Вы находитесь на странице: 1из 22

Journal of Family Violence, Vol. 13, No.

3, 1998

Coordinated Community Intervention for Domestic Abusers: Intervention System Involvement and Criminal Recidivism
Christopher M. Murphy,1,2 Peter H. Musser,1 and Kenneth I. Maton1

This study examined prosecution and post-prosecution elements of a coordinated community intervention approach to male perpetrators of adult domestic violence. In a sample of 235 cases, recidivism was assessed from official criminal justice data during a 12- to 18-month period after cases were initially handled by the Baltimore, Maryland State's Attorney's Domestic Violence Unit. Court orders for domestic violence counseling were associated with significantly lower criminal recidivism for battery or violation of a civil order of protection. Lower criminal recidivism was also associated with the cumulative effects of successful prosecution, probation monitoring, receiving a court order to counseling, attending counseling intake, and completion of counseling. Individuals with greater involvement in this intervention system had lower recidivism rates, even though offenders with more extensive abuse histories experienced more intervention. Results provide qualified support for coordinated community intervention for domestic violence perpetrators. KEY WORDS: community intervention; domestic violence; legal system; counseling; recidivism.

Domestic violence is a widespread crime that has devastating social and public health costs. Population surveys suggest that about 4% of married and cohabiting women in the U.S. experience serious physical violence from a male partner each year. The prevalence of domestic violence is considerably elevated in high stress urban environments and economically dis'Department of Psychology, University of Maryland, Baltimore County, 1000 Hilltop Circle, Baltimore, Maryland 21250. ^Tb whom correspondence should be addressed.
263
0885-7482/98/0900-0263115.00/0 1998 Plenum Publishing Corporation

264

Murphy, Musser, and Maton

advantaged ethnic minority communities (Straus and Gelles, 1990). Effective strategies to detect domestic violence and limit future violence are sorely needed. Toward this end, considerable research has been conducted on the effects of arrest and other police procedures on domestic violence recidivism (e.g., Berk et al, 1992; Dunford et al., 1990; Sherman and Berk, 1984). Yet, very few studies have examined post-arrest legal sanctions and their associations with recidivism. Research on the specific deterrent effects of arrest has yielded mixed results due, in part, to individual variation in response to arrest. It is also quite likely that police interventions exert relatively small and inconsistent effects because they are just one component of a complex intervention system that includes prosecution strategies, probation monitoring, court-ordered counseling for domestic violence perpetrators, and victim services (Edleson, 1991; Pence, 1983). Coordination of various interventions may be as important as the specific procedures used within each system component, because the intervention system can break down at many points. Effective arrest policies, for example, may have only limited effects on future behavior if charges are dropped or prosecution is otherwise ineffective. Limited data are available on the effects of coordinated community intervention for domestic violence. Steinman (1988) reported on prosecution and postprosecution variables as predictors of recidivism in a system wide community intervention program in Lincoln, Nebraska. The results were generally discouraging. The degree of involvement in the criminal justice system was not significantly correlated with reductions in recidivism, nor were the specific sanctions studied, which included citations, charges, pretrial diversion, probation or counseling, fines, or jail time. With the exception of formal charges, however, very few individuals in this study received each specific sanction. Legal sanctions may have been associated with small reductions in recidivism that were not detected as statistically significant due to the modest sample size and the wide array of system variables examined. More promising results were obtained by Syers and Edleson (1992) in their evaluation of the community intervention program in Minneapolis, Minnesota. They found that arrest alone was not significantly associated with reductions in recidivism, but that arrest along with a court-order for domestic violence counseling was associated with significantly lower recidivism in the ensuing 12 months. No other prosecution outcomes were assessed in this investigation, however, and no data were provided on perpetrators' compliance with the court orders for counseling. The current investigation examined whether key prosecution and postprosecution variables were associated with lower recidivism rates during a 12- to 18-month period after a case was handled by the Domestic Violence

