The Criminalization of Domestic Violence: Promises and Limits. Series: NIJ Research Report Published: January 1996 58 pages 134,576 bytes The Criminalization of Domestic Violence: Promises and Limits by Jeffrey Fagan Presentation at the 1995 conference on criminal justice research and evaluation Jeffrey Fagan, Ph.D., is visiting professor in the School of Public Health at Columbia University. This paper is an adaptation of the presentation Dr. Fagan made on July 10, 1995, at "What To Do About Crime," the annual conference on research and evaluation sponsored by the National Institute of Justice, the Bureau of Justice Assistance, and the Office of Juvenile Justice and Delinquency Prevention, and held in Washington, D.C. ----------------------------------------------- Abstract During the past 30 years, the criminalization of domestic violence has developed along three parallel but generally separate tracks: criminal punishment and deterrence of batterers, batterer treatment, and restraining orders designed to protect victims through the threat of civil or criminal legal sanctions. Each of these policy tracks has been informed, advanced, and supported by advocacy groups for battered women. Victim advocacy groups have worked vigorously for legislative and policy change, monitored and corrected the implementation of law and policy, and intensively supported expanded resources for victim services. Several jurisdictions have attempted to integrate these policies in systemwide approaches within the justice system. Thus far, however, research and evaluation on arrest and prosecution, civil or criminal protection orders, batterer treatment, and community interventions have generated weak or inconsistent evidence of deterrent effects on either repeat victimization or repeat offending. For every study that shows promising results, one or more show either no effect or even negative results that increase the risks to victims. Several factors have influenced the current state of policy and practice. Domestic violence and partner assault are complex behaviors. The range of sanctions for offenders has been limited, their deterrent effects mitigated by social and contextual factors, and their implementation constrained by practical operational contingencies. The social organization of courts and local legal cultures tend to devalue domestic violence cases. Perhaps most important, theories of violence have not been integrated with theories of domestic violence, and research and evaluation designs thus far have been weak. A program of research and development is recommended to advance the current state of knowledge on the effects of legal sanctions for partner violence. Theory is essential to this effort. Testable ideas should be identified from theoretical advances, formative evaluations of innovative practices, and qualitative studies of battering careers. A stable and sufficient resource stream will be required to support developmental, evaluation, and research efforts. Overview Beginning in the 1970's, social policy toward female victims of domestic assaults has focused on improving legal responses to protect women and punish offenders. The primary focus has been the mobilization of societal institutions to increase the range of formal and informal controls at their disposal. During this time, social control through law dominated theories on how best to reduce domestic violence, focusing on the effects of increasing the risks and punishment costs of violence toward intimate partners. Rooted in assumptions of specific deterrence, social control in this context emphasized the application of legal sanctions through arrest and prosecution of assailants or the threat of legal sanctions through civil legal remedies that carried criminal penalties if violated. Legal action was designed to exact a retributive cost, and to the extent that further violence was not evident, the suppression of violence was attributed to the intrinsic deterrent effects of legal sanctions (Dutton, 1995; Sherman, 1992a; Fagan and Browne, 1994). At the same time, reforms in civil legal protection also have expanded nationwide (Grau et al., 1984; Harrell, Smith, and Newmark, 1993). Until the legal reforms of the late 1970's, a woman could not obtain a restraining order against a violent husband unless she was willing to file for divorce at the same time (U.S. Commission on Civil Rights, 1982). When protective or restraining orders were available, their enforcement was weak, the penalties for violations were minor, and they were not available on an emergency basis. Reforms in protective and restraining order legislation enabled emergency, ex parte relief that included not only no contact provisions but also economic and other tangible reliefs for battered women. These reliefs, and the application of criminal laws as well, were extended to women in unmarried cohabiting couples and to divorced or separated women.1 These reforms held great promise. The "criminalization" of domestic violence cases beginning in the 1970's sought to increase the certainty and severity of legal responses, thereby correcting historical, legal, and moral disparities in the legal protections afforded to battered women (Zorza, 1992). For many years, societal responses to domestic violence excluded legal intervention. Advocates for battered women claimed that male batterers were rarely arrested, prosecuted, or sentenced as severely as other violent offenders. Research showed that these claims were accurate. Police often exercised discretion in avoiding arrest in responding to domestic violence incidents where there was probable cause (Black and Reiss, 1967). In many departments, policies for domestic "disputes" actively discouraged arrest, focusing instead on alternative responses such as family crisis intervention or counseling for batterers with alcohol abuse problems (Bard and Zacker, 1971). Prosecutors failed to actively pursue cases where victims and offenders had intimate relationships, fearing that women might drop charges (Parnas, 1967). Sentences often were less serious for males convicted of domestic violence. For example, Davis and Smith (1982) showed that the presence of a victim/offender relationship led to less serious case assessments in prosecutorial screening, even after controlling for victim injury and weapon use. The result of these processes was a higher dismissal rate for domestic cases at the prosecution stage compared to other violence cases, and less serious sentences. These concerns led to a wide variety of reforms in law and criminal justice, aimed primarily at increasing the likelihood that sanctions would be forthcoming in domestic violence cases. They ranged from the elimination of organizational, eligibility, and systemic complications that limited access to criminal legal remedies for battered women; to law changes mandating arrest for domestic violence; to mandatory treatment and supervision of men convicted of assaults against female intimate partners; to the reorganization of court structures to create special forums for the adjudication of these cases (Hilton, 1993). Resources, both political and fiscal, were mobilized to improve civil and criminal legal responses through training, technical assistance, and dissemination of innovations and reforms. Such efforts were undertaken with both symbolic and substantive goals. Certainly, the passage of legislative mandates for criminal sanctions symbolizes public contempt for the actions of persons who are violent toward adult partners. The mobilization of resources and the passage of (strong) laws signaled societal rejection of domestic violence and communicated important cultural messages rejecting norms supporting battering. Such responses are particularly appropriate when domestic violence is defined as a "cultural" problem. The symbolic component of criminalization policies also may be intended as a general deterrent, by conveying the message that legal consequences are likely and severe if a man assaults his wife. The substantive goals were to implement both specific and general deterrent threats to reduce the incidence of domestic violence. Here, however, the goals were less clear. Whether criminalization was designed for retributive, incapacitative, or other social control purposes or whether it was designed to assist victims or target batterers for criminal punishment was not clear in legislation and policy. These reforms did not question the deterrence underpinnings of legal interventions--that increases in the certainty and severity of criminal (and also civil) legal sanctions would reduce domestic violence was assumed. Policy goals addressed the legal and systemic problems that stood in the way of the full application of the law for domestic violence. With few exceptions, research and evaluation on legal reforms and innovations focused on their operational goals and on their outcomes but almost never on their effects.2 The narrow range of studies on the deterrent effects of legal sanctions for domestic violence stands in contrast with the extensive efforts of activists, victim advocates, and criminal justice practitioners to mobilize law and shape policy to stop domestic violence.3 This paper presents a review of the promises and limitations of the criminalization efforts in domestic violence. The reforms of the past 20 years have not been adequately evaluated with respect to their deterrence goals, despite the institutionalization of law and policy to criminalize domestic violence. With the exception of the influential Minneapolis Domestic Violence Experiment (Sherman and Berk, 1984a, 1984b), criminalization policies have proliferated without consideration of the empirical evidence of their effectiveness or their unintended consequences. Only one experiment on prosecution has been completed (Ford, 1993), yet policies to expand prosecutorial involvement proliferated quickly even in the absence of consistent deterrent effects. In fact, evaluations of prior efforts have not been influential in shaping the direction and content of criminal justice reform in general. In addition, basic research on interpersonal violence, domestic violence, and the effectiveness of legal sanctions also has been segregated from the development and evaluation of criminalization reform efforts. This paper addresses these concerns. It begins with a brief history of the development of modern legal reforms in domestic violence and examines their theoretical underpinnings. Next, it reviews the empirical evidence on the deterrent effects of criminal and civil legal sanctions for domestic violence. Then, the unique contexts of domestic violence are examined to identify factors that influence the deterrent effects of criminal justice reforms. These include both exogenous influences in communities and legal institutions and endogenous factors unique to the context of domestic violence. The paper concludes with an agenda for building an empirical base for knowledge and policy to control domestic violence. Modern Policies and Their Effectiveness Historical and Modern Origins of Legal Interventions Legal interventions in family violence have both modern and historical origins. Historically, female victims of domestic violence have sought help and protection from a variety of institutions, including family, church, and community. However, the involvement of legal institutions in domestic violence has been inconsistent throughout U.S. history. The historical origins of laws and legal interventions for child protection, for example, have roots in both the Puritan and Progressive Eras. Early laws concerning violence against wives and children were symbolic affirmations of Biblical principles demarking the line between legitimate physical force and "beating." However, Pleck (1989) reported that only 12 cases of domestic violence were prosecuted in the Plymouth Colony between 1633 and 1802. Legal interest and mobilization have occurred cyclically since that time, coinciding with State interests in enforcement of laws concerning "public morality" or increased fear of crime generally (Pleck, 1989). In 1871, Alabama became the first State to rescind a husband's right to beat his wife, noting that the "wife had the right to the same protection of the law that the husband can invoke for himself..." (Fulgham v. State, 46 Ala. 146-147). However, there were few prosecutions under these or any other State laws for violence toward wives, and the doctrine of family privacy continued to prevail over these largely symbolic statutes. Over the next century, attention from the police and the courts to violence against women has peaked at different times, influenced in part by broader concerns with the status of women and the legal control of social problems (Dobash and Dobash, 1979; Pleck, 1989; Gordon, 1988). The 1960's. In the 1960's, society began paying attention to violence within families. What historically was a private family matter became an appropriate target for State intervention. Violence toward wives and intimate partners was raised as a social problem within the context of violence against women--the result of the work of feminist activists, rape crisis counselors, clinical researchers working with women, and the earliest workers in battered women's shelters. Victims of domestic violence presented themselves to feminist grassroots organizations via rape crisis programs and hotlines as well as to newly created victim-assistance programs. The newly created victim-witness programs in the 1970's became magnets within the criminal justice system for victims of intimate or domestic violence; battered women quickly became a major portion of their workload. The Law Enforcement Assistance Administration of the U.S. Department of Justice took the lead in promoting a broader response, funding 23 programs between 1976 and 1981 for services including shelters, special prosecution units, treatment programs for wife beaters, mediation units, and civil legal interventions (Fagan, Friedman, Wexler, and Lewis, 1984). The publication of nationwide epidemiological data on violence within adult couples (Straus et al., 1980; Gaguin, 1977-78) reinforced for the general public the scale of domestic violence and prompted further activism. Unlike the rush to criminalize child abuse, legal institutions responded with ambivalence. In the early stages of reform, concerns about domestic violence were limited to married couples. It was identified as a recurring criminal justice problem, especially for the police (Bard and Zacker, 1971). Criticism of the inadequacy of police responses led to programs and training to improve police responses. However, these efforts were designed to do everything possible to avoid formal legal processing of men who beat their wives or partners. Police officers were trained in crisis intervention to look for comorbidity with drunkenness, to mediate on the scene to defuse the immediate crisis, and to make appropriate referrals for longer term interventions (Bard, 1970). They were trained to do anything except arrest violent husbands. For many years, police culture portrayed domestic violence as the most potentially dangerous situation for police officers, with elevated risks of serious injury or death; subsequent data proved this false (Garner and Clemmer, 1986). If not dangerous, spouse abuse was viewed by the police and the courts as an intractable interpersonal conflict unsuited for police attention and inappropriate for prosecution and substantive punishment (Parnas, 1967). In fact, many police departments had "hands off" policies prior to the 1970's, and police training manuals actually specified that arrest was to be avoided whenever possible in responding to domestic disputes (IACP, 1967). The 1970's. Other legal barriers prevented women from obtaining civil legal remedies as well. Until the legal reforms of the late 1970's, women could not obtain a restraining order against a violent husband unless they were willing to file for divorce at the same time (U.S. Commission on Civil Rights, 1982). When protective orders were available, enforcement was