Click for next page ( 67


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



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

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

OCR for page 66
4 Prevention and Delerrencei This chapter reviews research on prevention and deterrence efforts undertaken at various levels of government and by private organi- zations to reduce violence against women. It also identifies research needed to better understand what does and does not work in preventing and deterring these violent behaviors. AN EPIDEMIOLOGICAL MODEL Public health models can be applied to the problem of violence against women using the traditional framework of primary, secondary, and ter- tiary prevention. Primary prevention effects a change in risk factors. Its purpose is to prevent disease or injury, which in the case of violence against women means preventing the assault or abusive behavior from happening in the first place through, for example, the strengthening of protective factors. Secondary prevention is early identification. Its pur- pose is to reduce the risk of death or disability, for example, through screening and follow-up care. Tertiary prevention is a response to an event, such as a domestic violence assault. Its purpose from a public health point of view is to extend life after the event; an example in the case of Prevention involves programs and services available to victims and offenders that aim to decrease the number of new cases of assault or abusive behavior, reduce the risk of death or disability from violence, and extend life after a violent event; deterrence involves the use of sanctions to prevent offenders from using violence. 66

OCR for page 66
PREVENTION AND DETERRENCE 67 violence against women is psychological counseling to deal with the trauma of abuse. Most of the funding available for the prevention of intimate-partner violence and sexual assault is spent on tertiary prevention, or treatment. In public health, tertiary prevention is designed to effect major behavior changes in individuals. A public health prevention research agenda would seek to learn enough about risk factors to evaluate interventions starting in childhood and adolescence, and identify and target social norms con- doning violence. Primary Prevention In primary prevention, strategies and programs should be based on plausible and stable theories or causes and should be rigorously evalu- ated. Current primary prevention strategies focusing on the first occur- rence of violence include offering school-based training in conflict resolu- tion, educating the public about healthy relationships, providing early sex education, limiting children's exposure to sexually violent and ag- gressive media images, targeting specific interventions for children wit- nessing partner violence, and integrating training in parenting skills into school health curricula. There has been no evaluation of the effectiveness of most of these strategies in reducing violence; experimental studies are needed for this purpose. Secondary Prevention One objective of secondary prevention of violence against women is moderating the level of violence. Among suggested strategies for this pur- pose in the case of intimate-partner violence are the use of screening mechanisms for abuse, including psychological abuse, in health care and social service settings and training of health care workers to identify physical and sexual violence. Another objective of secondary prevention is to moderate or eliminate exposure to the violence, for example, by pro- viding protective orders and relocation for victims. A third objective is to modify the relevant characteristics of the hazard through such services as supportive counseling and crisis hot lines, aimed at persons at risk of be- coming offenders to reduce aggression. Finally, secondary prevention can be aimed at making victims more resistant to damage, through such ef- forts as counseling, self-defense classes, and strengthening of community mental health services for victims and their families. Research is still needed in many of these areas, as discussed later in the chapter.

OCR for page 66
68 RESEARCH ON VIOLENCE AGAINST WOMEN Tertiary Prevention The first objective of tertiary prevention is to counter the damage done by an assault. Means of accomplishing this objective include the criminal justice response, as well as medical, legal, and advocacy services to vic- tims. Batterer education programs might also be included in this category. Evaluating the effectiveness of these strategies requires information from both victims and abusers. It is important to note that coordinated commu- nity responses to violence against women should be expanded to treat the psychological impact of the violence and should ensure that victims have access to the services they need. OTHER RESEARCH ON PREVENTION Specific information about violent crimes including victim, offender, and situation or place (see Chapter 3) is critical to the design of appro- priate research programs that can support prevention efforts (Block, 2002~. Reliance on aggregate statistics to understand intimate-partner homicide has been shown to mask important differences and lead to ineffective tar- geting of services. Specific information such as marital status, age, and prior victimization experience, which have been shown to be risk factors- is important for prioritizing prevention efforts and targeting them to spe- cific groups (Bureau of Justice Statistics, 2002~. The design of projects targeting batterers and potential batterers must be sensitive to different cultural backgrounds, including the importance assigned to having an intact family in a given culture, and to different kinds of family situations and arrangements. It is also important to look at patterns of violent behavior and the risk of violence over the life span of both offenders and their victims in different settings (see also Chapter 3~. One example of an important contextual issue is the role of guardians or potential guardians in establishing safety for victims. Potential guard- ians friends, neighbors, faith communities, the health community, and the criminal justice community often do not know what to do when someone asks for help. Very few women talk to counselors only about 15 percent according to Block (2002) and those that do so turn first to guardians because they are the gatekeepers. Yet no evaluative research has been done on what gatekeepers and guardians need to know for suc- cessful interventions, and little research exists on what defines a safe place. Changing Attitudes, Beliefs, and Behaviors Among At-Risk Populations In their review of efforts to prevent violence against women, Maxwell and Post (2002) note that since the mid-1980s, colleges and universities