Community Intervention for Domestic Abusers

265

Unit of the Baltimore, Maryland State's Attorney's Office. Recidivism was assessed through criminal justice system data, and was defined as generation of new charges for battery, violation of a stay away order, or other offenses that are often associated with domestic abuse (e.g., assault, telephone abuse, malicious destruction of property). Although the problem of domestic violence is complex, the current investigation began with a simple explanatory model. We predicted that each specific component of community intervention would be associated with small reductions in recidivism, while no specific component would have an overwhelming effect on recidivism. This prediction was made in recognition that domestic abuse is a relatively intransigent problem, with both legal and psychological dimensions (Dutton, 1995). We further predicted that significant, cumulative reductions in recidivism would be apparent in the combined effects of coordinated intervention. The assumption was that the common legal system interventions for domestic abusers do impact behavior, yet are low intensity interventions. Very few offenders are incarcerated, and other interventions, such as probation or counseling, involve at most a few hours per week of direct contact with the intervention system. Their impact on day to day activities, attitudes, and relationship functioning are likely to be modest, but additive. When no intervention is forthcoming upon adjudication, the abuser not only receives the message that his behavior will be ignored by the legal system, but may also come away believing that abusive behavior is justified or within his rights. The working assumption is that the greater the degree of intervention system involvement, the more likely an abuse perpetrator will receive the message that abuse is socially unacceptable and unjust. In addition, due to individual differences, only a proportion of offenders may respond to each specific intervention component. Someone who has never been in trouble with the law and who has a reasonable capacity for self-control, for example, may be sufficiently intimidated by a courtroom appearance and probation to initiate behavior change. Someone with a prior criminal history may be less impressed by a guilty verdict or probation, but may find counseling to be a unique experience that stimulates personal reflection and behavior change. Differential responsiveness to intervention system components would be observed as small, cumulative effects because individuals who receive more intervention are more likely to be affected by at least one of the intervention components. Finally, our model of small, cumulative effects corresponds to our clinical experience and the impressionistic reports by probation agents and other domestic violence counselors of success with a modest proportion of offenders. Specifically, we hypothesized that lower recidivism rates would be associated with the degree of sanctions levied by the court and compliance

266

Murphy, Musser, and Maton

with court sanctions. The following specific variables were hypothesized to exert small, cumulative effects on recidivism. A brief rationale for the inclusion of each variable is also provided: a) a verdict that indicates successful prosecution (guilty verdict or probation before judgment) was included because it is assumed to facilitate further intervention and to communicate the message that the abusive behavior is illegal and unjustified; b) a suspended jail sentence is assumed to provide a threat of further loss of freedoms; c) probation monitoring is assumed not only to indicate the possibility of further sanctions for violating the probation order, but probation also introduces contact with a probation agent who may intervene in various ways such as talking things over with the offender and monitoring compliance with conditions of probation; d) an order to stay away from the victim may limit opportunity to re-offend and may reinforce the victim's legal right not to be abused; e) an order for domestic violence counseling may place further responsibility for change on the offender, indicating that the court believes he has a behavioral or psychological problem; and f) compliance with court ordered counseling may stimulate self-reflection and provide guidance with behavior change. Rather than postulating a more formal model of behavior change, these hypotheses reflect the implicit assumptions of the intervention system as revealed in discussions with prosecutors, probation agents, and domestic violence counselors. A secondary goal of the investigation was to explore whether incident and background variables predict recidivism, and whether involvement in the intervention system predicts recidivism independent of background factors. The archival nature of the investigation limited the range of variables which could be examined as well as the adequacy of measurement for background and incident factors. Sufficient data were available to explore recidivism associations for the offender's age, race, legal representation (private attorney, public defender, or no representation), frequency of past domestic violence incidents reported by the victim, severity of injuries, and time and day of the index offense. These variables were analyzed in an exploratory, hypothesis generating framework.

METHOD Sample The sample was obtained from the files of the Baltimore, Maryland State's Attorneys' Domestic Violence Unit, and contained cases from three police districts. Men who were charged with domestic violence-related offenses between January and August 1994 constituted the sample, with fol-

Community Intervention for Domestic Abusers

267

low-up data available through September 1995. Cases were excluded if their files were missing the victim interview or the statement of charges. Female perpetrators were also excluded (less than 1% of the case files examined). Cases that went to jury trial were excluded because their files had been transferred to a different court, and were not accessible to the researchers. The number of such cases could not be precisely estimated from the available data. The final sample consisted of 235 cases with adequate data on case disposition and follow-up. Cases ranged in age from 18 to 81 years, with a mean of 34.6 (SD = 10.6). The sample was 82% African American, 16% Caucasian, and 2% Asian American. Although no standard indicator of socioeconomic status was consistently available, a majority of the sample had low incomes, qualifying them for public defense, with 50% represented by public defenders, 23% represented by private attorneys, and 27% unrepresented. Precise statements about the representativeness of this sample are not possible with the limited demographic data available. The sample, however, is consistent with the urban population of Baltimore, which is majority black and of low SES. Higher SES offenders and the most severe offenses may be under represented in the sample due to the inability to include jury trial cases, which tend to involve greater legal expenses and more serious charges. Available data, unfortunately, did not allow for further estimation of sample biases that may limit the generality of conclusions. Background and Offense Variables Background and offense variables were coded from the State's Attorney's files. The extent of domestic violence history was coded from a structured victim interview conducted by a staff member of the State's Attorney's office. Research assistants used a 5-point scale (detailed in Table I) to code the frequency of prior domestic violence. The police (incident) report or arrest warrant was used to code victim injuries on a 3-point severity scale, and the day and time of the index offense. Data on the time of the offense were missing for a large number of subjects (19%), but this did not affect analysis of the primary research questions involving system intervention variables. Prosecution and Intervention Variables The State's Attorney's files were also used to obtain information about the participants' legal representation, the case disposition or outcome, probation, stay away orders (usually given as special conditions of probation),