weak, penalties for violations were minor, and use in emergencies was not possible. The convergence of the interests of feminists, victim advocates, and some conservative politicians interested in expanding the use of the law to enforce "public morality" led to a series of reforms beginning in the late 1970's to strengthen criminal justice responses to domestic violence. Political pressure by feminists sought to frame solutions to domestic violence in a legal context (Lerman, 1981; Zorza, 1992). The 1980's. By 1980, 47 States had passed domestic violence legislation mandating changes in protection orders, enabling warrantless arrest for misdemeanor assaults, and recognizing a history of abuse and threat as part of a legal defense for battered women who killed their abusive husbands.4 Police departments changed their procedures not only in response to these pressures but also pursuant to successful litigation by women against police departments for their failure to enforce criminal laws and to protect them from violent partners. (See, for example, Scott v. Hart U.S. District Court for the Northern District of California, C76-2395; Bruno v. Codd 47 N.Y. 2d 582, 393 N.E. 2d 976, 419 N.Y.S. 2d 901 [1979]; and Thurman v. City of Torrington 595 F. Supp. 1521 [1984].) The array of statutory, procedural, and organizational reforms covered nearly every aspect of the legal system. Police departments adopted proarrest or mandatory arrest policies. Domestic violence units were formed in prosecutor's offices, and treatment programs for abusive husbands were launched in probation departments and among community-based groups. Reforms in protective and restraining order legislation enabled emergency, ex parte relief that included not only no-contact provisions but also economic and other tangible reliefs for battered women (Grau, Fagan, and Wexler, 1984). These reliefs, and the application of criminal laws as well, were extended to women in unmarried, cohabitating couples and to divorced and separated women. A small number of jurisdictions developed coordinated, systemic responses that brought to bear the full range of social controls and victim supports for battered women. Institutionalization Looking back over 30 years, legal reforms in domestic violence have developed along three parallel but generally separate tracks: criminal punishment and deterrence of batterers, batterer treatment, and restraining orders in the civil court designed to protect victims through the threat of civil or criminal legal sanctions. The promise of criminalization was straightforward: the symbolism of public statements valuing the safety of battered women and condemning batterers, substantive expanded social control of wife beaters, and the political mobilization of legal resources and institutions to protect victims. Additional reforms included recognition of violence risks to cohabitants and broad interpretation of laws on domestic violence to include gay and lesbian couples. Other efforts have been designed to increase funding for battered women. Surcharges on marriage licenses, for example, created a funding stream for services to battered women. Each of these policy tracks has been informed, advanced, and supported by advocacy groups for battered women. These advocacy groups have worked aggressively for legislative and policy change, monitored and corrected the implementation of law and policy, and lobbied intensively for funding for victim services. Collaboration with the growing community of victim services agencies in the 1970's created powerful alliances that moved for procedural reforms in the criminal justice responses to domestic violence. By 1990, many States had developed sweeping and strong legislation that corrected historical wrongs such as warrantless arrests in misdemeanor cases or requiring women to file for divorce before receiving protective orders (Zorza, 1992). These reforms also made accessible a wide range of criminal and civil remedies that recognized the reality of domestic violence and the complexity of its criminalization (Lerman, 1992). These efforts were institutionalized in law and policy with significant changes achieved in statutes, the organization of investigative and prosecutorial agencies, and the allocation of court services and resources. Legal Sanctions and Domestic Violence Two types of reforms have characterized criminal justice responses to family violence. One set of reforms was aimed at procedures and jurisprudential issues. These reforms attempted to increase and simplify the participation of battered women in the criminal justice process. Many sought to rectify procedural barriers, whereas others relaxed evidentiary standards for initiating criminal prosecution. Underlying these reforms was the theory that family violence could be stopped through legal sanctions and that legal sanctions were effective in reducing violence. Another set of reforms was aimed at specific measures to stop the violence. Although subsuming many of the procedural reforms, these reforms embodied explicit measures to legally sanction offenders. These interventions to control violence against adult intimate partners reflect several different policy goals and separate but parallel tracks: criminal punishment and deterrence of batterers, batterer treatment, and protective interventions designed to insure victims' safety and empowerment. Because of the interdependency of legal institutions in pursuing these policy goals, the discussion of legal interventions is not organized according to their separate responsibilities but on the basis of policy goals they pursue with respect to reducing violence toward intimates. Criminal Legal Sanctions Minneapolis research on arrest as a deterrent. Efforts to deter domestic violence have focused primarily on the police. However, this burden may have unfairly fallen on the police. Although arrest may have independent effects in reducing the risks of further violence, sanctions ultimately result from the actions of prosecutorial and judicial actors who mete out criminal penalties. Nevertheless, a rich criminological literature acknowledges the general and specific deterrent effects of police actions independent of substantive punishments. Police actions in domestic violence have been widely evaluated; this also has been the arena of legal interventions where experimental designs have been used most often. Although the results have been the strongest overall, they have also been equivocal. The Minneapolis Domestic Violence Experiment is perhaps the most widely cited and influential criminal justice experiment in recent criminological and policy literature. The Minneapolis Domestic Violence Experiment (Sherman and Berk, 1984a, 1984b) was a critical event in changing public and scholarly perceptions of domestic violence from a "family problem" amenable to mediation and other informal, nonlegal interventions (Bard and Zacker, 1971) to a law violation requiring a formal criminal justice sanction. In that experiment, street-level police officers' selections of the most appropriate response to misdemeanor domestic violence were determined by an experimental design, i. e., random assignment to one of three treatments: (1) arresting the suspect, (2) ordering one of the parties out of the residence, and (3) advising the couple. Using victim interviews and official records of subsequent police contact, Sherman and Berk (1984a:267) reported that the prevalence of subsequent offending--assault, attempted assault, and property damage--was reduced by nearly 50 percent when the suspect was arrested. On the basis of the results from what they emphasized was the "first scientifically controlled test of the effect of arrest on any crime," Sherman and Berk (1984b) concluded that: (T)hese findings, standing alone as the result of one experiment, do not necessarily imply that all suspected assailants in domestic violence incidents should be arrested. Other experiments in other settings are needed to learn more. But the preponderance of evidence in the Minneapolis study strongly suggests that the police should use arrest in most domestic violence cases (1984b:1). In the decade since the preliminary results were announced in the "Science" section of the New York Times (Boffey, 1983, p. L1), the study's findings were reported in over 300 newspapers in the United States, broadcast on three major television networks in prime-time news programs or documentaries, and featured in numerous nationally syndicated columns and editorials (Sherman and Cohn, 1989). The Attorney General's Task Force on Family Violence endorsed the study's findings and recommended that State and local agencies adopt a proarrest policy toward domestic violence (U.S. Attorney General, 1984). Following the attention given to this study's results, a dramatic change in formal policy consistent with the study's proarrest findings has been reported by police departments in both large and small U.S. cities (Sherman and Cohn, 1989). The Minneapolis experiment was designed, funded, and implemented as a test of specific deterrence theory (Sherman, 1980) and was a direct response to the call for such tests by the National Academy of Sciences (Blumstein et al., 1978). The Minneapolis experiment is atypical for its innovative experimental design, test of theory, extensive public visibility, focus on a controversial policy issue, and apparent effect on public policy. This experiment is atypical for another reason: It was replicated. Replications of the Minneapolis experiment. The initial reports of deterrent effects in the Minneapolis experiment were tempered by later criticisms of its designs and claims of the overreach of its conclusions given the limitations in the design (Binder and Meeker, 1992). Replications of the Minneapolis experiment in five jurisdictions failed to produce consistent results. Garner, Fagan, and Maxwell (1995) showed inconsistency in the directions, effect sizes, and statistical power in experiments based on both official arrests and victim interviews over a 6-month followup period. Thus, where there was once one experiment with two consistent findings (prevalence and time to failure), there are now seven similar experiments in which the findings on the specific deterrent effect of arrest on the prevalence of reoffending--the central finding of the Minneapolis experiment--differ internally by data source and externally by site. The replications are noteworthy also because of two important corollary findings. First, Sherman et al. (1992a) and Berk et al. (1992) in multisite analyses and Pate and Hamilton (1992) in a single-site analysis reported on the interaction of arrest with two measures of "stake-in-conformity" (Toby, 1957), the marital and employment status of the suspect. This line of analysis suggests that arrest increases violence for unmarried and also unemployed suspects and deters it for married and employed suspects. Sherman et al.'s (1992a:687) evaluation of the published findings from Colorado Springs, Dade County, Milwaukee, and Omaha is that: All four experiments that have examined this hypothesis report an interaction with unemployment consistent with the stake-in-conformity hypothesis, at least in the official data. The interaction effects suggest the importance of subgroup differences in the specific deterrent effects of arrest. However, because the cases in each of the experiments disproportionately came from neighborhoods with concentrations of the common risk factors for violence--high unemployment, poverty, and divorce rates--deterrent effects may have been confounded with broader contextual effects. The analysis of these hierarchically structured processes requires methods to disentangle social structural factors from the individual factors that contribute to violence generally and that may undermine the deterrent effects of legal sanctions. The availability of informal social controls--the potential for job or relationship loss or for social stigmatization from neighbors or relatives-- may also undermine the effectiveness of legal controls. In fact, the absence of a systematic deterrent effect across sites in the arrest experiments and evidence of the escalation of violence among men who were unemployed and unmarried led Sherman (1992a, 1992b) to conclude that mandatory and proarrest policies were inadvisable. Second, one of the experiments included an "offender-absent" group in which arrests were not made, but warrants were issued for the absent batterers (Dunford, 1990b). The deterrent effects of police intervention were clearer and more consistent across different outcome measures in this experiment than in any of the other conditions. The continuing threat of legal sanctions evidently has a stronger deterrent effect than the actual imposition of a sanction through the arrest process. This is a theme discussed later in terms of prosecution programs. Unfortunately, the Omaha offender-absent experiment was not replicated elsewhere in the Spouse Assault Replication Program (SARP) cohort. However, conclusions about the ineffectiveness of deterrence based on the police experiments may in fact be inaccurate, because the actual implementation of deterrence was minimal in these experiments. Most offenders were not prosecuted once arrested. Legal sanctions were limited to booking for most of those arrested. Few were handcuffed, most spent only a few hours in custody, and only a small number were jailed overnight (Sherman, 1992a). Only misdemeanor cases were included in the experiment, a selection bias that reduces the generalizability of the experiments. Only in the Omaha offender-absent experiment-- the experiment with consistent evidence of deterrence-- were felony cases included. Accordingly, the limitations of the SARP experiments led Garner et al. (1995) to conclude that the effects of arrests remain unknown. Prosecution Experiments The low rate of prosecution in domestic violence cases undermines deterrence by neutralizing the actions of the police and reducing the likelihood of legal sanctions following arrest. Although some level of sanction may result from arrest, deterrence ultimately results from the actions of prosecutorial and judicial actors whose actions lead to substantive punishment.5 Historically, like the police, prosecutors were accused of disinterest in family violence cases, failing to file cases presented by the police or discouraging willing victims from pursuing criminal complaints. Whether discouraged by the evidentiary problems in these cases, the ambivalence and lack of cooperation from victims to press forward with prosecution, patriarchal notions about family privacy and male prerogatives, or signals from a disinterested judiciary who were unwilling to respond to prosecution with meaningful sanctions, prosecutors had little incentive to follow through with vigorous presentation of domestic violence cases (Elliott, 1989; Ford, 1993). In many communities, the rate of prosecution remains extremely low, less than 10 percent for misdemeanor cases (Ford, 1993; Schmidt and Steury, 1989). Fagan (1989) found that fewer than 5 percent of 270 cases involving women with injuries were criminally prosecuted. Dutton (1995) reported that conviction and sentencing are even rarer. Special prosecution units. The advent of special prosecution units created an atmosphere and organizational context in prosecutors' offices in which domestic violence cases had high status. These units created incentives for vigorous prosecution without competing with other units for scarce trial or investigative resources with other high-visibility cases (Forst and Hernon, 1985). These units also created an atmosphere in which prosecutorial screening could include a wider range of factors than simply the evidentiary strength of the case or the severity of the victim's injuries. Thus, for example, prosecutors can entertain criteria that may reflect the likelihood of future (severe) abuse, such as the history of violence in the relationship or the past frequency of victim injuries. Although such factors may be less concerned with the likelihood of obtaining