Systemic Obstacles to the Criminal Prosecution of a Battering Partner: A Victim Perspective

18
Violence and Victims, Vol. 14, No. 4, 1999 © 1999 Springer Publishing Company Legal Report Obstacles to Victims' Cooperation With the Criminal Prosecution of Their Abusers: The Role of Social Support Lisa Goodman Boston College Lauren Bennett University of Maryland, College Park Mary Ann Dutton George Washington University Medical Center and George Washington University Law Center Over the last 20 years, diminishing tolerance for domestic violence has triggered sig- nificant criminal justice reforms designed to facilitate the prosecution of abusers. Prosecutors, for example, have adopted policies requiring that cases go forward even if the victim later has second thoughts. Although increasingly common, these "no drop" policies reflect a profound irony about domestic cases that is well known but little under- stood: the most formidable problem in prosecuting such cases is often the victim's own unwillingness to bring the abuser to justice. This prospective study explored a range of factors potentially predictive of domes- tic violence victims' cooperation with the prosecution of their abusers. Although the study focused on interpersonal and institutional social support, it also investigated the influence of violence severity, victim demographic factors, and victim mental health characteristics, including the presence of depressive symptoms, emotional dependence on the abuser, and substance abuse. Findings showed that tangible support, severity of violence in the relationship, and the presence of children in common with the abuser all significantly predicted victims' cooperation with the prosecution of their abusers. Substance abuse significantly predicted victims' noncooperation with prosecution. The research and policy implications of these findings are discussed. Over the last 20 years, diminishing tolerance for domestic violence has triggered significant criminal justice reforms designed to facilitate the prosecution of abusers (Cahn, 1992; Cahn & Lerman, 1991). 1 Police officers who once might have granted a victim's request to keep the matter "in the family" now are frequently required by law to make an arrest or to issue a cita- tion requiring a judicial appearance (Sherman, Schmidt, & Rogan, 1992). Prosecuting author- ities likewise have adopted policies requiring that cases go forward, even if the victim later has 427

Transcript of Systemic Obstacles to the Criminal Prosecution of a Battering Partner: A Victim Perspective

Violence and Victims, Vol. 14, No. 4, 1999© 1999 Springer Publishing Company

Legal ReportObstacles to Victims' Cooperation With theCriminal Prosecution of Their Abusers:The Role of Social Support

Lisa GoodmanBoston College

Lauren BennettUniversity of Maryland, College Park

Mary Ann DuttonGeorge Washington University Medical Center andGeorge Washington University Law Center

Over the last 20 years, diminishing tolerance for domestic violence has triggered sig-nificant criminal justice reforms designed to facilitate the prosecution of abusers.Prosecutors, for example, have adopted policies requiring that cases go forward even ifthe victim later has second thoughts. Although increasingly common, these "no drop"policies reflect a profound irony about domestic cases that is well known but little under-stood: the most formidable problem in prosecuting such cases is often the victim's ownunwillingness to bring the abuser to justice.

This prospective study explored a range of factors potentially predictive of domes-tic violence victims' cooperation with the prosecution of their abusers. Although thestudy focused on interpersonal and institutional social support, it also investigated theinfluence of violence severity, victim demographic factors, and victim mental healthcharacteristics, including the presence of depressive symptoms, emotional dependenceon the abuser, and substance abuse. Findings showed that tangible support, severity ofviolence in the relationship, and the presence of children in common with the abuser allsignificantly predicted victims' cooperation with the prosecution of their abusers.Substance abuse significantly predicted victims' noncooperation with prosecution. Theresearch and policy implications of these findings are discussed.

Over the last 20 years, diminishing tolerance for domestic violence has triggered significantcriminal justice reforms designed to facilitate the prosecution of abusers (Cahn, 1992; Cahn& Lerman, 1991).1 Police officers who once might have granted a victim's request to keep thematter "in the family" now are frequently required by law to make an arrest or to issue a cita-tion requiring a judicial appearance (Sherman, Schmidt, & Rogan, 1992). Prosecuting author-ities likewise have adopted policies requiring that cases go forward, even if the victim later has

427

428 L Goodman et al.

second thoughts about prosecution (Hanna, 1996). Although increasingly common, these"mandatory arrest" and "no drop" policies reflect a profound irony about domestic violencecases that is well known but little understood: the most formidable problem in prosecutingsuch cases is often the victim's own unwillingness to bring the perpetrator to justice (Ford &Regoli, 1992). In Washington, DC, alone—to take one example—approximately 50% of bat-tered women who initially tell authorities they want to press charges subsequently fail to coop-erate with the prosecution of their abusers (Robert Spagnoletti [Chief of the DomesticViolence Unit of the United States Attorney's Office in Washington, DC], personal communi-cation, November, 1995). Consistent with this statistic, the American Prosecutors' ResearchInstitute (APRI) conducted a national study of local prosecutor practices in domestic violencecases and found that 33% of those who responded reported that in over half of their cases bat-tered women failed to cooperate with criminal prosecution (Rebovich, 1996). In this study,cooperating in the prosecution of an abuser involved providing needed information to prose-cutors and being willing to appear in court to testify if necessary.

Although it is common knowledge that domestic violence victims frequently refuse tocooperate in the prosecution of their abusers, few researchers have gathered any empiricaldata—let alone developed a theoretical framework—to help explain why this is so. The fewstudies that have attempted to explore factors related to domestic violence victims' coop-eration with the criminal prosecution of their abusers provide, at best, a partial picture.Ford (1983) found that none of the demographic or event-related factors examined in hisstudy predicted follow-through, but the sample used was small (N = 25). McLeod (1983),using archival data, found that cooperation with prosecution was associated with the abuser'suse of a deadly weapon in the precipitating incident and with preexisting separation or divorce.

Without a more nuanced understanding of factors associated with victim cooperation, police,prosecutors, and advocates have no way of knowing what measures might decrease obstaclesfor victims who wish to assist prosecution. Perhaps influenced by a research literature thatemphasizes pathology in battered women (Campbell, Miller, Cardwell, & Belknap, 1994; Kosset al., 1994), criminal justice officials seeking to explain why victims do not cooperate withprosecution frequently derive psychological explanations, citing domestic violence victims asdepressed, emotionally dependent, or suffering from learned helplessness (Dutton, 1988).Less thought has been given to the possible impact of the social isolation of domestic violencevictims, and the lack of emotional and tangible support they receive from support networks.These may play an equally important—if not more important—role than victims' psycholog-ical characteristics and may be more accessible to intervention.

This study used a prospective design, standardized measures, and information elicitedfrom domestic violence victims themselves to explore the relationship between victim coop-eration in the criminal prosecution of their abusers and the social and institutional supportvictims receive during the process. It also investigated the influence of violence severity,victim demographic factors, and victim mental health characteristics, including the pres-ence of depressive symptoms, emotional dependence on the abuser, and substance abuse.