268

Murphy, Musser, and Maton

Community Intervention for Domestic Abusers

269

270

Murphy, Musser, and Maton

suspended sentences, orders to drug and alcohol counseling, and orders to domestic violence counseling. Although of potential interest to the explanatory model under investigation, reliable data on the extent of incarceration (e.g., number of hours spent in jail on the initial charge) were not consistently available. Relatively few cases were subsequently incarcerated as a result of the court disposition, so this variable was not further analyzed. All cases were prosecuted by a special domestic violence unit operating with a pro-prosecution approach. The prosecutors made consistent efforts to bring cases to trial even without victim testimony. Domestic violence counseling was normally ordered as a condition of probation, and, therefore, generally required a verdict of guilty or probation before judgment. Unlike the counseling referral process in some jurisdictions, deferred prosecution was seen as an undesirable alternative from the prosecutor's standpoint because of the problems inherent in monitoring cases within a large urban criminal justice system and because probation monitoring was seen as an important concomitant of court ordered counseling. In line with this perspective, successful prosecution was therefore defined as a verdict of guilty or probation before judgment, which was generally a necessary condition for further involvement in the intervention system. Compliance with Orders for Domestic Violence Counseling A computer search at the private nonprofit agency offering counseling for domestic abusers was used to determine whether the individuals who were court-ordered to domestic violence counseling attended an intake session, and whether they completed the 22-session counseling program. Recidivism Measures The outcome measures involved recidivism for battery and related offenses as evident in criminal justice data. The entire criminal history in the State of Maryland up to the follow-up date of August 31,1995 was obtained for each participant. Charges for all violations during the follow-up period were recorded from the criminal history. Variations in the duration of follow-up interval, which ranged from approximately 12 to 18 months, may have introduced random variance into the analysis of recidivism, which would most likely mitigate against finding significant effects. Inspection of the data suggested that most of the individuals who generated new charges did so within the first 12 months after adjudication. The nature of court sanctions and interventions appeared to be quite consistent across the 6

Community Intervention for Domestic Abusers

271

month adjudication interval sampled, and therefore no further efforts were made to control for the duration or timing of the follow-up interval. At the time of the study, no offense codes or criminal justice data codes specific to domestic abuse were available in the automated criminal justice data on recidivism. Even if such codes had been available, however, the domestic nature of certain crimes can be quite difficult to discern. For example, charges such as trespassing or breaking and entering are sometimes related to an estranged partner situation. Clinical experience suggests that some offenders generate new assault or battery charges by attacking the original victim's family members, friends, or dating partners. Such offenses are related to the domestic abuse, but may not fit a precise definition of partner violence. These ambiguities were addressed by constructing two recidivism measures on the basis of charges received during the follow-up period. The primary recidivism measure (RECID1) used a narrow definition of offenses that are most likely to be domestic abuse related. Subjects were considered to be recidivists if they generated any new battery charge (the most common charge applied to incidents of domestic violence in the jurisdiction studied) or violation of an ex parte order (civil order of protection) during the follow-up period. Using this measure, 37 of the 235 participants (15.7%) were recidivists. A second recidivism measure (RECID2) was based on a broad range of charges that are sometimes received as a result of domestic abuse incidents. Most of these charges are violent in nature. Due to the imprecise nature of the available recidivism data, it is likely that some of the charges included in this broader recidivism measure were unrelated to domestic abuse. In the current sample, participants coded as recidivists on this variable had received the following charges: battery, violation of an ex parte order, assault, assault with intent to murder, rape, child abuse, malicious destruction of property, telephone abuse, illegal handgun possession, disorderly conduct, and murder. Other charges received during the follow-up period were considered less likely to result from domestic abuse situations, and were not used to categorize individuals as recidivists. These included drug possession or distribution, theft, burglary, and fraud. Using the broader recidivism measure, 60 of the 235 participants (25.5%) were recidivists.