a conviction, sorting cases in this way provides for the allocation of legal sanctions based on the priority of victim protection rather than the typical pragmatic case-sorting factors. The likelihood of prosecution also reflects the interaction of prosecutorial case-screening decisions with victim choices. Some research suggests, however, that victims may have wider interests in mind than legal sanctions when filing for prosecution. For example, Ford (1991) suggests that victims' goals are instrumental: obtaining money or property, coercing partners to obtain counseling, or protecting themselves or their children. Thus, their evaluations of prosecution may involve a more complex determination about how prosecution, in combination with other factors, will increase their safety and well-being. To avoid these complications, some prosecutors have adopted no-drop policies that avoided the last-minute withdrawal of charges that frustrated police and judges. However, critics of no-drop policies suggest that they provide further disincentives for women to interact with the legal system to create a "context of deterrence" within the relationship, based on the possible conflict between victims' and prosecutors' goals or interests, as well as victims' perceived costs of prosecution vis-a-vis their actual goals. Thus, victims' efforts to end the violence may involve strategies in which choices are constrained by both the context of legal institutions and the contexts of their relationships. Despite the development of special prosecution units, few studies have documented the effects of prosecution on the control or recurrence of spouse or partner assault. Most studies of prosecution of partner violence have focused on prosecutorial decisionmaking regarding the sorting and selecting of cases for prosecution (e.g., Schmidt and Steury, 1989). Fagan (1989) found subgroup differences similar to those reported by Sherman et al. (1991) for arrests. Men with prior arrest records or who had lengthy histories of severe violence toward their partners were more likely to reoffend if prosecuted compared with men not prosecuted. Again, evidence of iatrogenic or counterdeterrent effects raises serious questions not only about the deterrent effects of legal sanctions but also about the interactions of violent men with legal institutions that may produce this effect. Indianapolis experiment. The most comprehensive prosecution study has been the Indianapolis Domestic Violence Prosecution Experiment (Ford, 1991, 1993). There was no significant protective effect from prosecution in the experiment, but there was a significant reduction in "severe" violence when victim-initiated prosecutorial actions were compared with the traditional summons-and-prosecution procedure (Ford, 1993). The results suggest small marginal gains in deterrence from the use or threat of prosecution, gains that are mitigated by small effect sizes coupled with small sample sizes.6 These studies raise the question of victim empowerment and the hypothesis that the threat of prosecution, placed in the hands of the victim to use in her efforts to end her partner's violence, may have deterrent effects. Several jurisdictions have encouraged the use of warrants by victims in cases in which arrests were not made. The "Sword of Damocles" model of deterrence may invoke deterrence processes through the elimination of the mediating effects of legal actors and the allocation of legal power to the woman-victim (Dunford, 1990; Ford, 1993). When coupled with informal sources of social control, the threat of prosecution may have a greater deterrent effect compared with the more typical deterrence model in which threats are contingent on the dynamics and processes of legal institutions. This indirect deterrent effect is an important avenue for future research. Batterer Treatment Treatment interventions for batterers vary in several respects. Most are court-mandated programs, some are self-help, and others operate under the aegis of social service or private agencies. Their underlying assumptions about the causal and restraining mechanisms for intimate violence vary. Their operational characteristics vary as well, including the duration and frequency of contacts and the objectives of treatment. Most address the need for anger control or dissipation techniques and recognize the relationship of power and control to the use of violence. Most do not allow for "relapse" in the way that clinical trials do in some substance abuse treatment or pharmacological research. They integrate victim safety with the offender's behavioral changes as central components of program development. There are several typologies of batterer treatment programs. Edelson and Syers (1990) distinguish programs into three categories: self-help programs that emphasize anger management strategies and personal responsibility, educational programs that teach through passive learning about the sources of violence and the techniques of anger control, and some combination of the two methods. Harrell (1991) studied programs in Baltimore County that embraced three types of underlying assumptions about causal and cessation processes: feminist theories about power relations, social casework models that emphasize reduction of external stressors and interpersonal dynamics, and cognitive-behavioral models that stress anger management. The latter models also vary in their attention to psychopathological variables versus cognitive deficits in anger control. What is not in evidence in these programs is recognition of different types of batterers or efforts to match batterer profiles to specific treatment types (Saunders and Azar, 1989). However, there may be considerable need to address these concerns of responsivity (Andrews et al., 1990). For example, violence toward intimates is more intractable to treatment interventions for men with longer and more serious histories of violence toward intimates, longer criminal records of violence toward strangers, and traumatic violence exposure as children (Fagan et al., 1984; Hamberger and Hastings, 1993). Experimental evidence. There is very little nonexperimental or quasi-experimental evidence to evaluate the effectiveness of batterer interventions. For example, Hamberger and Hastings (1993) reviewed 19 studies published in the 1980's. Only one (Dutton, 1995) used an untreated control group and reported sizeable differences in recidivism between treated and untreated batterers. In this study, only 4 percent of offenders reported subsequent violence to the police, whereas 16 percent of victims reported subsequent violence. Based on police reports during a followup period of 6 months to 3 years, the recidivism rate was 40 percent. Harrell (1991) compared outcomes across treatment programs and reported no differences, with recidivism rates of over 60 percent. However, most studies are not useful for assessing the effects of batterer treatment because they have no comparison group. The few that do have such groups rely on comparisons of completers with noncompleters, a selection bias that presents serious obstacles to the assessment of treatment effectiveness. Followup times varied as well, from 1 month to 3 years. Analyses that examine the hazards of renewed violence controlling for differential time periods were not evident in any of the studies. Measurement varied as well, ranging from self- and police reports of subsequent violence to outcome measures related to the treatment intervention but not the violence. For example, Hawkins and Beauvais (1985, cited in Edelson and Tolman [1993]) reported only on pre-post differences in Symptom Check List (SCL)-90 scores but did not report on subsequent violence.7 Sample sizes in these studies ranged from 9 to 170, indicating a range of statistical power estimates and strength of results. Among the treated groups, cessation rates varied from 46 percent to 91 percent, a range that more likely reflects the diversity in intervention methods and research designs than the true range of effects in treatments. In a similar review, Davis (1995, personal communication) found that 6 of 27 studies of batterer treatment used quasi-experimental designs of varying construction. Experimental designs using control groups of men who were under other forms of legal supervision were not evident in the literature. But such designs are of obvious importance, both to assess treatment effects substantively and to assess the marginal effects of treatment compared with other social or legal interventions. Current research by Robert Davis and colleagues at the New York City Victim Services Agency involves randomized trial of batterer treatment that addresses many of the limitations of the previous studies. Although recidivism rates in batterer treatment are similar to recidivism rates for criminal cases in general, the absence of systematic controls makes it difficult to conclude that there are marginal gains from treatment compared with either incarceration or untreated probation supervision. Systemic Responses: The Domestic Violence Court Recent innovations have focused on the creation of specialized courts to process domestic violence cases and intensive systemic reforms designed to align the components of the civil and criminal legal systems to ensure consistent application of sanctions and reliefs in cases involving domestic violence. Systemic programs, such as the comprehensive systems of coordination among legal and community-based programs in Duluth, Minnesota, and San Francisco, embed legal sanctions in a dense web of social control that reinforces the messages of treatment and the threats of criminal punishment. However, these programs are difficult to evaluate. Establishing comparison conditions internally or across communities is difficult, making it tough to sort out the effects of prosecution or advocacy from the effects of treatment. This makes it hard to answer the question of whether or how legal sanctions create a deterrent effect, and again the question of the deterrent effects of legal sanctions is unanswered. Even within these programs, recidivism rates among treated batterers are comparable to rates for protective order and arrest studies: Recidivism in Duluth ranged from 40 percent (Shepard, 1992; cited in Edelson and Tolman, 1993) to over half (Edelson and Syers, 1990) and were invariant over short and long followup periods (Hamberger and Hastings, 1993). Failures reflect the familiar correlates of lengthy prior record and abuse history in the batterer's family of origin (Fagan et al., 1983; Harrell, 1991). The creation of specialized courts for family violence cases responds to the devaluation of these cases in mainstream courts. The "stream of cases" argument suggests that cases are prioritized for processing and allocation of punishment resources according to their relative severity compared with other cases in the same context (Emerson, 1983; Jacob, 1983). This framework suggests that domestic violence cases may be assigned a lower priority for prosecution and punishment when placed alongside other violence cases involving strangers. These courts provide substantive dispositions, often batterer treatment programs coupled with probation supervision, that create incentives for prosecutors to complete prosecution. The advantages of specialized prosecution units apply to specialized courts as well. Cases are evaluated not in comparison with external priorities but in the narrow light of other similarly structured cases. Dade County Domestic Violence Court. The Dade County, Florida, Domestic Violence Court (DCDVC) illustrates these ideas. DCDVC is a criminal court with a civil component designed by a team of representatives from every segment of the criminal justice system to serve as a coordinated, systemic response to the treatment of domestic violence cases in the courts. DCDVC, which commenced in November 1992, represents an innovative, interdisciplinary, and integrated systemwide approach of a team of criminal justice system professionals to the treatment of domestic violence misdemeanor cases, civil protection orders, and violation of civil protection order cases. Currently, only misdemeanor cases are processed in the court. The members of the court, led by the judiciary, work together as a team toward a shared goal of reducing family violence. From arrest to completion of sentence, only judges trained in family violence handle the cases in a court that recognizes the necessity of expanding traditional roles and limits in an effort to create court reform in a system that has proven in the past to be ineffective and unresponsive. The founders of DCDVC believe that the combination of intensive victim services, treatment for batterers, and an active judicial role in the social contexts of the community can improve the control of misdemeanor domestic violence and avoid its escalation to more serious violence and injuries. The court is based on the following principles: o The administration of therapeutic jurisprudence creates an expansion of the traditional role of the criminal justice system, which historically has been concerned with punishment but has failed to consider the role of treatment in domestic violence cases. Defendants are required to successfully complete a batterer's treatment program based on the Duluth model (Edelson and Syers, 1990), complete substance abuse treatment, and meet other case-specific requirements such as mental health counseling. All cases are monitored by the court after imposition of the sentence, and the defendant is required to return to court periodically during probation to discuss progress in counseling and compliance with the sentence. o There is an emphasis on the needs of children who live in violent homes. Parents are educated about the effects of domestic violence on their children. The court--in partnership with a facility associated with the University of Miami School of Medicine (The Mailman Center for Child Development, which has developed a 10-week age-specific counseling program for children who have witnessed domestic violence)--makes completion of the group counseling by the defendant's children a condition of the defendant's probation. o The members of the court acknowledge and accept the responsibility, both in the courtroom and in the community, to educate the public about domestic violence and the fact that domestic violence is a crime. The role of "judge as teacher" in the courtroom is tested, and judges have a responsibility to make public appearances at community meetings and in the popular media and to educate the public about the court and about domestic violence. o The court serves as a catalyst for change as a community leader by coordinating a communitywide approach and communitywide participation in a local campaign to combat family violence. o Judicial education and training in family violence is mandatory for all judges and prosecutors and some public defenders assigned to DCDVC. Victim advocates are employed in the court to facilitate the victim's participation in the court process and to make services available and accessible to the victim. DCDVC is also noteworthy because it is currently being evaluated using an experimental design. The use of the "gold standard" in court evaluations is rare, often the result of legal and ethical complications (Zeisel, 1972; Fagan, 1990). Evaluation data are not yet available for the DCDVC experiment. What will be learned from this experiment? First, the evaluation will indicate whether legal sanctions are more likely and severe in a court dedicated to domestic violence cases. Second, the evaluation will assess the effectiveness of sanctions fashioned in the context of the broader concerns of victim and child safety plus treatment intervention for assailants. Accordingly, comprehensiveness will be a component of the court response not forthcoming in courts of general jurisdiction where domestic violence misdemeanors are docketed alongside other misdemeanor cases. Some questions will not be answered by this design, questions that are nevertheless important in evaluating the