PROSECUTION AS A STRESSOR ANDTHE ROLE OF SOCIAL SUPPORT

Domestic violence victims who prosecute their abusers face a host of stressors that vic-tims of other crimes do not (e.g., Elliott, 1989; Ford, 1991). For example, a batteredwoman may be forced to testify about intimate details of her relationship. She may

429

forfeit the loyalty of family members who ally themselves with the abuser. She may bedeprived of tangible support if her home belongs to the abuser or if he is a main sourceof her livelihood. If children are involved, the stakes may become higher still: she mayfear losing the children to the abuser or having to raise them on her own, perhaps with-out child support (Hart, 1996).

Domestic violence victims also are more vulnerable than other victims to new acts ofviolence while the prosecution is under way (Langen & Innes, 1986). The abuser is rarelyjailed while awaiting trial (Lerman, 1980) and may return home over the victim's objec-tions, or sometimes even with the victim's permission. While a case makes its way throughthe criminal justice system, the abuser may destroy the victim's property or assault her, herchildren, or other family members in retaliation for her pressing charges, or in an attemptto force her to return to the relationship (Hart, 1993; Lerman, 1986). Prosecution, there-fore, should be viewed as a period of intense stress for victims of domestic violence.

Researchers have recognized the importance of social support as a weapon against stress.Hobfoll (1988) posits that people in difficult circumstances have an intense need to con-serve the resources they possess. These resources may be objects like a home, personal char-acteristics like self-esteem, conditions or roles like employment or being a member of anintimate relationship, or energies like time. When an individual either fears or experiencesthe loss of these resources, or anticipates an investment of resources without a resulting netgain of resources, the individual feels stress. In the face of this stress, social support is aunique external resource that a person may draw upon. Members of a social support net-work may directly replace lost resources, for example, by loaning money or materialgoods; they may reaffirm positive personal qualities of which one has lost sight, for exam-ple, by buoying injured self-esteem; or they may provide a sense of belonging when otherparts of one's support system are threatened. Domestic violence victims, faced with thepotential loss of their partners (through prosecution or other efforts to end the violence) andthe resources that accompany their relationships with their partner, may perceive the poten-tial loss of tangible resources, self-esteem, and sense of identity (for example, as wife orgirlfriend) as insurmountable obstacles. For these victims, then, social support may makethese losses seem less insurmountable and therefore may increase victims' ability to coop-erate with the criminal prosecution of their abusers.

There are at least three types of social support for domestic violence victims seekinghelp from the justice system, each of which was investigated in this study: emotional sup-port from members of one's social network; tangible or material support from members ofthat network; and institutional support from police and victim advocates within the crimi-nal justice system.

To understand the potential contribution of social support within a broader context, wealso investigated the predictive ability of demographic, violence severity, and mentalhealth variables. Potentially relevant demographic variables were identified based on con-versations with advocates and prosecutors within the criminal justice system. We chose toexamine the role of severity of violence in the relationship because prior empirical researchhas demonstrated its association with domestic violence victims' willingness to seek helpin general (Gondolf & Fisher, 1988; Wauchope, 1988) and, more particularly, with theirwillingness to seek help from police (Bachman & Coker, 1995; Bowker, 1984; Straus &Kantor, 1990) and lawyers (Bowker, 1987). Victim mental health characteristics, includ-ing the victim's levels of depression, substance abuse, and emotional dependence on theabusive partner, were also included because many advocates and prosecutors with whomwe spoke believed them to be central contributors to domestic violence victims' participa-tion in the criminal justice system.

Prosecution ofAbusers

430 L. Goodman et al.

METHODS

Participants

In Washington, DC, police are required to make an arrest in every domestic violence casewhen there is probable cause to believe a crime has been committed (Skolnik, 1997). Whenan arrest is made, the arresting officer tells the victim to report to the Domestic ViolenceIntake Center associated with the DC Superior Court the next morning in order to provideadditional information. Although only about one-third of the victims actually do appear atthe Intake Center (Clarke, 1998), it nevertheless remains the primary point of entry for alldomestic violence complainants in criminal cases in Washington, DC. The study samplewas recruited from this subgroup of domestic violence victims. It included only those vic-tims whose abusers were charged with a misdemeanor (simple assault, threats, and destruc-tion of property) rather than a felony (a crime punishable by more than 1 year in prison)because this group comprises the vast majority (80%-85%) of those who appear at the IntakeCenter. Also, including both misdemeanor and felony victims in the sample would not haveallowed for a uniform measure of the outcome. In misdemeanor cases, victims are uniformlyasked to return for the first time roughly 12 weeks after the initial intake, when trials areroutinely scheduled to begin. In felony cases, however, there is no corresponding set pro-cedure. Felony victims are typically required to cooperate with the prosecutors' office atmany different points and in many different ways before a trial date is even scheduled. Thus,it was anticipated that the study of felony domestic violence arrests would involve a dif-ferent set of dependent (and possibly independent) variables and necessitate a differentseries of measures.

Potential participants included 119 women over the age of 18 who appeared at theIntake Center from June 1 to December 31 of 1997, following the arrest on a misdemeanorcharge of a current or former male partner who assaulted them. (The roughly 5% of intakeswho were men were excluded from the study because different dynamics were anticipatedto underlie these cases [e.g., Koss et al., 1994].) Of these, 14 (11.8%) refused to partici-pate. Thirteen (12.4%) of the women who did participate by completing questionnaires wereexcluded from the final analysis because either information on cooperation with prosecu-tion was unobtainable (neither prosecutor nor victim ratings were available) (5), the casejackets had disappeared from the system so that the individual cases could not be followed(3), charges were never officially filed (2), the cases were upgraded to felony status frommisdemeanor (2), or the defendant had died (1). The final sample consisted of 92 partici-pants. Most of these participants (69.6%) had called the police themselves or directlyrequested that the police be called. For a small proportion (13.0%), a child or neighbor hadcalled the police, while the source of the call was not available for the remaining partici-pants. Neither predictor nor outcome scores were related to the source of the call to thepolice.

Procedure

When victims first arrive at the Intake Center, they meet with a legal advocate whoexplains their options and helps them to make a decision regarding their course of action.Participants were recruited for the study directly after meeting with an advocate. The pro-ject director for this study (second author), a clinical psychology graduate student, wason site 3 to 5 days a week for 7 months to administer questionnaires. After the nature ofthe study was explained to them and they signed an informed consent form, participants

Prosecution ofAbusers 431

completed a set of questionnaires in a semiprivate corner of the waiting room or in a quiethallway within the courthouse (no private rooms were available). In those few cases whereparticipants expressed concern about their ability to read the materials, the project directoradministered the protocol orally. The questionnaires required 45 - 60 minutes to complete.Participants were paid $10.00 for their time.

