Manhattan DV Court 04 by Administrator

CJA

NEW YORK CITY CRIMINAL JUSTICE AGENCY

Jerome E. McElroy Executive Director

THE IMPACT OF MANHATTAN’S SPECIALIZED DOMESTIC VIOLENCE COURT

Richard R. Peterson, Ph.D. Project Director and Director, Research Department

FINAL REPORT November 2004

52 Duane Street, New York, NY 10007

(646) 213-2500

THE IMPACT OF MANHATTAN’S SPECIALIZED DOMESTIC VIOLENCE COURT

Richard R. Peterson, Ph.D. Project Director and Director, Research Department

Research Assistance: Raymond Caligiure Graphics and Production Specialist Elyse J. Revere Junior Research Analyst Elizabeth Walton Senior Research Assistant Systems Programming: Barbara Geller Diaz Associate Director, Information Systems Wayne Nehwadowich Senior Programmer/Analyst Administrative Support: Bernice Linen-Reed Administrative Associate

November 2004

This report can be downloaded from www.nycja.org\research\research.htm

ACKNOWLEDGEMENTS

This report could not have been completed without the assistance of colleagues at CJA, Judges, members of the District Attorneysâ&#x20AC;&#x2122; offices and others. The author acknowledges colleagues at CJA who provided advice, information and editorial suggestions: Jerome E. McElroy, Executive Director of CJA, as well as Mari Curbelo, Barbara Geller Diaz, Marian Gewirtz, Peter Kiers, Mary T. Phillips, Frank Sergi, Qudsia Siddiqi and Freda F. Solomon. The author also thanks those who provided research and other assistance for this study: Justin P. Bernstein, Raymond Caligiure, Elyse J. Revere and Elizabeth Walton. Barbara Geller Diaz did the programming to extract re-arrest data from the CJA database and Wayne Nehwadowich extracted the third quarter 1998 and first quarter 2001 case-level data used in this study. The author would also like to thank Liberty Aldrich, Esq., Deirdre Bialo-Padin, Esq., Abena Darkeh, Esq., Hon. Matthew J. Dâ&#x20AC;&#x2122;Emic, Prof. Jo Dixon, Peter Glick, Esq., Darlene Haywood, Scott Kessler, Esq., Karen Kleinberg, Esq., Elisa Koenderman, Esq., Melissa Labriola, Sharon Lastique, Hon. John M. Leventhal, Wanda Lucibello, Esq., Audrey Moore, Esq., Hon. Esther M. Morgenstern, Joseph Muroff, Esq., Nora Puffett, Michael Rempel, Terri Roman, C.S.W., Deborah Tuerkheimer, Esq. and Amanda Voytek for their assistance with this and other projects. Finally, thanks to the New York State Division of Criminal Justice Services (DCJS) and the New York City Police Department (NYPD) for providing supplemental data. The methodology, findings and conclusions of the study, as well as any errors, omissions and misinterpretations are the responsibility of the author.

TABLE OF CONTENTS

Introduction ...................................................................................................... 1 A. Specialized Domestic Violence Courts in New York City .............................. 1 B. Review of Research on Specialized Domestic Violence Courts ................... 2 C. Research Questions ..................................................................................... 9

II.

Methodology................................................................................................... 11 A. Overview of the CJA Database and the Third Quarter 1998 and First Quarter 2001 Datasets................................................................................ 11 B. Identifying Domestic Violence Cases.......................................................... 12 C. Identifying Re-arrests for Domestic Violence Offenses............................... 15 D. Selection of the Crimes Against Persons and Property Subsample ........... 16 E. Case-based vs. Defendant-based Data Files ............................................. 18 F. Plan of Analysis .......................................................................................... 19

III.

Overview of Differences between 1998 and 2001 in the Processing of Domestic Violence Cases in Manhattan....................................................... 21 A. Case Dispositions, Sentence Outcomes, Length of Jail Sentences, Re-arrest Rates .......................................................................................... 21 B. Defendant and Case Characteristics .......................................................... 28 C. Summary and Discussion of Findings......................................................... 37

IV.

Models Predicting Likelihood of Conviction................................................ 41 A. Comparison of Models for Third Quarter 1998 and First Quarter 2001....... 41 B. Summary and Discussion of Findings......................................................... 44

Models Predicting Likelihood of Incarceration............................................ 47 A. Comparison of Models for Third Quarter 1998 and First Quarter 2001....... 47 B. Summary and Discussion of Findings......................................................... 50

VI.

Models Predicting Length of Jail Sentence ................................................. 53 A. Comparison of Models for Third Quarter 1998 and First Quarter 2001....... 53 B. Summary and Discussion of Findings......................................................... 56

Table of Contents Continues on Next Page

TABLE OF CONTENTS, CONTINUED

VII. Models Predicting Likelihood of Re-Arrest .................................................. 57 A. Comparison of Models for Third Quarter 1998 and First Quarter 2001....... 57 B. Summary and Discussion of Findings......................................................... 61 VIII. Conclusion ...................................................................................................... 63 A. Major Findings ............................................................................................ 63 B. Discussion .................................................................................................. 65 1. Faster Processing of Domestic Violence Cases .................................... 65 2. Improved Identification of Domestic Violence Cases............................. 67 3. Improved Monitoring of Domestic Violence Defendants ........................ 68 C. Conclusions ................................................................................................69 IX.

References ..................................................................................................... 71

Appendix A: Statistical Methods.......................................................................... 75 1. Logistic Regression Analysis ...................................................................... 75 2. Interaction Effects....................................................................................... 78 3. Correcting for Selection Bias ...................................................................... 78 A. Models of Incarceration ......................................................................... 78 B. Models of Length of Jail Sentence......................................................... 79 4. Linear Regression Analysis ........................................................................ 80 Appendix B: Coding of Variables for Regression Models .................................. 83

LIST OF FIGURES

Figure 3-1

Case Dispositions in Criminal Court ....................................................... 23

Figure 3-2

Sentence Outcomes in Criminal Court..................................................... 24

Figure 3-3

Length of Jail Sentence in Criminal Court................................................ 26

Figure 3-4

Post-Disposition Re-Arrest Rates ........................................................... 27

iii

LIST OF TABLES

Table 3-1 Arraignment Charge Penal Law Article by Year for Domestic Violence Cases in Manhattan .................................................................................................28 Table 3-2 Defendant-Victim Relationship by Year for Domestic Violence Cases in Manhattan .....................................................................................................29 Table 3-3 Defendantâ&#x20AC;&#x2122;s Demographic Characteristics by Year for Domestic Violence Cases in Manhattan ......................................................................................30 Table 3-4 Defendantâ&#x20AC;&#x2122;s Criminal History by Year for Domestic Violence Cases in Manhattan .................................................................................................... 31 Table 3-5 Arrest and Arraignment Charge Characteristics by Year for Domestic Violence Cases in Manhattan ...................................................................... 32 Table 3-6 Case Processing Characteristics by Year for Domestic Violence Cases in Manhattan .................................................................................................... 34 Table 4-1 Logistic Regression Model Predicting Likelihood of Conviction for Domestic Violence Cases in Manhattan ...................................................................... 42 Table 5-1 Logistic Regression Model Predicting Likelihood of Incarceration for Convicted Domestic Violence Defendants in Manhattan.............................. 48 Table 6-1 Regression Model Predicting Length of Sentence for Convicted Domestic Violence Defendants Sentenced to Jail in Manhattan .................................. 54 Table 7-1 Logistic Regression Model Predicting Likelihood of Re-arrest for a Domestic Violence Offense: Domestic Violence Offenders in Manhattan ................... 58