OCR for page 66
PREVENTION AND DETERRENCE 69 across the United States have begun to offer educational programs that address sexual violence (see also Roark, 1987~. This effort was enhanced in the early 1990s with the passage of two federal laws providing financial resources to rape crisis centers, nonprofit entities, and schools to support educational programs designed to increase awareness and help prevent sexual assault, intimate-partner violence, and sexual harassment. Unfortunately, the substantial variation among education interven- tion models has led to little cumulative or synthesized knowledge about effective methods of delivering such education and offered limited un- derstanding of both its proximal and distal preventive effects (Breitenbecher, 2000; McCall, 1993~. Maxwell and Post (2002) believe this may be because preventive education programs have been designed to provide a complete package of treatments to cover all possible causes of violence against women (Schewe and O'Donohue, 1993:668~. Neverthe- less, while assessing the research on these interventions, Maxwell and Post note several themes: The nature and quality of the evaluation designs varied signifi- cantly. Evaluations ranged, for example, from a pre- and post-test design involving a single treatment group (see Ellis et al., 1992) to more rigorous methods, such as a study that used a Solomon four-group design (see Fonow et al., 1992~. Only weak associations were found between sexual assault-related attitudes and behavior. Two of the three studies measuring actual behav- ior (e.g., dating behavior) found that the programs were not effective in reducing the incidence of victimization (Breitenbecher, 2000~. Except for a handful of evaluations, most studies did not provide long-term follow-up by testing or retesting their subjects on either atti- tude or behavior. Additional research suggests that the growth of these prevention programs has not had a contagious effect on the targeted population. Woods and Bower (2001) note that students in 2001 answered questions about date rape very similarly to those in the early 1980s and 1990s. Changing the Opportunities for Violence Through "Target Hardening" Maxwell and Post (2002) note that during the 1990s, theoretical work was done on why victims, offenders, and crime are environmentally con- centrated (see, e.g., Eck and Weisburd, 1995; Miethe, 1994; Felson, 1987; Garofalo, 1987~. Scholars investigating violence against women demon- strated that sexual assault (LeBeau, 1994; Schewe and O'Donohue, 1993; Sanday, 1981) and intimate-partner violence (Lemon, 1996; Campbell, 1992) are not equally distributed spatially or temporally (see, e.g., Kayitsinga and Maxwell, 2000; Baron and Straus, 1989~.

OCR for page 66
70 RESEARCH ON VIOLENCE AGAINST WOMEN Thus it has been argued that environmental strategies designed to make crime more difficult to commit can facilitate prevention. Such strat- egies include, for example, "target hardening" and the concentration of limited law enforcement resources. According to Schewe and O'Donohue (1993), the prevention technique most often used involves activities self- initiated by women and informed by the temporal and spatial patterns of rape to reduce opportunities for an assault. A majority of the women in the study, regardless of their expressed fear of rape, reported taking steps to reduce the risk of rape, such as carrying keys in their hand. More re- cently, sexual assault prevention programs and the literature have also emphasized situational precautions a woman can take to prevent the in- troduction of a "date rape" drug into her drink without her knowledge. From the public health perspective of primary prevention, these types of efforts are about strengthening the woman's ability to resist environmen- tal stressors (Bloom, 1981~. In addition to such self-initiated steps taken by women, both public and private agencies have undertaken efforts to reduce the opportunities for rape. Universities, police departments, private property managers (Shenkel, 1989), and other public agencies have, for example, installed emergency phones, increased lighting, tightened security in residence halls, provided escort services, and increased the number of police in cer- tain areas. Congress also acknowledged the importance of this prevention model by including two sections in the Violence Against Women Act of 1994 that authorized spending $25 million for capital improvement projects to increase safety in urban parks, recreation areas, and public transportation areas. Unfortunately, in their review, Maxwell and Post (2002) could find no published evaluation of efforts to prevent violence against women by reducing opportunities. Self-Defense Models Self-defense techniques defined as "psychological, verbal, and physical resistance in dangerous situations when they arise" (Rozee and Koss, 2002:12) are taught in programs that typically focus on increasing women's physical self-defense skills and their use of physical and psy- chological strategies to set boundaries and boost self-esteem (Thompson, 1991~. Indeed, self-defense techniques may be among the most widely pro- moted tools for preventing violence against women. Maxwell and Post (2002) found more than 20,000 websites that discussed, promoted, or sold goods and services for women's self-defense. An overabundance of empirical research on outcomes of self-defense measures taken by women is noted in Understanding Violence Against Women. Ullman's (1997) review of self-defense studies includes 31 pub-