While obtaining consent for the study, the project director also requested permissionfrom participants to make follow-up phone calls to each participant to learn more about herexperience with the legal system. The project director, along with several trained under-graduate research assistants, made these phone calls just after the first scheduled trial date(about 12 weeks after intake) whether or not the trial actually occurred on that date. At thatpoint, 51 (55.4%) of the 92 cases had been continued, 26 (28.3%) were over (in 9 casescharges were dropped, in 4 cases the defendant was convicted, in 1 case the defendant wasacquitted, and in 12 cases, the defendant entered into a plea bargain), for 4 (4.3%), case sta-tus information was unobtainable because the case files were missing, and in 11 cases(12.0%), the U.S. Attorney's Office entered into a deferred sentencing agreement with thedefendant. These agreements stipulate that if the defendant abides by a set of conditions(e.g., counseling, no new arrests, staying away from the victim) for 9 months, the case isdropped. The main purpose of the follow-up calls was to elicit information from the par-ticipants themselves about whether or not they had cooperated with prosecutors up to thepoint of the first trial date. To ensure the participants' safety, phone interviewers asked ifthey were alone and if it was a good time to talk. Interviewers did not leave messages.Additionally, as described below, the prosecutor assigned to each case rated each partici-pant's willingness to proceed with prosecution at the time of the first scheduled trial date.

Measures

Demographic Information. The questionnaire elicited information on the following par-ticipant demographic variables: age, ethnicity, housing status, current relationship statuswith the abuser, length of that relationship, number of dependent children, children in com-mon with the abuser, employment status, economic dependence on the abuser, and whetherthe participant had called the police or cooperated in pressing charges against him before.Only those variables with sufficient variability were included as covariates. Sufficiency ofvariability was based on the percentage of participants in the largest response category ofa variable. If the percentage was less than 80%, variability was deemed sufficient.

Violence Severity. Violence severity was measured using the Revised Conflict TacticsScales (CTS2-Form A) (Straus, Hamby, Boney-McCoy, & Sugarman, 1996), a recentlyrevised version of the original Conflict Tactics Scales (Straus & Gelles, 1990), the mostwidely used measure of intra-family violence. The three violence-related subscales of theCTS2 were used: Physical Assault (12 items), Sexual Coercion (7 items), and Injury(6 items). Items elicit information on types of abuse (e.g., whether respondents have beenslapped or forced to have oral sex) and ramifications of the abuse (e.g., whether there werevisits to the doctor) experienced in the past year at the hands of the arrested partner. Participantsprovide frequency ratings on a 7-point ordinal scale, with response choices ranging from"never or not in the past year" to "more than 20 times in the past year." Items are summedto produce an index of violence severity over the last year. According to Straus and his col-leagues (1996), severe violence would include at least one instance of being kicked, punched,hit with something, slammed against a wall, choked, burned, "beaten up," or assaulted witha knife or gun. Severe sexual violence would include being forced to engage in oral, anal,or vaginal sex by the use of actual or threatened physical coercion. Severe injury would

432 L. Goodman et al.

include passing out from being hit on the head, needing to go to, or actually going to,a doctor because of a fight, or having a broken bone from a fight. Preliminary psychome-tric analyses of the CTS2 have demonstrated good internal consistency for all subscales,as well as adequate construct and discriminant validity (Straus et al., 1996). Cronbachalpha coefficients for participants in this study ranged from .75 for the Injury Subscale to.90 for the Physical Assault Subscale.

Social Support. The Interpersonal Support Evaluation List(ISEL) (Cohen, Mermelstein,Kamarck, & Hoberman, 1985), slightly modified by the authors of this report for use witha low-income population, was used to measure interpersonal social support.2 The ISEL isa 40-item true-false self-report instrument that measures the perceived availability ofsocial resources in four categories: tangible support (e.g., availability of material aid suchas help with transportation, childcare, or financial crises); appraisal support (e.g., avail-ability of someone to talk to about one's problems); self-esteem support (e.g., availabilityof positive comparisons to others); and belonging support (e.g., availability of people withwhom one can do things). Cohen and Hoberman (1983) reported that the ISEL was mod-erately correlated with a previously validated measure of social support, the Inventory ofSocially Supportive Behaviors (Barrera, Sandier, & Ramsay, 1981), suggesting adequateconstruct validity. As suggested by House and Kahn (1985), due to the high intercorrela-tions among the appraisal, belonging, and self-esteem scales (ranging from .57 to .69), thethree scales were combined by summing the individual scores to form one measure of emo-tional support. Cronbach alpha coefficients for participants in this study were .91 forEmotional Support and .80 for Tangible Support.

Developed in conjunction with staff from the Domestic Violence Intake Center, aJustice System Support Scale (Goodman, Dutton, & Bennett, 1997) was also administeredat intake as a measure of initial perceived institutional social support. The measure consistsof two subscales. The Police Subscale assesses perceived support from the police (withwhom the participant had contact the night before intake and, generally, during the intakeprocess itself). The Advocate Subscale assesses perceived support from the victim advo-cate (with whom the participant had just met prior to filling out the research questionnaires).Each subscale consists of five questions eliciting information on how participants felt theyhad been treated by the advocate or the police officer. For example, participants were askedabout the extent to which they felt protected from their abuser or informed about theiroptions. Each response was given on a 5-point scale, with response choices ranging from"not at all" to "very much," and item scores were summed. Cronbach alpha coefficientsfor the two measures were .77 (for the Police Subscale) and .75 (for the AdvocateSubscale).

Mental Health Variables. The Center for Epidemiologic Studies Depression Scale (CES-D) (Radloff, 1977) was used to measure current depressive symptomatology. On a 4-pointscale ranging from "rarely or none of the time" to "most or all of the time," respondentsreported the number of times they had experienced each of 20 depressive symptoms overthe week prior to the intake interview. The total score reflects severity of depression. Thisscale has been used extensively with community samples of low-income women (Belle,1982; Goodman, 1991). It has a high degree of construct and concurrent validity when com-pared to clinical diagnostic criteria, as well as to other self-report scales of depression (Belle,1982). In a study of the CES-D comparing a community sample and a sample of psychi-atric inpatients, 21% of the general sample scored above a cutoff score of 16, while 70%of psychiatric inpatients scored above 16 (Radloff, 1977). This score has come to representthe cutoff score, with scores above the cutoff considered indicative of clinical depression.The Cronbach alpha coefficient for the scale in this sample was .91.

Prosecution of Abusers 433

The Attachment Scale (Dutton, 1988). The Attachment Scale was used to assess emo-tional dependence on the abusive partner. The Attachment Scale is a 20-item scale devel-oped from the Kitson Scale (Kitson, 1982), designed to measure attachment duringdivorce, and from a scale by NiCarthy (1982), used to measure idealization of one's part-ner. Each item is rated on a 5-point Likert-type scale indicating how much (from "not atall" to "very much") the statement reflects the participant's feelings. For this sample, theCronbach alpha coefficient was .91 for this scale.