deterrent effects of criminal sanctions for violence toward intimates. First, the exclusion of felony cases limits generalizations from this study. Felony cases are important not only because of their severity but also because of the potential escalation from repeat serious offenses to potentially lethal cases (see, for example, Browne, 1987). Other forms of serious partner violence may be excluded, including "marital" rape and other sexual crimes between cohabitating or former intimates. Second, treatment interventions are provided in the context of legal sanctions, introducing legal coercion as a potential confounding factor in treatment effectiveness. The DCDVC experiment will not allow for a test of the deterrent effects either of legal sanctions or punishment on intimate violence or of the effects of treatment independent of legal threats. Factorial experiments may be needed to sort out these potentially competing and confounding factors in controlling domestic violence. Civil Legal Sanctions: Protective Orders and the Prosecution of Violators Reforms in the concept of restraining orders for battered women preceded reforms in arrest and criminal law. Beginning with the passage of the Pennsylvania Protection from Abuse Act in 1976, every State now provides for protection orders in cases of domestic violence (Klein, n.d.). In many locales, the barriers and complications of criminal arrest and prosecution have made protective orders the primary source of legal sanction and protection for battered women. In contrast to the reactive arrest and criminal prosecution processes, protection orders are victim-initiated and timely. They also allow a relaxed standard of proof, focus on the victim's protection, and prescribe a wide range of specific interventions or reliefs that address extralegal concerns of safety and economic well-being. However, few studies have examined the effectiveness of restraining orders in reducing the incidence of domestic violence, and those few studies have been nonexperimental or quasi-experimental with designs that weaken any conclusions about their effectiveness. Moreover, there are little data on the extent to which protection orders are used in conjunction with criminal prosecution. How effective are protective orders in stopping domestic violence? Harrell, Smith, and Newmark (1993) found that 60 percent of 300 women interviewed twice in 1 year after receiving a protective order suffered abuse at least once. Over 1 in 5 reported threats to kill; severe violence was reported by 29 percent. Other acts of violence were reported by 24 percent, and property damage was reported by 43 percent of the women. Threats and violence did not subside over time, and there were no significant differences in the percentage reporting subsequent violence in the first 3 months of the year compared with the final 9 months of the year. Klein (n.d.) used official records (new arrests for domestic violence, new restraining orders against the same defendant issued by the same victim) to measure reabuse in 644 cases in which temporary restraining orders were issued. Nearly half (48.8 percent) of the men reabused their victims within 2 years of the issuance of a restraining order. Moreover, over half (54.5 percent) were rearrested for other crimes as well. Neither of these studies reported results for comparison or control groups. Grau, Fagan, and Wexler (1984) found no significant differences in subsequent abuse between women receiving restraining orders and women receiving other interventions. Moreover, they reported that subsequent violence was more likely among men with histories of severe domestic violence or prior records of stranger crime. Limitations and Contradictions in the Criminalization of Domestic Violence There is little conclusive evidence of either deterrent or protective effects of legal sanctions or treatment interventions for domestic violence. A closer reading of this literature suggests several issues that may lead to a better understanding of why past research has failed to locate deterrent effects and whether and how law influences the control of domestic violence. The issues fall into three general domains: the embedment of domestic violence in complex social and individual contexts, weak research designs and limitations on policy experiments, and the theoretical issues in male violence. Legal and Social Control of Domestic Violence The experiments on the effects of arrest on domestic violence raised important hypotheses on the interaction of legal controls (such as arrest and prosecution) with informal social controls (such as social bonds). Sherman et al. (1992a) and Berk et al. (1992a) in multisite analyses and Pate and Hamilton (1992) in a single-site analysis reported significant interactions of arrest with two measures of "stake-in-conformity" (Toby, 1957): Arrest increases the risk of violence for unmarried and unemployed suspects and deters it for married and employed suspects. Results from four of the replications concluded that ". . .all four experiments that have examined this hypothesis report an interaction with unemployment consistent with the stake-in-conformity hypothesis, at least in the official data" (Sherman et al., 1992a:687). Fagan (1989) reported similar interactions between sanctions and social position among prosecution cases, and Harrell (1991) reported the same patterns for batterers in treatment. The lesson of these studies is that formal (legal) sanctions are effective when reinforced by informal social controls and weakened when those informal controls are absent (Tittle and Logan, 1973). Williams and Hawkins (1989a, 1989b) suggest that deterrence of domestic violence is contingent on reciprocity of formal and informal social controls. They suggest the deterrent effects of arrest will be greater for batterers who perceive higher social costs associated with the act of violence and with arrest (Bowker, 1983, 1984). These costs include loss of job, relationship and children, social status in the neighborhood, and whatever substantive punishment they receive. Accordingly, the social and structural position of batterers, including their prior punishment experiences and the meaning they attach to them, will mediate the deterrent effects of sanctions. When batterers perceive that punishment is not a cost worth avoiding, legal sanctions alone are unlikely to induce compliance with the law. The sources of informal social control of domestic violence may lie either within the individual, in the form of internalized beliefs and social bonds, or may be externally reinforced through normative behaviors within neighborhoods and other social contexts (Williams and Hawkins, 1989b; Fagan, 1992).8 That is, social and economic conditions may shape the motivations and perceptions of batterers regarding the salience of legal sanctions and the extralegal costs that accompany them (Zimring, 1973). Both in Minneapolis and the other arrest replication sites, the cases disproportionately came from neighborhoods in which risk factors were high: high unemployment, poverty, and divorce rates. The availability of informal social controls--the potential for job or relationship loss or for social stigmatization from neighbors or relatives--in those neighborhoods may have attenuated the development of informal social controls and, in turn, undermined the effectiveness of legal controls. This suggests that opportunity structures at the neighborhood or community level have direct effects on the availability of informal (or extralegal) individual-level controls that are critical reinforcers of legal sanctions. Legal Contexts of Criminalization and Reform The legal context in which case processing and sentencing decisions are made may influence the severity of sanctions handed down in domestic violence cases. For example, norms within local legal "cultures," such as practitioner beliefs about whether victims will drop charges or patriarchal views of domestic violence, influence the likelihood of official action when there is discretion. Differences in court "cultures" regarding domestic violence may explain the variation across communities in the rates and severity of sanctions for these offenses.9 Like many other legal reforms, criminalizing domestic violence may have unintended consequences, reflecting the social organization of the courts and processual contexts, rather than legal statute. The relatively recent entry of domestic violence cases into court calendars also challenges the existing calculus of how cases are evaluated. Emerson (1983) argues that decisionmakers evaluate cases against a backdrop of other cases and that assessments of seriousness are relative and occur in comparison with other cases. If criminal authorities in different jurisdictions classify objectively similar cases differently because of their respective caseload contexts, then the sanctions in each jurisdiction should reflect the comparative position of domestic violence relative to the other cases before those judicial authorities (Hassenfeld and Cheung, 1985). Accordingly, the criminalization of domestic violence will influence the reactions of officials in the working groups that exist within each court. The criminal court may produce less stability in processing domestic violence cases because bringing battered women into criminal court entails a change in standard operating procedures. Jacob (1983) suggests that criminal court participants, particularly prosecutors who possess more information than other courtroom personnel and who have a disproportionate influence over the disposition of cases, will behave inconsistently when faced with a new class of offenders (Mohr, 1976). This may lead to less cohesiveness among the working group members of the criminal court whose social organization is geared to case attributes of stranger crimes and less oriented to the special circumstances of cases involving intimate relationships. Prosecutorial screening continues to reflect the strength of the case, with prosecutors taking their cue from the actions of police as well as the likelihood of conviction (Schmidt and Steury, 1989). Similar processes may influence the construction of restraining orders in the civil (family) court. There, the presentation of cases involving family members is consistent with the historical stream of cases around which the legal cultures and working groups have evolved. However, courts accustomed to divorce and custody proceedings may evaluate domestic violence cases as different and perhaps less weighty compared with the majority of their dockets. Accordingly, the "going rate" for crimes involving violence between intimate partners may not be any higher compared with stranger violence cases in the criminal court or divorce and child custody cases in the family court, and in fact may be lower (Emerson, 1983; Mohr, 1976; Mather, 1979). The Complexity of Domestic Violence Compared with many other violent crimes, the legal and social dimensions of domestic violence present several complications for effective legal control. Domestic violence differs significantly from other forms of violence in several important ways. First, there are strong emotional ties between victims and assailants. The parties often love one another, or at least the victim may love the assailant. The bond may be traumatic (Dutton and Browning, 1988), complicating victim resolve to enter into a lengthy adversarial proceeding to invoke punishments and creating internal conflict regarding separation. The victim may be financially dependent on the assailant or may face a severely diminished standard of living if separated. Arguably, she faces an economic life at or below the official threshold of poverty upon leaving the relationship (Sidel, 1986). These ties to assailants may lead victims toward rational objectives in invoking legal sanctions. Thus, they may be less concerned with deterrence than they are with using legal institutions to guarantee their own safety, survive economically, protect their children, or get counseling help for their assailants (Ford, 1991). They may also see the threat of prosecution as a means of terminating the relationship and escaping the violence (Lerman, 1992). Thus, victim choices about invoking legal sanctions may be less concerned with punishment and deterrence and ultimately seek to use the law for other goals. The social control functions of the law are compromised in this context even when victim choices and well-being are optimized. Second, domestic violence often is a recurring event between individuals in daily contact, usually without the forms of guardianship and surveillance that are available in public spaces. Unlike robberies, in which victims and offenders often are unacquainted, or other assaults involving acquaintances, victims and assailants often occupy the same space, share and compete for resources, and have emotional ties. In this context, threats are readily conveyed and quite believable. On the other hand, it is extremely difficult to mount and maintain a deterrent threat within a context of ongoing and unsupervised contact between victim and assailant. Third, the scale of domestic violence makes it difficult to control solely through legal sanctions and deterrent threats. The base rates remain quite high relative to other violent crimes, with self-reported domestic assault prevalence rates of at least 10 percent for both men and women (Straus and Gelles, 1986). Prevalence rates exceed 30 percent for some subgroups. Domestic violence rates are highest among subgroups who also have high rates of stranger violence, further burdening limited police resources within spatial areas where assaults are concentrated (Fagan, 1993). Many cases are unreported, and estimates of the extent of reporting to the police are as low as 20 percent (Dutton, 1995). However, those who do report appear to be individuals who have few nonlegal resources for protection or deterrence (Bowker, 1983). Even if reporting were not increased, the high rates of domestic violence make it difficult for police departments to arrest every man who commits a misdemeanor or felony assault against his partner, much less to arrest him every time he does it, without paralyzing their own agencies and the courts. In the face of a high base crime rate, police departments are challenged to maintain a credible deterrent threat in cases where arrests do not occur. Finally, the deterrence logic of criminalization assumes a rational offender-actor who weighs the costs of offending--costs associated both with the act itself and the legal actions that ensue-- against whatever benefits that may accrue from the behavior (Miller and Anderson, 1986; Fagan, 1992). This logic is strained in the context of domestic violence. Although domestic violence has been interpreted as a goal-oriented and implicitly rational behavior (Tedeschi and Felson, 1995), episodes of rage during more serious assaults often obviate rational calculations and perceptions of costs (Browne, 1987). Studies with batterers in treatment suggest conditions of impaired cognition or mental disorder (Dutton, 1995). The logic of deterrence is compromised among batterers whose behavior is patterned over time and for whom rational calculations are not possible during the arousal of a violent assault. Domestic violence is unique in the concentration of risk factors and absence of formal controls for violence. Only the reciprocity between legal and informal social controls makes possible the control of domestic violence in general. Among violent men whose behaviors are increasingly spiraling out of control, the threat of punishment may be remote and inconsequential under conditions of arousal and cognitive distortion. Weak Research and Evaluation Designs Few empirical studies use research or evaluation designs that can detect deterrent effects of legal sanctions. Accordingly, it is difficult to determine the extent and magnitude of effects from legal sanctions. The empirical literature is littered with weak evaluation designs. Experiments are rare, as are clinical trials for treatment