improve battered women's safety has yet to be resolved. (The research on arrest is discussed in more detail in the section on interventions with offenders.) Women are more likely to call the police and police are more likely to make an arrest for the first battering incident or if a battering incident results in injury (Bachman and Coker, 1995). Whether arrest of perpetrators makes battered women safer is still not known, but recent analysis indicates arrest does not increase their danger more than any other police intervention (Garner et al., 1995). Willingness of the police to arrest batterers may improve women's satisfaction with police response and increase their willingness to call on the police in the future (Jaffe et al., 1993).
Protective orders, both temporary and permanent, also appear to serve specific functions for women in addition to protecting them from their batterer. Based on in-depth interviews with women who had applied for protective orders, Fischer and Rose (1995) found that women felt protective orders told their batterers that society did not approve of their behavior. Getting a protective order also served as documentation of the abuse—making it a matter of public record
should anything happen to the woman and serving as proof to police that she was serious about ending her abuse. Finally, taking the step of getting a protective order enhanced women's sense of self-efficacy and control. Although Fischer and Rose interviewed only 83 women in one midwestern city, their findings are suggestive of the ways in which criminal justice system interventions affect victims of violence.
Many of the reforms in rape laws were promoted as efforts to make it easier to prosecute and convict offenders (see discussion under interventions with offenders). Changes in rape shield laws, however, were predicted to make it easier for women to report rapes and less onerous for them to testify in court. (Rape shield laws determine how much of a victim's sexual history can be used as evidence in court.) The few studies that have looked at the impact of rape law reform have found no clear-cut evidence that rape shield laws had the intended effect (Horney and Spohn, 1991). There is evidence that the rate of reporting rape has increased nationally since the early 1970s, when rape laws began to be reformed (Bourque, 1989), but because of the variety of state laws, it is impossible to attribute that increase to legal reform. Furthermore, there are no good data on how often a complainant's sexual history had actually been used in cases prior to the enactment of rape shield laws (Horney and Spohn, 1991). Overall, a great deal of judicial discretion remains in deciding how much evidence about a victim's sexual history is allowed in court.
Other Services and Service Seeking
In order to better understand what services are important to battered women, Gondolf and Fisher (1988) studied a sample of 1,482 battered women in shelters and 650 battered women using nonresident shelter-based programs in Texas. They looked at seven categories of services offered to shelter residents: referral, transportation, medical care, child services, counseling, legal services, and employment services.
Gondolf and Fisher concluded that shelters play a pivotal role in facilitating help seeking. These women used the shelter for much more than just refuge. An average of 3.3 services were used: 45 percent of the Texas women used only one or two services, 37 percent used three or four services, and 20 percent used five, six, or seven services. The most used services were counseling (85 percent), transportation (73 percent), and referral (67 percent). On average, the women anticipated using two services after leaving the shelter; nearly one-third said they would use three or more services. The services they anticipated using the most were the crisis hotline (70 percent), counseling (61 percent), and referral services (52 percent).
Taylor (1995) surveyed 250 victims of battering in New York City who had called 911 for help. The most common services sought were help with seeking an order of protection (58.5 percent), counseling (43.1 percent), and discussion of troubling feelings (42 percent). Nearly 80 percent of those who wanted help with protection orders or other legal assistance received that help. Table 4.1 details the services sought and received by these battered women.
Various studies of rape victims reveal an important finding: a majority do not seek crisis or advocacy services or treatment in spite of service availability. Koss (1988) found that only 4 percent of college student victims had contacted rape crisis centers. In a study conducted by Kimerling and Calhoun (1994), victims were assessed prospectively over 1 year following their rapes. Although victims continued to experience more psychological symptoms than a comparison group of nonvictims, only one-fifth sought mental health treatment even though they had access to rape crisis services. This lack of use of service contrasts with an increased use of medical services during the period.
Treatment outcome studies report that among the victims serving as subjects, many had waited years to seek treatment. For example, in both the Koss (1988) and Kimerling and Calhoun (1994) studies, in spite of the fact that all of the
TABLE 4.1 Services Sought and Received by Victims of Battering in New York City Who Called 911
Percentage Who Sought Service
Percentage Who Received Servicea
Help with order of protection
Discussion of troubling feelings
Help with other practical problems
Help with emotional and practical problems for both victim and perpetrator
Help with housing
a Of those who sought the service.
SOURCE: Taylor (1995:Table D.2). Reprinted by permission of Victim Services.
victims met diagnostic criteria for PTSD, the average length of time since the assault was about 6 years.
The components and characteristics of a service response have been identified (Koss and Harvey, 1991) and are often available in many communities. Yet these services for rape victims are often not used, even though at least some of these victims continue to suffer from rape-related effects and the available treatment could be helpful. It would be important to understand the barriers and facilitators to seeking the current array of crisis intervention, advocacy, and treatment services, and determine whether there are other types of services
that victims want. Devising strategies to provide early assistance to victims who have significant mental health consequences might prevent or reduce needless additional suffering. It may be that the most effective approaches to reaching and serving victims have yet to be identified or fully developed.
Similarly, the majority of battered women may not seek services. In New York City, it is estimated that less than 2 percent of all battered women go through the shelter system, which is the entry point for many services (Friedman, 1995). There is some evidence to suggest that women who seek shelter services experience more frequent victimization than nonservice seekers. In studies by Giles-Sims (1983) and Okun (1986), women in shelters reported an average of between 65 and 68 assaults per year. The National Family Violence Survey of 1985 found an average of 6 assaults per year for all victimized women in the sample, and 15.3 assaults per year for women who had sought shelter services (Straus, 1990a). Women in shelters may have quantitatively different experiences from those of nonservice seekers associated with the difference in frequency of assaults. Yet since most of what is known about battered women and battering comes from samples of women drawn from shelter populations, lack of information about women who do not seek services represents a major gap in knowledge about violence against women. For victims of both rape and battering, it is not known to what extent they do not know about services, are unwilling or unable to use them, or choose not to use them.