The Short Michigan Alcohol Screening Test (SMAST) (Selzer, Vinokura, & van Rooijen,1975). The SMAST was used to assess both alcohol and drug abuse. (It applies to bothdespite its name.) It is a 13-item scale with response options of yes and no. Items ask aboutparticipants' own and others' perceptions of their drug or alcohol use, and about their his-tory of substance use. The Cronbach alpha coefficient for this sample was .88 for this scale.

Cooperation With Prosecution. The major measure of victim cooperation was a ratingprovided by the prosecutor for each case on the day of the first scheduled trial date, whetheror not the trial actually occurred. The rating was made following a brief conversation withthe victim, either by phone or in person, before the hearing actually began or was originallyscheduled to begin. The first trial date was chosen because it occurs at roughly the samepoint in time (approximately 12 weeks after intake) in every case, and it is a time when theprosecutor typically makes contact with the victim. On a sticker attached to the front ofeach related prosecution file, the prosecutor checked off a box indicating whether the vic-tim was perceived as "cooperative" with the prosecution at that point. Cooperation wasdefined as the victim's perceived willingness to come into court and testify for the prose-cution if necessary. The prosecutor had three response options. For those victims who refusedto come to court, or indicated they would not testify or would lie if they did testify, the pros-ecutor was instructed to give a rating of "not cooperative." Victims who were willing tocome to court and testify truthfully were rated as "cooperative." Toward the end of thisstudy, the Domestic Violence Unit at the Intake Center began moving to a "no drop" pol-icy, which meant that some victims were told that they would have to testify whether theywanted to or not. Subsequently, a few prosecutors issued subpoenas to compel victims' tes-timony. Thus, a third response option was "only if compelled" for those victims who cameinto court to testify but only because they felt compelled to do so. The nine participantswho fell into the "only if compelled" category (who were not significantly different on anyof the variables we examined) were excluded from the study.

In 18 cases, prosecutors neglected to provide a cooperation rating. For these participants,information from the follow-up phone call between researcher and victim was usedinstead. Using the same behavioral criteria adopted by prosecutors, interviewers rated par-ticipants' phone responses with respect to their willingness to cooperate with the prosecu-tion of their abusers. Thus, if participants said they came into court at the first scheduledtrial date, or would have been willing to do so if necessary, they were rated cooperative,whether or not they were satisfied with the process. If participants reported that they weregenerally committed to cooperating with prosecutors but did not appear in court on theappropriate day because of ambivalence or logistical problems, they were rated as unco-operative. In those 37 cases (40.2%) where both prosecutor and interviewer ratings wereavailable, the two ratings were perfectly correlated.

Data Analysis

Logistic regression analysis was used to explore the relationship between the predictorvariables and the dichotomous outcome: whether or not participants cooperated with thecriminal prosecution. Following Hosmer and Lemeshow's (1989) guidelines, the initial step

434 L. Goodman et al.

taken was to use univariate logistic regressions to assess unadjusted odds ratios for eachvariable, independent of the others. Using variables that were significant at the level ofp < .30 in these univariate analyses, we examined the data by means of multivariate logis-tic regression. Using the Wald chi-square statistic, potential predictors were retained ordiscarded depending on whether or not the associated odds ratio remained statisticallysignificant when controlling for the contributions of the other variables.

RESULTS

Sample Description

Demographics. Almost 90% of the participants were African American. Their ages rangedfrom 18 to 46, with a mean age of 29.0 (SD = 7.0), and the average length of their rela-tionship with the abusive partner was about 31/2 years. Table 1 shows additionaldemographic characteristics of the sample. Participants reported a mean number of 1.9 chil-dren in the home, and almost half (45.8%) had children in common with the abuser. Closeto half the women (40.0%) were employed. Of these, 86.7% held a full-time job. Somebackground variables did not have sufficient variability to be included as covariates in theanalyses of cooperation. These included the participant's ethnicity, housing status (home-less or housed), and financial dependence on the abuser, and whether the participant wascurrently pregnant by the abuser, or had previously cooperated with pressing charges againsthim.

Violence Severity. As Table 2 shows, participants generally reported high levels ofphysical violence during the past year. Ninety percent had experienced at least one instanceof severe physical violence at the hands of the defendant. Although none of the participantsreported an instance of severe sexual violence during the past year, over half reported aninstance of "minor" (i.e., less severe) sexual violence. And over half reported severe injurylevels resulting from physical or sexual assault.

Social Support. Table 3 displays means and standard deviations for each continuous pre-dictor variable both for the total sample and for each of the two subgroups ("cooperating"and "noncooperating"). Participants reported comparable levels of tangible and emotionalsupport.3 Interestingly, both emotional and tangible support scores were similar to thosefound in a sample of predominantly Caucasian, middle-income battered women (Dutton,1992), though they were lower than scores for other community samples (Weinberger,Tierney, Booher, & Hiner, 1990). It may be that the isolation brought on by the experienceof domestic violence washes out any potential differences in tangible support that can usu-ally be found across social classes. With respect to institutional support, there are no normsagainst which to compare this sample since we developed the scales. On the advocate sup-port scale, where the maximum possible score was 20, 71.1% of the sample scored 17 orhigher. On the police support scale, where the maximum possible score was also 20,77.1%scored 17 or higher. Such high scores suggest that participants were highly satisfied withthe support offered by police and advocates at the initial intake, an unsurprising resultgiven that all the participants chose to heed the advice of police and report to the IntakeCenter the day after the abusive event when the questionnaire was also administered. It isalso possible, however, that participants were afraid of the negative consequences shouldthey rate criminal justice personnel poorly. Yet another explanation, to be described laterin this article, has to do with the nature of the Justice System Support Scale itself.

Prosecution ofAbusers 435

TABLE 1. Demographic Characteristics

Characteristic Number Percent

Level of Violence

NoneMinorSevere

Number

53

75

Percent

6.03.6

90.4

Number

36450

Percent

44.455.60

Number

82649

Percent

9.631.359.0

!Two participants did not complete the sexual abuse section of the CTS. Therefore, for thisvariable n = 81, rather than 83.

Mental Health Characteristics. The vast majority of participants (79.5%) exceeded thecutoff of 16 on the CES-D, used by Radloff (1977) to indicate clinical depression, and anonly slightly smaller percentage (72.3%) exceeded the score of 21 used by Weissman andLocke (1975) as the criterion to identify acute depressives.