THE IMPACT OF MANHATTAN’S SPECIALIZED DOMESTIC VIOLENCE COURT

I. INTRODUCTION The criminal justice system’s response to domestic violence has changed dramatically over the last decade in many jurisdictions in the United States. Domestic violence incidents are now being taken more seriously through more frequent arrests, vigorous prosecution and harsher sentencing laws. Federal legislation, notably the Violence Against Women Act (VAWA) of 1994, has encouraged new efforts to combat domestic violence. Under this Act, the Department of Justice has provided funding and technical assistance to state and local law enforcement agencies. The national trends are reflected in changes in New York State and New York City. Recent state legislation includes mandatory arrest requirements for certain family offenses, enhanced penalties for violating an order of protection, a “primary physical aggressor” statute, and anti-stalking legislation. In conjunction with the new legislation, New York City has promoted changes in the criminal justice response to domestic violence. Specialized police units, prosecution bureaus, and courts have been created to focus more attention on domestic violence cases, and to develop a consistent approach to handling these cases. New York City is now in the forefront of efforts to combat domestic violence. These changes in New York City’s approach to domestic violence raise significant research questions. How are domestic violence cases processed in the courts? Are there differences among the boroughs? How often are offenders re-arrested on domestic violence charges? What is the impact of specialized domestic violence courts? To address these questions, the New York City Criminal Justice Agency (CJA) developed a research agenda on domestic violence. The current report is one of a series of CJA reports on domestic violence. Previous reports have compared the case processing of domestic violence (DV) and non-domestic violence (Non-DV) cases in New York City’s Criminal Courts (Peterson 2001), examined borough differences in the processing of domestic violence cases (Peterson 2002), and examined rates of re-arrest for defendants in DV cases (Peterson 2003a). A summary report synthesized the research findings and policy implications (Peterson 2003b), and a “Research Brief” updated the findings with data from 2001 (Peterson 2003c). The current report examines the impact of the specialized misdemeanor domestic violence court established in Manhattan in 2000. A. Specialized Domestic Violence Courts in New York City Specialized DV courts have become the central focus of the courts’ efforts to combat domestic violence in New York City and throughout New York State. All five boroughs of New York City now have specialized DV courts to hear misdemeanor

2 cases. Three of the five boroughs (Brooklyn, Queens and the Bronx) also have specialized DV courts to hear cases sustained as felonies. The most recent innovation is the development of Integrated Domestic Violence (IDV) courts, which have been established in Queens and the Bronx. IDV courts are designed to integrate the processing of multiple cases that would previously have been processed independently in different courts (Mazur and Aldrich 2002). Under the principle of “one family, one judge,” an IDV court brings together misdemeanor DV cases, felony DV cases, DV cases in Family Court and matrimonial, divorce, custody and visitation cases involving the same family. Plans are currently under way to establish IDV courts in the remaining three boroughs of New York City, as well as in every judicial district throughout New York State. Although IDV’s are the latest innovation in the courts’ approach to DV cases, they will not replace the specialized misdemeanor and felony DV courts. IDV’s will only handle cases for families that have simultaneous cases in two or more courts. The majority of DV cases, which involve defendants whose only case is a misdemeanor DV case, will continue to be processed in the specialized misdemeanor DV courts. The current report focuses attention on the operation of these specialized misdemeanor DV courts, which will continue to play an important role. Specifically, the current report examines the impact of a specialized misdemeanor DV court on case outcomes and re-arrests. This report builds on our earlier report examining borough differences in the processing of DV cases in the third quarter of 1998 (Peterson 2002). In that report, we compared DV cases processed in Brooklyn and the Bronx, which had specialized DV courts, to DV cases processed in Manhattan, which did not have a specialized DV court. We found that boroughs using specialized domestic violence courts were more likely to monitor convicted DV offenders through batterer intervention programs, whereas convicted DV offenders processed in mixed-docket courts were more likely to be sentenced to jail. This difference reflected the emphasis of the specialized courts on monitoring DV offenders as a means of controlling their behavior. The current study examines the impact of the specialized DV court established in Manhattan in June 2000. It compares DV cases before and after the establishment of the DV court in Manhattan using data from 1998 and 2001. Whereas we previously compared boroughs with and without specialized DV courts, the current study examines one borough before and after the establishment of a DV court. This provides a different method of assessing the impact of a specialized DV court. Throughout this study, we will discuss our “before” and “after” findings in the context of our previous findings from the cross-borough study. To provide background for our study, we begin by reviewing the literature on specialized DV courts. B. Review of Research on Specialized Domestic Violence Courts Many jurisdictions established specialized DV courts to respond to problems encountered in processing DV cases in courts with mixed dockets. Three problems

3 have been identified. First, DV cases are treated more leniently than Non-DV cases when they are processed in mixed-docket courts (Fritzler and Simon 2000, Miethe 1987). This lenient treatment may be the result of several factors. Judges, prosecutors and others may hold stereotypes that DV offenses are less serious than other offenses, or that the victim provoked the offender. DV cases may be seen as having weaker evidence than Non-DV cases; specifically, victims may be less cooperative in DV cases. DV cases may also be viewed as more complex than other cases, requiring resources (such as access to batterer intervention programs) that courts with mixed dockets may lack. Second, DV cases have unique characteristics not associated with similar NonDV cases (Fritzler and Simon 2000, Newmark et al. 2001, Steketee et al. 2000). In DV cases, the victim and defendant often have emotional and economic ties that continue during the processing of the case and after case disposition. Under these circumstances, victims are often more concerned about their own safety and less concerned with punishing the defendant. In DV cases, the victim is in greater danger of facing renewed violence. The defendant has greater access to the victim, and greater motivation to intimidate the victim. Furthermore, domestic violence, more than other types of violence, often occurs in a private location and is therefore more difficult to detect and to prevent. These unique characteristics of DV cases are often ignored when cases are heard in mixed-docket courts. DV cases therefore appear to be weaker and more problematic than Non-DV cases. Third, some have argued that mixed-docket courts are not sufficiently focused on victim safety (Fritzler and Simon 2000). In a mixed-docket setting, the ongoing risks faced by the victim may not receive serious attention, since these risks are not typical of the majority of cases on the docket. As a result, the court may not routinely issue orders of protection in DV cases and may not appropriately warn defendants to refrain from intimidating the victims. The traditional adversarial approach may anger the defendant while failing to provide protection for the victim. Furthermore, in mixeddocket courts the reasons for victim non-cooperation and the need for non-traditional types of evidence may not be addressed adequately. Specialized courts were designed to take a different approach to DV cases. In New York, for example, the specialized DV courts have been part of a broader effort to introduce “problem-solving courts … [which] attempt to reach beyond the immediate dispute to the underlying issue, and then to involve community agencies and others in resolving it …” (Kaye 2001, p. 4). By addressing the underlying problem, these courts seek to reduce “revolving door” justice, where defendants return to the courts repeatedly for committing similar offenses (Berman and Feinblatt 2001). While no one philosophical approach underlies the problem-solving approach, specialized DV courts have been guided by the principles of therapeutic jurisprudence, preventive law and restorative justice (Fritzler and Simon 2000). Therapeutic jurisprudence encourages positive therapeutic consequences for victims through coordination of the efforts of the criminal justice system. Preventive law focuses attention on preventing future violence

4 through judicial monitoring of the defendant. Restorative justice focuses on providing assistance to victims as well as rehabilitation for offenders. Drawing on these general principles, specialized domestic violence courts were established to achieve three goals: increase defendant accountability, promote victim safety, and coordinate the activities of criminal justice agencies that respond to domestic violence (Kaye and Knipps 2000). These goals are explicitly aimed at recognizing the special characteristics of domestic violence cases. Because the defendant’s relationship with the victim poses a risk of future violence against the same victim, the specialized DV courts monitor defendants’ behavior closely for any evidence of further violence. To enhance victim safety, these courts provide victims with links to social services and alternative housing. To encourage consistency in the approaches of police, DA’s, probation, corrections and the courts, the specialized courts work to coordinate the institutional responses to domestic violence. Among the key features of specialized DV courts are: assigning cases to a specialized calendar, screening for related prior cases, providing advocacy and services to victims, providing treatment programs for defendants, coordinating efforts with partner agencies, providing specialized training to court personnel, improving case management through information technology, and monitoring the defendant’s behavior (Buzawa et al. 1999, MacLeod and Weber 2000, Mazur and Aldrich 2002, Newmark et al. 2001, Steketee et al. 2000). Specialized DV courts usually seek to require defendants to complete a batterer intervention program. Although there is mixed evidence on the effectiveness of batterer intervention programs (Chalk and King 1998, Davis and Taylor 1999, Watson 2000), the courts use them as a means of monitoring defendants (Casey and Rottman 2003, Mazur and Aldrich 2002, Puffett and Gavin 2004, Tsai 2000). Jail sentences are usually imposed on chronic offenders or those who inflict serious injuries. Furthermore, victims often prefer that the defendant receive treatment rather than a jail sentence, and it may be easier to obtain cooperation from the victim if the sentence does not include jail time. It is also often easier to reach a plea agreement with the defendant if the sentence does not include jail time. Early descriptive research has found that the unique features of specialized DV courts change the way cases are processed (Casey and Rottman 2003, MacLeod and Weber 2000, Newmark et al. 2001). This research has shown that victims receive more services and have more information when cases are heard in specialized DV courts than they do when the cases are heard in mixed-docket courts. Court personnel develop a better understanding of the unique features of DV cases, and are more responsive to victims. Because only DV cases are heard, these cases are handled more consistently and procedures to enhance victim safety are used routinely. The frequent monitoring of defendants improves enforcement of the conditions of release and sentencing (Puffett and Gavin 2004). In addition to descriptive studies, several studies of the impact of specialized DV courts have also been completed. Newmark et al. (2001) studied the first specialized DV court established in New York City: a Brooklyn felony DV Court, which was