Criminal Justice Interventions With Offenders
Throughout U.S. history, women's advocates have sought laws and law enforcement to prevent violence against women. The modern women's movement succeeded in bringing federal attention to the problem in the 1970s.2 In 1984 the U.S. Attorney General's Task Force on Family Violence published
recommendations for justice system action and for needed research on family violence, including battering against women. More recently, the Violence Against Women Act of the Violent Crime Control and Law Enforcement Act of 1994 promotes a number of efforts in the criminal justice arena, such as arrest of batterers and those who violate protective orders; coordination of police, prosecutor, and judicial responsibilities for battering cases; and coordination of computer tracking systems for communication among police, prosecutors, and courts.
The women's movement in the 1970s called not only for the creation of services for women who had suffered violent victimization, but also for the criminal justice system to treat rape and battering as it did other crimes. This call entailed pressure for changes in laws, particularly with respect to rape, and better enforcement of the laws pertaining to all forms of violence against women. Interest in treatment for sex offenders and batterers also grew during the 1970s and 1980s.
A growing body of research has examined the impact of criminal justice interventions. There is mixed evidence about the ability of various criminal justice interventions to protect individual women or change individual men's behavior. The impact on the rate of battering or sexual assault at the community level remains unknown. In discussing criminal justice responses, it must be remembered that there really is not a single criminal justice system in the United States; rather there are at least 51 models, determined by state and federal laws, and numerous other practices associated with local jurisdictions. In addition, one cannot discuss criminal justice interventions in terms of individual- and community-level interventions as did the previous section. Many of the rehabilitation and counseling efforts that would be individual-level interventions are tied to the criminal justice system—community-level interventions—for offenders. This section, therefore, first summarizes the research on interventions for batterers and sexual offenders and then summarizes the research on rehabilitation (treatment) efforts.
Whether or not police should arrest perpetrators on the scene of batterings tends to dominate the discussions of appropriate police interventions. Warrantless arrest has been the subject of extensive research to evaluate its effectiveness in preventing battering. The Minneapolis Domestic Violence Experiment pioneered the use of randomized designs to compare the consequences of arrest to alternative police actions (Sherman and Berk, 1984). In the Minneapolis experiment, police officers' discretion was replaced with randomized treatments of arrest, separation, or offering advice before leaving the victim and offender together. Arrest was found to result in a reduced chance of new violence within the next 6 months. These findings encouraged many jurisdictions across the country to institute either mandatory or presumptive warrantless arrest for battering cases.3
Attempts to replicate the Minneapolis findings have resulted in less clear-cut outcomes. Known collectively as SARP (the Spouse Assault Replication Program of the National Institute of Justice), experiments testing arrest and alternative police interventions were performed in Omaha, Nebraska (Dunford et al., 1990), Charlotte, North Carolina (Hirschel et al., 1992), Milwaukee, Wisconsin (Sherman et al., 1992), Dade County, Florida (Pate and Hamilton, 1992), and Colorado Springs, Colorado (Berk et al., 1992). Because the experimental designs differed among the sites, comparing results proved difficult and confusing. Recently, Garner et al. (1995) reanalyzed the data for comparative purposes and concluded that arrest was no better or worse than other police actions much of the time. When arrest did make a difference, it reduced the chance of new violence relative to alternative police treatments. Contrary to some interpretations, Garner et al. (1995) found no reliable evidence that arrest increases the likelihood of repeat violence; rather, it may be less effec-
tive in decreasing violence than other practices. However, if arrest is an option, one must wonder if the failure to arrest is perceived by a batterer as implicit permission to continue his violent behavior (Buzawa and Buzawa, 1990).
The experience with the arrest studies indicates how difficult it can be to convert even well-designed studies into policy recommendations. The initial Minneapolis findings were widely circulated and led to arrest policies in many jurisdictions, even after the replication studies were less supportive of arrest (Sherman, 1992). Making policy on the basis of one unreplicated study can be risky, but research is not the only impetus for policy. It may be that the Minneapolis results simply reinforced other social pressures at the time, such as successful lawsuits against police departments that had failed to respond to battering incidents.
As in evaluations of services for victims, controversies arise over what outcomes should be measured and how to measure them. One may wish to measure subsequent violence, but the use of official records may not give an accurate portrayal of new incidents of violence. Victims may not report repeat violence to police for a number of reasons, including dissatisfaction with the police handling of the case. Estimates from National Crime Victimization Survey data indicate that only 56 percent of battering incidents are reported to police (Bachman and Coker, 1995). Other research has estimated that as few as 7 percent to 14 percent of battering incidents are reported (Kantor and Straus, 1990). When victims are pleased with police response, they may be more likely to report repeat violence (Davis and Taylor, 1995). In either event, what is being measured is not repeat violence, but willingness to call the police. Reliance on interviews with victims may catch repeat incidents of violence that were not reported to police, but victims are often difficult to locate or may not want to be interviewed. For example, of the 330 cases in the Omaha experiment, 67 (20.3 percent) of the women did not complete initial interviews and an additional 21 women (6.4 percent) could not be located for 6-month
follow-up interviews (Dunford et al., 1990). Another issue in outcome measurement is how long the follow-up period should be.