OCR for page 66
PREVENTION AND DETERRENCE 71 fished studies of situational and behavioral factors related to rape avoid- ance. Maxwell and Post's (2002) assessment of these studies is similar to that of other reviewers: there is sufficient indication that women who re- sist a rape attempt are more likely to thwart the attack and not incur more serious injuries (Rozee and Koss, 2002; National Research Council, 1996~. No systematic assessments of whether self-defense programs increase the effectiveness of self-defense measures over what women can achieve with- out such training were found, however. Evaluation of Preventive Programs Schematic models help in visualizing all of the elements involved in an act of violence and the many possible points of intervention. Consider a simple three-part model, for example, based on the assumption that vio- lence causes injuries because it involves a transfer of energy from perpe- trator to victim via a weapon: victim-weapon-perpetrator. This simple model suggests two points of intervention between perpetrator and weapon and between weapon and victim. Laws controlling access to guns are an example of interventions in the former category, while safe houses for battered women are among interventions in the latter. Such simple models can be elaborated more fully as research helps us understand what factors are involved at each point in the process. Knowledge of the situa- tion in which perpetrator and victim are likely to find themselves is an example of the kind of information that can inform the design and imple- mentation of interventions that are likely to be effective. Determining whether a specific intervention works depends on evalu- ation that assesses the process of program implementation as well as final outcomes. Formative evaluation that yields a rich description of the pro- cess of intervention is helpful to the field because it provides information about possible interactions. Regardless of whether the final outcome proves statistically significant, such evaluation can aid in formulating new hypotheses and better-informed plans for intervention. Assessments of program outcomes that specify and operationalize desired outcomes are critical to scientific research and helpful to program designers in determining whether a program can achieve the desired goals. As the principle of evidence-based practice gains currency, how- ever, the demand for evaluation studies that meet high standards for both design and implementation is likely to grow. The National Task Force on Community Preventive Services, like other groups that assess the evidence base for practice, uses an algorithm to assess the quality of published stud- ies and reports. A crucial factor for meeting such standards is the use of comparison groups and substantial follow-up. Both of these requirements raise the level of scientific expertise needed for the design of an interven-