5 established in June 1996. Based on a comparison of felony DV cases processed before and after the establishment of this court, the study concluded that the specialized court resulted in several changes in the way DV cases were processed. The District Attorney’s (DA’s) office became more aggressive, charging defendants with felonies who previously would have been charged with misdemeanors. All victims were assigned advocates by the DA’s office in the new DV court, compared to only 55% of victims who were assigned advocates before the court was established. Defendants were more likely to be released pending a disposition, but virtually all released defendants were required to attend a batterer intervention program or substance abuse treatment program. Case processing time increased. Multivariate analyses indicated that the new DV court did not significantly increase the conviction rate (about 90%) or change the sentencing patterns in DV cases. Re-arrest rates were higher after the establishment of the DV court, presumably because of better identification of DV cases and the closer monitoring of offenders, which increased the likelihood of detecting a repeat offense. Taken together, these findings suggest that the felony DV court changed the way DV cases were processed, but these changes did not affect case disposition or sentencing. Interestingly, the closer monitoring of offenders led to an increase in the re-arrest rate. The findings from the study of the Brooklyn felony DV court provide a detailed look at the impact of a specialized DV court. However, most DV cases are charged as misdemeanors, and the impact of a DV court may be quite different in misdemeanor cases. In a felony case, defendants are often facing the possibility of a long prison sentence, and plea bargaining can be used to induce defendants to agree to reduced sentences under the condition that they successfully complete a program. Defendants in misdemeanor cases rarely face significant incarceration time, and have fewer incentives to agree to plea bargains. Although specialized misdemeanor DV courts are relatively new, there are also several completed studies describing their impact. Three studies have focused on the impact of misdemeanor DV courts in New York City. Peterson (2002) used 1998 data to examine non-felony DV cases using multivariate models. He compared two boroughs with specialized DV Criminal Courts (Brooklyn and the Bronx) to a borough that did not have a specialized DV Criminal Court (Manhattan). The use of specialized DV courts did not appear to have any consistent effect on conviction rates. Instead, conviction rates appeared to be more influenced by whether the District Attorney’s office screened out weak DV cases (as in the Bronx, where the conviction rate was relatively high) or prosecuted virtually all DV cases (as in Brooklyn and Manhattan, where the conviction rates were relatively low). Convicted DV offenders in the two boroughs with specialized DV courts (Brooklyn and the Bronx) were less likely to be incarcerated than convicted defendants in Manhattan, where cases were heard in mixed-docket courts. This difference reflected the stronger emphasis on the use of batterer intervention programs to monitor defendants in boroughs with specialized DV courts. The specialized DV courts had no consistent effect on the length of jail sentences imposed. Jail sentences were longer in Manhattan and the Bronx, and shorter in Brooklyn. Finally, the study found that cases were processed more quickly in boroughs that use

6 specialized DV courts. Assistant District Attorneys (ADAâ&#x20AC;&#x2122;s) reached plea agreements more quickly in specialized DV courts because they were less likely to insist on a jail sentence as part of the agreement. Peterson (2003a) followed up on the earlier study by examining re-arrests of DV offenders for new DV offenses. The two boroughs with specialized Criminal Court DV parts (Brooklyn and the Bronx) had a higher re-arrest rate than Manhattan, which did not have a specialized DV court, even after controlling for legal, case processing and demographic variables. This difference does not suggest that specialized DV courts increase recidivism. Rather, Peterson (2003a) concluded that re-arrests for a new DV offense were less likely to be classified correctly as DV re-arrests in Manhattan than in the other two boroughs. DV re-arrests in Manhattan were more likely to be incorrectly identified as Non-DV re-arrests. The establishment of the specialized DV courts in Brooklyn and the Bronx led to better efforts to identify DV cases in those boroughs so that they could be tracked for appearances in the specialized parts. Miller (1999) examined the impact of the simultaneous establishment of a specialized DV prosecution bureau and a specialized DV Criminal Court in Queens, New York. The specialized court handled cases of misdemeanor or lesser severity. The conviction rate increased from 30% to 60%, even as the volume of cases increased from 3,500 per year to 4,700 per year. Miller does not report any data on changes in case processing time, sentencing patterns, or re-arrest rates. There is also some research on the impact of specialized DV courts operating in other jurisdictions in the U.S. (see Keilitz (2000) for the results of a survey of these courts, and Tsai (2000) for a description of DV courts in several jurisdictions). One early study found that the dismissal rate declined from 42% to 37% after the establishment of a misdemeanor DV court in Miami in 1993 (Goldkamp et al. 1996). No comparative data were available on case processing time, jail sentences, or re-arrest rates. Davis et al. (2001) examined the impact of a specialized misdemeanor DV court in Milwaukee in 1994. By reducing case processing time, Milwaukee officials hoped to reduce the opportunity for the defendant to intimidate the victim, thereby increasing the number of victims willing to testify. The specialized DV court cut case processing time in half, from 166 days (when cases were heard in the general misdemeanor courts) to 86 days (when cases were heard in the specialized DV court). As expected, defendants were less likely to intimidate the victim before the case was disposed. Bivariate analyses indicated that more victims cooperated with the prosecution and the conviction rate increased from 56% to 69%. There was a decline in the percentage of convicted defendants sentenced to jail, from 75% to 39%. Recidivism within 6 months of case disposition declined from 30% to 16%, as reported in victim interviews. Angene (2000) examined the introduction of a specialized DV court to hear cases with misdemeanor charges in San Diego, California. This study compared the outcomes of DV cases before and after the introduction of a specialized DV court in the

7 late 1990â&#x20AC;&#x2122;s. Case processing time declined from 57 days to 15. There was no change in the proportion of defendants who were convicted (about 93%) or in the proportion of convicted defendants assigned to a batterer intervention program (about 85%). However, there was a decline from 61% to 33% in the proportion of convicted defendants who were incarcerated (some convicted defendants were incarcerated and assigned to the batterer intervention program), although the length of the sentences increased from a median of 45 to 60 days. Recidivism (defined as a new police contact for domestic violence within one year of conviction) declined from 21% to 14%. The San Diego study suggests that introducing a specialized DV court does not affect the likelihood of conviction, reduces reliance on incarcerative sentences, and reduces recidivism rates. Eckberg and Podkopacz (2002) evaluated the impact of the Minneapolis domestic violence court, which handles misdemeanor DV cases. They found that case processing time for DV cases declined after the court was established. The conviction rate increased by 18 percentage points. Pre-trial re-arrests for a new domestic assault case were essentially unchanged, while post-disposition re-arrests for a new domestic assault case declined slightly, from 18% before the court was established to 14% after. Eckberg and Podkopacz (2002) did not conduct multivariate analyses to account for the changes in conviction and recidivism rates. Gover et al. (2003) evaluated the impact of establishing a specialized DV court to hear all non-felony battery cases of domestic violence in Lexington County, South Carolina. The court was established in November 1999 in this predominantly rural county as part of a coordinated response among law enforcement and mental health agencies. The study found that the number of arrests for domestic violence increased significantly after the establishment of the court. Gover et al. (2003) suggest that the court increased the responsiveness of the police, who realized that DV arrests would be taken more seriously. The study does not report information about case processing time, conviction rates or sentences. The recidivism rate of DV offenders declined after the court was established, from 19% to 12%. A multivariate model of the likelihood of re-arrest found that this decline could not be attributed to differences in legal, case processing or demographic characteristics. Gover et al. (2003) concluded that the specialized DV court was effective in deterring domestic violence. Although they do not examine the impact of a misdemeanor DV court, two additional studies of DV courts deserve mention in this review. These studies do not have comparison data from the period before the DV court was established or from a comparable jurisdiction without a DV court. Buzawa et al. (1999) studied the Quincy District Court (QDC), a DV court that handles misdemeanors and certain felony cases in Quincy, Massachusetts. The conviction rate during the study period was about 67%, and 35% of those convicted were sentenced to jail. About 25% of the offenders were re-arrested for a new DV offense within one year of their arraignment. Puffett and Gavin (2004) studied the results of various program mandates required of defendants convicted in the Bronx misdemeanor DV court. Although their report focuses primarily on the impact of programs on convicted defendants, it is possible to pool data for all the