Apart from the question of specific prevention, it remains to be demonstrated that arrest is a general deterrent to battering. Williams and Hawkins (1989) analyzed perceptual data from males interviewed as part of the second National Family Violence Survey (Straus and Gelles, 1986) to test the relative effects of formal and informal controls on potential aggression. They found that the greatest deterrent effect was associated with informal controls implied by moral disapproval of assault and by social attachments. A lesser, but significant, general deterrent operated through the perceived risk of arrest. Assuming that the perceived risk is enhanced by knowledge of actual arrests when police respond to incidents of battering, one can expect a proarrest policy to inform and then deter men who would be inclined to batter. There may be differential deterrence effects of arrest among various groups of men. For example, the extremely high arrest and incarceration rate (for all crimes) among urban African American young men may make arrest less of a deterrent for them. Yet African American women are more likely than white women to call the police, and African American men are more likely than white men to be arrested (Bachman and Coker, 1995).
In jurisdictions that have not paid special attention to or were apparently not concerned about domestic violence, prosecution rates of battering cases typically have been low (Ford, 1983; Dutton, 1988). In jurisdictions that are supportive of criminal justice interventions for domestic violence, prosecution rates have been higher (Ford, 1993).
Whether a case begins through arrest or a victim-initiated complaint, it is the prosecutor's office that acts as the gatekeeper in bringing a case to the courts. There is great
variety across the country in policies pertaining to prosecution of battering cases. Some jurisdictions encourage victim-initiated complaints, while others discourage or outright refuse to prosecute such cases. There is some evidence to suggest that allowing victim-initiated complaints serves a protective function for the victim. Ford (1993) found a decrease in violence of at least 47 percent for any victim-initiated complaint policy, according to victim interview reports of incidents before and after prosecution.
A major issue in prosecution is whether or not a victim should be allowed to drop the charges. In general, it has been assumed that prosecution can only protect victims under policies meant to keep victims in the criminal justice system while state power is used to control offenders (U.S. Attorney General's Task Force on Family Violence, 1984; Goolkasian, 1986; Lerman, 1986). Foremost among recommended policies is a ''no-drop" policy, promoted as making it impossible for a defendant to pressure a victim to drop charges, thereby possibly protecting the victim from coercive violence (Lerman, 1981). Furthermore, a no-drop policy signifies that the prosecution is committed to acting against battering, which may serve as a general deterrent. Others argue that a no-drop policy disempowers victims who may be able to use the possibility of dropping charges to negotiate self-protective arrangements (Elliott et al., 1985; Ford, 1991). Findings from the Indianapolis Domestic Violence Prosecution Experiment revealed that a victim who was permitted to drop charges and whose defendant had been arrested on a warrant had the lowest chance of suffering new violence within 6 months of the case's being settled (Ford, 1993). There is an important difference between a policy of allowing charges to be dropped and the actual dropping of charges. Ford (1993) also found that women who were permitted to drop charges and did so were no better off than those whose cases were prosecuted under a no-drop policy.
Protective orders include any of a number of court orders demanding that a potentially dangerous suspect stay away from an identified individual. Typical orders include restraining orders issued in the course of a civil matter, orders of protection issued at the request of a victim who feels endangered, and no-contact orders issued by a criminal court during its proceedings or as a condition of probation. The terms of a protective order can vary from no violent contact to no contact whatsoever with the petitioner, her friends, coworkers, etc. Orders may specify terms of child visitation or use of property. Most states today have criminalized the violation of protection orders and give police officers the authority to make warrantless arrests for their violation.
The Violence Against Women Act of 1994 created a new crime of crossing a state line to violate a protective order. The law also gives full faith and credit to protective orders, so that an order issued in one state must be enforced in another state. Whether or not this federal law improves the enforcement of protection orders will depend in large part on how local police coordinate with federal authorities in enforcement.
A few studies have addressed the effectiveness of protective orders in preventing violence. All are limited in their findings by problems in implementing and enforcing their terms. In multiple sites around the United States, Grau et al. (1984) found that, although victims felt protective orders to be effective, there was no difference in rates of victimization between those with and without orders. Indeed, the study revealed evidence that protective orders may aggravate violence when they are not enforced by the police. Buzawa and Buzawa (1990) report that samples of police in Massachusetts rarely arrested people even when they had probable cause to act against violations of protective orders. Harrell et al. (1993) found that protective orders in Denver and Boulder were effective for at least a year in preventing violence against the
petitioners in comparison with victims without orders. In other respects, however, the orders were largely ineffective: 75 percent of offenders contacted the victims and 21 percent stalked them. As in other studies, arrest for violations was rare, and, thus, compromised the potential of a protective order for preventing violence.
Courts and Judicial Processing
Judges and judicial processing are the least studied aspects of criminal justice, yet judges' discretion and court procedures can have profound effects on the actions of other parts of the criminal justice system and on victim safety. For example, Ferraro (1989) found that judges in Phoenix were critical of police who made arrests under the department's presumptive arrest policy, and, therefore, police were reluctant to make arrests.
Judges' decisions can directly affect battering cases: judges can determine pretrial release conditions, such as whether or not a defendant should be released, how long he should be held prior to release, the nature and amount of bond, and whether a protective order should be issued. In victim-initiated cases, judges decide whether to issue a warrant for the man's arrest or a summons for his appearance. During the course of a trial, judges make decisions about continuances and the types of evidence that will be allowed. If a defendant is found guilty, judges often exercise considerable discretion in sentencing. Little research exists on the effects of these various judicial decisions. In response to low rates of prosecution for batterers, some jurisdictions have created special courts to deal with cases involving intimate partner violence. Although no evaluations have been completed on the effects of such courts, the Dade County (Florida) Domestic Violence Court is currently being evaluated using an experimental design to see if legal sanctions are more likely and more severe under this system and if victims and their children are safer (Fagan, 1996).
Court-mandated counseling for batterers is a popular option from the perspective of both victims and criminal justice practitioners. It satisfies victims' interests that batterers receive help and seems to give judges a feeling of constructive action to rehabilitate offenders. Hamberger and Hastings (1993) describe two primary types of court mandates. The first is pretrial diversion, whereby the batterer can have his arrest record cleared or the charge reduced on successful completion of treatment. The second is a direct court order to participate in treatment as part of a sentence imposed after conviction. Little research has been done on the effect of court mandates on treatment completion or on whether sanctions are imposed for noncompliance.