OCR for page 66
72 RESEARCH ON VIOLENCE AGAINST WOMEN lion, as well as its cost. Nonetheless, it has become apparent from the work of the National Task Force on Community Preventive Services and other evidence-based review groups that the systematic review of a body of literature makes an important contribution to determining what works. The Centers for Disease Control and Prevention (CDC) has begun to use such information both to encourage the use of effective programs and to promote research in areas in which there is insufficient evidence to make decisions about specific interventions. This judicious use of resources is likely to speed the development and widespread implementation of effec- tive practice. Legal Reforms Although sanctions against wife abuse have historically been part of many state criminal codes, criminal justice institutions have reacted to- ward violence against women with ambivalence until relatively recently. During the 1970s, the influence of the modern feminist movement led to a series of reforms designed to strengthen legislation and practice in re- sponse to physical and sexual assaults against women by intimate part- ners and others (National Research Council and Institute of Medicine, 1998~. Since then, many legal sanction strategies have been implemented in an effort to reduce violence against women (see Gosselin, 2000; Na- tional Research Council and Institute of Medicine, 1998; Pagan, 1996; Pleck, 1989, 1987~. These interventions cover nearly every aspect of the legal system from the creation of special law enforcement, prosecution, and victim service units for intimate-partner violence and rape; to the adoption of pro-arrest policies; to the adoption of new kinds of protection orders for women being abused or stalked and the development of spe- cial courts for handling cases of intimate-partner violence. Most legal strategies have not been rigorously evaluated, however. The evaluation literature on legal interventions is characterized by small study samples, legal problems in implementing some reforms, and diffi- cult-to-overcome ethical problems in implementing experimental designs, among other problems (National Research Council and Institute of Medi- cine, 1998~. The National Research Council's Committee on the Assess- ment of Family Violence Interventions found that no standard existed for conducting research on the immediate effects of legal reforms or on levels of subsequent abuse by offenders. The committee concluded that "accord- ingly, policies and procedures reflect ideology and stakeholder interests more than empirical knowledge" (National Research Council and Insti- tute of Medicine, 1998:159~. For example, few of these strategies, or their evaluations, have been guided by an explicit formulation of deterrence theory.

OCR for page 66
PREVENTION AND DETERRENCE 73 In a review of research assessing the impact of statutory reforms on responses by the criminal justice system, Maxwell and Post (2002) found that in general, statutory reforms increased official compliance across a number of domains, including reporting behavior, arrests, and conviction rates, and victims received improved treatment by criminal justice offi- cials. As noted above, however, very little research has been conducted on the relationship between legislative changes and measurable reduc- tions in the incidence of violence against women. While analyses of data from both the National Crime Victimization Survey and the Uniform Crime Reports show a decline for all violent victimization and an increase in the reporting of rape, it is not clear how the drop in rape or the increase in the reporting of rape is related to specific efforts to prevent rape or to other factors leading to general decreases in crime. Legal Reforms for Sexual Assault Maxwell and Post (2002) identify three areas in which legislative changes to address sexual assault have occurred since enactment of the 1970s rape statutes: sex offender notification laws, sexually violent preda- tor laws, and laws on the use of intoxicants that facilitate sexual assault. Since 1975, all 50 states have changed their rape laws (Loin, 1981~. Since the 1980s, the most broadly implemented changes have been the sex of- fender community notification laws, which require certain convicted sex offenders to register with law enforcement agencies and provide for com- munity notification of the offender's presence, depending on a predefined degree of risk that these persons will reoffend. Yet little research exists on the general or specific deterrent effects of these laws. Results of a study assessing the State of Washington's notification law suggest that commu- nity notification had "little effect on recidivism as measured by new ar- rests for sex offenses or other types of criminal behavior" (Schram and Milloy, 1995:3~. Maxwell and Post caution, however, that longer follow- up periods are needed, as well as a more qualitative assessment of changes in law enforcement and community behavior as a result of these laws. By 1998, 12 states had enacted statutes authorizing the confinement and treatment of highly dangerous sex offenders following completion of their criminal sentence (Lieb and Matson, 1998~; by 2000, an estimated 900 sex offenders were confined in the United States based on a court deter- mination of the future danger they posed (Goldberg, 2001~. Maxwell and Post (2002) were unable to find any published studies demonstrating the incapacitative effects of these predator confinement laws. A final set of legal reforms has addressed incidents involving the use of intoxicants that facilitate sexual assault, known collectively as "date rape" drugs. By 2001, 20 states had passed legislation criminalizing the