8 programs to provide an overview of all convicted defendants. When the data are pooled, about 40% of convicted DV defendants were sentenced to jail, and about 43% of convicted DV offenders were re-arrested for a new offense (not necessarily a DV offense) within one year of their release. In summary, research on specialized misdemeanor DV courts suggests that these courts are more responsive to victims than mixed-docket courts and provide a more coherent approach to processing DV cases. Four studies found that specialized misdemeanor DV courts reduce case processing time (Angene 2000, Davis et al. 2001, Eckberg and Podkopacz 2002, Peterson 2002). Most studies report an increase in conviction rates or a decline in dismissals (Goldkamp et al. 1996, Miller 1999, Davis et al. 2001, Eckberg and Podkopacz 2002), although two studies (Angene 2000, Peterson 2002) found no impact on conviction rates. As expected, DV courts reduced the use of jail sentences (Angene 2000, Davis et al. 2001, Peterson 2002), relying more heavily on monitoring defendants through batterer intervention and drug and alcohol treatment programs (Goldkamp et al. 1996, Peterson 2002). There are mixed results on the impact of DV courts on the length of jail sentences for incarcerated defendants. Angene (2000) reported longer sentences, while Peterson (2002) found no consistent difference. Four studies report declines in recidivism (Angene 2000, Davis et al. 2001, Eckberg and Podkopacz 2002, Gover et al. 2003), while one reports an increase in recidivism. The latter study (Peterson 2003a) suggests that the increase reflects improved identification of DV cases and closer monitoring of DV offenders for new DV offenses (see also Newmark et al.â&#x20AC;&#x2122;s 2001 study of a felony DV court for a similar finding and interpretation). Taken together, previous research studies suggest that DV courts reduce case processing time, increase conviction rates, reduce the use of jail sentences, and may reduce recidivism. Our previous studies of the impact of specialized DV courts on case outcomes (Peterson 2002) and re-arrests (Peterson 2003a) compared Manhattan, which did not have a specialized DV court in 1998, to Brooklyn and the Bronx, which did. Although we believe that these comparisons provided valuable insight into the impact of a specialized misdemeanor DV court, there were some issues that were difficult to address. Notably, the conviction rates in Brooklyn and the Bronx diverged widely, a finding which we interpreted as indicating that DV courts have no consistent effect on conviction rates. Because the policy for screening DV cases apparently had a stronger effect on the conviction rate, it was difficult to draw definitive conclusions about the impact of the DV courts. A second issue that was difficult to address was the impact of the DV courts on re-arrest rates. Our findings strongly suggested that the re-arrest rate is higher in DV courts because their procedures are better at identifying DV cases. However other borough differences might account for our findings. These limitations of our previous work suggested that further research is warranted using a different approach. The current study uses the approach taken in other studies of the impact of DV courts: a comparison of findings before and after the introduction of a specialized DV court.

9 C. Research Questions This study examines the impact of establishing a specialized domestic violence court in Manhattan. It compares DV cases processed in Manhattan before the establishment of the specialized DV court with DV cases processed after the DV court was established. We address two research questions: 1) What are the effects of the use of a specialized domestic violence court on case outcomes in DV cases? 2) What is the effect of the use of a specialized domestic violence court on the re-arrest rate for DV offenders? As we address the first question, we consider whether there were changes in DV case outcomes between 1998, when DV cases were processed in courts with mixed dockets, and 2001, when DV cases were processed in specialized DV courts. Three case outcomes are considered. First, we examine case dispositions. We do not expect to find that the introduction of the specialized DV court affected the conviction rate. Although most studies have reported an increase in the conviction rate, our own comparative study in New York found that the conviction rate was not consistently related to the use of specialized DV courts. Second, we examine sentence outcomes. We expect to find that jail sentences were imposed less often after a conviction in 2001 than in 1998. Specialized courts focus more on requiring convicted DV offenders to complete a batterer intervention program, usually as an alternative to incarceration. Several studies have found that in mixed-docket courts, convicted DV offenders were more likely to receive jail sentences. Furthermore, in courts with mixed dockets, it may be more difficult to identify appropriate programs and to enroll defendants in them. Finally, we examine the length of jail sentences. We expect to find that jail sentences were of shorter duration in 2001 than in 1998, since specialized DV courts are less focused on the use of jail as a sanction. To address the second research question, we examine the post-disposition rearrest rate for new DV offenses. We compare defendants whose cases were processed in 1998, before the DV court was established in Manhattan, to those whose cases were processed in 2001, after the court was established. We expect to find that the re-arrest rate increased, since we believe that the establishment of the specialized DV court led to improved identification of DV cases. In our analyses of 1998 data (Peterson 2001, 2002) we found that the volume of DV arrests was lower than expected in Manhattan. We believe that when the specialized DV court was established in Manhattan, greater efforts were made to identify DV cases to be processed in the new court.

[This page intentionally left blank]

11 II. Methodology A. Overview of the CJA Database and the Third Quarter 1998 and First Quarter 2001 Datasets The data for this study were drawn primarily from the CJA database. This database contains information about the arrest, case processing and case outcomes of most New York City arrestees. The CJA database includes data from three sources: CJA’s pre-arraignment interview,1 the New York City Police Department’s Online Booking System (OLBS) Database, and the New York State Office of Court Administration (OCA). Information concerning demographic characteristics and the community ties of the offenders is taken from the CJA pre-arraignment interview. Information about the arrests is based on the OLBS data. Detailed Criminal Court and Supreme Court processing and outcome data on each of the arrests are drawn from the OCA data. This report is based on analyses of two datasets. The Third Quarter 1998 Dataset consists of data collected on a three-month cohort of arrests made from July 1, 1998 through September 30, 1998 (Eckert and Curbelo 2000). The dataset includes information on 89,524 arrests where the district attorney elected to bring charges and where a docket number was assigned.2 The First Quarter 2001 dataset consists of data collected on a three-month cohort of arrests made from January 1, 2001 to March 31, 2001. The dataset includes information on 91,729 arrests where the district attorney elected to bring charges and where a docket number was assigned. In addition to information in the CJA database, the two datasets also include information provided by the New York State Division of Criminal Justice Services (DCJS).3 DCJS data were used to supplement and check the reliability of criminal history information that was routinely collected by CJA interviewers. For cases that had multiple dockets, case-processing information in this study is based on the docket that had the most severe arraignment charge (based on Penal Law 1

CJA conducts pre-arraignment interviews to measure the defendant’s community ties and to serve as the basis for making a recommendation as to whether or not the defendant should be released on recognizance at his or her first court appearance. Defendants who are arrested on a bench warrant, given a Desk Appearance Ticket (DAT), or who are held for arraignment on prostitution charges in the downtown Manhattan Criminal Court, are not interviewed by CJA. CJA collects police arrest and Criminal Court information for all arrestees, and they were included in the Third Quarter 1998 Dataset whether or not they were interviewed (Siddiqi 1999). 2 For a more detailed discussion of how the sample was drawn, see Eckert and Curbelo 2000, Appendix A. The sample includes both Summary Arrests and DAT’s. Arrests that had no docket number were retained in the sample if they appeared to be either “A” docket cases in Manhattan (cases given the suffix “A” to distinguish them from other cases that have the same docket number) or direct indictments. Arrest information for these two types of cases was supplemented with defendant information and court processing information when available. 3 DCJS, OCA, and the NYPD are not responsible for the methods or conclusions of this report.