One study of court-mandated treatment found that men who completed the program were significantly more likely to have been in treatment by court order (Hamberger and Hastings, 1989). About two-thirds of the court-mandated treatment group completed treatment, compared with about two-fifths of men whose treatment was not court ordered. Saunders and Parker (1989) found that other factors may have also been at work. In their study, younger, less well-educated men were more likely to complete treatment if it was court ordered; men over 25 with college educations were more likely to complete treatment if it was voluntary. Saunders and Parker (1989) also looked at the type of court mandate. Although significant differences in rates of attrition from treatment were not found, there was a trend for those men with deferred prosecution to have greater completion rates.
Much remains unknown about the impact of court-mandated treatment. Of particular concern is the effect of court-mandated treatment on reducing violent behavior. A few studies have found lower recidivism (as measured by repeat arrests) among men whose treatment was court mandated in comparison with men who received other court sanctions (Steinman, 1990; Palmer et al., 1992; Syers and Edleson, 1992). However, Harrell (1991) found higher recidivism among men
in court-mandated treatment than among those in a nontreatment group.
Coordinated Community Responses
As changes came about in various components of the criminal justice system, both battered women's advocates and their colleagues in the various parts of the justice system recognized a need for more coordinated efforts (Hart, 1995). Multiple models of coordinated community approaches have been developed, including community intervention projects, coordinating councils, community partnering, technical assistance projects, and community organizing. Community intervention projects have been the most evaluated of the various models. These projects involve a grassroots organization that sets up procedures with the local criminal justice agencies to establish and monitor policies related to battering, to be informed of batterers' arrests so that advocates can be provided to the battered woman to help her navigate the legal system and locate services for herself and her children, and to provide information on batterers' treatment options to the arrested man; see Brygger and Edleson (1987) for a full description of a coordinated intervention project.
Although initial evaluations seem to indicate that coordinated community efforts may decrease repeat violence, research is sparse. Gamache et al. (1988) retrospectively studied three communities in which community intervention projects had been established. Using archival police and court records, they found that the projects had a significant impact on increasing the levels of arrests for battering, convictions, and court mandates to treatment. In a comparison of cases of battering that occurred prior to the establishment of a coordinated community effort to those after it was established, Steinman (1990) found that arrests prior to the coordinated effort increased repeat violence, while police action, particularly arrest, in coordination with other criminal justice efforts deterred further violence. Syers and Edleson (1992) collected
data on all woman abuse cases during a 13-month period in two police precincts in Minneapolis. Besides looking at official police and legal advocate records, they interviewed as many female victims as could be reached by telephone contact immediately following the police visit and 6 and 12 months later. At 12 months, they found a trend for the least repeat violence among men who were arrested and ordered to treatment, followed by the men who were arrested and not ordered to treatment, with the highest repeat violence among men who were not arrested.
Coordinated community responses that involve activities other than domestic violence programs and the criminal justice system have not been evaluated. Many practitioners recognize the need for interactions and coordination among various systems that might intervene with victims and offenders. The Violence Against Women Act of 1994 includes demonstration grants for coordinating domestic violence programs with the criminal justice system, the social service system, the health care and mental health systems, the education community, the religious community, the business community, and other pertinent community groups and activities.
Rape Statute Reform
While much of the effort in criminal justice surrounding batterers concerned better enforcement of existing laws, for rape much of the effort was directed at reforming the rape laws. These reforms were expected to make it easier to obtain convictions of sex offenders as well as to improve the victims' experience in court so that women would be more willing to report sexual assaults.
Before the mid-1970s, most states' rape laws required prompt reporting of the crime; corroboration by other witnesses; physical resistance by the victim; and cautionary instructions to the jury about the difficulty of determining the
truth of a victim's testimony. Most of these state statutes also narrowly defined rape as sexual intercourse with a woman not one's wife by force or against her will (Bienen, 1980). Beginning in the mid-1970s and continuing throughout the 1980s, most states changed their rape laws in a number of ways, including: definition of acts included; gradation of offenses; focus on behavior of offender rather than consent of victim; inclusion of rape shield provisions; elimination of witness corroboration; elimination of prompt reporting; elimination of cautionary instructions; and elimination of marital exclusion.
Most states moved to a gender-neutral definition of rape that includes vaginal, anal, and oral penetration by body parts or other objects. For example, the federal code (18 U.S.C. § 2245) defines sexual act as:
(A) contact between the penis and the vulva or the penis and anus, and for purposes of this subparagraph contact involving the penis occurs upon penetration, however slight; (B) contact between the mouth and the penis, the mouth and the vulva, or the mouth and the anus; or (C) the penetration, however slight, of the anal or genital opening of another by a hand or finger or by any object, with an intent to abuse, humiliate, harass, degrade, or arouse or gratify the sexual desire of any person.
Many states and the federal code have also dropped the term "rape" and substituted a series of graded offenses such as "aggravated sexual abuse (or assault)," "sexual abuse," and "abusive sexual contact" that more closely parallel other criminal offenses. The graded offenses are differentiated by aggravating circumstances, such as use of a weapon, and have different levels of penalties attached. The intent of replacing the single offense of rape with a series of graded offenses was to make it easier to obtain convictions. It was predicted that the availability of lesser appropriate offenses would mean more convictions through plea bargaining and through providing juries options of lesser charges. However, conviction
rates do not seem to have improved subsequent to these definitional changes (Horney and Spohn, 1991).