RESULTS Background and Offense Variables: Associations with Recidivism Table I presents data on the prediction of recidivism from background variables and variables related to the initial offense. Subjects' age was sig-

272

Murphy, Musser, and Maton

nificantly and negatively correlated with both recidivism measures. A consistent downward trend in recidivism was observed across the age spans studied on the narrow recidivism measure (battery or ex parte violations), with the exception of the 46- to 50-year-old group, who had a relatively high recidivism rate. The broader recidivism measure appeared to follow a more curvilinear association with age. Those in their 30's had lower recidivism rates than those in their 20's and 40's. Given the relatively small sample size at each age reported, however, no firm conclusions can be drawn other than the general decline in recidivism with age indicated in the significant negative correlation. Participants' race was not significantly associated with recidivism. The type of representation in court, however, was significantly associated with the broad recidivism measure. Those represented by a public defender had the highest recidivism rates, while those represented by a private attorney had recidivism rates that were slightly lower than those who had no representation. The legal representation variable was included as a crude indicator of socioeconomic status, but legal representation may also be confounded with the seriousness of the charges and strength of evidence. In the current sample, however, use of a public defender was not significantly associated with the extent of domestic violence history reported in the victim interview nor with the severity of injuries coded from the statement of charges. In general, individuals who qualified for and used public legal defense had higher recidivism rates than individuals who retained private attorneys, which is consistent with the idea that those who have fewer economic resources would have higher recidivism. There were no significant associations between the time or day of the incident and the propensity to re-offend. These variables were included to be consistent with prior work on this topic (Steinman, 1988), but no specific hypotheses were made about their implication for recidivism. Surprisingly, indicators of the severity of the offense (victim injuries and medical attention as recorded from the police report) and the frequency of past domestic violence incidents (as reported by the victim) were also not significantly associated with recidivism, as assessed by point-biserial correlations. The lack of association between the frequency of prior domestic violence incidents and recidivism may in part reflect the effects of coordinated intervention. Individuals with more extensive histories of abuse, on average, received more sanctions from the court, and the number of court sanctions in turn was associated with lower recidivism (described below).

Community Intervention for Domestic Abusers

273

Prosecution and Intervention Variables: Associations with Recidivism Table II displays data on the prediction of recidivism from prosecution and post-prosecution variables. Case outcome was not significantly associated with recidivism, although the findings were in the expected direction with individuals who received a guilty verdict or probation before judgment (successful outcome for the prosecution) having somewhat lower recidivism rates than individuals whose cases were dropped or deferred from prosecution. Similarly, those who were ordered to probation had rates that were 37% lower on the narrow recidivism measure and 33% lower on the broad recidivism measure than their no-probation counterparts, but these differences were not statistically significant. Suspended sentences, ex parte orders, and orders to drug and alcohol counseling were not associated with lower recidivism. A statistically significant association between orders to domestic violence counseling and the narrow measure of recidivism was found. Individuals who received a court order to domestic violence counseling were 56% less likely than other offenders to generate a new charge for battery or ex parte violation. On the broad measure of recidivism, those ordered to counseling had recidivism rates that were 36% lower than other offenders, but this result was not statistically significant. Two indicators of compliance with court orders, attendance at a counseling intake and completion of the 22-session counseling program, were associated with low recidivism, but these effects were not statistically significant. Cumulative Effects of Prosecution, Probation, and Counseling on Recidivism The underlying assumption of most community intervention programs is that coordinated interventions will produce more effective results than will isolated and unsystematic interventions. This assumption was tested by examining the cumulative effects of successful prosecution, probation monitoring, and domestic violence counseling, which are three crucial aspects of intervention from the point of prosecution onward. Table III displays very promising initial findings on the cumulative effects of coordinated interventions. Small, cumulative reductions in recidivism were associated with greater involvement in the intervention system. The recidivism rates for those who were successfully prosecuted (N = 106), as indicated by a court verdict of guilty or probation before judgment, were slightly lower than the recidivism rates in the overall sample of 235 cases. As a further illustration