12 severity) in Criminal Court.4 When the most severe arraignment charges on 2 or more dockets are of equal Penal Law severity, the top charge is determined according to guidelines developed by OCA. These guidelines provide a consistent set of rules for determining which of two arraignment charges of equal severity will be identified as the top arraignment charge. In New York Stateâ&#x20AC;&#x2122;s two-tiered court system for handling criminal cases, the Criminal Courts only have trial jurisdiction over cases having a most serious charge of misdemeanor or lesser severity. Most defendants charged with felonies are first arraigned in Criminal Court. Cases sustained at the felony level must be brought for prosecution in Supreme Court. In felony cases where the DA decides not to prosecute the case in Supreme Court (or the Grand Jury fails to return an indictment), the case may be disposed in Criminal Court by dismissal or by a plea to a reduced charge less severe than a felony, or by a transfer to another courtâ&#x20AC;&#x2122;s jurisdiction (e.g., Family Court).5 The cases selected for inclusion in the analyses in this report include only cases that reached a final disposition in Criminal Court. The overwhelming majority (about 96%) of domestic violence cases in Manhattan in both 1998 and 2001 were disposed in Criminal Court. The Third Quarter 1998 Dataset includes case processing information in Criminal Court through final disposition (and sentencing, if there was a conviction), or until August 6, 1999, if the case was not yet disposed. The First Quarter 2001 Dataset includes case processing information in Criminal Court through final disposition (and sentencing, if there was a conviction), or until November 30, 2001. Information about the final disposition in Criminal Court beyond these cutoff dates was not included in the datasets. B. Identifying Domestic Violence Cases Social scientific and legal definitions of domestic violence have changed over the last 30 years. We reviewed the history of these changes in our first report (Peterson 2001, p. 11). In the current report, we review how New York City courts identified DV cases and the differences between 1998 and 2001 in how DV cases were handled in Manhattan.

New York State Penal Law categorizes most offenses according to their severity. The most serious crimes are A felonies, followed by felonies classified as being of severity B through E. Misdemeanors are less severe than felonies, and are classified as A or B misdemeanors (A misdemeanors are more severe). Violations are less severe than misdemeanors, and are not considered crimes, although they can result in jail sentences. No distinctions of severity are made within the category of violations. 5 The Family Courts have concurrent jurisdiction over certain domestic violence cases (Aldrich and Domonkos 2000). Some DV cases are heard only in Criminal Court, some are heard in both Criminal Court and Family Court, and others are heard only in Family Court. We do not have access to data on DV cases that are heard in Family Court, and our report draws no conclusions about these cases.

13 In New York State the statutory definition of domestic violence approximates what has come to be known in the social scientific literature as “family violence.” Under New York State’s Criminal Procedure Law (CPL) §530.11 (as amended by the 1994 Family Protection and Domestic Violence Intervention Act), family offenses are defined as offenses committed against a member of the same family or household, where “family or household” are defined as: (1) persons related by consanguinity or affinity, (2) persons legally married to each other, (3) persons who were formerly married, and (4) persons who have a child in common, whether or not they have ever been married or lived together. New York State’s statutory definition of domestic violence excludes unmarried partners, unless they have a child in common. However, the New York City Police Department (NYPD) operates with an expanded definition of domestic violence that includes individuals who are not married, but who are cohabiting or have previously lived together. This NYPD definition of “family” expands on New York State law by including “common-law” marriages, same-sex couples, and registered New York City domestic partners (NYPD 2000). By citywide agreement, the DA’s offices and the Criminal Courts in all five boroughs also use this expanded definition to identify DV cases, whether or not the relationship between the victim and defendant meets the New York State statutory requirements contained in CPL §530.11. To identify domestic violence cases, Assistant District Attorneys (ADA’s) use information collected by the police about the relationship between the victim and the defendant, if any. ADA’s also often ask victims about their relationship with the defendant. When this information indicates that the defendant-victim relationship meets the NYPD expanded definition of domestic violence, the case is identified as a DV case. DV case files in all 5 boroughs are then given beige “backs” (special color-coded back sheets) to distinguish them from other case files. At Criminal Court arraignment, Court Clerks assign an arraignment hearing type of “DV” to domestic violence cases, and this designation is entered in OCA’s computerized court records. At the time the defendants in the Third Quarter 1998 Dataset were arrested, all cases in Manhattan with a DV hearing type were sent to mixed-docket courts for post-arraignment appearances. At the time the defendants in the First Quarter 2001 Dataset were arrested, most cases in Manhattan with a DV hearing type at arraignment were sent to the specialized DV Criminal Court for post-arraignment appearances. There were exceptions—some cases with a DV hearing type (especially those that did not involve intimate partner violence or child abuse) were sent to mixed-docket courts. Finally, some cases that did not have a DV hearing type at arraignment in 2001 were also sent to the specialized DV courts, presumably because information that these cases involved domestic violence became available only after arraignment. In this study, we identified “domestic violence” cases by relying on the court’s identification of these cases. We used information about both hearing type and court part, since not all DV cases were assigned a DV hearing type. We identified cases as domestic violence cases if the Office of Court Administration reported that:

14 (1) the case had a domestic violence hearing type at Criminal Court arraignment, and/or, (2) the case had one or more appearances in a specialized domestic violence Criminal Court.6 Using these criteria, we identified 990 domestic violence cases in the Third Quarter 1998 Dataset and 1,249 DV cases in the First Quarter 2001 Dataset. In the Third Quarter 1998 Dataset, all the DV cases were identified on the basis of a domestic violence hearing type at arraignment, since there was no specialized DV court in Manhattan in 1998. In 2001, 72% of DV cases had both a DV hearing type at arraignment and at least one appearance in the specialized DV court. An additional 18% of cases had a DV hearing type at arraignment but no appearance in the specialized DV court. Finally, 10% of the cases had at least one appearance in a specialized domestic violence court, but did not have a domestic violence hearing type at Criminal Court arraignment. Our method of identifying domestic violence cases is appropriate for the purposes of this study. The cases that we identified as DV cases were clearly known to the courts as DV cases. Since this report examines how the processing of DV cases affected case outcomes and re-arrest rates, it is important that we examine only cases where the DAâ&#x20AC;&#x2122;s, judges and other key personnel were aware that the case was a DV case. As we noted in a previous report (Peterson 2001), the measure identifying DV cases does have some limitations. First, there may be instances where a DV case was not identified as such in court records (i.e., it did not receive a DV hearing type at Criminal Court arraignment and did not have any appearances in the specialized DV parts). The measure did not identify these as DV cases but instead categorized them as Non-DV cases. In the current study, this limitation affected our analyses in two ways. First, it reduced the sample size of DV cases on which we report. Nevertheless, we have an adequate sample size for our analyses, and were able to draw valid conclusions about DV cases from our samples. Second, we slightly overestimated the number of Non-DV cases in the sample. However, the number of DV cases misidentified as Non-DV cases is likely to be a very small proportion of the total number of Non-DV cases. A second problem with our identification of domestic violence cases is that it relies, in part, on identifying cases that appeared in specialized domestic violence courts. In 1998, Manhattan did not have a specialized domestic violence court while in 6

In the third quarter of 1998, the specialized domestic violence Criminal Court parts were AP-12 and AP-15 in Brooklyn, AP-10 and TAP-2 in the Bronx, AP-4 in Queens, and AP2-DV in Staten Island. In the first quarter of 2001, there were 2 additional DV Criminal Courts: Parts D and JP13 (for DV jury trials) in Manhattan. The specialized DV parts in Brooklyn were renamed DV1 and DV2 in January 2001. Although AP2-DV in Staten Island was identified as a separate court part in our data, it was actually a specialized DV calendar. DV cases on this calendar were heard in an all-purpose (i.e., mixed-docket) court part two days a week. We identified cases as DV cases if they had one or more appearances on this calendar.