interventions. Most studies have small samples and limited experimental power. Followup periods are too short (usually 6 months), making it impossible to see longer term effects that may accrue. Budgetary limitations for many grant programs, including the National Institute of Justice, historically have constrained the results of research and evaluation on domestic violence. If the development and testing of theory is a cumulative process from repeated experiments, the foundations of empirical knowledge to advance theory and practice are not available for domestic violence. In intervention research, sanctions are conceptualized and measured using dichotomous variables, a strategy that fails to account for variation in the delivery of interventions. The implementation of the underlying theories of interventions also is overlooked in most intervention research. These strategies lead to two sources of error. First, we may falsely reject theory when in fact the theory was not adequately implemented in the interventions. This is a case of program failure, in contrast to a theory failure, that reflects a valid experiment. This source of error can be addressed by careful attention to the measurement of treatment .strength and integrity" (Sechrest et al., 1979). Program content and ideology are critical elements of intervention that often are not captured adequately in batterer treatment research. Second, the absence of continuous measures of interventions or sanctions may fail to detect incremental or marginal effects. That is, "dosage" is a critical yet unstudied dimension of research on legal sanctions generally and particularly in domestic violence (Sherman, 1992a). Most studies have conceptualized legal sanctions as discrete variables with limited range. Few studies have analyzed the effects of legal sanctions within a framework of increasing severity. Thus, different forms of arrest may constitute qualitatively different levels of sanction. Legal sanctions may involve different levels or conditions of probation supervision or treatment regimens of various intensity. Restraining or protective orders for battered women involve a wide range of the reliefs, and the number and types of reliefs vary extensively across cases (Fagan, Maxwell, Macaluso, and Nahabedian, 1995). Accordingly, the limited effects of legal sanctions may reflect analytic and measurement strategies that may conceal important differences in the extent and severity of legal sanctions. A related concern is the narrow range of sanctions in most experiments on legal interventions. Since the 1970's, the range of legal sanctions for batterers has expanded at a glacially slow pace. Although arrests have increased, the substantive sanction in most cases remains simply the process of arrest. The range of sanctions in the arrest experiments expanded in small increments. Few arrestees were handcuffed, and a relatively low percentage of the arrested group spent anywhere from a few hours to overnight in jail following arrest. Prosecution rates remain low. Pretrial and postconviction treatment regimens vary widely in the intensity of the treatment and the burdens they place on assailants. There is limited evidence of the use of incarceration or more intensive forms of supervision unless injuries are serious. When the range of punishments or sanctions is narrow, the validity of tests of deterrence is intrinsically weak, and the likelihood of detecting a reduction in violence appears remote. There are other limitations on research design that weaken empirical findings. Although experiments represent a "gold standard" of social research, there are many circumstances in which random assignment is neither practical nor ethically justifiable. In particular, untreated control groups are not tenable when victim safety is at risk. However, there are alternatives that do not get used: Factorial and bootstrapping designs, in particular, are valid design options that are absent in this literature. Censoring (exclusion) of cases and attrition are sources of selection bias. Analysis of subgroup differences often is not reported, even though it is reasonable to expect different outcomes among population subgroups. Treatment effects are likely to decay over time, yet analytic strategies rarely examine the effects of sanctions on the temporal dimension of recidivism (Visher, Lattimore, and Linster, 1991; Lattimore, Visher, and Linster, 1995). Finally, there are exogenous influences on sanction effects that often are not addressed in research on domestic violence (Fagan, 1993). For example, subcultural influences may overwhelm the effects of legal sanctions or treatments in motivating domestic violence. That is, high divorce rates may devalue marriage or coupling, weakening the informal controls on violence that work reciprocally with legal sanctions. Residential mobility, high rates of poverty, and weak social cohesion are dimensions of social disorganization that weaken informal social controls on violence generally and undermine motivations for compliance with the law. Problems in Designing Policy Conflicts in policies about legal interventions have led to ambiguous findings in the domestic violence literature. Ambivalence best describes the policy goals and theoretical under- pinnings of criminal justice interventions. For example, many of the developments in prosecution of domestic violence have been designed to increase the involvement of prosecutors, but strategies diverge on the identification of goals. Although some reforms are designed to punish offenders and establish a specific deterrent effect, other policies aim to protect victims by threatening prosecution while achieving other important goals such as economic relief, victim safety, or coerced counseling. These are competing alternatives that are not necessarily compatible. They involve focusing resources on victims versus offenders and using resources (especially prosecution resources) for goals that do not involve conviction and punishment. The critical evaluation question is whether legal institutions organized around the goal of detecting and punishing crimes can effectively shift toward a more flexible and preventive set of activities. Many of the reforms and innovations in arrest and prosecution have been designed to empower victims and afford them a greater role in decisionmaking on the use of legal resources. However, the social organization and legal "culture" of these institutions is challenged by this additional focus. These policies may raise internal contradictions, offer risks and tradeoffs (for example, victim autonomy versus prosecution of violent offenders), and complicate the evaluation of "effectiveness" of legal interventions. Moreover, criminal justice institutions are asked to make linkages to social service agencies in domestic violence cases that they are not asked to make in other types of crimes. Finally, do these efforts control violence, especially the repeat violence that may escalate toward lethality? Most policy statements would include the reduction or cessation of violence, yet current evidence from these reforms is not promising. Most important, current research and evaluation efforts have not asked these questions. Practical Limitations in Mounting Deterrence Criminal and civil legal reforms over the past 20 years have raised the priority of domestic violence cases within legal institutions. However, implementation of policies to control domestic violence competes with other crime and violence problems for limited resources. Several criminal justice problems have competed for resources over this time, each with an urgency that demanded a share of a fixed pool of resources. For example, special drug courts have been established in several jurisdictions, prosecutors have developed special units to prosecute gang violence, and investigative resources have been allocated to child abuse cases in the wake of increasing child fatalities in the late 1980's. The result is uneven or weak implementation of newly developed policies for domestic violence with the unintended consequence of weakening the criminal justice response. On a more practical and day-to-day level, domestic violence cases compete with violence and other patrol priorities for immediate attention by the police. Consider the following example. It's Friday night, and there are five cars on patrol in a city of approximately 125,000 people. One car is investigating an injury accident, another is responding to a fight in a bar, another to a report of a man with a gun, another is directing traffic at a corner where a traffic light is not working, and another is investigating a report of a robbery. At 10:30 p.m., three domestic violence calls come in. What does the dispatcher do? Consider a second scenario: The officer in the fourth car leaves the traffic light scene to respond to one of the domestic violence calls. She is about to make the arrest when she receives a radio call about a robbery in progress. What does she do? What is she ordered to do? These are not uncommon scenarios. The reality of competing priorities for sorting cases for arrest and prosecution suggests that domestic violence cases, especially low-injury or noninjury domestic violence, will not receive a higher priority than other events. Similar scenes can be imagined in prosecutors' offices. It is not uncommon that within many "legal cultures" and working groups in prosecutors' offices, prosecuting domestic violence cases is not a pathway to recognition and promotion even when resources are organized in a way that makes such prosecution possible. A related concern is the implementation of rapidly proliferating treatment programs for domestic violence assailants. Consider first that there is virtually no methodologically sound evidence of effective treatment interventions for domestic violence. Yet many jurisdictions have mandated a wide variety of treatment interventions of varying lengths, behavioral orientations, qualifications of providers, and level and type of criminal justice supervision. In addition, these programs vary extensively on such important dimensions as victim safety planning. In one State, a statewide program mandates weekend treatment regimens of approximately 36 hours, including showing of films and testimonials or confessions from batterers. There is little chance of success in such atheoretical efforts with minimal implementation. Implementation of law and policy suffers from these types of real-world constraints. What appear to be weak policies in fact may reflect weakly implemented policies or policies whose goals are undercut by resource limitations and organizational constraints. The Failure to Recognize the Importance of Differentiated Responses for Different Types of Battering Analyses of SARP data from Milwaukee and other sites suggest the possibility of interactions between formal and informal sources of social control. What has not been tested is the possibility of differences in the effects of legal sanctions for different types of batterers. Yet there is ample reason to proceed in this way. For example, Fagan et al. (1984) reported consistent differences in the recurrence of recidivism between subgroups of batterers defined by the severity of their prior violence. Just as domestic violence is best understood from the characteristics of batterers (Hotaling and Sugarman, 1986), so too may the effects of legal sanctions be best understood based on the battering careers of violent males. Yet most research on the effects of legal sanctions for domestic violence has treated batterers as a homogeneous group. This obscures potentially important subgroup differences in the effects of legal sanctions. Moreover, failure to distinguish analytically among subgroups may mask potential iatrogenic effects from legal sanctions that elevate risks for victims of more serious assaults. Several studies have suggested typologies of batterers that distinguish them along several dimensions. Holtzworth-Munroe and Stuart (1994) reviewed 19 empirically derived typologies based on either rational-deductive strategies or empirical-deductive strategies. They identified three dimensions that distinguish among subtypes of batterers: severity of marital violence, generality of violence (toward strangers as well as intimates), and psychopathology or personality disorders. Based on these dimensions, three types of batterers are hypothesized: family only, generally violent, and dysphoric or borderline personality batterers. Each of these types is hypothesized to be involved in different levels of severity of domestic violence. There is utility in typologies such as these to predict responses to legal sanctions. For example, impulsivity and low self-control characterize generally violent batterers, personality variables that may complicate the rational logic underlying deterrence theory. The family-only batterer engages in the least severe forms of violence and also exhibits the lowest levels of impulsivity and may be most amenable to legal sanctions. Analyses by Fagan et al. (1984) confirm the different reactions to legal sanctions for family-only batterers versus generally violent males. These dimensions are rarely considered in research on domestic violence yet should be an important component of sanction and treatment research. The failure to consider these dimensions is a failure to identify factors that may mitigate the effects of sanctions. For now, the empirical literature does not include any studies that examine the relationship between types of batterers and the effects of legal interventions. Treatment research with offenders generally has recognized the importance of "responsivity" of different types of individuals to various interventions (Andrews et al., 1990). Understanding the effects of legal sanctions for batterers must account for the different responses of different types of batterers to both types of sanctions and their "doses." These factors also may be important and useful in sorting cases for prosecution or in determining the extent to which sanctions may risk victim safety. We know, for example, that when there is a lengthy history of prior calls for service, stronger legal intervention may be needed compared with cases in which there is a shorter history (Fagan et al., 1984). The Segregation of Theories of Interpersonal Violence From Theories of Domestic Violence A corollary concern is the extent to which theories of violence generally inform research on domestic violence, including the effects of legal sanctions. The typology suggested by Holtzworth-Munroe and Stuart (1994) is based on a range of personality and developmental variables that were derived not only from research on batterers but from the literature on violence generally. Yet theory and research on domestic violence have segregated theories of violence from theories of battering. The social and ideological constructions of battering have limited the types of variables considered in research on domestic violence. Assuming that patriarchy and power relations alone cause domestic violence leads us toward conclusions that do not consider a full array of explanatory variables from other disciplines (Fagan and Browne, 1994). However, assuming that domestic violence is caused by a more complex set of hierarchical influences--for example, weak social controls, situational arousal, or even psychopathology--may lead us in quite another direction. The importance of recognizing factors from theories of violence that may influence the effects of legal sanctions is evident from the types of variables that define the typologies of batterers. Their inclusion offers a significant advance over the current level of empirical knowledge. The Role of Legal Institutions in the Control of Domestic Violence The criminalization of domestic violence proceeded from two perspectives. For advocates of battered women, mobilizing legal institutions was designed to have symbolic and generally deterrent effects. But these reforms also included goals to protect women victims of domestic assaults through the mobilization of