OCR for page 66
74 RESEARCH ON VIOLENCE AGAINST WOMEN use of such drugs (State Legislature, 2001), and the U.S Congress had passed two acts prohibiting their use. Legal Reforms for Intimate-Partner Violence Most state legislatures have presumptive arrest laws, and many po- lice departments prescribe that officers arrest a suspect whenever they have probable cause that a misdemeanor or felony assault has occurred between two intimate partners, even if they have not witnessed the event (Zorza, 1992; Hirschel et al., 1991; Sherman and Cohen, 1989~. By 1997, more than $1 billion was being spent annually by the federal government to encourage local authorities to use formal interventions, such as arrests and prosecutions, rather than informal interventions to address intimate- partner violence (National Institute of Justice, 1997~. Between 2000 and 2002, the government spent an additional $492 million to help local gov- ernments strengthen their law enforcement, prosecution, and victim ser- vices to address violence against women. Maxwell and Post (2002) were unable to find any published research evaluating changes in the incidence of intimate-partner violence as a result of the general deterrent effects of such legislation, however. It is clear that to mount programs that work to prevent both initial and subsequent acts of violence against women, the nation must do a bet- ter job of supporting high-quality, scientific evaluation of prevention strat- egies and programs. Current funding levels are inadequate to support the experimental and prospective designs that would move the field forward in these areas. The committee recommends that Congress provide ad- equate funds to support rigorous research designs and long-term evalua- tions of prevention and treatment programs. Because of inherent conflicts of interest (no program wants to be found ineffective), funds for program evaluation should be independent of program funds so that the ability to evaluate interventions will not be constrained by legislative or other ad- ministrative requirements or by political interests affecting program funds or agencies. The committee also believes it is important to link health, education, and criminal justice prevention efforts, and recommends that the National Institutes of Health and the National Institute of Justice (NIT) collaborate to develop an integrated program of rigorous evaluations of prevention, intervention, and control strategies in the area of violence against women. Summary There are now many educational programs designed to address the distal causes of violence against women, but there is little evidence on

OCR for page 66
PREVENTION AND DETERRENCE 75 long-term positive attitude changes induced by these programs or on the programs' ability to reduce the incidence of violence against women. This lack of evidence of efficacy exists because evaluations cannot use victim- ization rates or offending patterns as measures given the poor data avail- able on prevalence (see Chapter 2~. The few evaluations that have included behavior measures have shown no difference between treatment groups. Thus at this time, it is fair to conclude that we do not know whether these programs have directly reduced violence against women. At the same time, however, evaluations of similar efforts focused on crime in general have revealed that strategies aimed at reducing opportunity are promis- ing preventive tools. Moreover, a significant amount of research has shown that self-defense measures have been effective in preventing vio- lence against women. In the long run, however, a permanent reduction in violence against women must rely on changing the motivations for such behaviors. Finally, in accordance with principles of evidence-based practice, the committee recommends that evaluation of initiatives and programs to re- duce violence against women be required. These evaluations should meet the scientific standards required for impact studies, including those of experimental designs. Such studies should explore whether issues exist that are particular or distinctive to evaluations of programs for violence against women. While rigorous evaluation may pose risks to a program's continued operation, assessment of outcomes is necessary to show which approaches work and whether ineffective strategies can be turned into programs that work. Moreover, such evaluations can ensure that well- intentioned programs are not actually doing harm. Government agencies such as CDC and NIT should continue to promote research in areas in which there is insufficient evidence for making decisions about specific interventions. RESEARCH ON DETERRENCE The general literature on crime control provides substantial evidence that criminal justice sanctions have a deterrent effect on a wide range of behaviors. Both Nagin (1998) and Cook (1980) have found that the collec- tive actions of the criminal justice system have a large and important de- terrent effect on crime. This finding also appears to hold true at least for the effects of arrest for violence against women (Maxwell et al., 2001~. However, Nagin (1998) notes that there is almost no empirical knowledge base regarding the extent to which any specific policy, added to the exist- ing sanctions designed to address a specific problem, can be expected to have the desired preventive effect. He posits that in general, the response of crime rates to a change in sanction policy will depend on a number of