15 2001, there was a specialized DV Criminal Court. In 1998, we identified 990 DV cases in Manhattan, based solely on DV hearing type at arraignment. Since not every domestic violence case received a DV hearing type at arraignment, we almost certainly failed to identify some domestic violence cases in 1998. It is difficult to gauge the magnitude of the problem, but it is possible to generate some rough estimates. Specifically, by examining DV cases that we identified in 1998 in Brooklyn, the Bronx, Queens and Staten Island, we can determine what proportion of these cases had one or more appearances in a specialized domestic violence court but did not have a DV hearing type at arraignment. About 21% of the DV cases in Brooklyn, the Bronx, Queens and Staten Island were identified as having an appearance in a specialized DV court, but did not have a DV hearing type at arraignment. A second estimate is based on Manhattan data in 2001. About 9% of DV cases in Manhattan in 2001 had appearances in the specialized DV court but did not have a DV hearing type at arraignment. Assuming that similar patterns hold for Manhattan in 1998, between 9% and 21% of DV cases in Manhattan were not identified by our measure. Stated another way, we have identified about 79% to 91% of the DV cases in Manhattan during the third quarter of 1998. A third problem with our identification of domestic violence cases is that not all DV cases were sent to the specialized DV court in 2001. Some DV cases (especially those that did not involve intimate partner violence or child abuse) were sent to mixeddocket courts. In the first quarter of 2001, about 18% of DV cases in Manhattan were sent to the mixed-docket courts. We included these cases in our 2001 sample because similar cases were also included in our 1998 sample. It was impossible to determine which cases would have been sent to the specialized DV court in 1998 if one had been in existence then. To keep the samples comparable we included in our 2001 sample all DV cases identified by the court, even if they were sent to mixed-docket courts. Although not every case in 2001 was heard in the specialized DV court, the vast majority (82%) were, and our findings for 2001 primarily reflect the outcomes of cases in the specialized DV court. C. Identifying Re-arrests for Domestic Violence Offenses We measured recidivism in this study by examining re-arrests for DV offenses over the 18-month period following the disposition of the defendantâ&#x20AC;&#x2122;s case. Unfortunately, re-arrest rates are likely to underestimate recidivism. New DV offenses may not lead to re-arrest, since many victims do not call the police when a new offense occurs, and police may not make an arrest even when they are called. We compensated for this problem in part by using a long 18-month at-risk period. Since recidivists are likely to re-offend multiple times, using a long at-risk period increases the chances that at least one of the new offenses will lead to re-arrest during the term of our study. To overcome the limitations of re-arrest, some studies measure recidivism using interviews with the victim. Interviewers can learn about incidents that did not result in calls to the police and re-arrest of the defendant. Rates of recidivism based on victim

16 interviews are generally higher than rates based on re-arrest. Victim interviews also have weaknesses, however. It is often very difficult to reach victims and to complete interviews with them. Furthermore, victim interviews ignore the possibility that the defendant has re-offended with a new victim. Because both types of data have strengths and weaknesses, we would have preferred to measure recidivism using both victim interviews and re-arrest data. We used re-arrest data for practical reasonsâ&#x20AC;&#x201D;it was the only measure available to us. Although re-arrest may underestimate recidivism, it has two advantages over victim interviews. Data are potentially available for all defendants, not just those for whom victim interviews were completed. Also, it measures recidivism against new victims as well as against the same victim. Our measures of re-arrest are also affected by the problems described above regarding the identification of DV arrests. To the extent that the courts fail to identify DV arrests as such, our measures of re-arrest underestimated the actual number of DV rearrests. In our study, we expect to find that re-arrests were more accurately identified in 2001 than in 1998, for three reasons. First, we expect that more cases were identified with a DV hearing type at arraignment in 2001 than in 1998. After the specialized DV court was established in 2001, it was more important for cases to be accurately identified as DV cases at arraignment so that subsequent appearances would be scheduled in the specialized court. Second, cases that were not identified with a DV hearing type at arraignment, but were identified as DV cases during or after arraignment were treated differently in 1998 and 2001. In 1998, such cases were sent to mixeddocket courts, and were not identifiable as DV cases in our dataset. In 2001, most such cases were sent to the specialized DV court and were identified as DV cases in our dataset. Third, the specialized DV court monitored defendants more closely and defendants were probably more likely to be re-arrested for new offenses in 2001. In 1998, when cases were heard in mixed-docket courts, defendants who committed new DV offenses were probably not as likely to be re-arrested. For these reasons, we expect that our measure of re-arrest is more accurate in 2001 than in 1998, and that the re-arrest rate will be higher in 2001 than 1998. This expectation is based not on the assumption that DV offenders were more likely to reoffend in 2001 than 1998, but on the assumption that new DV offenses were more likely to lead to re-arrest in 2001. D. Selection of the Crimes Against Persons and Property Subsample The current study used data from the Crimes Against Persons and Property (CAPP) Subsample described in a previous report (Peterson 2001).7 This subsample 7

In our first report (Peterson 2001), we also discussed results for an Assaults Subsample, which included all cases where the top arraignment charge was assault (PL 120). We used the Assaults Subsample to provide a more focused comparison of DV cases to similar Non-DV cases. In this study, we focus primarily on DV cases, and make only a few comparisons of DV to Non-DV cases. We therefore use only the Crimes Against Persons and Property Subsample

17 was selected so that we could examine domestic violence cases across a wide range of charges. We began by selecting cases for this subsample where there was an alleged attempt to cause injury or where an overt threat of injury was made (Weis 1989). We used the most severe arraignment charge (based on Penal Law severity) to determine the nature of the offense, since this charge determines how the case is handled in the court system. We did not use the most severe arrest charge, which reflects charging decisions made by the police. We initially selected all cases that had a top (i.e., most severe) arraignment charge from any of the following New York State Penal Law articles: PL 120 (Assault), PL 130 (Sex Offenses), PL 160 (Robbery), PL 260 (Crimes Against Children), or PL 265 (Weapons). Unfortunately, we were not able to include cases disposed in Criminal Court that had top arraignment charges from PL 125 (Homicide), PL 150 (Arson), and PL 135 (Kidnapping).8 Cases charged with offenses in these Penal Law articles were excluded since there were too few domestic violence cases with top arraignment charges in each of these Penal Law articles for reliable multivariate analysis. Recognizing that domestic violence often includes offenses that result in financial and psychological harm, rather than just physical harm, we also selected cases if they had a top arraignment charge from one of the following Penal Law articles: PL 140 (Burglary), PL 145 (Criminal Mischief), PL 155 (Larceny), PL 205 (Escape and Resisting Arrest), PL 215 (Criminal Contempt),9 and PL 240 (Public Order Offenses). Within each Penal Law article, we selected only those cases that had charges that could plausibly include elements comparable to those found in domestic violence cases. For example, cases with a top arraignment charge of Assault in the Third Degree (PL §120.00) were included in the subsample. However cases of Gang Assault in the First Degree (PL §120.07) or Gang Assault in the Second Degree (PL §120.06) were excluded, since it is unlikely that a domestic violence case would include a gang assault charge. Similarly, in PL 240, prostitution charges (PL §240.37) were excluded. We then narrowed the subsamples further to identify an appropriate group of cases for the analysis. First, we included only Summary Arrests (i.e., cases in which the defendant was held in custody pending Criminal Court arraignment), excluding cases where the defendant was issued a Desk Appearance Ticket (DAT) and released by the arresting officer. DAT’s are rarely issued in DV cases. We also excluded cases with juvenile defendants (under age 16), cases that did not reach a final disposition by the cutoff date, and cases that were missing data on defendant’s criminal history or sex. We also excluded cases that were disposed in Criminal Court on Vehicle and Traffic Law (VTL) or Administrative Code (AC) charges. After these exclusions, the Crimes Against Persons and Property Subsample included 7,043 cases in Manhattan in the third quarter of 1998, of which 990 (about 14%) were domestic violence cases. In the first quarter of 2001, the Crimes Against Persons and Property Subsample included 8,486 Manhattan cases, of which 1,249 (about 15%) were DV cases. for our analyses. This subsample includes information about the full range of DV offenses, including assaults as well as violations of orders of protection, crimes against children, etc. 8 Most, but not all, of these cases were sustained as felonies and disposed in Supreme Court. 9 Penal Law article 215 includes violations of orders of protection.

E. Case-based vs. Defendant-based Data Files The datasets used to analyze case outcomes in this study are case-based data files that include information on all prosecuted arrests that were held for arraignment. Some defendants were arrested two or more times during the quarter. The 1998 dataset includes information about 990 DV cases in Manhattan for 955 defendants. Over 97% of these defendants had only one case in the data file, about 3% had two cases initiated during the Third Quarter of 1998, and only 1 defendant had 3 cases in the data file. The 2001 dataset includes information about 1,249 DV cases in Manhattan for 1,188 defendants. Over 93% of these defendants had only one case in the data file, about 6% had two cases initiated during the First Quarter of 2001, and only 6 defendants had 3 cases in the data file. The analyses of case outcomes use the case-based data files in order to present a comprehensive picture of the processing of all cases that were initiated during the sample period. Information about each case is included in the case-based files. While the vast majority of defendants have only one case in the case-based data file, it is important to remember that the descriptions and analyses of case outcomes summarize information about cases, not about defendants. The use of a case-based file presents some minor problems for statistical analysis. First, the statistical procedures used in this report assume that the units of analysis in the data file are statistically independent of each other. This requires that the characteristics of each case in the data file should not have a fixed relationship to the characteristics of any other case in the data file. When a defendant has two or more cases in the data file, the defendantâ&#x20AC;&#x2122;s ethnicity, sex, age, criminal history, community ties, etc. for one case do have a fixed relationship to that defendantâ&#x20AC;&#x2122;s characteristics in the other case(s). Using a case-based data file therefore violates one of the assumptions of the statistical techniques employed in the report. However, the impact of this violation of assumptions is likely to be minimal because the number of defendants with multiple cases is relatively small. Second, when defendants have multiple cases, the outcome of one case may be affected by the outcome of another. For example, a defendant may plead guilty in one case in return for having charges in another case dropped. Prior research suggests that once prior record and case characteristics are controlled for, there is no difference in case outcomes between defendants who had multiple cases and those who did not (Klein et al. 1991). Since our statistical models control for prior record and case characteristics, we expect the impact of this problem on our results to be minimal. The models of re-arrest, unlike the models of case outcomes, use a defendantbased file. For defendants who had multiple arrests within the quarter, we only included in the defendant-based file information about the first DV arrest in the quarter. The sample sizes for the defendant-based files are therefore slightly lower than for the casebased files used to analyze case outcomes.