extralegal services and the development of referral linkages. Also, by lending the political authority of legal institutions in efforts to prevent domestic violence, the moral authority of messages from women's groups and community-based organizations was reinforced. The inclusion of legal sanctions in a network of services helped to expand the web of social control designed to protect women victims. All these focused on women. But for actors in legal institutions, these reforms were focused on a class of offenses and offenders now prioritized for adjudication. The advent of services for domestic violence within legal institutions offered the promise of addressing violent crimes that previously had eluded the mechanisms of sanctions and legal control. Goals and expectations differed in these two perspectives, particularly in terms of the focus of the policies. In pursuing victim protection goals, legal institutions, especially criminal justice system agencies, were asked to refocus their efforts on the protection of victims and the coordination of extralegal and legal services. This perspective differs from the traditional goals of criminal justice institutions to focus on the detection and punishment of crimes. In this view, assailants are the focus of the efforts of criminal justice agencies. Accordingly, for domestic violence, policy goals coexisted within legal institutions to both punish offenders and protect victims. These goals may be reciprocal as policy but may be in conflict at the operational level. Placing expectations for police and prosecutors to invoke informal social controls in which legal sanctions play an indirect role may require tasks and roles for personnel for which they are not well trained or that may contradict the roles and expectations in their jobs with respect to other types of crimes. It may require legal actors to pursue goals in domestic violence cases that they do not pursue in other types of crimes. Also, for crimes of the scale of domestic violence, it may be unrealistic to expect legal institutions to effectively control crimes that affect significant portions of the population. These contradictions raise concerns because they may undermine the effectiveness of legal institutions in stopping domestic violence. Role and policy ambiguity can affect the performance of agencies with respect to their missions; in this case, it may undermine their effectiveness in pursuing either victim protection or offender sanctioning roles. There is no doubt that linkages between legal institutions and services for domestic violence victims are critical to stopping violence. However, these linkages may best be accomplished through a strategic division of roles among institutions that tap the strengths of each organization. Domestic violence is best explained by the characteristics of men. Social control is most effective when legal controls interact reciprocally with extralegal social controls. This suggests that the role of legal institutions in stopping domestic violence may most effectively focus on the detection and punishment/control of batterers and indirectly on the coordination of extralegal services to protect battered women. Although legal systems should be open and accessible to battered women, these institutions should not take on the role of managing the coordination of services that involve social service, shelter, and other interventions. Pursuing goals for specific types of cases that may conflict with the primary mission of legal institutions raises the danger of marginalizing those cases. It was the historical marginalization or denial of domestic violence cases that motivated contemporary reforms to increase criminal justice system involvement in domestic violence. The question now, after two decades of reform, is what type of involvement of criminal justice agencies works best to control domestic violence? By emphasizing the deterrence and punishment of domestic violence, legal institutions focus their efforts in directions that may permit them to maximize their effects. Advancing Knowledge and Policy Through Research and Development For over 20 years, research on the effects of increased criminal justice involvement on domestic violence has emphasized systemic reforms and efforts to increase the rate at which legal sanctions are applied. Yet there remains inconsistent and inconclusive knowledge about the effectiveness of criminalizing domestic violence on controlling repeat victimization. Research on the effects of legal sanctions has been limited by weak research designs, a narrow range of theories, poor conceptualization of potential interaction effects and subgroup differences, weak interventions and sanctions, and implementation problems. We simply do not know what the effects of legal sanctions for domestic violence are, whether there are differences in these effects for specific population groups, what the theoretical bases are for their effects or noneffects, and what the risks and limitations of a policy of "criminalization" are. This dilemma is partially the result of a strategy for knowledge development in which well-intentioned reforms were mounted but with weak evaluation designs that often were introduced after programs were designed and launched. The dilemma also reflects a reluctance to ask the more difficult questions of the utilitarian effects of these reforms with respect to the control and cessation of subsequent violence. But this state of knowledge has left us without an adequate basis for formulating policy or practice. Given the current state of affairs, a reviewer a decade from now may well conclude again that we still do not know whether legal sanctions can effectively control domestic violence. Another reason has been the segregation of evaluation research from basic, theoretically driven research. In domestic violence and many other social policy areas, evaluation has been an enterprise quite separate and apart from basic research. But recall that the Minneapolis experiment and the replications in SARP have been the most influential studies in the development of legal policy on domestic violence. These were theoretical studies, not demonstration evaluations or policy experiments. They were tests of deterrence theory sponsored by the National Institute of Justice Crime Control Theory Program. The lesson is simple: The greatest gains in knowledge and policy have come from theoretically driven studies. This lesson should form the foundation for developing a research strategy that will begin to examine the effects of the criminalization of domestic violence and foster gains that will lead to more effective policies and greater safety for victims. To begin the development of a cumulative body of theoretical and empirical knowledge to inform policies, a research program is needed that addresses the concerns and limitations of existing research. Such a process can be translated to other criminological problems and form the basis of a "model" for building knowledge and policy. There are several steps to this effort. Establish a Framework To Develop and Organize Knowledge To Inform Policy Over the past three decades, disparate voices have called for the creation of an experimenting society (Campbell, 1969; Riecken et al., 1974; Boruch, 1994). These essays were noteworthy for their endorsement of experiments to inform policy. But they also suggested a dynamic process in which theory, method, and practice should converge to inform policy. Knowledge is cumulative, and these "social experimenters" sought to influence and rationalize how contemporary policies are developed. The point here is not necessarily that all research should proceed from an experimental base. Experiments are difficult to mount and often in domestic violence are ethically unsound. Rather, knowledge and policy will advance when built on a cumulative foundation of empirical evidence, practical wisdom, and theory. All three elements are needed to move policy forward; none is sufficient by itself. The threshold for what constitutes knowledge should be high, and knowledge should be cumulative. Leaders in the development and testing of innovation in criminal justice should commit their agencies and organizations to this model of knowledge development. The sources of ideas to fuel this process for domestic violence must come from a variety of efforts. Qualitative research, often providing the context for "discovery" of social processes, is critical to develop testable hypotheses. The studies of Bowker (1983, 1984) are examples of how theory can be constructed from a "thick description" of the processes that women invoke to end domestic violence, including the complex interactions of legal and social sources of control. Research on interpersonal violence generally also contributes to the formation of theories about how legal sanctions might work. For example, recent ethnographic work on male violence highlights the role of "hypermasculinity" in the genesis of bar fights, fights that often are conflicts over women and status (Oliver, 1994). In addition, formative evaluation of innovative and effective practices can provide testable ideas. When formed in the context of theory, these studies also can drive the design of rigorous studies to test their effects under controlled conditions. For example, pilot studies on the effects of an experimental program of police-social service interactions in New York suggest promising results (Davis and Taylor, 1995). Replication and extension of these experiments are needed to inform policy. Build a Foundation of Theories Although criminalization of domestic violence has proceeded apace for over two decades, only recently and perhaps after the fact did theoretical research begin on its crime control effects. Before that, important formative research had focused on mechanisms to mount deterrence: increasing the certainty of arrest, developing policies for increasing the involvement of prosecutors, creating treatment programs. The few studies that have been built on a theoretical foundation have identified important interactions and contingent effects that need further elaboration and testing. The question is, however, which theories? And, whose theories? Theories about the motivations and control of male violence generally and specifically toward intimates should be integrated with deterrence and social control theories that guide criminal sanctions. Recognition of subgroups and their differences along key theoretical dimensions (e.g., developmental backgrounds, cognition, mental disorder, embedment in violent social networks) should be part of the conceptual development and testing of interventions--whether legal sanctions or treatment regimens. Theories about the reciprocity between informal and formal social control should be part of the foundation for testing the effects of criminal sanctions. Theories about the contextual effects of neighborhoods and communities that influence the salience of sanctions also should be part of the testing of legal sanctions. Perhaps most important, the development of theory within a framework of cumulative research suggests that no research or evaluation should proceed without a theoretical foundation. It is theory that is generalizable, not practice in the absence of a conceptual framework for its effects. While cautiously avoiding "decontextualizing" the complexity of domestic violence (Lerman, 1992), we need to examine the interfaces of theories of violence, domestic violence, and social control in the context of the dynamics of domestic violence. Theoretical perspectives on violent events can complement theories on violent persons, providing unique perspectives on how social contexts shape the onset of domestic assaults (Tedeschi and Felson, 1995). Develop a Program of Research and Development To Test Theoretically Driven Interventions and Policies To accomplish these goals, evaluation must be institutionalized as part of a framework for assessing policy. Although experiments are preferred, controlled testing involves a variety of designs that do not necessarily require the randomization of people. In fact, there is growing interest in experiments in which programs, organizational units, or communities are the units of control and analysis (Boruch, 1994). Nevertheless, experiments involving individuals continue to be important, although they are ethically challenged. Alternatives that maintain experimental design but avoid no-service controls are available. For example, yoked designs involve random allocation of subjects to different combinations of interventions, thereby avoiding the problems of no-service control groups. A variety of other alternative design options are available for constructing control groups. One design may include case controls from other programs or from a group receiving a competing intervention. Multiple baseline comparisons, prior program cohorts, and other alternatives to random assignment can produce results with high internal validity (Rothman, 1986). Other dimensions of evaluation should include careful measurement of implementation and "therapeutic integrity" within programs as well as "dosage" to individuals within the program. Baseline and postprogram measurement of violence should be specific with respect to time, action, location, circumstances, and outcomes (injuries). The limitations of official records suggest that multiple measures of postprogram violence should be recorded. Considerable effort is needed to avoid sample attrition of individuals with the highest risks of reabuse or reinjury. This will allow for the measurement not only of multiple outcomes but also of multiple dimensions of recidivism necessary for deterrence research (Blumstein et al., 1978). Results should be disaggregated among population subgroups, if sample sizes permit, to examine offender-intervention interactions. This will ultimately contribute to knowledge about responsivity of different individuals to different forms of treatment (Andrews et al., 1990). Followup periods should be sufficiently lengthy to determine the decay rates of treatment and the factors that bear on postprogram failure. Statistical power must be measured and reported as part of the evaluation of significance and effect size. Evaluation should be made a requirement for fiscal support. Ongoing assessment of programs is good management, and control of risks and improvement of effectiveness are two dimensions of that assessment. Although programs may rightfully fear the withdrawal of funds when programs are ineffective, there are two reasons to take that risk. First, ethical standards mandate that programs ensure they are not doing harm, and the costs of harm in a violence intervention are quite high. Second, poor results should be a cue for refinement of program design not a sign to abandon efforts at improvement. Funders must be educated similarly that political risks are necessary for the evolution of successful and effective programs. Finally, basic research on domestic violence and violent offenders should become routinized within programs and ongoing services. The need for basic research on violent offenders and violence is evident from the recommendations of three major commissions on violence (American Psychological Association, 1993; Reiss and Roth, 1993; Centers for Disease Control, 1993). Yet many programs, whether private or public, see basic research on the causes and correlates of domestic violence as a task for others. Some see research as a burden, others as a distraction from their mission, and still others see it as exploiting their clients. Even when there is recognition of its importance, knowledge generated from research often is assigned a lesser value than folk knowledge gained from anecdotes and the reflections of staff and administrators. Programs must realize the opportunities for knowledge development from their interactions with violent offenders. There is critical information from research that can inform both theory and practice in interventions. Construct a Stable Infrastructure for Supporting Research Research support is