274

Murphy, Musser, and Maton

Community Intervention for Domestic Abusers

275

276

Murphy, Musser, and Maton

of this point, those who were dropped or deferred from prosecution or found not guilty generated recidivism rates of 19% and 29% on the narrow and broad measures, respectively, when compared to rates of 13% and 23% for those who were successfully prosecuted. When probation was added to successful prosecution (N = 101), the recidivism rates decline to 12% and 21%, respectively, and for those who received orders to domestic violence counseling along with successful prosecution and probation (N = 56), the recidivism rates were lower still (9% and 18%). The small group of individuals (N = 9) who were successfully prosecuted, on probation, ordered to counseling, and completed counseling had no recidivist charges (see Table III). It is important to note also that self-selection biases may have influenced results for the counseling compliance variables if individuals who are more motivated to stay out of trouble are more likely to comply with the court order for counseling. The statistical significance of these cumulative intervention effects was tested by developing a measure of system involvement. Each participant was assigned a score from 0 to 5 based on how many of the intervention components in Table III the individual experienced. This 0 to 5 scale was then correlated with the two recidivism measures, yielding significant pointbiserial correlations of modest magnitude, r = -.12, p < .05 and r = -.12, p < .05 for the narrow and broad recidivism measures, respectively. As

Table III. Cumulative Effects of Prosecution, Probation, and Counseling on Recidivism

N
Overall sample Not successfully prosecuted0 Cumulative interventions 1. Guilty verdict or PBJd 2. Guilty/PBJ + probation 3. All of the above + ordered to D. V counseling 4. All of the above + attended counseling intake 5. All of the above + completed counseling program

RECID1" 15.7% 19.2% 13.2% 11.9%

RECID2fc 25.5% 29.2% 22.6% 20.8% 17.9%

235 120 106 101

56 19
9

8.9% 5.3%

5.3%

0%

0%

"RECID1 = battery or ex pane violation. *RECID2 = battery, ex parte violation, assault, assault with intent to murder, rape, child abuse, malicious destruction of property, telephone abuse, illegal handgun possession, disorderly conduct or murder. Significance tests are described in the text. Tound not guilty, dropped, or deferred from prosecution. d PBJ = Probation Before Judgment.

Community Intervention for Domestic Abusers

277

predicted, those who were more involved with the intervention system had lower recidivism rates.

Statistical Control of Offender Severity and Background Variables One possible explanation for these findings is that less severe cases, on average, may be more involved with the intervention system. For example, this could happen if judges apply interventions more aggressively in less severe cases, perhaps under an implicit assumption that these individuals will be more responsive to intervention. The data, however, suggest the opposite. More severe cases were more likely to receive court sanctions and counseling orders. Individuals with a more extensive history of domestic violence, as coded from the victim interview, were significantly more likely to be successfully prosecuted (guilty verdict or probation before judgment vs. other case outcomes; pt. biserial r = .19, p < .01), more likely to receive probation (pt. biserial r = .17, p < .01), and more likely to receive an order to domestic violence counseling (pt. biserial r = .19, p < .01). Those with more extensive domestic violence histories were also more likely to complete an intake for domestic violence counseling (pt. biserial r = .13, p < .05). Although these correlations are of modest magnitude, they suggest that offender severity is not a likely explanation for the effects of cumulative intervention system involvement on recidivism. Multiple regressions were conducted to examine the unique contribution of intervention system involvement in predicting recidivism after statistically controlling for case severity and background variables. Four variables were entered on the first step of the regression equations: severity of injury as coded from the statement of charges, extent of domestic violence history as coded from the victim interview, age, and legal representation (public defense). These latter two background variables were chosen because they had significant associations with at least one of the two recidivism variables in the initial bivariate analyses. As displayed in Table IV, with statistical controls for offender severity and background factors, the degree of intervention system involvement still contributed a significant degree of unique variance to the prediction of both recidivism variables. The effects of intervention system involvement were relatively independent of offender severity and background variables. Cumulative intervention was significantly associated with lower recidivism on both measures despite the fact that individuals with more extensive histories of domestic violence tended to be more involved in the intervention system.

278

Murphy, Musser, and Maton Table FV Unique Prediction of Recidivism from Intervention System Involvement Controlling Background Variables and Case Severity Predicting Battery or Ex Parte Violation (RECID1)

Variable Step 1

P
-.14

/ -2.05" 1.24 ns .73ns 1.14 ns

R2 Change

F Change

Age
Legal rep. Viol. hist. Injury sev. Step 2 System Involvement Step 1

.08 .05 .08


-.16

.194
-2.30

.038

2.16 ns
5.29

.246 .023 Predicting the Broad Recidivism Measure (RECID2) -.11


-1.60 ns 2.17" -.30 ns .59ns

Age
Legal rep. Viol. hist. Injury sev. Step 2 System Involvement

.15
-.02

.04
-.16

.197
-2.30*

.039

2.21 ns

.239
"p < .05.