19 F. Plan of Analysis This study will begin with a description of domestic violence cases in Manhattan in the Third Quarter 1998 and First Quarter 2001 datasets. We will compare DV cases in these two datasets in terms of four important outcomes: case disposition, sentence outcome, length of jail sentence, and the re-arrest rate for DV offenses. We also will examine a variety of defendant and case characteristics: arraignment charges, demographic characteristics, defendantâ&#x20AC;&#x2122;s criminal history, charge characteristics, release recommendation and case processing characteristics. After this overview, we will present statistical models predicting the likelihood of conviction, likelihood of an incarcerative sentence, length of jail sentence and likelihood of re-arrest for a DV offense. These models will assess the influence of defendant and case characteristics on conviction rates, incarceration rates, average sentence length, and re-arrest rates. We analyze outcomes separately for 1998 and 2001 to determine whether the influence of specific characteristics on outcomes was different in 2001 than in 1998. The analysis of outcomes focuses on the two research questions posed in the Introduction to this report: What are the effects of the use of a specialized domestic violence court on case outcomes in DV cases? What is the effect of the use of a specialized domestic violence court on the re-arrest rate for DV offenders? To address these questions, this report will begin by providing information on case outcomes, re-arrests, and defendant and case characteristics in Section III. In Sections IV, V, VI and VII, we present models predicting the likelihood of conviction, likelihood of incarceration, length of jail sentence and likelihood of re-arrest, respectively. Section VIII summarizes the findings and discusses their implications.

[This page intentionally left blank]

21 III. OVERVIEW OF DIFFERENCES BETWEEN 1998 AND 2001 IN THE PROCESSING OF DOMESTIC VIOLENCE CASES IN MANHATTAN This section of the report provides an overview of differences between 1998 and 2001 in the processing of DV cases disposed in Manhattan’s Criminal Courts. As noted earlier, the Crimes Against Persons and Property Subsample includes most of the DV cases we identified in the datasets, and provides a comprehensive picture of the full range of DV cases. To determine whether differences between 1998 and 2001 are real differences, or are due to chance alone, we use tests of statistical significance. Tests of Statistical Significance Statistical significance tests assess the likelihood that the percentage differences that were observed in the sample could have occurred by chance alone. The tests take into account the size of the sample and the magnitude of the differences observed. Larger percentage differences and percentage differences based on larger samples are more likely to be statistically significant. In this report, following standard convention, significance levels less than .05 were considered statistically significant. This means that the statistically significant differences found in this study had less than a 5% chance of being due to chance alone. A. Case Dispositions, Sentence Outcomes, Length Of Jail Sentences, Re-arrest Rates We now turn our attention to reviewing data on four major outcomes: case disposition, sentence outcome, length of jail sentence, and re-arrest rates for new DV offenses. In New York State, cases disposed in Criminal Court can result in one of several final dispositions:10 a plea of guilty, a finding of guilty after trial, an acquittal after trial, dismissal,11 or an adjournment in contemplation of dismissal (ACD). In this report, convictions are defined to include pleas of guilty and findings of guilty after trial, including pleas or findings of guilty to a violation. (Although violations are not considered crimes under New York State Penal Law, they can result in a jail sentence.) Acquittals, dismissals and ACD’s are categorized in this report as non-convictions. Although ACD’s are not convictions, they sometimes have conditions attached. Based on interviews with the DA’s office, we learned that in Manhattan in both 1998 and 2001

Cases whose last disposition in Criminal Court resulted in a transfer to Supreme Court or Family Court were excluded from the analyses because these cases continued in another court. A small number of cases were briefly sent to Supreme Court before being returned to Criminal Court for a final disposition. We count these cases as disposed in Criminal Court. 11 We include as dismissals cases that are disposed as “sealed” as well as cases that are “dismissed—not sealed” but are never subsequently brought back to court.

22 ACD’s sometimes required completion of a batterer intervention program.12 Defendants who failed to comply may have had their cases restored to the calendar. The use of ACD’s was one of a variety of strategies employed to achieve the goal of monitoring defendants in DV cases. When these conditions are not fulfilled, or if the defendant is re-arrested within 6 months (12 months in the case of a family offense), the case can be re-opened and restored to the calendar for another, possibly more severe, disposition. However ACD’s do not result in a conviction unless the defendant violates the conditions of the ACD and subsequently is returned to court and convicted. In both 1998 and 2001, the conviction rate in DV cases was about 29% (see first row of Figure 3-1). Furthermore, there was virtually no difference in the proportion of cases that ended in dismissal or an ACD. Slightly over half of DV cases were dismissed in both years and about 15% were disposed with an ACD. This pattern of results suggests that the establishment of a specialized DV court did not affect the distribution of case dispositions. The specialized DV court was not designed to increase the conviction rate and we did not expect to find that the conviction rate would be higher after it was established. As noted earlier, Fritzler and Simon (2000) suggested that there is a better understanding of the unique nature of DV cases in specialized courts, and that this might lead to higher conviction rates in DV courts. Their argument was not supported by our findings. Figure 3-1 also includes information about case dispositions for comparable NonDV cases. Conviction rates in Non-DV cases were more than twice as high as conviction rates in DV cases. The conviction rate for Non-DV cases was higher in 2001 than in 1998 (71% vs. 65%). Similarly, the dismissal rate was lower in 2001 than in 1998 (19% vs. 25%). The percentage of ACD’s was the same in Non-DV cases in both years (10%). The shift toward higher conviction rates and lower dismissal rates in NonDV cases was not evident in the results for DV cases reported above. DV case dispositions were essentially the same in 1998 and 2001. Taken together, the findings on case dispositions by year suggest two major conclusions. First, the use of a specialized DV court seems to have had little effect on dispositions of DV cases in Manhattan. There was no difference in conviction rates for DV cases in Manhattan between 1998 and 2001. Furthermore, there were no differences between 1998 and 2001 in the rate at which cases were dismissed or disposed as ACD’s. Second, the equal conviction rates for DV cases in 1998 and 2001 do not reflect the pattern of conviction rates for Non-DV cases in those years. The conviction rate increased for Non-DV cases between 1998 and 2001, and the dismissal rate declined.

In cases where the victim did not sign a supporting deposition, a defendant whose case was disposed with an ACD generally could not be required by the court to complete a batterer intervention program.

FIGURE 3-1 CASE DISPOSITIONS IN CRIMINAL COURT Prosecuted Cases in Manhattan DV Cases 3rd Quarter 1998

DV CASES 1st Quarter 2001

(N = 990)

(N = 1,249)

Convicted 29%

Convicted 29% Dismissed 55%

Dismissed 56%

ACD 16%

ACD 15%

NON-DV CASES 3rd Quarter 1998

NON-DV CASES 1st Quarter 2001

(N = 6,053)

(N = 7,262) Dismissed 25%

Convicted 65%

ACD 10%

Dismissed 19% ACD 10% Convicted 71%

Next, we considered differences in sentence outcomes for those cases that ended in a conviction (see first row of Figure 3-2). The differences in sentence outcomes between 1998 and 2001 were statistically significant. The proportion of convicted defendants in DV cases who received a jail sentence was lower in 2001 than in 1998 (27% vs. 31%). (Jail sentences include both “time served” sentences and definite sentences, i.e., sentences for a specified number of days). In both years, most convicted defendants who did not receive a jail sentence were given a conditional discharge. The proportion of defendants who received a conditional discharge was higher in 2001 than in 1998 (67% vs. 54%). The greater use of conditional discharges in 2001 reflects the specialized DV court’s greater emphasis on requiring convicted defendants to complete batterer intervention programs as a condition of the conditional discharge. To monitor compliance with the requirements of batterer intervention programs, cases are transferred to a DV compliance part for regular appearances.