OCR for page 66
76 RESEARCH ON VIOLENCE AGAINST WOMEN important variables: (1) the specific form of the policy, (2) the context of its implementation, (3) the process by which people come to learn of it, (4) differences among people in their perceptions of the changes in risks and rewards resulting from the policy, and (5) feedback effects triggered by the policy itself (Nagin 1998:4~. Measuring Declarative and Deterrent Effects Declarative effects of laws and their enforcement involve establishing normative social boundaries and distinguishing behaviors that are toler- ated from those that are not. Bonnie (1981) found that laws against drug use may generate declarative effects because such laws express social dis- approval. Similarly, the very existence of laws against intimate-partner violence and other violence against women symbolizes and reinforces so- cial norms against such behavior. By shaping beliefs and attitudes toward these kinds of acts, these laws may have an effect on assaultive behavior toward women, separate from the deterrent effect of actually imposing sanctions. Such effects are difficult to isolate and therefore to measure because they often cannot be distinguished from deterrent effects or disentangled from the effects of preexisting social norms (National Research Council, 2001~. For example, some studies show that people worry about the shame and embarrassment of violating certain strongly held social norms that are supported by laws requiring compliance (see Institute of Medicine, 1999; Grasmick et al., 1991~. The imposition of formal sanctions can some- times increase compliance with a strongly held social norm beyond the levels dictated by declarative effects. However, some studies show that formal-sanction policies may lose their preventive power over time (Sherman, 1990; Ross, 1982~. The committee believes that future research should explore how to gauge declarative effects and disentangle them from other types of effects. General Deterrence General deterrence is the impact of legal sanctions on the larger popu- lation that is, those who have neither violated the law nor been pun- ished. Presumably, sanctions against committing violence against women should depress prevalence and incidence, but because of continuity and definition problems in the prevalence literature, it is difficult to measure trends in this area. Nagin (1998) reviews a large literature (the data for which are assembled from surveys) focused on the links between percep- tions of the risk and severity of sanctions and self-reported crime and

OCR for page 66
PREVENTION AND DETERRENCE 77 delinquency. He notes that individuals who perceive high sanction risks and costs generally report lower criminality. With respect to violence against women, Williams and Conniff (2002) examine a number of related studies of general deterrence bearing on the perceived risk of arrest for wife assault, as well as the costs of that legal sanction, as perceived by respondents. Data were collected from a nationally representative sample of households. The authors conclude that: A higher perceived risk of arrest is associated with a lower likeli- hood of wife assault (Lackey and Williams, 1995; Williams and Hawkins, 1989a; Carmody and Williams, 1988~. The perceived costs of arrest appear to be linked more strongly to the indirect and negative social and personal consequences of arrest than to the direct legal consequences of this sanction, such as time in jail (Will- iams and Hawkins, 1989b). Fear of arrest is a function more of the social and personal costs associated with this legal sanction itself than of the costs connected with perpetrating an act of wife assault (Williams and Hawkins, 1992~. The perceived social costs of arrest not only are inversely related to participation in wife assault, but also are a function of other theoreti- cally relevant factors, such as gender, race, socioeconomic status, nor- mative approval of violence, power imbalances in heterosexual intimate relationships, and isolation from community resources of social control (Williams, 1992~. These findings provide some empirical support for a theory of gen- eral deterrence and its application to the understanding and prevention of violence against women. Williams and Conniff (2002:7) report, "threats of legal sanctions may well deter would-be male perpetrators of intimate- partner violence; yet this conclusion must be tempered by other implica- tions of these findings." They believe that the nexus between legal sanc- tions (e.g., arrest) and informal sources of control (e.g., social costs of arrest) must be thoroughly understood. "Integrated into a larger network of social control, sanction threats may reveal their power," they note (Wil- liams and Conniff, 2002:7~. For example, a man who is enmeshed in social networks of intimate relationships, strong familial and friendship ties, work and financial gain, and other accoutrements of "conventional" liv- ing has much to lose by having his violence dramatized publicly through an arrest. This life-disrupting event could send shock waves rippling through his life, with damaging results. However, a man on the margins of conventional life may have less to lose and may well perceive arrest for wife assault, or other infractions of the law for that matter, as something that just happens, meaning little stigma is attached to getting arrested.