EthnicityAfrican Americanother

Relationship with abusive partnermarriedboyfriendseparated/divorced/ex-boyfriend

Childrenwith abuseronly with someone other than abuserno children

Currently pregnant by abuseryesno

Financially dependent on abuseryesno

Permanent place to liveyesno

Employedfull-timepart-timeunemployed

Called the police about this person beforeyesno

Cooperated with pressing chargesagainst this person before

yesno

729

122744

382223

479

1467

781

264

45

6419

1766

88.811.2

14.532.553.0

45.826.527.7

4.895.2

17.382.7

98.71.3

34.75.3

60.0

77.122.9

20.579.5

TABLE 2. Levels and Types of Violence and InjuryAs Reported in the Conflict Tactics Scales

Type of Violence and Injury

Physical Sexual1 Injury

436 L. Goodman et al.

TABLE 3. Means, Standard Deviations and Ranges for Predictor Variables

Total sample(n = 83)

Predictor

ISELTISELEMSSADVASSCOPCESDATTACHSMASTCTSAGEKIDHOMERELLONG

M

17.150.117.517.928.525.2

1.535.828.9

1.941.6

SD

2.77.23.23.3

14.016.22.2

25.87.02.1

42.1

Range

11-2032-60

8-205-213-570-650-104-117

18-460-141-204

Cooperating(n = 44)

M17.650.517.618.327.722.5

.941.228.6

1.646.2

SD

2.47.33.13.1

14.015.4

1.329.8

7.11.4

41.0

Range

12-2037-6012-205-213-571-580-64-117

19-440-51-180

Noncooperating(n = 39)

M

16.449.617.417.529.428.3

2.229.729.2

2.236.4

SD2.87.13.43.5

14.216.62.8

19.06.92.7

43.3

Range

11-2032-60

8-207-203-570-650-100-79

18-460-143-204

Note. The four dichotomous predictor variables (relbatt, job, kidbatt, and copcall) were not includedin this table but are included in Table 1.ISELT = Interpersonal Support Evaluation List, Tangible Subscale; ISELEM = InterpersonalSupport Evaluation List, Emotional Subscales; SSADVA=Social Support from the Victim Advocate;SSCOP = Social Support from the Police; CESD = Center for Epidemiological Studies DepressionScale; ATTACH = Attachment Scale; SMAST = Short Michigan Alcohol Screening Test; CTS =Conflict Tactics Scales; AGE = Age of Participant; KIDHOME = number of kids the participanttakes care of at home; RELLONG = the length of the relationship between the participant and theabuser in months.

With regard to emotional dependence or attachment, scores for participants in this studyare similar to those reported by Dutton and Painter (1993) in a study of women who hadrecently left a physically and/or emotionally abusive relationship (M = 28.5, SD = 18.0).Of the 54 participants who filled out the SMAST, 12 (22.2%) received a score of 3 or more,indicating likely substance abuse (Selzer et al., 1975).

Cooperation. Of the 83 participants, 39 (47.0%) were rated as noncooperative, while 44(53.0%) were rated as cooperative.

Multivariate Predictive Model

Table 4 shows the bivariate correlations among all predictor variables and the outcome vari-able. A positive correlation represents a positive relationship between the predictor andcooperation. As Table 4 demonstrates, four variables—tangible support, severity of vio-lence experienced by the participant (at the hands of the abuser) in the past year, substanceabuse, and relationship with the abuser—were significantly correlated with cooperation.Table 5 displays the unadjusted odds ratios obtained in univariate logistic regressions. Forcontinuous variables, odds ratios were calculated with reference to an increment of onestandard deviation. As with the bivariate correlations, tangible support, severity of violence,and substance abuse were significantly related to the outcome in this set of analyses, as wasthe variable reflecting the contrast between boyfriend vs. divorced/separated/ex-boyfriend(with the latter more often associated with cooperation), though the contrast betweenmarried and divorced/separated/ex-boyfriend was not. Additionally, the following vari-ables were marginally (p < .30) associated with the outcome: police support, emotionaldependence on the abuser, length of the relationship with the abuser, having children incommon with the abuser, and having called the police in the past.

Prosecution ofAbusers 437

TABLE 4. Bivariate Correlations Between Predictors and Outcome

Predictor

ISELTISELEMSSADVASSCOPCTSCESDATTACHSMASTAGEKIDHOMERELLONGKIDBATTJOBRELBATT1RELBATT2COPCALL

Correlation With Outcome

.22

.06

.03

.12

.22-.06-.18-.29-.05-.16.12.14

-.09-.04.27

-.12

P.05.56.81.27.04.58.10.01.68.16.30.21.42.74.02.28

Outcome = Cooperated with the prosecution - yes/no; ISELT = InterpersonalSupport Evaluation List, Tangible Subscale; ISELEM = Interpersonal SupportEvaluation List, Emotional Subscales; SSADVA = Social Support from the VictimAdvocate; SSCOP = Social Support from the police; CTS = Conflict TacticsScales; CESD = Center for Epidemiological Studies Depression Scale; ATTACH= Attachment Scale; SMAST = Short Michigan Alcohol Screening Test; KID-HOME = the number of children the participant takes care of at home; RELLONG= the length of the relationship between the participant and the abuser in months;KIDBATT = does the participant have children in common with the abuser -yes/no; JOB = does the participant have a job; RELBATT1 = relationship withthe abuser - married versus ex-partner; RELBATT2 = relationship with the abuser- boyfriend versus ex-partner; COPCALL = has the participant called the policeabout the abuser in the past.

Using only those variables that were significantly or marginally associated with the out-come at the univariate level, a multivariate logistic regression model was developed thatconsidered these variables simultaneously. Table 6 presents the final predictive model.Participants with a higher level (one standard deviation increase) of tangible support wereabout twice as likely to cooperate with the prosecution; those who reported more severeviolence were almost two-and-a-half times as likely to cooperate. Participants with chil-dren in common with the abuser were three-and-three-quarter times more likely to coop-erate with prosecution; and particants who reported higher levels of substance abuse weretwo- and-a-quarter times less likely to cooperate. The chi-square goodness-of-fit value forthe final model was 23.81, p - .0001, indicating that this is a useful model for predictingcooperation with domestic violence prosecution.

DISCUSSION

Results of this study suggest that women who are domestic violence victims in urban crim-inal justice systems struggle with a range of difficulties. Over half of the respondents wereunemployed. Approximately 90% had endured severe physical violence at the hands of apartner during the preceding year, and over half had suffered severe injuries from the