inconsistent and generally at levels too low for thorough testing of policies and practices in domestic violence. The complexity of followup with victims and offenders alone will consume time and resources. A strategy for stable funding will need to address several dimensions to overcome the structural limitations of the current research context. First, we must carefully consider the infrastructure for funding both programs and research on interventions for domestic violence. Funds for research, like program funding, tends to be driven by streams tied to specific agencies or problem definitions. In communities with serious violence problems and extensive service networks, a more rational and need-driven basis for supporting evaluations should be constructed. Accordingly, a "superfund" for evaluation could be constructed with contributions from specific government entities (that support services), private foundations concerned with developing effective violence prevention and interventions, and research agencies concerned with developing basic knowledge or evaluation data on domestic violence. Evaluation block grants from Federal agencies could provide a Federal share for local or State evaluation "superfunds." Funders of services, programs themselves, or researchers could request evaluation support from the fund. Second, the structure of research should be considered--longer studies of broader scope are needed to pursue certain questions. We need to set realistic goals for research; we cannot realistically expect answers to complex domestic violence questions in 2 years and $250,000. Reforms to the peer review process should be part of this effort. For example, stable and continuous review panels could work closely with applicants to refine and revise promising proposals. Third, research and evaluation must be supported externally and at appropriate funding levels as part of funding for program operations. Programs should not be confronted with choices between services and research. Part of the institutionalization of research and evaluation should be the creation of a stable funding stream independently from services funding. This will support an uninterrupted research agenda that is funded at a level to create valid information. Independence of researchers from programs is necessary to ensure that programs receive an impartial assessment. The principle of risk-responsivity (Andrews et al., 1990) should apply as well to funding: the higher levels of funding should be allocated to the programs that deal with the highest level of risk or threat. The production of valid and generalizable research knowledge is not cheap and may cost as much if not more than interventions. Political "shyness" over this reality must be set aside. Finally, collaboration should be encouraged between universities and domestic violence services and intervention programs, whether they be community supervision, legal sanctions, or residential treatment. For example, doctoral programs that emphasize research can establish field placements or internships with intervention agencies to initiate either basic or evaluation research agendas. Violence is a complex phenomenon, not well explained by the traditionally separate disciplines of the behavioral and health sciences. Because it involves theoretical knowledge from several disciplines, the creation of internships within programs can foster interdisciplinary research and the advancement of knowledge beyond the limitations of single disciplines. The proliferation of commissions and legislative actions suggests that the control of violence has become a national priority. Funding for basic research, evaluation, and intervention programs should reflect that priority. One reason for the inadequate knowledge base about violence or its interventions has been the traditionally low level of funding for violence research (Reiss and Roth, 1993; American Psychological Association, 1993). Reductions in violence, like progress in the fight of disease and technological advancement, will begin when there are investments in knowledge development commensurate with the urgency of the problem. Two Additional Concerns Support development of methodological tools for consistent research. Measurement error and design inconsistencies make studies often noncomparable. There is much controversy surrounding measurement and design in family violence (Weis, 1989). A program of support for the development of validated measures and methodological innovations would provide a compatible body of knowledge for synthesis and theory building. Translate research findings to inform policymakers and practitioners. Policymakers, practitioners, and advocates rightfully complain that academics produce alien and unreadable documents that are not helpful in their work. But asking academics to recast their work in nontechnical language may require skills that they may not have. Let academics and policy researchers be technical; do not ask them to direct their efforts toward a different audience than their peers who are the gatekeepers on theoretical and substantive knowledge. Instead, enterprises are needed that create multiple products for diverse audiences from these technical reports. This form of social science journalism will provide an invaluable bridge from social and behavioral science to the audiences who will implement policies and ideas. A Final Note Without meaningful change in the structure of research and evaluation in domestic violence, a reviewer 5 or 10 years from now will likely reach the same conclusions reached in this review: "We just don't know, the evaluation data aren't very good." We could have said all this 5 years ago and actually did say it 10 years ago (Boruch, 1994). Let's not be embarrassed or embarrass ourselves by continuing on this frustrating path of fad-driven and nonsystematic policies with weak after-the-fact evaluations. Collaborative research to develop and test theoretically driven interventions and policies will make a significant contribution to the development of policies for legal interventions to protect battered women. A continuation of the research efforts of the past two decades will not. Notes 1. However, throughout this period in nearly every State, civil and criminal legal sanctions have remained separate though parallel remedies with divergent underlying legal theories and behavioral assumptions. Although legislatures have acted to increase the use of both civil and criminal legal sanctions to control domestic violence, there is continuing discussion of how the court system can most effectively protect domestic violence victims. 2. For example, a significant portion of the Family Violence Services and Prevention Act of 1984 provided funding for 23 law enforcement training projects across the country from 1986-1992 (Newmark, Harrell, and Adams, 1995). The goals of the training effort were to improve the quality of responses of police officers to female victims, improvements that will encourage their use of the law in future incidents. Whether the increased quality or quantity of police response made a difference in the lives of battered women was not addressed. 3. In fact, only a handful of studies have examined the effects of legal sanctions, both civil and criminal, on the recurrence of domestic violence. Although domestic violence has been a longstanding concern at the National Institute of Law Enforcement and Criminal Justice, NIJ's predecessor, the priority and resources assigned to the evaluation of legal reforms in domestic violence have varied, as have the types of questions and designs to answer them. Accordingly, the empirical evidence to assess the effectiveness of legal reforms has been narrow and methodologically weak. The Minneapolis Domestic Violence Experiment (Sherman and Berk, 1984a, 1984b) arguably has been NIJ's most influential research effort. But the Minneapolis experiment was noteworthy not because it was an evaluation of arrest policy for domestic violence. In fact, it was a test of deterrence theory, and domestic violence was not its primary concern. The replication experiments, collectively known as Spouse Assault Replication Program (SARP), were concerned with tests of deterrence theory also (Sherman, 1992a; Garner, et al., 1995). 4. The "battered woman's" defense was applied not only in cases in which the woman killed the man during an attack but also in cases in which the man was not actively threatening or abusing the woman at the time of the incident (Browne, 1987). 5. Historically, like the police, prosecutors were accused of disinterest in family violence cases by failing to file cases presented by the police or discouraging willing victims from pursuing criminal complaints. Whether discouraged by the evidentiary problems in these cases or the signals from a disinterested judiciary that was unwilling to respond to prosecution with meaningful sanctions, prosecutors had little incentive to follow through with vigorous presentation of domestic violence cases (Elliott, 1989; Ford, 1993). For example, Fagan (1989) found that fewer than 5 percent of 270 cases in 5 criminal justice systems were criminally prosecuted. 6. These conditions reduce the statistical power of the experiment and limit its effect on theory and policy. Statistical power is an estimate of the probability of falsely rejecting a null hypothesis- -that is, detecting a significant effect when in fact it may be valid (Cohen, 1988). In this case, the small effect size and limited sample sizes suggest that these findings may well result from chance. 7. They did report, however, on subsequent calls to the police for domestic disturbances, an imprecise measure of domestic violence with fairly high measurement error. 8. Williams and Hawkins (1986, 1989b) specify three types of costs that create informal controls: attachment costs (e.g., the loss of valued relationships), stigma (e.g., social opprobrium, embarrassment), and commitment costs (e.g., loss of job or economic opportunity) (Carmody and Williams, 1987). Thus, Williams and Hawkins (1986, 1989b) are consistent with other deterrence theorists in suggesting a reciprocal and complementary relationship between formal and informal controls for domestic violence. They state, for example, that ". . . persons (may) anticipate that others will disapprove of their arrest for committing a certain act, and they (may) refrain from that activity because they fear the stigma of being caught" (1986:562-563). Thus, for all these types of costs, extralegal punishment may be contingent on legal sanction. 9. Local legal culture describes the local patterns of practice that reflect in part the informal norms and expectations that regular players in the system have developed and have come to accept as "how we do things" (Kritezer and Zemmans, 1993). For the purpose of this proposed research, "local legal culture" includes the norms and attitudes, formal rules, and social relations that influence case outcomes. Criminal courtroom proceedings reflect both formal externally imposed rules and informal procedures and unspoken rules and customs. "The gap between the formal and informal rules--what the public expects and what actually occurs in practice--is largely the product of local legal culture" (Schiller and Manikas, 1987). Schiller and Manikas (1987) suggest that the local legal culture is shaped by the fragmented nature of the criminal justice system and the many participants involved in reaching consensus on "going rates" of sanctions for specific types of cases. Formal rules are often not enforced because to strictly adhere to the rules would contradict the values and expectations of the legal culture. Schiller and Manikas (1987) suggest that the courts' formal rules and informal customs be reconciled so that the reality in a criminal justice courtroom reflects the theoretical underpinnings of the justice system. References American Psychological Association (1993). Violence and Youth: Psychology's Response. Volume I: Summary Report of the American Psychological Association Commission on Violence and Youth. Washington, DC: American Psychological Association. Andrews, Don A., Ivan Zinger, R.D. Hoge, James Bonta, Paul Gendreau, and Francis T. Cullen (1990). "Does Correctional Treatment Work? A Clinically Relevant and Psychologically Informed Meta-analysis." Criminology 28:369-397. Bard, Morton (1970). Training Police as Specialists in Family Crisis Intervention. Washington, DC: U.S. Department of Justice. Bard, Morton, and Joseph Zacker (1971). "The Prevention of Family Violence: Dilemmas of Community Interaction." Journal of Marriage and the Family 33:677-682. Berk, Richard A., Alec Campbell, Ruth Klap, and Bruce Western (1992a). "Bayesian Analysis of the Colorado Springs Spouse Abuse Experiment." Journal of Criminal Law and Criminology 83:170-200. Berk, Richard A., Alec Campbell, Ruth Klap, and Bruce Western (1992b). "The Deterrent Effect of Arrest in Incidents of Domestic Violence: A Bayesian Analysis of Four Field Experiments." American Sociological Review 57:698-708. Binder, Arnold, and James W. Meeker (1992). "Implications of the Failure to Replicate the Minneapolis Experimental Findings." American Sociological Review 58:886-888. Black, Donald (1980). The Manners and Customs of the Police. New York: Academic Press. Black, Donald, and Albert J. Reiss, Jr. (1967). "Patterns of Behavior in Police and Citizen Transactions." Pp. 1-139 in Studies of Crime and Law Enforcement in Major Metropolitan Areas, Field Surveys III, Volume 2, President's Commission on Law Enforcement and the Administration of Justice. Washington, DC: U.S. Government Printing Office. Blumstein, Alfred, Jacqueline Cohen, and Daniel Nagin (eds.) (1978). Estimating the Effects of Criminal Sanctions on Crime Rates. Washington, DC: National Academy Press. Boffey, Phillip M. "Domestic Violence: Study Favors Arrest." New York Times. April 5, 1983. Boruch, Robert F. (1994). "The Future of Controlled Randomization Experiments: A Briefing." Evaluation Practices 15:265-274. Bowker, Lee (1983). Beating Wife-Beating. Lexington, MA: D.C. Heath. Bowker, Lee (1984). "Coping With Wife Abuse: Personal and Social Networks." In Battered Women and Their Families, edited by A.R. Roberts. New York: Springer. Browne, Angela (1987). When Battered Women Kill. New York: Free Press. Campbell, Donald T. (1969). "Reforms as Experiments." American Psychologist. 