OCR for page 66
78 RESEARCH ON VIOLENCE AGAINST WOMEN Specific Deterrence Most empirical studies evaluating the effectiveness of actual experi- ences with sanctions in preventing or reducing violence were not explic- itly designed to test the theory of specific deterrence. Nonetheless, spe- cific deterrence is implied by the research focus on actual sanctioning experiences. Virtually all of the specific deterrence research on violence against women is on repeat intimate-partner violence. Arrest Research Few studies have had as much impact on violence prevention and control policies as the Minneapolis Domestic Violence Experiment (see, e.g., Sherman and Cohen, 1989~. Using police records and victim inter- views, the investigators in this study found that arrested suspects were significantly less likely to repeat their assaults in the 6-month follow-up period than those asked to leave the residence or those offered advice. The Spouse Assault Replication Program, funded by NIT, replicated this research in five locations. Two studies reported evidence generally con- sistent with the findings of the Minneapolis study (Berk et al., 1992; Pate and Hamilton, 1992~; the others reported evidence of escalation rather than deterrent effects of arrest (Hirschel et al., 1991; Sherman et al., 1992; Dunford et al., l990~. Moreover, some studies found that estimated effects were contingent on the "social bonding" of suspects, with arrest decreas- ing repeat violence for those who were married and employed, but in- creasing such violence for those who were unmarried and unemployed (for summaries, see Sherman, 1992a; Schmidt and Sherman, 1993~. Interpreting these mixed results has been problematic (see, e.g., Gar- ner et al., 1995; Garner and Maxwell, 2000~. Recent research involved pooling data from all five sites to standardize the intervention, outcome measures, and statistical models (Maxwell et al., 2001~. Using victim in- terviews as outcome measures, this reanalysis found that independent of site, length of time between initial and follow-up interviews, and suspect characteristics. offenders in the arrested proud were significantly less likely to repeat their aggression than those in the nonarrest group. However, no statistically significant effects of arrest were found when prevalence and frequency measures were based on officially recorded aggression. Other Policing Research Other recent research on the relationship between arrest and violence against women has focused on issues of policing rather than on the spe-

OCR for page 66
PREVENTION AND DETERRENCE 79 cific deterrence of police action or inaction. Feder (1998) found police were more likely to make an arrest when the assault did not involve an inti- mate relationship, and therefore concluded that police generally underenforce domestic violence laws. Connolly and colleagues (2000) found that arrest was actually more likely to occur in intimate situations. Chaney and Saltzstein (1998) report that direct orders requiring arrest in the form of state and local laws were effective at increasing the likelihood that officers would arrest, while Kane (1999) notes that risk to the victim was the most important factor affecting an arrest decision. Prosecution Research Most research on the prosecution of violence against women in recent years has focused on the issue of gaining the cooperation of victims (Robbing, 1999; Blair, 1996; Hanna, 1996~. Studies have examined factors influencing the decision to prosecute (e.g., Hirschel and Hutchison, 2001; Goodman et al., 1999; Cretney and Davis, 1997; Ford, 1991~. However, none of this research builds on the earlier studies of prosecution that tracked behavioral outcomes (e.g., Ford and Regoli, 1992; Fagan, 1989~. As was found for arrest, Fagan (1989) reports that in cases where prosecu- tion was attempted, convictions obtained, and offenders sentenced, re- ductions in subsequent violence were greatest for less-severe offenders. However, for offenders with "more severe histories of violence, the impo- sition of more severe sanctions was associated with increased incidence of violence" (Fagan, 1989:384~. Ford and Regoli (1992) report similar find- ings. They note that these mixed results do suggest that serious offenders are less influenced by legal sanctions, and indeed if such sanctions have any influence at all, they may actually increase violence against women (Mills, 1998~. Summary The literature on repeat intimate-partner violence demonstrates that legal sanctions do have deterrent effects, although modest in magnitude, but that these effects vary by the characteristics of perpetrators, their rela- tionship with their partners, their stake in social conformity, and factors influencing the decision to impose sanctions. Future research on this topic should seek to fill gaps in the research on general deterrence that may have implications for more-successful policies and practices. Factors that may moderate the effects of sanction risks and experiences on the violent victimization of women include neighborhood and community character- istics (see, e.g., Fagan, 1996), and especially the life circumstances of women victims (e.g., Bowman, 1992; Frisch, 1992; Lerman, 1992; Zorza,