438 L. Goodman et al.

TABLE 5. Unadjusted Odds Ratios for Predictors

Predictors

ISELTISELEMSSADVASSCOPCTSCESDATTACHSMASTAGEKIDHOMERELLONG**KIDBATTJOBRELBATT1***RELBATT2COPCALL

B

.17

.02

.02

.08

.02-.01-.02-.34-.01-.17.01.56

-.36.67

-1.20-.57

SE

.09

.03.07.07.01.02.01.14.03.12.01.45.44.66.49.53

WALD

3.85*.34.06

1.213.91*.31

2.615.61*

.181.841.091.58.67

1.035.70*1.16

UnadjustedOdds

1.581.141.051.291.641.131.452.091.101.421.071.751.441.943.331.77

95%Confidence

Interval

1.00,2.4.74, 1.76.68, 1.63.82, 2.02

1.01,2.66.73, 1.75.92, 2.27

1.14,3.83.71,1.69.86, 2.34.94, 1.22.73,4.21.60, 3.42.54, 7.04

1.27,8.77.63, 4.98

ISELT=Interpersonal Support Evaluation List, Tangible Subscale; ISELEM = InterpersonalSupport Evaluation List, Emotional Subscales; SSADVA = Social Support from theVictim Advocate; SSCOP = Social Support from the Police; CTS = Conflict TacticsScales; CESD = Center for Epidemiological Studies Depression Scale; ATTACH =Attachment Scale; SMAST = Short Michigan Alcohol Screening Test; AGE = Age ofParticipant; KIDHOME = the number of children the participant takes care of at home;RELLONG = the length of the relationship between the participant and the abuser, inmonths; KIDBATT = does the participant have children in common with the abuser -yes/no; JOB = does the participant have a job; RELB ATT = relationship with the abuser- married, boyfriend, or ex-partner; COPCALL = has the participant called the policeabout the abuser in the past.*/?<.05.**In order to increase ease of understanding, an increment of 12 months was used forthe calculation of this odds ratio, rather than the standard deviation of 42.29 months.***RELBATT1 and RELBATT2 are contrasts, such that RELBATT1 = married versusdivorced/separated/ex-boyfriend, and RELBATT2 = boyfriend versus divorced/separated/ex -boyfriend.

TABLE 6. Final Model

Predictor

KIDBATTCTSSMASTISELT

B

1.32.03-.37.26

SE

.58

.01

.14

.10

WALD

5.17*7.78**6.66**6.55**

UnadjustedOdds

3.762.422.261.98

95%Confidenc

Interval

1.20,11.751.30,4.511.22,4.181.17,3.33

KIDBATT = have children in common with the abuser - yes/no; CTS = Conflict TacticsScale; SMAST = Short Michigan Alcohol Screening Test; ISELT = Interpersonal SupportEvaluation List, Tangible Subscale.*/?<.05. **p<.01.

Prosecution ofAbusers 439

abuse. About three-quarters reported symptoms consistent with a diagnosis of clinical depres-sion and about one-fifth met criteria for likely substance abuse. It is perhaps remark-ablethat despite these difficulties, over half of the participants were cooperative with theprosecution at the time of the first scheduled trial date. As noted above, such cooperationprobably did not come without substantial personal costs.

Interpersonal and Institutional Support

Tangible support significantly predicted cooperation with prosecution in the final model:those who reported a higher level of availability of people to help with a range of practicalissues (e.g., someone to help with daily chores, take care of the participant's children, orprovide an emergency loan) were about twice as likely to cooperate with the prosecution.There are several possible explanations for this finding. Although most of the women inthis study (82.7%) reported that they were not financially dependent on the abuser, infor-mal discussions with participants indicated that for some, the partner provided other formsof support such as childcare, transportation, food, or emergency help. By pursuing crimi-nal prosecution, these women stand to lose whatever support their abusive partners pro-vided them. Moreover, many of these women may need additional forms of tangiblesupport, such as housing or help finding a new job in a new location, or to help them reor-ganize their lives to safeguard against continuing violence from their partners. Finally,material help in the form of transportation and babysitting may be necessary for domesticviolence victims to pursue the very simplest tasks that are part of the criminal prosecutionprocess, such as attending hearings or meeting with the prosecutors. The more that tangi-ble support from family and friends is available to replace contributions from the abusivepartner and to help create new options, the more likely it may be that those women whoseek help from the criminal justice system are willing or able to cooperate with the crimi-nal prosecution of their abusers.

Contrary to expectation, however, the relationship between emotional support andcooperation with prosecution was not significant. It may be that some of the support thewomen received emphasized loyalty to the abusive partner or to the African Americancommunity above all else, and led them to choose not to cooperate with the prosecution.Thus, high emotional support scores could have represented a mix of encouragement tocooperate and to not cooperate—thereby eviscerating a potential relationship betweenemotional support and cooperation with prosecution.

Institutional support, whether from the police or from the victim advocates at court, wasalso unrelated to cooperation in the final model. It is possible, however, that a ceiling effectwas at work for both the advocate and the police support scales, with the institutional sup-port scales lacking sufficient sensitivity to capture participants' reactions to advocates andpolice. It is also important to note that support received was measured during the initialintake process; disillusionment with the institutional support provided likely manifests itselflater in the criminal justice process (Bennett, Goodman, & Dutton, in press).

Mental Health Factors

Neither depression nor emotional dependence on the abusive partner was related to coopera-tion in the final model. This finding is surprising. It contradicts the common perception of thebattered woman as unable to cooperate with the prosecution out of depression, helplessness,or attachment to the abuser. Instead, it suggests that many domestic violencevictims press on in the face of depression and despite sometimes complex emotional ties to

440 L. Goodman et al.

their partner. It is also possible, however, that these results are simply artifacts of the timing ofquestionnaire administration. Participants had been assaulted less than 24 hours before partic-ipating in this study. Thus, their emotional states and feelings for the abuser were likely in astate of great flux. It may be that if depression and emotional dependence had been measuredcloser to the time of the first scheduled trial date, different results would have been obtained.

Those who reported more symptoms of substance abuse, however, were over two timesless likely to cooperate with prosecution. The finding that substance abuse predicts a lackof cooperation is not surprising. Women with alcohol or drug problems have difficulty struc-turing their lives to carry out simple tasks, much less the complex process of criminal pros-ecution (Steele & Southwick, 1985). Moreover, victims who were abusing substances dur-ing the assault that initiated the arrest may believe that the abuse was partially their fault,or that a judge or prosecutor will not take them seriously. Some may fear that their sub-stance abuse will come to light in the course of these proceedings and lead to negative reper-cussions such as criminal charges or the loss of custody of their children. This fear maylead them to resist criminal charges against their partners to avoid this possibility.

Violence Severity

Greater severity of violence in the relationship between the victim and her partner was asso-ciated with a greater likelihood of cooperation in the final model. This finding is consistentwith earlier studies showing that violence severity is related to greater levels of helpseekinggenerally (Gelles & Straus, 1988; Gondolf & Fisher, 1988), and that the severity of the pre-cipitating incident is related to cooperation with criminal prosecution in particular(McLeod, 1983). It is also consistent with studies showing that victims who suffer less severeabuse are more likely to remain with their abusive partners (Snyder & Fruchtman, 1981).Far from creating passivity in victims, greater severity of abuse seems to spur action. It shouldbe noted, however, that this study explored misdemeanor cases only. It may be that for somefelony victims, who endure the most severe and injurious assaults, passivity may be moreprevalent. For such victims, staying in the relationship may be viewed as safer than leavingit, particularly if leaving is believed to increase the risk of lethal violence.

Demographics

Interestingly, among the demographic covariates investigated, only having a child or chil-dren with the abuser was related to cooperation in the final model. Those participants whoshared children with the abusive partner were almost four times as likely to cooperate withthe prosecution. Perhaps these women perceive the criminal justice system as a vital toolfor managing the ongoing relationship that shared children necessitate. For example, vic-tims who share children with their abusive partner may be motivated by the hope that thissystem will force the abuser into treatment.