24:409-429. Carmody, Diana Cyr, and Kirk R. Williams (1987). "Wife Assault and Perceptions of Sanctions." Violence and Victims 2:25-38. Centers For Disease Control (1993). Injury Control in the 1990s: A National Plan for Action. Atlanta: Centers for Disease Control. Cohen, Stanley (1988). Statistical Power for the Behavioral Sciences, 2nd Edition. Hillsdale, NJ: Lawrence Erlbaum Associates. Davis, Robert M. (1995). Personal Communication. Davis, Robert C., and Barbara Smith (1982). "Crimes Between Acquaintances: The Response of the Criminal Courts." Victimology: An International Journal 8:175-187. Davis, Robert C., and Bruce Taylor (1995). "A Joint Social Service and Police Response to Domestic Violence: The Results of a Randomized Experiment." Unpublished. New York: Victim Services Agency. Dobash, Rebecca E., and Russell P. Dobash (1979). Violence Against Wives: A Case Against the Patriarchy. New York: Free Press. Dunford, Franklyn W. (1990). "Victim Initiated Warrants for Suspects of Misdemeanor Domestic Assault: A Pilot Study." Justice Quarterly 7:631- 653. Dunford, Franklyn W. (1992). "The Measurement of Recidivism in Cases of Spouse Assault." Journal of Criminal Law and Criminology 83:120-136. Dunford, Franklyn W., and Delbert S. Elliott (1990a). "The Role of Arrest in Domestic Assault: The Omaha Experiment." Criminology 28:183-206. Dutton, Donald G. (1988). "Profiling of Wife Assaulters: Preliminary Evidence for a Trimodal Analysis." Violence and Victims 3:5-30. Dutton, Donald G. (1995). The Domestic Assault of Women (second edition). Boston: Allynn-Bacon. Dutton, Donald G., and J.J. Browning (1988). "Concern for Power, Fear of Intimacy, and Aversive Stimuli for Wife Abuse." Pp. 163-175 in Family Abuse and Its Consequences: New Directions for Research, edited by Gerald T. Hotaling, David Finkelhor, J.T. Kilpatrick, and Murray Straus. Beverly Hills, CA: Sage. Dutton, Donald, Stephen G. Hart, Leslie W. Kennedy, and Kirk R. Williams (1991). "Arrest and the Reduction of Repeat Wife Assault." In Domestic Violence: The Changing Criminal Justice Response, edited by E. Buzawa and C. Buzawa. Westport, CT: Greenwood. Edelson, Jeffrey, and M. Syers (1990). "The Relative Effectiveness of Group Treatments for Men Who Batter." Social Work Research and Abstracts 26:10-17. Edelson, Jeffrey L., and Richard M. Tolman (1993). Intervention for Men Who Batter: An Ecological Approach. Newbury Park, CA: Sage. Elliott, Delbert S. (1989). "Criminal Justice Procedures in Family Violence Crimes." Pp. 427-480 in Family Violence, Volume 11 of Crime and Justice: An Annual Review of Research, edited by Lloyd Ohlin and Michael Tonry. Chicago: University of Chicago Press. Emerson, Robert E. (1981). "On Last Resorts." American Journal of Sociology 87:1-22. Emerson, Robert E. (1983). "Holistic Effects in Social Control Decision Making." Law and Society Review 17:425-455. Fagan, Jeffrey (1989). "Cessation from Family Violence: Deterrence and Dissuasion. Pp. 357-426 in Family Violence, Volume 11 of Crime and Justice: An Annual Review of Research, edited by Lloyd Ohlin and Michael Tonry. Chicago: University of Chicago Press. Fagan, Jeffrey (1990). "Natural Experiments." Pp. 108-137 in Measurement Issues in Criminology, edited by Kimberly Kempf. New York: Springer-Verlag. Fagan, Jeffrey (1992). "Social Control and Spouse Assault." Pp. 187-234 in Advances in Criminological Theory, Volume 4, edited by Freda Adler and William Laufer. New Brunswick, NJ: Transaction. Fagan, Jeffrey (1993). "Social Structure and Spouse Assault." Pp. 209-254 in The Socio-Economics of Crime and Justice, edited by Brian Forst. Toronto: ME Sharpe. Fagan, Jeffrey, and Angela Browne (1994). "Violence Against Spouses and Intimates." Pp. 115-292 in Understanding and Preventing Violence, Volume 3, edited by Albert J. Reiss, Jr., and Jeffrey A. Roth. Washington, DC: National Academy Press. Fagan, Jeffrey, Douglas K. Stewart, and Karen V. Hansen (1983). "Violent Men or Violent Husbands: Background Factors and Situational Correlates of Severity and Location of Violence." Pp. 49-68 in The Dark Side of Families, edited by D. Finkelhor, M. Straus, G. Hotaling, and R. Gelles. Beverly Hills, CA: Sage. Fagan, Jeffrey, Elizabeth Friedman, Sandra Wexler, and Virginia O. Lewis (1984). Final Report: National Family Violence Evaluation, Grant 80-JN- AX-0004, Office of Juvenile Justice and Delinquency Prevention. Washington, DC: U.S. Department of Justice. Fagan, Jeffrey, Christopher Maxwell, Lisa Macaluso, and Cynthia Nahabedian (1995). Evaluation of the Domestic Violence Hearing Officer Pilot Program: Final Report. Trenton, NJ: Administrative Office of the Courts. Farrell, Graham, and Ken Pease (1993). Once Bitten, Twice Bitten: Repeat Victimization and its Implications for Crime Prevention. London: Police Research Group. Crime Prevention Unit Series: Paper No. 49. Finn, Peter (1989). "Statutory Authority in the Use and Enforcement of Civil Protection Orders Against Domestic Abuse." Family Law Quarterly 23:43-67. Finn, Peter (1991). "State-by-State Guide to Enforcement of Civil Protection Orders." Response to the Victimization of Women and Children 14(1):3- 12. Finn, Peter, and Sara Colsen (1990). Civil Protection Orders: Legislation, Current Court Practice, and Enforcement. Washington, DC: U.S. National Institute of Justice. Ford, David A. (1991). "Prosecution as a Power Source: A Note on Empowering Women in Violent Conjugal Relationships." Law and Society Review 25:313-334. Ford, David A. (1993). The Indianapolis Domestic Violence Prosecution Experiment. Final Report, Grant 86-IJ-CX-0012 to the National Institute of Justice. Indianapolis, IN: Indiana University. Forst, Brian E., and J.C. Hernon (1985). "The Criminal Justice Response to Victim Harm." National Institute of Justice Research in Brief. Washington, DC: U.S. Department of Justice. Gaguin, Deidre A. (1977-78). "Spouse Abuse: Data from the National Crime Survey." Victimology: An International Journal 2:632-642. Garner, Joel, and Elizabeth Clemmer (1986). "Danger to Police in Domestic Disturbances: A New Look." National Institute of Justice Research in Brief. Washington, DC: U.S. Department of Justice. Garner, Joel, Jeffrey Fagan, and Christopher D. Maxwell (1995). "Published Findings from the NIJ Spouse Assault Replication Program: A Critical Review." Journal of Quantitative Criminology 8(1):1-29. Goolkasian, Gail A. (1986). Confronting Domestic Violence: A Guide for the Criminal Justice Agencies. National Institute of Justice Issues and Practices, May. Gordon, Linda (1988). Heroes of Their Own Lives: The Politics and History of Family Violence. New York: Viking. Grau, Janice, Jeffrey Fagan, and Sandra Wexler (1984). "Restraining Orders for Battered Women: Issues in Access and Efficacy." Women and Politics 4:13-28. Hagan, John, and Kirsten Bumiller, 1983. "Making Sense of Sentencing: A Review and Critique of Sentencing Research." In Alfred Blumstein et al. (eds.), op. cit. Hamberger, L. Kevin, and J. E. Hastings (1993). "Court-Mandated Treatment of Men Who Assault Their Partners: Issues, Controversies, and Outcomes." Pp. 188-229 in Legal Responses to Wife Assault: Current Trends and Evaluation, edited by N. Zoe Hilton. Newbury Park, CA: Sage. Harrell, Adele V. (1991). "Evaluation of Court-Ordered Treatment for Domestic Violence Offenders." Final Report, Grant 90-12L-E-089, State Justice Institute. Washington, DC: The Urban Institute. Harrell, Adele V., Barbara Smith, and Lisa Newmark (1993). Court Processing and the Effects of Restraining Orders for Domestic Violence Victims. Final Report to the State Justice Institute. Washington, DC: The Urban Institute. Hassenfeld, Y., and P. Cheung (1985). "The Juvenile Court as a People Processing Organization: A Political Economic Perspective." American Journal of Sociology 90:801-825. Hilton, N. Zoe (1993). Legal Responses to Wife Assault: Current Trends and Evaluation. Newbury Park, CA: Sage. Hirschel, J. David, Ira W. Hutchison, III, and Charles W. Dean (1992a). "The Failure of Arrest to Deter Spouse Abuse." Journal of Research in Crime and Delinquency 29:7-33. Hirschel, J. David, Ira W. Hutchison, III, and Charles W. Dean (1992b). "Female Spouse Abuse and the Police Response: The Charlotte, North Carolina Experiment." Journal of Criminal Law and Criminology 83:73-119. Holzworth-Munroe, Amy, and Gregory L. Stuart (1994). "Typologies of Male Batterers: Three Subtypes and Differences Among Them." Psychological Bulletin 116(3):476-497. Hotaling, Gerald T., and David B. Sugarman (1986). "An Analysis of Risk Markers in Husband to Wife Violence: The Current State of Knowledge." Violence and Victims 1:101-124. International Association of Chiefs of Police (1967). "Training Key 16: Handling Domestic Disturbance Calls." Gaithersburg, MD: IACP. Jacob, Herbert (1983). "Courts as Organizations." In Empirical Theories About the Courts, edited by Lynn Mather and Kenneth Boyum. New York: Longman. Kaci, Judy Hails (1992). "A Study of Protective Orders Issued Under California's Domestic Violence Prevention Act." Criminal Justice Review 17(1):61- 76. Klein, Andrew R. (n.d.). "Re-abuse in a Population of Court-Restrained Male Batterers." Unpublished. Quincy, MA: Quincy Court Domestic Violence Program. Kritzer, Herbert M. and Frances Kahn Zemmans (1993). "Local Legal Culture and the Control of Litigation." Law and Society Review 27(3):535-557. Lattimore, Pamela K., Christy Visher, and Richard L. Linster (1995). "Predicting Rearrest for Violence among Serious Youthful Offenders." Journal of Research in Crime and Delinquency 32:54-83. Lerman, Lisa G. (1981). "Criminal Prosecution of Wife Beaters." Response to Violence in the Family 4(3):1-19. Lerman, Lisa G. (1992). "The Decontextualization of Domestic Violence." Journal of Criminal Law and Criminology 83:217-240. Mather, Lynn (1979). Plea Bargaining or Trial? Lexington, MA: D.C. Heath. McCord, Joan (1993). "Deterrence of Domestic Violence: A Critical View of Research." Journal of Research in Crime and Delinquency 29:229-239. Miller, J.L., and A.B. Anderson (1986). "Updating the Deterrence Doctrine." Journal of Criminal Law and Criminology 77(2):418-438. Mohr, L.B. (1976). "Organizations, Decisions and Courts." Law and Society Review 14:621-642. Newmark, Lisa, Adelle Harrell, and Barbara Smith (1995). Evaluation of the Domestic Violence Services in the Family Violence Prevention Services and Training Program. Washington, DC: The Urban Institute. Oliver, William (1994). The Violent Social World of Black Men. New York: Lexington Books. Parnas, Raymond I. (1967). "The Police Response to the Domestic Disturbance." Wisconsin Law Review 31:914-960. Pate, Anthony M., and Edwin E. Hamilton (1992). "Formal and Informal Deterrents to Domestic Violence: The Dade County Spouse Assault Experiment." American Sociological Review 57:691- 697. Pleck, Elizabeth (1989). "Criminal Approaches to Family Violence, 1640-1980." Pp. 19-57 in Family Violence, Volume 11, Crime and Justice: An Annual Review of Research, edited by Lloyd Ohlin and Michael Tonry. Chicago: University of Chicago Press. Reiss, Albert J., Jr. (1985). "Selective Strategies of Social Control Over Organizational Life." In Enforcing Regulation, J. Thomas and K. Hawkins (eds.). Boston: Kluwer-Nijhoff. Reiss, Albert J., Jr., and Jeffrey A. Roth (1993). Understanding and Preventing Violence. Washington, DC: National Academy Press. Riecken, H.W., R.F. Boruch, D.T. Campbell, N. Caplan, T.K. Glennan, J.W. Pratt, A. Rees, and W. Williams. (1974). Social Experimentation. New York: Academic Press. Rothman, Kenneth J. (1986). Modern Epidemiology. Boston: Little Brown. Sampson, Robert J. (1992). "Family Management and Child Development: Insights from Social Disorganization Theory." Pp. 63-92 in Facts, Forecasts, and Frameworks, edited by Joan McCord. New Brunswick, NJ: Transaction. Sampson, Robert J., and W. Byron Groves (1989). "Community Structure and Crime: Testing Social Disorganization Theory." American Journal of Sociology 94:774-802. Saunders, Dan, and Sandra Azar (1989). "Treatment Programs for Family Violence." Pp. 481-546 in Family Violence, Volume 11, Crime and Justice: An Annual Review of Research, edited by Lloyd Ohlin and Michael Tonry. Chicago: University of Chicago Press. Schiller, Stephen A., and Peter M. Manikas (1987). "Criminal Courts and Local Legal Culture." DePaul Law Review 36:327-341. Schmidt, Janell, and Ellen Hochstadler Steury (1989). "Prosecutorial Discretion in Filing Charges in Domestic Violence Cases." Criminology 27:487- 510. Sechrest, Lee, Susan O. White, and E.D. Brown (1979). The Rehabilitation of Criminal Offenders: Problems and Prospects. Washington, DC: National Academy Press. Sherman, Lawrence W. (1980). "Specific Deterrent Effect of Spouse Assault." Proposal submitted to the National Institute of Justice, Crime Control Theory Program. Washington, DC: Crime Control Institute. Sherman, Lawrence W. (1992a). "The Influence of Criminology on Criminal Law: Evaluating Arrests for Misdemeanor Domestic Violence." Journal of Criminal Law and Criminology 83:1-45. Sherman, Lawrence W. (1992b). Policing Domestic Violence. New York: Free Press. Sherman, Lawrence W., and Richard A. Berk (1984a). "The Specific Deterrent Effects of Arrest for Domestic Assault." American Sociological Review 49:261-272. Sherman, Lawrence W., and Richard A. Berk (1984b). The Minneapolis Domestic Violence Experiment. Washington, DC: Police Foundation Reports, No. 1. Sherman, Lawrence W., and Ellen G. Cohn (1989). "The Impact of Research on Legal Policy: The Minneapolis Domestic Violence Experiment." Law and Society Review 23:117-144. Sherman, Lawrence W., Janell D. Schmidt, Dennis P. Rogan, Patrick Gartin, Ellen G. Cohen, Dean J. Collins, and Anthony R. Bacich (1991). "From Initial Deterrence to Long-Term Escalation: Short Custody Arrest for Poverty Ghetto Domestic Violence." Criminology. 29:821-850. Sherman, Lawrence W., Douglas A. Smith, Janell D. Schmidt, and Dennis P. Rogan (1992a). "Crime, Punishment, and Stake in Conformity: Legal and Informal Control of Domestic Violence." American Sociological Review 57:680-690. Sherman, Lawrence W., Janell D. Schmidt, Dennis P. Rogan, Douglas A. Smith, Patrick R. Gartin, Ellen G. Cohn, Dean J. Collins, and Anthony R. Bacich (1992b). "The Variable Effects of Arrest on Crime Control: The Milwaukee Domestic Violence Experiment." Journal of Criminal Law and Criminology 83:137-169. Sidel, Ruth (1986). Women and Children First. New York: Viking. Straus, Murray A., and Richard Gelles (1986). "Societal Change in Family Violence from 1975 to 1985 as Revealed by Two National Surveys." Journal of Marriage and the Family 48:465-479. Straus, Murray A., Richard Gelles, and Suzanne Steinmetz (1980). Behind Closed Doors: Violence in the American Family. Garden City, NY: Anchor/Doubleday Press. Tedeschi, James, and Richard Felson (1995). Violence, Aggression and Coercive Actions. Washington, DC: APA Press. Tittle, Charles R., and Charles H. Logan (1973). "Sanctions and Deviance: Evidence and Remaining Questions." Law and Society Review 7(3):371-392. Toby, Jackson (1957). "Social Disorganization and Stakes in Conformity: Complementary Factors in the Predatory Behavior of Hoodlums." Journal of Criminal Law, Criminology, and Police Science 48:12-17. U.S. Attorney General's Task Force on Family Violence (1984). Final Report. Washington, DC: U.S. Government Printing Office. U.S. Commission on Civil Rights (January 1982). Under the Rule of Thumb, Battered Women and the Administration of Justice. Washington, DC. Visher, Christy, Pamela Lattimore, and Richard Linster (1991). "Predicting Recidivism of Serious Youthful Offenders Using Survival Models." Criminology 29:329-366. Weis, Joseph G. (1989). Family Violence Research Methodology and Design. In Family Violence, Volume 11, Crime and Justice, An Annual Review of Research, edited by Lloyd Ohlin and Michael Tonry. Chicago: University of Chicago Press. Williams, Kirk R., and Richard Hawkins (1986). "Perceptual Research on General Deterrence: A Critical Review." Law and Society Review 20:545- 572. Williams, Kirk R., and Richard Hawkins (1989a). "The Meaning of Arrest for Wife Assault." Criminology 27:163-181. Williams, Kirk R., and Richard Hawkins (1989b). "Controlling Male Aggression in Intimate Relationships." Law and Society Review 23:591-612. Williams, Kirk R., and Richard Hawkins (1992). " Wife Assault: Costs of Arrest and the Deterrence Process." Journal of Research in Crime and Delinquency 29:292-310. Zeisel, Hans (1972). "Reflections on Experimental Techniques in the Law." Journal of Legal Studies 2(1):107-124. Zimring, Franklin E. (1973). Deterrence: The Legal Thread in Crime Control. Chicago: University of Chicago Press. Zimring, Franklin E. (1989). "Toward a Jurisprudence of Family Violence." Pp. 547-570 in Family Violence, Volume 11, Crime and Justice: An Annual Review of Research, edited by Lloyd Ohlin and Michael Tonry. Chicago: University of Chicago Press. Zorza, Joan (1992). "The Criminal Law of Misdemeanor Domestic Violence, 1970-1990." Journal of Criminal Law and Criminology 83:240-279. For more information on the National Institute of Justice, please contact: National Criminal Justice Reference Service P.O. Box 6000 Rockville, MD 20849-6000 800-851-3420 e-mail: askncjrs@ncjrs.aspensys.com You can view or obtain an electronic version of this document from the NCJRS Bulletin Board System (BBS), or the NCJRS Justice Information Center World Wide Web site. To access the World Wide Web site, go to http://ncjrs.aspensys.com:81/catalog.html If you have any questions, call or e-mail NCJRS. The National Institute of Justice is a component of the Office of Jutice Programs, which also includes the Bureau of Justice Assistance, Bureau of Justice Statistics, Office of Juvenile Justice and Delinquency Prevention, and the Office for Victims of Crime.