.019

4.32"

DISCUSSION The results provide a basis for cautious optimism regarding the effectiveness of coordinated community interventions for male domestic violence perpetrators. Court orders for domestic violence counseling were associated with significantly lower probability of battery charges or an ex parte violation during a 12- to 18-month period after prosecution on a domestic abuse-related charge. This finding replicates the results of a community intervention study in Minneapolis, Minnesota, which also indicated that court orders for domestic violence counseling were associated with significantly lower recidivism (Syers and Edleson, 1992). Most importantly, the current results provide evidence that coordinated interventions may have a cumulative effect on recidivism risk. The combined effects of prosecution, probation, and court-ordered counseling were associated with significant reductions in recidivism. Unfortunately, however, only about 4% of cases actually experienced the highest level of intervention system involvement, which included a verdict of guilty or pro-

Community Intervention for Domestic Abusers

279

bation before judgment, probation, a court order to domestic violence counseling, and completion of counseling. Contrary to initial expectations, the extent of prior domestic violence history, as reported by victims in an interview with the state's attorney, did not predict recidivism. The extent of prior domestic violence, however, was significantly associated with intervention system involvement. Individuals with more extensive histories of domestic violence were more likely to be successfully prosecuted, to receive probation, and to be ordered to domestic violence counseling. These findings help to rule out the possibility that cumulative intervention effects were a spurious result of case severity. In brief, more serious offenders tended to experience more intervention; yet, greater system intervention was associated with lower recidivism. Unfortunately, however, very few individuals received the highest level of intervention. These findings suggest that enhanced efforts to coordinate and monitor intervention compliance for domestic abusers may reduce criminal recidivism. The significantly lower level of recidivism associated with a court order for domestic violence counseling was somewhat puzzling given the low compliance with court-ordered counseling. In the current sample, less than 20% of those who received the court-order actually completed the 22-session counseling program. Recidivism rates for this self-selected group were very low, indicating that counseling may have helped them to avoid further trouble with the law. Low rates of official recidivism (averaging less than 10%) have been reported among those who complete domestic violence counseling (Rosenfeld, 1992), although controlled empirical studies have yielded only mixed evidence of counseling efficacy with court mandated abusers (Dunford, 1997; Dutton, 1986; Harrell, 1991; Palmer et al, 1992). It is also possible that the court order to counseling itself motivates behavior change through the implication that the offender has a problem and is being held responsible to change his behavior. Judges' decision processes also influence the observed associations, as assignment to counseling is not random. Prior research has provided evidence that domestic abuse perpetrators are more likely to be ordered to counseling if they are married to, or cohabiting with, the victim than if the relationship is more casual (Chen et al., 1989; Harrell, 1991). In the current study, those with more extensive histories of domestic violence reported by the victim were more likely to receive an order to counseling, but other factors, such as age and legal representation, were not associated with counseling orders. Thus, the effects of counseling orders may actually be underestimated in the current study, because those in more committed relationships with more extensive histories of violence may be at increased risk for re-offense.

280

Murphy, Musser, and Maton

Two offender characteristics also predicted recidivism. Those who were represented by public defenders had higher recidivism rates on a broad measure of offenses often related to domestic abuse, but they did not have significantly higher recidivism on a narrow measure that included only battery charges or ex parte violation. Younger offenders, in general, had higher recidivism for battery or ex parte violations. There was some indication, however, of a curvilinear age association with the broader recidivism measure such that offenders in their 20's and 40's had relatively higher recidivism than that observed for offenders in their 30's. This curvilinear pattern should be interpreted cautiously given the relatively small number of individuals in each age category. This result remains puzzling, as prior survey studies consistently documented a linear association between age and domestic violence in the U.S. population, with prevalence rates of physical marital aggression for individuals in their 20's about twice as high as the rates for individuals in their 30's and three times as high as the rates for individuals in their 40's (Suitor et al., 1990). The recidivism risk associated with public legal defense most likely reflects socioeconomic factors. No adequate indicator of SES was consistently available in case files, so this issue could not be further explored. Note also, however, that individuals who had no legal representation had lower recidivism risk than those represented by public defenders. The seriousness of charges or strength of evidence was probably confounded with legal representation such that offenders facing less serious may have been mpre likely to forego legal representation. Similarly, individuals with criminal records for activities unrelated to domestic abuse, or those who were already on parole or probation for other offenses may have been more likely to use public defender services, which could explain their higher risk for recidivism through a broad range of charges. The present study reflects both the challenges inherent in modeling complex system interventions and the nascent stage of this type of research on domestic violence. While the results suggest that multi-system intervention lowers recidivism, a number of important limitations must be considered. First, the recidivist offenses were legal charges, some of which were probably not domestic abuse related, particularly on the broad recidivism measure. The significantly lower recidivism associated with cumulative system intervention may represent a specific deterrent effect on domestic abuse crimes, a general deterrent effect on criminal behavior, or some combination of the two. In either case, however, reduced criminal recidivism associated with the relatively low-cost interventions of successful prosecution, probation, and domestic violence counseling represents an important finding worthy of further investigation.