Another major shift in rape law reform has been the move from focusing on the consent of the victim to focusing on the behavior of the offender. Prior to the rape law reform movement, most statutes required that the victim prove nonconsent by physical resistance to the attack. Rape was the only crime in which resistance by the victim had to be proven. Many statutes now focus on the force or threat of force made by the offender rather than the consent of the victim. Laws differ as to whether consent can be used as a defense after charges have been filed.
Linked to the issue of consent is the admissibility of the victim's past sexual history. A history of sexual activity with the offender or with others had been used to show a pattern of behavior that implied consent. Feminist reformers pushed hard for rape shield provisions that would disallow grilling victims about their sexual history. Rape shield laws have been enacted in every state except Utah (Epstein and Langenbahn, 1994), although they vary from state to state in how much and under what circumstances evidence of a victim's previous sexual behavior can be admitted. Some consider rape shield provisions one of the most important aspects of reform to victims, increasing their comfort with prosecution (Goldberg-Ambrose, 1992; Epstein and Langenbahn, 1994). Yet as noted above, there is no evidence that these changes have had an impact on the actual evidence that is allowed in court (Horney and Spohn, 1991).
Other aspects of rape laws that served to call into question the veracity of the victim have also been changed. Most of the laws have dropped the necessity to have corroborating witnesses. Since rape is most often committed by lone offenders in an isolated setting, the need for witnesses made prosecution of rape extremely difficult. Convictions were often also said to be hampered by the requirement of cautionary instructions to the jury, in which the judge told the jury prior to deliberations that they must consider how easy it was
to bring false accusations of rape. The use of cautionary instructions has been dropped in most states.
Although national rates of reporting of sexual assaults increased between 1970 and 1980 (Bourque, 1989), research has found only minor effects in some locations on rates of arrest, prosecution, or conviction as a result of legal reforms. The reforms may not have resulted in changes in these rates, because, in some instances, practices were already changing prior to the legal reform (Horney and Spohn, 1991), and in others, the amount of decision-making discretion available to prosecutors and judges maintained previous practices (Loh, 1980; Goldberg-Ambrose, 1992). Other changes in the criminal justice system, as well as social changes during that time, make it difficult to attribute any rate changes in reporting, arrest, prosecution, or conviction to legal reform alone (Goldberg-Ambrose, 1992). Furthermore, Goldberg-Ambrose (1992) reports that some of those most involved in the rape law reform movement identified symbolic and educational effects of reform as the most important consequences of legal change (Nordby, 1980; Bienen, 1983; Chappell, 1984). From that perspective, legal reforms could be thought of as a preventive intervention.
Rape within marriage presents special legal problems. For most of Western history, marital rape was not considered a crime. Its recognition as a crime today is by no means universal and remains controversial, despite evidence that rape within marriage is often repeated and extremely brutal (Browne, 1987; Pagelow, 1988). The marital exemption to rape is a carryover from the days of coverture—when married women had no legal identity separate from their husbands and the marriage vow was considered equivalent to the woman's consent to sexual access. A few states have completely eliminated the marital exemption from their law; others have enacted legislation that makes some rapes within marriage a crime; and in other states court actions have essentially overturned the marital exemption (Russell, 1991). Assessing the effects of changes in marital rape laws may be
even more difficult than assessing the effects of the other legal reforms. The National Clearinghouse on Marital and Date Rape compiled statistics on the number of marital rape arrests between 1978 and 1985 and found that 56 percent of the arrests resulted in prosecution and 88 percent of the prosecuted cases resulted in conviction (Russell, 1991). Russell (1991) speculates that this high conviction rate may be attributed to wives only reporting extremely violent cases of marital rape.
The Uniform Crime Reports (Federal Bureau of Investigation, 1993) do not keep track of offender-victim relationships, so it is difficult to ascertain how many marital rapes are reported to police. Twenty-six percent of the rapes and sexual assaults reported on the 1992-1993 National Crime Victimization Survey (Bachman and Saltzman, 1995) were committed by an intimate (i.e., husband, ex-husband, boyfriend, or ex-boyfriend).
Law Enforcement and Prosecution
Institutional changes have taken place both in law enforcement and prosecutors' offices with regard to sexual assaults. Many large city police forces have established specialized sex crime units, while smaller forces may give specialized training to individual investigators (Epstein and Langenbahn, 1994). These specialized units can help ensure that appropriate forensic evidence is collected. Prosecutors' offices also have specialized units that handle sex crimes. Both police and prosecutors may make use of in-house victim advocates or advocates from rape crisis centers to work with victims and serve as a liaison between victims and the legal system.
For cases that go to trial, the jury's preconceived notions about rape may interfere with conviction. Legal reforms excluding cautionary instructions to the jury and limiting the evidence about the victim's sexual history have been instituted, but their effects are not well known and may not be as anticipated. In fact, in one study that played tapes of real rape
trials to simulated jurors (reported in Temkin, 1987), the jurors were more likely to convict in a case with a cautionary instruction than in one without. Besides eliminating prejudicial testimony, a more recent tact by prosecutors has been the introduction of expert testimony on rape myth acceptance, and on posttraumatic stress disorder or rape trauma syndrome. While there is a growing amount of legal literature on the admissibility of such testimony in rape trials, no research has actually tested the impact of this testimony on jurors (Goldberg-Ambrose, 1992).
Another aspect of implementing legal sanctions for sexual assault that has received some research attention is the effect of race of both the victim and the offender. LaFree (1989) noted that virtually all research done prior to 1970 found that black defendants receive more serious official sanctions than white defendants, while most research conducted after 1970 found little evidence of discrimination. However, much of the recent research has not examined the entire range of criminal justice processing. LaFree (1989) argues that examining only one or two processing outcomes may show no evidence of discrimination, although the cumulative effect of race over all processes may be considerable. For that reason, LaFree (1989) studied the impact of race of offender and victim from reporting to arrest through prosecution and sentencing. Looking at all forcible sex offenses reported to the Indianapolis police in 1970, 1973, and 1975, LaFree (1989:145) concluded that ''black offender-white victim rapes resulted in substantially more serious penalties than other rapes, even controlling for the case characteristics that are most often included in this type of research. Moreover, black intraracial assaults consistently resulted in the least serious punishment for offenders."