In sum, the results of this study suggest that, although substance abuse appears to be animportant obstacle to cooperation, mental health characteristics such as emotional depen-dence and depression do not explain domestic violence victims' reluctance to cooperatewith the prosecution of their abusers. Instead, contextual factors such as violence severity,shared children, and tangible support are more important in shaping the decisions of suchvictims in the criminal justice system.

Limitations and Directions for Future Research

It is important to note several limitations of this study. First, we measured participants'willingness to cooperate with prosecution at the first scheduled trial date; it is possible that

Prosecution ofAbusers 441

some women who were willing to cooperate at that point chose to drop out later in theprocess, which could last anywhere between a few months to a year or more. Second, weassessed only a limited number of institutional factors potentially related to cooperation.Informal conversations with participants indicated that their perceptions of how they aretreated within the criminal justice system are critical to their willingness to continue theirinvolvement. Future research might examine a wide range of system-related factors suchas, for example, the length of the process, the actions and demeanor of the judge at pretrialhearings, and the extent to which advocates and prosecutors remain in contact with victims(Bennett, Goodman, & Dutton, in press).

Finally, given this study's small sample size and the specific characteristics of thepopulation from which it was drawn (urban and African American), caution should be takenin generalizing the results. It would be useful to replicate this study (with larger samples) inother jurisdictions and with other populations in order to cross-validate this study's findings.

Despite these limitations, this study represents one of the first efforts to explore factorsrelated to domestic violence victims' cooperation with the criminal prosecution of theirabusers. As such, it presents important findings for researchers and advocates to consider.

Implications for Intervention

The results of this study do suggest that there is a need for further education of those in thelegal system who have contact with domestic violence victims, particularly with regard tothe causes of prosecution "dropout." Training should underscore that the more severe theviolence domestic violence victims experience, the more likely they are to seek help ingeneral and to cooperate with prosecution in particular. This finding might reduce the skep-ticism and frustration of police and prosecutors who believe that domestic violence victimswho have called the police repeatedly, or who have entered the system before only to dropout, will never cooperate. Instead, they might see victims' use of the criminal justice sys-tem as a process, during which time it is important to continue to encourage their use ofthis system, particularly so that a response can be forthcoming quickly as the level of vio-lence increases.

Additionally, as part of intake and follow-up, advocates and prosecutors should assessvictims' social support systems and consider how to help them to enhance and utilize theirtangible social support network during the stressful prosecution process. Concretely, theymight offer help with transportation and even childcare for times when the victim needs tocome to court or to the prosecutor's office. More broadly, advocates and prosecutorsshould explore the kinds of tangible support a domestic violence victim will be risking byprosecuting the abuser, and how these might be replaced, at least partially. Hart (1996)goes even further, proposing that domestic violence advocates engage and educate mem-bers of the victims' social network so that they can provide needed support. Instead of rely-ing on stereotypes of victims of domestic violence as passive or emotionally dependent,advocates, police, prosecutors, and judges who identify contextual obstacles to victims'cooperation in the criminal justice process have a far better chance of helping women pur-sue a course of action to end the abuse in their lives.

NOTES

lrThe vast majority of domestic violence or partner violence cases involve male perpe-trators and female victims. Further, this article describes a study in which all victims werewomen and all abusers were men. Thus, although there are certainly cases in which the

442 L. Goodman et al

gender roles are reversed or which involve same-sex intimate abuse, this article focusesexclusively on male violence against women.

2Only a few modifications were made to this scale, none of which was likely to dam-age its integrity. For example, for one item, the phrase "If needed a ride to the airport veryearly in the morning . . ." was replaced with the phrase, "If I needed a ride to somewhereimportant very early in the morning...."

3Although the mean tangible support score appears to be much lower than the mean emo-tional support score, this is because the latter is based on three subscales combined. Whenthe number of subscales involved is taken into account, the average score on the two scalesare quite similar.

REFERENCES

Bachman, R., & Coker, A. L. (1995). Police involvement in domestic violence: The inter-active effects of victim injury, offender's history of violence, and race. Violence andVictims, 10(2), 91-106.

Barrerra, M., Sandier, I., & Ramsay, T. B. (1981). Preliminary development of a scale ofsocial support: Studies on college students. American Journal of Community Psychology,9(4), 435-447.

Belle, D. (1982). Social ties and social support. In D. Belle (Ed.), Lives in stress: Womenand depression (pp. 133-144). Beverly Hills: Sage Publications.

Bennett, L., Goodman, L. A., & Dutton, M. A. (in press). Systemic obstacles to the criminalprosecution of a battering partner: A victim perspective. Journal of Interpersonal Violence.

Bowker, L. H. (1984). Battered wives and the police: A national study of usage and effec-tiveness. Police Studies, 7(2), 84-93.

Bowker, L. H. (1987). Battered women as consumers of legal services: Reports from anational survey. Response, 10(1), 10-17.

Cahn, N. R. (1992). Innovative approaches to the prosecution of domestic violence crimes:An overview. In E. S. Buzawa & C. G. Buzawa (Eds.), Domestic violence: The chang-ing criminal justice response (pp. 162-177). Wesport, CT: Greenwood.

Cahn, N. R., & Lerman, L. G. (1991). Prosecuting woman abuse. In M. Steinman (Ed.),Woman battering: Policy responses (pp. 97-109). Cincinnati, OH: Anderson PublishingCo., Criminal Justice Division.

Campbell, J. C., Miller, P., Cardwell, M., & Belknap, R. A. (1994). Relationship status ofbattered women over time. Journal of Family Violence, 9(2), 99-111.

Clarke, C. (1998, April 6). Written testimony for the Mayor's Commission Against DomesticViolence. Unpublished manuscript.

Cohen, S., & Hoberman H. M. (1983). Positive life events as buffers of life change stress.Journal of Applied Social Psychology, 13(2), 99-125.

Cohen, S., Mermelstein, R., Kamarck, T., & Hoberman, H. M. (1985). Measuring the func-tional components of social support. In I. G. Sarason & B. R. Sarason (Eds.), Socialsupport: Theory, research, and applications (pp. 73-94). The Hague, The Netherlands:Martinus Nijhoff.

Cohen, S., & Wills, T. A. (1985). Stress, social support, and the buffering hypothesis.Psychological Bulletin, 98(2), 310-357.

Dutton, D. G. (1988). The domestic assault of women: Psychological and criminal justiceperspectives. Boston: Allyn and Bacon, Inc.

Prosecution of Abusers 443

Dutton, M. A. (1992). Empowering and healing the battered woman: A model for assess-ment and intervention. New York: Springer Publishing Co.

Dutton, M. A., & Painter, S. (1993). The battered woman syndrome: Effects of severity andintermittency of abuse. American Journal of Orthopsychiatry, 63(4), 614-622.