Community Intervention for Domestic Abusers

281

The use of official criminal justice data to measure recidivism is an important limitation in its own right. A relatively small proportion of domestic violence incidents result in official charges. In studies of court-ordered counseling, the average recidivism rates from criminal justice data are about 7%, whereas the average rates gathered from victim interviews are about 36% (Rosenfeld, 1992). Population surveys suggest that only about 3% of minor violence incidents (e.g., pushing, grabbing, or slapping) and about 14% of severe violence incidents (e.g., hitting with a fist or objects, or beating up) result in calls to the police (Kantor & Straus, 1990). Further, these figures are consistent across urban, rural, and suburban settings. Local data from Baltimore indicate that arrests are made in about 24% of the cases in which the police issue reports for domestic violence (Baltimore Domestic Violence Coordinating Committee Report, 1992). Obviously, a large number of incidents go undetected hi criminal justice data. Alternative measures of recidivism, especially victim interviews, may yield different associations with intervention system variables. For example, the intervention system may reduce a victim's willingness to involve the police, which could yield reductions in criminal justice recidivism without actual reductions in abusive behavior. Conversely, however, important enhancements in victim safety, such as reduced injury risk, reduced lethality risk, lower victim fear levels, or greater victim empowerment, may be associated with intervention system involvement yet may not be readily apparent in criminal recidivism data. Other important concerns involve the correlational nature of the findings, which cannot provide definitive support for a causal explanation. Variables related to attendance and completion of counseling have an inherent subject selection bias reflecting motivation to change. Judges' decision processes also exert uncertain effects on the outcomes of this type of study. The current data suggest that judges are more prone to refer cases for counseling who have more extensive histories of abuse. Prior studies have found that judges are more prone to refer individuals who maintain a cohabiting or marital relationship with the victim, and who may therefore have increased risk for reoffense (Chen et al., 1989; Harrell, 1991). The most likely effect of these judges' decision processes would be to reduce the observed impact of system interventions, which are more likely to be used with higher risk offenders. When feasible and ethical, future investigations may use experimental designs, for example random assignment to court-ordered counseling versus no counseling, which provide the best method to isolate the effects of specific intervention components (e.g., Palmer et al., 1992). Unfortunately, however, for many variables, such as successful prosecution, random as-

282

Murphy, Musser, and Maton

signment is not possible. Sophisticated correlational methods are available to test theoretical models of the combined and isolated effects of a complex intervention system, but these methods require very large samples, highly reliable measures (with multiple indicators of key constructs), and well-formulated explanatory models, all of which are more appropriate after initial studies have demonstrated intervention effects. Given these caveats, the current findings should be seen as initial and exploratory in nature. The results provide important suggestions for future research on coordinated community interventions for domestic violence. Future investigations will require very large samples for several reasons. Official justice system data yield relatively low recidivism rates, categorical outcome variables require the use of low power statistical techniques, the effects of specific components of community intervention appear to be modest in magnitude, and certain legal sanctions or combinations of events are relatively infrequent, yet important to study. As a simple illustration of this point, using the broad measure of recidivism in the current study, individuals ordered to probation had a 20% recidivism rate and individuals not ordered to probation had a 30% recidivism rate. This one third difference in recidivism rate looks promising. Yet, a sample size of over 600 participants would be needed to have an 80% chance of detecting this probation effect as significant (Cohen, 1988). It may also be important to oversample cases who receive specific interventions, such as special probation monitoring or a particular type of domestic violence counseling. Obtaining more comprehensive measures of recidivism that include victim reports would also greatly strengthen the conclusions of this type of research. Finally, the effects of pre-prosecution interventions, in particular arrest and initial incarceration, should be modeled as additional components of coordinated intervention, as these are very common practices that facilitate further interventions. In conclusion, the current study provides qualified support for the effects of a coordinated community response to domestic abuse that involves effective prosecution, probation, and court-ordered counseling. Court orders to counseling were associated with lower recidivism risk on one of the two recidivism measures. As originally hypothesized, specific intervention system components appeared to exert small effects on recidivism risk. More substantial, cumulative reductions in recidivism were observed from successful prosecution, probation, and court-ordered counseling, and these reductions in recidivism were not explained by case severity or background variables. The results offer encouraging data on coordinated interventions, and suggest important areas for further research.