Stalking has only recently been recognized by the law as a separate crime, and it is a concern in both intimate partner
violence and sexual assault. Anecdotal evidence suggests that battered women who attempt to leave their batterers are often stalked by them (e.g., Walker, 1979; Browne, 1987). California passed the first stalking law in 1990. By 1994, 47 other states and the District of Columbia had passed stalking laws, many of them patterned after the California statute (Sohn, 1994). These antistalking laws prohibit a person from following or harassing another person with the intent to cause the victim to fear imminent death or serious bodily injury. Critics of antistalking legislation contend that the laws are over broadly written and that harassment and other types of threatening behavior were already illegal in most states (Sohn, 1994). Because these laws have so recently been passed, there has not yet been an opportunity to study whether they will make victims of stalking safer or not.
Intervening with batterers in order to change their behavior emerged as a reaction to frustration among shelter workers who saw women revictimized after returning home to an unchanged male partner or saw the batterer move on to a new victim. Although batterers were being seen in individual and marital therapy, shelter workers and other women's advocates believed a different type of program was needed and began to develop group programs for male batterers in the late 1970s. Programs such as EMERGE in Boston (Adams and McCormick, 1982) and the Duluth model (Pence and Paymar, 1993) have been emulated in many locations (Feazell et al., 1984; Pirog-Good and Stets-Kealey, 1985). As more and more communities began to arrest batterers, mandating them to a group intervention became popular. At the same time as group programs for batterers were expanding, the family therapy model of counseling was also growing. This treatment focused on couples and family counseling efforts with men who abused their partners (Geller and Walsh, 1977-1978; Coleman, 1980).
While court-mandated treatment for batterers may be recommended in lieu of incarceration, treatment of convicted rapists and other sex offenders usually occurs as part of incarceration. Nearly all state correction systems offer some type of counseling or treatment for incarcerated sex offenders, and 19 states have separate facilities for at least some sex offenders (Maguire and Pastore, 1995). Treatments include insight-oriented, cognitive-behavioral, behavioral modification, and pharmacological therapies (Prentky, 1990).
Counseling or educational groups for men who batter have emerged as the most common form of intervention for offenders in the United States (Feazell et al., 1984; Pirog-Good and Stets-Kealy, 1985; Gondolf, 1987; Caesar and Hamberger, 1989; Gondolf, 1995). Group treatment is believed by many to be more appropriate than individual treatment because it expands the social networks of batterers to include other men who are supportive of being nonabusive. It also provides men an opportunity to reinforce their own learning by helping teach others (Edleson and Tolman, 1992). Furthermore, men have reported group support as being one of the most beneficial aspects of treatment (Gondolf, 1985; Tolman, 1990). However, there is a danger that the group may reinforce negative behavior rather than change. In a study of batterers' groups, Tolman (1990) received reports of negative effects from some of the female partners of the men.
Batterers' groups vary in length of time, although most are relatively short term, ranging from 6 to 32 weeks (Eisikovits and Edleson, 1989; Tolman and Bennett, 1990). Programs are generally structured in format and use some variation of cognitive-behavioral or social learning approaches (Tolman and Edleson, 1995; for a more complete discussion of structure of groups, see Edleson and Tolman, 1992). Many programs also include a gendered analysis of battering: that is, battering is viewed as a tool of male control of women in intimate rela-
tionships. The focus of the program is as much on changing men's view of their entitlement to control their partners, through whatever means, as it is on stopping the actual physical battering (see, e.g., Pence and Paymar, 1993).
A number of researchers have studied batterers' groups to see how effective they are in stopping physical violence (e.g., Purdy and Nickle, 1981; Edleson et al., 1985; Dutton, 1986; Hamberger and Hastings, 1986; Rosenbaum, 1986; Waldo, 1986; DeMaris and Jackson, 1987; Edleson and Grusznski, 1989; Tolman and Bhosley, 1989; Edleson and Syers, 1990; Palmer et al., 1992; Saunders, 1994). The findings are somewhat positive, but they must be cautiously interpreted due to many methodological shortcomings. Most of the studies did not use control groups, making it impossible to ascribe outcomes to the treatment; sample sizes were frequently small; and attrition in both the programs and the study follow-up periods was high. It is also difficult to compare results across studies because outcomes were measured differently: some studies relied on official records, for example, looking at rearrest rates; others relied on information reported by the batterers; still others gathered information from the partners of the batterers. The length of the posttreatment follow-up period varied from a few months to several years. Because of these many methodological shortcomings and differences, some analysts have concluded that there is little that one can conclude about the effectiveness of these treatment programs (Hamberger and Hastings, 1993). Others are more optimistic, noting that there are consistent findings that a large proportion of men stop their physically abusive behavior after involvement in a program (Tolman and Edleson, 1995). Studies that relied on partners' reports of physical abuse have found between 53 percent and 85 percent success rates in follow-up periods ranging from 4 to 26 months (Edleson and Tolman, 1992).
There is little information about interventions for batterers who are racial or ethnic minorities. In a recent survey of 142 batterers' programs, the majority did not provide ser-
vices that targeted minority clients. For example, only 25 percent of the programs provided education and training in minority communities. Even fewer programs incorporated specific program elements designed to encourage minority participation (Williams and Becker, 1994). Anecdotal evidence from programs designed specifically for African American men suggest that they are more successful with that population than other batterers' programs (O.J. Williams, 1992, 1994, 1995). The development and dissemination of culturally relevant programs and policies would benefit from program descriptions and outcome evaluations.