Elliott, D. S. (1989). Criminal justice procedures in family violence crimes. In L. Ohlin &M. Tonry (Eds.), Family violence. Chicago: University of Chicago Press.

Ford, D. A. (1983). Wife battery and criminal justice: A study of victim decision-making.Family Relations, 32, 463-475.

Ford, D. A. (1991). Prosecution as a victim power resource: A note on empowering womenin violent conjugal relationships. Law and Society Review, 25(2), 313-334.

Ford, D., & Regoli, M. J. (1992). The preventive impact of policies for prosecuting wifebatterers. In E. S. Buzawa & C. G. Buzawa (Eds.), Domestic violence: The changingcriminal justice response (pp. 181-207). Westport, CT: Greenwood.

Gelles, R. J., & Straus, M. A. (1988). Intimate violence. New York: Simon & Schuster.Gondolf, E. W., & Fisher, E. R. (1988). Battered women as survivors: An alternative to

treating learned helplessness. Lexington, MA: Lexington Books.Goodman, L. A. (1991). The prevalence of abuse among homeless and housed poor moth-

ers: A comparison study. American Journal of Orthopsychiatry, 61(4), 489-500.Goodman, L. A., Dutton, M. A., & Bennett, L. (1997). Justice System Support Scale.

Unpublished scale.Hampton, R. L., & Gelles, R. J. (1993). Violence toward Black women in a nationally rep-

resentative sample of Black American families. In R. L. Hampton, T. P. Gullotta,G. R. Adams, E. Potter, II, & R. P. Weissberg (Eds.), Family violence: Prevention andtreatment (pp. 113-139). Thousand Oaks, CA: Sage Publications.

Hanna, C. (1996). No right to choose: Mandated victim participation in domestic violenceprosecutions. Harvard Law Review, 109, 1850-1909.

Hart, B. (1993). Battered women and the criminal justice system. American BehavioralScientist, 36(5), 624-638.

Hart, B. (1996). Victim issues [On-line]. Available: [email protected], S. E. (1988). The ecology of stress. New York, NY: Hemisphere.Hosmer, D. W., & Lemeshow, S. (1989). Applied logistic regression. New York: John Wiley

& Sons.House, J. S., & Kahn, R. L. (1985). Measures and concepts of social support. In S. Cohen &

S. L. Syme (Eds.), Social support and health (pp. 83-108). New York: Academic Press.Kitson, G. C. (1982). Attachment to the spouse in divorce: A scale and its application.

Journal of Marriage and the Family, 44, 379-393.Koss, M. P., Goodman, L. A., Browne, A., Fitzgerald, L. F, Keita, G. P., & Russo, N. F.

(1994). No safe haven: Male violence against women at home, at work, and in thecommunity. Washington, DC: American Psychological Association Press.

Langen, P. A., & Innes, C. A. (1986). Preventing domestic violence against women: Discussionpaper. Bureau of Justice Statistics Special Reports. Washington, DC: Bureau of JusticeStatistics.

Lerman, L. G. (1980). Prosecution of spouse abuse: Innovations in criminal justice response.Washington, DC: Center for Women Policy Studies.

Lerman, L. G. (1986). Prosecution of wife beaters: Institutional obstacles and innovat-ions. In M. Lystad (Ed.), Violence in the home: Interdisciplinary perspectives(pp. 250-295). New York: Brunner/Mazel, Inc.

444 L. Goodman et al.

McLeod, M. (1983). Victim noncooperation in the prosecution of domestic assault: Aresearchnote. Criminology: An Interdisciplinary Journal, 27(3), 395-416.

NiCarthy, G. (1982). Getting free: A handbook for women in abusive relationships. Seattle,WA: Free Press.

Radloff, L. S. (1977). The CES-D Scale: A self-report depression scale for research in thegeneral population. Applied Psychological Measurement, 7(3), 385-401.

Rebovich, D. J. (1996). Prosecution response to domestic violence: Results of a survey oflarge jurisdictions. In E. S. Buzawa & C. G. Buzawa (Eds.), Do arrests and restrain-ing orders work? (pp. 176-191). Thousand Oaks, CA: Sage Publications.

Selzer, M. L., Vinokura, A., & van Rooijen, L. (1975). A self-administered Short MichiganAlcoholism Screening Test (SMAST). Journal of Studies on Alcohol, 36, 117-132.

Sherman, L. W., Schmidt, J. D., & Rogan, D. P. (1992). Policing domestic violence:Experiments and dilemmas. New York: Free Press.

Skolnik, S. (1997). Confronting domestic violence. Legal Times, 20(14), 1, 14-16.Snyder, D. K., & Fruchtman, L. A. (1981). Differential patterns of wife abuse: A data-based

typology. Journal of Consulting and Clinical Psychology, 49(6), 878-885.Steele, C. M., & Southwick, L. (1985). Alcohol and social behavior I: The psychology of

drunken excess. Journal of Personality and Social Psychology, 48, 18-34.Straus, M. A., & Gelles, R. J. (1990). Physical violence in American families: Risk factors

and adaptation to violence in 8,145 families. New Brunswick, NJ: Transaction.Straus, M. A., Hamby, S. L., Boney-McCoy, S., & Sugarman, D. B. (1996). The Revised

Conflict Tactics Scales (CTS2): Development and preliminary psychometric data.Journal of Family Issues, 77(3), 283-316.

Straus, M. A., & Kantor, G. K. (1990). Response of victims and the police to assaults onwives. In M. A. Straus & R. J. Gelles (Eds.), Physical violence in American families:Risk factors and adaptation to violence in 8,145 families (pp. 473-487). New Brunswick,NJ: Transaction.

Wauchope, B. A. (1988.) Help-seeking decisions of battered women: A test of learned help-lessness and two stress theories. Unpublished master's thesis, University of NewHampshire, Durham, NH.

Weinberger, M., Tierney, W. M., Booher, P., & Hiner, S. L. (1990). Social support, stress,and functional status in patients with osteoarthritis. Social Science and Medicine, 30(4),503-508.

Weissman, M., & Locke, B. (1975). Comparison of a self-report symptom rating scale(CES-D) with standardized depression rating scales in a psychiatric population: A pre-liminary report. Paper presented at the Society for Epidemiological Research, Albany,New York.

Acknowledgments. Support for this study was provided by the Psychology Department at the Universityof Maryland, College Park, from which the first author recently moved, and the University of MarylandWomen's Forum. We would like to thank the following people for their invaluable help on this study:Charlotte Clarke, Robert Spagnoletti, KathrynTurman, Kevin Weinfurt, Leily Farhat-Sepahi, and thedomestic violence victim advocates, prosecutors, and support staff at the U.S. Attorney's Office inWashington, DC.

Offprints. Requests for offprints should be directed to Lisa Goodman, PhD, Campion 310, CounselingPsychology, School of Education, Boston College, Chestnut Hill, MA 02467.