OCR for page 66
80 RESEARCH ON VIOLENCE AGAINST WOMEN 1992) (see also Chapter 3~. Moreover, such research should focus on all forms of violence against women, not just intimate-partner violence. A BROADER FRAMEWORK FOR FUTURE RESEARCH While research shows that the collective actions of the criminal justice system exert a substantial deterrent effect on crime, this fact is of limited value in formulating policy for specific crime problems (Nagin,1998~. It is the committee's view that future research on deterring violence against women would be of most benefit if folded into broader efforts to study the decision making of potential perpetrators and the deterrence of crimi- nal behavior generally. This is a particularly important point given the scope and cost of program efforts aimed at deterring and preventing vio- lence against women. The committee identified four research gaps that impede progress in informing the development of effective policies and programs to deter and control crime. The committee believes that research in these four areas is critical to improving the ability to deter violence against women. Long-Term Effects of Sanctioning Policy There is a large amount of research on short-term effects of crime con- trol policies and programs, but almost nothing is known about long-term effects. If, for example, the principal deterrent effect of formal sanctions for violence against women (or other crimes) derives from fear of social stigma, the extent to which such penalties are actually meted out could either reinforce or erode such fear. Nagin (1998:5) posits that "policies that are effective in the short-term might erode the very basis for their effectiveness in the long-term if they increase the proportion of the popu- lation that is stigmatized." A criminal record cannot be socially isolating if it is commonplace. Thus, while arrest policies appear to work in the aggregate to reduce intimate-partner violence (Maxwell et al., 2001), some research studies indicate that they have no effect, or a criminogenic effect, on some offenders (Sherman, 1992b). The committee believes that research on the ways in which social stigma for acts of intimate-partner violence is generated and then either sustained or eroded would inform the develop- ment of more-effective policies and programs. Formation of Perceptions of the Risk of Sanctions There is a large body of research analyzing the links between percep- tions of the risk of sanctions and behavior (e.g., Lackey and Williams, 1995; Nagin and Paternoster, 1993; Bachman et al., 1992; Williams and

OCR for page 66
PREVENTION AND DETERRENCE 81 Hawkins, 1989a; Carmody and Williams, 1988; Paternoster et al., 1982; Grasmick and Bryjak, 1980~. One study on the deterrent effect of such perceptions with regard to intimate-partner violence was described above (Williams and Conniff, 2002~. However, very little is known about how these perceptions are formed. Behavioral decision theorists have found that human beings prefer more rather than less certainty about the prob- able outcomes of their choices. For example, where the perceived prob- ability of punishment is uncertain, there may be an initial deterrent effect. However, Sherman found that initial deterrent effects are eroded as of- fenders learn through experience that they have overestimated the chances of being caught and punished (Sherman, 1992b; Ross, 1982~. This finding supports Nagin's (1998) point that although the percep- tion of the risk that a sanction will be imposed may affect the decision to commit a crime, one cannot therefore conclude that a specific policy de- ters crime. To know that, one must have a more specific understanding of how risk perception is generated and sustained. For example, we do not know whether offenders perceive a risk of penalty because of the overall effectiveness of enforcement or because there is a crime-specific mecha- nism in place. In addition, we do not know how quickly policy changes influence perceptions of sanction risk. Better studies are also needed of the effects of the crime rate on actual sanction levels and of how those effects in turn influence the formation of perceptions of the risk of punish- ment. Finally, it is important to explore Sherman's (1990) theory that ini- tial deterrence can be made permanent by continually experimenting with novel police strategies, deployments, or enforcement priorities. How Responses to Crime Vary Across Time and Space Many evaluations of the deterrent effects of specific policies, such as increasing the number of police, are too broad measuring effects across the general population. Dugan and Apel's (2002) recent analysis of differ- ential risk for different groups of women underscores the need for an- other strategy. Nagin (1998) found that while there may be some credible estimates of the average deterrent effects of certain policies (e.g., effects of arrest policies on intimate-partner violence), the capacity to translate those effects into a prediction for a specific place or population is limited. For purposes of deterrence, he concludes, "It is the response of crime to policy in a specific city or state that is relevant to its population, not the average response across all cities and states" (Nagin, 1998:6~. This effect has been demonstrated by the approach to deterring gun violence among youth in Boston, as described and documented by Kennedy et al. (2001) and re- cently evaluated by Braga et al. (2001~.

OCR for page 66
82 RESEARCH ON VIOLENCE AGAINST WOMEN Links Between Actual and Intended Policy Finally, the link between intended and actual policy has not been well explored. In some jurisdictions, for example, police have reacted to man- datory arrest policies by arresting both partners involved in intimate-part- ner assault calls. With regard to policy changes in the early 1980s calling for increased penalties for gun carrying, the court in one city and the po- lice in another found ways to circumvent the change so the system could essentially maintain the status quo. Arrest or sentencing policies may be circumvented in other ways through plea bargaining, for example. There is a need for research on how sanctions are generated and implemented so their effects on crime and on perceptions of the risk of sanctions can be better understood.