The use of couples therapy when there is abuse in a relationship is highly controversial. Many practitioners think that the dynamics of the battering relationship do not lend themselves to joint therapy, especially in court-referred cases (e.g., Dobash and Dobash, 1992; Kaufman, 1992). Most important is the potential of putting the woman in greater danger as a result of the therapy. Those who advocate profeminist approaches to batterers' treatment think that couples therapy, particularly if it focuses on the couple as a system, may allow the batterer to continue to deny responsibility for his violence (Breines and Gordon, 1983; Bograd, 1984).
Little research has been done on the effectiveness of couples therapy in ending violent behavior. Several studies of couples counseling found a reduction or cessation of violence in a sizable proportion of the sample following couples counseling (Taylor, 1984; Neidig et al., 1985; Deschner et al., 1986). However, the samples were small (only 15 couples in one study), the follow-up time was short, and the studies did not specify how violent behavior after treatment was assessed. The last point is particularly important because it has been consistently found that men who batter often report their use of violence to be less frequent and less severe than is reported by their female partners (Szinovacz, 1983; Jouriles and
O'Leary, 1985; Edleson and Brygger, 1986; Bohannon et al., 1995). A recent study comparing couples and single-sex group therapies with couples who had experienced physical aggression—none of it severe—and who wanted to stay together found significant decreases in physical aggression from both therapies a year after therapy ended (O'Leary et al., 1995). Similarly, Harris and colleagues (1988) found no difference in the level of violence or participants' level of psychological well-being between couples who participated in individual couples counseling and those in a group program that consisted of single-sex group sessions followed by couples group sessions. However, participants were four times more likely to drop out of individual couples counseling than the group program. Although there is some evidence that couples therapy may be effective in reducing violence, philosophical disagreements about its use are bound to continue. There is, however, agreement among most practitioners that couples therapy is not appropriate for court-mandated cases or for severely violent men (who are likely to be more violent) (Gondolf, 1995).
The question of treatment effectiveness for rapists is the subject of significant controversy. Reviews of treatment outcome studies arrive at different conclusions. An influential paper by Furby et al. (1989) concluded that the methodological problems with most available studies precluded drawing any definitive conclusions. They note that most studies did not carefully describe the treatments, included a mixed group of offenders, and, most importantly, did not have a comparison group of untreated offenders. A recent review had a more optimistic assessment of the available studies, but concluded that even effective programs "do far better with child molesters and exhibitionists than with rapists" (Marshall et al., 1991:481).
The debate about whether the currently available body of
treatment outcome literature is informative on the question of treatment effectiveness revolves around the scientific merit of the studies. It is argued that without formal controlled clinical trials with untreated comparison groups, recidivism rates cannot be attributed to the intervention (Quinsey et al., 1993). But there are moral and ethical constraints—as well as community objections—to withholding treatment from known sex offenders that argue against trying to achieve this standard of scientific rigor. Marshall (1993) contends that it is possible to accumulate knowledge about treatment effectiveness by identifying appropriate comparison groups among those who refuse treatment or for some reason do not receive the specified treatment. He believes that any reduction in recidivism is worthwhile because of the enormous cost of even a single repeat offense to the victim and society.
There is one study that used a random assignment design with control groups that provides specific data on rapists (Marques et al., 1994). The treatment consisted of a comprehensive cognitive behavioral program that used a "relapse prevention" framework, in which offenders are taught to identify situations and emotional states in which they are likely to reoffend and the skills to avoid those situations and deal with the emotional states. Preliminary results for the 59 rapists (as opposed to child molesters) in the study revealed reoffense rates within 5 years of prison release (as measured by rearrest) for sex offenses of 9 percent for the treatment group, 28 percent for the offenders who volunteered but were not assigned to the treatment condition, and 11 percent for the nonvolunteer control group. The rates for other violent offenses were 23 percent for the treatment group, 33 percent for the volunteer controls, and 22 percent for the nonvolunteer controls. It is unfortunate that the total treatment sample includes so few rapists, and therefore, the power necessary for statistical analyses is reduced. It is worth noting that the overall reoffense rates are relatively modest considering current beliefs about the risk for recidivism among rapists. There was one striking finding: of the rapists who had to be re-
moved from the treatment group for seriously disruptive behavior, 100 percent committed a new sex offense. It is possible that offenders with a particularly high risk of reoffending can be identified as those who are too disruptive to participate in treatment programs.
Many questions remain about what treatment for rapists should entail and whether it is a worthwhile endeavor, as opposed to simply imposing long periods of incarceration. Efforts might be useful to identify those who are at highest risk for reoffending and imposing especially long sentences on them. An actuarial method for prediction of future sexual dangerousness has been developed (Quinsey et al., 1995): variables such as prior convictions, deviant sexual interests, and psychopathy have been shown to be associated with increased risk. However, there are major philosophical questions involved in sentencing on the basis of future predicted behavior.
Conclusions And Recommendations
There are few good evaluations of preventive or treatment interventions for either victims or perpetrators of violence against women. Existing evaluations are difficult to compare because different outcomes have been measured in different ways and at different times. One serious problem with studies of interventions is selecting and operationally defining outcomes to measure. For example, is the goal of a batterers' treatment program to reduce or completely stop batterers' violent behavior? Should prevention programs measure reduction in victimization rates or reduction in perpetration rates? Should services for battered women consider improvement in self-esteem, separation from the batterer, or physical safety as the outcome? Should treatment of rapists measure sexual arousal or behavior outside the laboratory as an outcome? At what intervals following the completion of the program should measurements be made? Are there unintended negative outcomes from prevention or treatment in-
terventions? Close collaboration between researchers and practitioners is necessary to carefully specify the basic assumptions underlying the intervention and to determine what outcomes constitute success.