Community Intervention for Domestic Abusers

283

ACKNOWLEDGMENTS

The authors wish to thank the members of the Mayor of Baltimore's Domestic Violence Coordinating Council (DVCC) for their input into the design and completion of this project, and to acknowledge the invaluable efforts of Deborah Bright, DVCC coordinator, Roni Young of the State's Attorney's Domestic Violence Unit, Janet Moses of the Criminal Justice Information Systems, and Alice Jordan of the House of Ruth flatterers' Program. This project was supported by Grant Number 93-DD-CX-0049 awarded by the Bureau of Justice Assistance, Office of Justice Programs, United States Department of Justice, to the Baltimore DVCC. Points of view or opinions in this document are those of the authors and do not represent the official position or policies of the United States Department of Justice. Portions of this work were presented at the Annual Convention of the American Psychological Association, Toronto, Ontario, Canada, August 1996.

REFERENCES
Berk, R. A., Campbell, A., KJap, R., and Western, B. (1992). The differential deterrent effects of an arrest in incidents of domestic violence: A Bayesian analysis of three randomized field experiments. Am. Sociological Rev. 57: 698-708. Baltimore Domestic Violence Coordinating Committee (1992). Baltimore City Domestic Violence Summit Report. Chen, H., Bersani, C, Myers, S. C., and Denton, R. (1989). Evaluating the effectiveness of a court sponsored abuser treatment program. J. Fam. Viol. 4: 309-332. Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences. Hillsdale, NJ: Lawrence

Erlbaum. Dunford, F. W. (1997, June). The research design and preliminary outcome findings of the San
Diego Navy experiment, Presented at the Fifth International Family Violence Conference, Durham, NH. Dunford, F. W, Huizinga, D., and Elliot, D. S. (1990). The role of arrest in domestic assault: The Omaha Police Experiment. Criminology 28: 183-206. Dutton, D. G. (1986). The outcome of court-mandated treatment for wife assault: A

quasi-experimental evaluation. Viol Viet. 3: 163-175. Dutton, D. G. (1995). The Domestic Assault of Women: Psychological and Criminal Justice
Perspectives, Vancouver, British Columbia, UBC Press. Edleson, J. L. (1991). Coordinated community responses to woman battering. In M. Steinman

(Ed.), Woman Battering: Policy Responses, Anderson, Cincinnati, OH, pp. 203-219. Harrell, A. (1991). Evaluation of Court-Ordered Treatment for Domestic Violence Offenders:
Final Report, The Urban Institute, Washington, DC. Kantor, G. K., and Straus, M. A. (1990). Response of victims and the police to assaults on wives. In Straus, M. A., and Gelles, R. J. (eds.), Physical Violence in American Families, Transaction, New Brunswick, NJ, pp. 473-488. Palmer, S. E., Brown, R. A., and Barrera, M. E. (1992). Group treatment program for abusive husbands: Long-term evaluation. Am. J. Orthopsychiatry 62: 276-283. Pence, E. (1983). The Duluth Domestic Abuse Intervention Project. Hamline Law Rev. 6: 247-275.

284

Murphy, Musser, and Maton

Rosenfeld, B. D. (1992). Court-ordered treatment of spouse abuse. Clin. Psychol. Rev. 12: 205-226. Sherman, L. W, and Berk, R. A. (1984). The specific deterrent effects of arrest for domestic assault. Am. Sociological Rev. 49: 261-272. Steinman, M. (1988). Evaluating a system-wide response to domestic abuse: Some initial findings. J. Contemp. Crim. Just. 4: 172-186. Straus, M. A., and Gelles R. J. (eds.) (1990). Physical Violence in American Families. Transaction, New Brunswick, NJ. Suitor, J. J., Pillemer, K., and Straus, M. A. (1990). Marital violence in life course perspective. In Straus, M. A., and Gelles, R. J. (eds.), Physical violence in American families, Transaction, New Brunswick, NJ, pp. 305-317. Syers, M. A., and Edleson, J. L. (1992). The combined effects of coordinated criminal justice intervention in woman abuse. J. Interpers. Viol. 7: 490-502.

Вам также может понравиться