Recommendation: Evaluations of preventive and treatment intervention efforts must clearly define the outcomes expected from the intervention. These outcome measures should derive from an explicit theory underlying each intervention. Defining outcomes requires close collaboration between researchers and service providers.
School-based programs to prevent date rape and intimate partner violence, as well as programs on conflict resolution and general violence prevention, have become popular in recent years. However, these programs have seldom been evaluated, and the evaluations that have been done usually look only at short-term attitudinal change. Their longer term impact on behavior, arguably what these programs are trying to effect, remains unknown. In light of studies that have detected negative outcomes with some program participants, a better understanding of long-term outcomes and differential effects of preventive efforts is needed.
Early childhood predictors have been identified for violence in adolescence and adulthood as well as for other undesirable behavior (e.g. substance abuse, teenage pregnancy). Some of the predictors that have been identified include impaired relationships with parents, witnessing violence, experiencing physical abuse and neglect, absence of prosocial role models, and early learning and behavior problems. It is possible that sexual and intimate partner violence have common precursors with other forms of violent and dysfunctional or undesirable behavior. Except for the few prevention programs targeted at preventing dating violence, most youth violence prevention programs have ignored the prevention of intimate
partner and sexual violence. Inclusion of the consideration of violence against women in these studies would greatly increase understanding of possible links between various types of violent behavior and might lead to better designed prevention programs.
Recommendation: Programs designed to prevent sexual and intimate partner violence should be subject to rigorous evaluation of both short- and long-term effects. Programs designed to prevent delinquency, substance abuse, teenage pregnancy, gang involvement, or general violence (including conflict mediation programs) should include evaluation of risk factors for and prevention of intimate and sexual violence. In addition, studies of at-risk children and adolescents should include an examination of the relationship of risk factors, such as poverty, childhood victimization, and brain injury, to outcomes of sexual and intimate partner violence.
Interventions with Victims
Services for victims of rape and domestic violence are available in many communities. Anecdotal evidence and the few evaluation studies that have been done seem to indicate that the services are helpful to the women who use them. Yet many women do not use those services. Research is needed to determine whether those women are unaware of the services; if the services offered fail to meet their needs; if they seek help from systems, either formal or informal, other than those specifically designed for victims of violence against women; if they do not want or need services; or if other factors play a part in their underutilitzation or nonutilization of services. Special attention should be paid to whether there are systematic differences in the types of services sought by different subpopulations of women, and, if so, the implications of those differences for providing services. A better understanding of both the effectiveness of current services
and the service needs that remain unmet could lead to better program designs. Furthermore, studying women who do not seek services may yield information on factors that affect resiliency and obviate the need for services. An understanding of the social, economic, and institutional barriers that may prevent women from seeking services is important to designing these alternatives. For example, services might be made available in primary health care, community clinics, or educational facilities. In spite of the proliferation of training and screening protocols, particularly for professionals in medical settings, violent victimization is frequently overlooked. The effectiveness of training programs needs to be assessed to determine the training models that best equip professionals to identify and assist victims.
Recommendation: Studies that describe current services for victims of violence and evaluate their effectiveness are needed. Studies to investigate the factors associated with victims' service-seeking behavior, including delaying seeking of services or not seeking services at all, are also needed. These studies should describe and evaluate innovative or alternative approaches or settings for identifying and providing services to victims of violence against women.
Interventions with Offenders
Interventions with sex offenders and batterers consist of a criminal justice response or specialized social service treatment programs for certain offenders, or both. Information is needed to determine the relative effectiveness of various interventions and to develop a means of matching offenders to interventions. Special attention should be paid to cultural and ethnic differences that may have a bearing on the effectiveness of interventions. Studies should not overlook possible informal surveillance and control mechanisms that may work independently or in conjunction with more formal con-
trols to deter offenders from repeating their violent behavior. Efforts are also needed to find effective interventions for the large subset of offenders for whom current approaches are ineffective.
The panel considers court-mandated programs for batterers to be ripe for randomized, controlled outcome studies. Such programs are gaining in number around the country. Criminal justice personnel, judges, and victims all seem to find mandated batterers' programs an attractive option. As these mandated programs proliferate, it is important to understand what features—including other community resources and concomitant sanctions, as well as program philosophy, structure, and length—make them effective, and for whom.
Recommendation: Randomized, controlled outcome studies are needed to identify the program and community features that account for the effectiveness of legal or social service interventions with various groups of offenders.
Criminal Justice System
The crimes of rape and battery by intimate partners have historically been handled and perceived differently than other person-to-person crimes. Legal reforms have been proposed and implemented to treat sexual assault and intimate partner violence similarly to other crimes, but little is known about how these reforms have affected actual practices or what differences, if any, they have made for victims. For example, many changes have been made in rape laws and rape trial procedures, but little is known about the impact of those changes on investigation, prosecutorial decision making, or jury behavior. Similarly, the use of expert witnesses on the results of trauma (e.g., rape trauma syndrome, posttraumatic stress syndrome, and battered woman's syndrome) is becoming more widespread, but few studies have been done on the
effects of such testimony on trial outcome. More broadly, there is little information on the way in which reforms interact with people and practices within the systems. The process by which reforms are implemented, or not implemented, may well be as important as the reforms themselves.
Recommendation: Studies are needed that examine discretionary processes in the criminal and civil justice systems, including implementation of new laws and reforms, charging and prosecutorial decision making, jury decision making, and judicial decision making. Legal research, which supplies the theoretical basis behind legal interpretations and reforms, is also needed.
The advocates used in this study were trained undergraduates; some researchers and practitioners speculate that the impact of advocacy may have been greater if professional advocates had been used.