FIGURE 3-2 SENTENCE OUTCOMES IN CRIMINAL COURT Convicted Defendants in Manhattan DV CASES 3rd Quarter 1998

DV CASES 1st Quarter, 2001

(N = 283) Jail 27%

Jail 31% Conditional Discharge 54% Other Sentence 15%

(N = 360)

Other Sentence 6%

Conditional Discharge 67%

NON-DV CASES 3rd Quarter 1998

NON-DV CASES 1st Quarter, 2001

(N=3,911)

(N = 5,183)

Jail 54%

Conditional Discharge 42%

Other Sentence 4%

Jail 54%

Conditional Discharge 43%

Other Sentence 3%

The changes in sentence outcomes for DV cases do not appear to reflect more general patterns of change that also affected Non-DV cases. There was no change between 1998 and 2001 in the proportion of defendants in Non-DV cases who received a jail sentence in Manhattan or in the proportion who received a conditional discharge sentence (see second row of Figure 3-2). Incarceration rates in Non-DV cases were 54% in both years, considerably higher than the rates in DV cases. The pattern of results for sentence outcomes suggests two general conclusions. First, after the introduction of the specialized DV court in Manhattan, convicted DV defendants were less likely to be sentenced to jail. One result of using specialized DV courts is less frequent use of incarceration. Second, there was greater reliance on

25 conditional discharges and batterer intervention programs in DV cases after the introduction of the specialized court. The third case outcome we examined was length of jail sentence. The average length of jail sentence received by incarcerated defendants in DV cases in Manhattan was 65 days in 1998 and 48 days in 2001 (see first row of Figure 3-3). This difference appears to suggest that the introduction of the specialized DV court in Manhattan reduced the length of jail sentences. However, the sample sizes on which these averages are based are quite small, and the difference between 1998 and 2001 is not statistically significant. Furthermore, our previous analysis of cross-borough differences found no impact of the use of specialized DV courts on length of jail sentences. Measuring Length Of Jail Sentences We measured length of sentence by determining how many days the defendant actually spent in jail for the sentence in the case. For “time served” sentences, we used information about release status to measure the amount of time the defendant was incarcerated between arrest and final disposition. For definite sentences, we used the number of days of jail imposed by the court. We then subtracted one-third of the length of the definite sentence to account for the time allowance that most defendants receive for “good behavior,” as provided by New York State Penal Law §70.30(4b). For example, a 30-day definite sentence was coded as 20 days in jail, after allowing for a 10-day reduction in the sentence. However, if the definite sentence was imposed after the defendant had already served more than two-thirds of the sentence, we used the actual time served as the sentence. For example, if a defendant who had been held for 25 days received a 30-day sentence, the sentence was coded as 25 days to indicate the actual time the defendant served. Sentences in Non-DV cases were, on average, shorter than for DV cases (see second row of Figure 3-3). Moreover, like sentences in DV cases, jail sentences in Non-DV cases were shorter in 2001 than in 1998. This suggests that the shift toward shorter sentences in DV cases may have reflected a general trend toward shorter sentences in both DV and Non-DV cases. Finally, we examined post-disposition re-arrests among DV offenders in Manhattan in 1998 and 2001 (Figure 3-4). (We do not report data on re-arrests among Non-DV offenders in Manhattan; citywide data for 1998 are available in Peterson 2003a). The re-arrest rate for DV offenses indicates the percentage of defendants who were re-arrested at least once for a DV offense within 18 months of case disposition. The re-arrest rate for all types of offenses indicates the percentage of defendants who were re-arrested for any new offense, whether it was a DV offense or not. To calculate

26 FIGURE 3-3 LENGTH OF JAIL SENTENCE IN CRIMINAL COURT Defendants Sentenced to Jail in Manhattan

Mean Number of Days Sentenced to Jail Incarcerated DV Defendants 75

65 48

50 25 0

3rd Quarter 1998 (N = 87)

1st Quarter 2001 (N = 98)

Mean Number of Days Sentenced to Jail Incarcerated NON-DV Defendants 75 50 25

3rd Quarter 1998

1st Quarter 2001

(N = 2,103)

(N = 2,788)

re-arrest rates, we used defendant-based files (see discussion in Section II E above). Each case identified in this table represents the first DV arrest during the third quarter of 1998 or first quarter of 2001 for each defendant arrested in the quarter. As discussed in Section II above, the establishment of the specialized DV court in Manhattan in June 2000 required the courts to more accurately identify DV arrests. We expected that more arrests would be flagged as DV arrests in 2001 than in 1998. We also expected that the specialized DV court would more closely monitor DV defendants for new offenses in 2001. As a result, we expected that the rate of DV rearrests would be higher for the first quarter 2001 dataset than for the 1998 dataset. As shown in Figure 3-4, our expectation was confirmed. The re-arrest rate for DV offenses was about 12% for the 1998 defendants and about 16% for the 2001 defendants. The re-arrest rate for all types of offenses was the same in both years: 37%. Since the overall rate of re-arrests was the same in 1998 and 2001, it seems likely that the higher re-arrest rate for DV offenses in 2001 than in 1998 is due to the more accurate identification of re-arrests as DV offenses in 2001. Stated another way, the data

FIGURE 3-4 POST-DISPOSITION RE-ARREST RATES Prosecuted DV Defendants in Manhattan

Re-arrest Rate for DV Offenses

50% 25%

12%

16%

3rd Quarter 1998

1st Quarter 2001

(N = 955)

(N = 1,188)

Re-arrest Rate for All Types of Offenses

50%

37%

3rd Quarter 1998

1st Quarter 2001

25% 0% (N = 955)

(N = 1,188)

suggest that the re-arrest rate for DV offenses in 1998 was underestimated. The identification of re-arrests for DV offenses improved after the establishment of the specialized DV court. Further evidence in support of this explanation was provided in our citywide study of re-arrest rates based on 1998 data (Peterson 2003a). We found that the likelihood of re-arrest was significantly lower for defendants whose cases were disposed in Manhattan than in other boroughs. While this might indicate a real difference among the boroughs, we believe it is more likely that the lower re-arrest rate for DV offenses in Manhattan was due to weakness in the measure of DV re-arrests in this borough. As discussed in Section II-C above, we believe that some DV re-arrests in Manhattan were incorrectly classified as Non-DV re-arrests.

28 B. Defendant and Case Characteristics We now examine differences between 1998 and 2001 in terms of defendant and case characteristics in DV cases. These characteristics will be used in subsequent sections of the report as predictors of case outcomes. Since we will be using these characteristics to determine if they can account for changes in case outcomes over time, this section of the report will highlight areas where there are differences between the third quarter of 1998 and the first quarter of 2001. Arraignment Charge We begin our overview with an examination of a variable that will be used as a control variable in our models: arraignment charge Penal Law article. The differences between 1998 and 2001 were small but statistically significant (see Table 3-1). The percentage of cases with assault and criminal contempt charges declined slightly from 1998 to 2001. The percentage of cases charged with harassment and â&#x20AC;&#x153;otherâ&#x20AC;? charges increased slightly. Differences in the distribution of offenses may affect DV case outcomes. Our models predicting case outcomes will therefore include arraignment charge Penal Law article as a control variable. TABLE 3-1 ARRAIGNMENT CHARGE PENAL LAW ARTICLE BY YEAR FOR DOMESTIC VIOLENCE CASES IN MANHATTAN Crimes Against Persons and Property Subsample ARRAIGNMENT CHARGE PENAL LAW ARTICLE *** Assault (PL 120) Criminal Contempt (PL 215) Harassment (PL 240) Crimes Against Children (PL 260) Other Total, all cases (N of cases)

***

3rd Quarter 1998

1st Quarter 2001

65%

63%

100% (990)

100% (1,249)

Differences between 1st Quarter 2001 and 3rd Quarter 1998 were statistically significant at p < .001.

29 Defendant-Victim Relationship Using data collected by the NYPD, we were able in many cases to determine how the defendant was related to the victim at the time of the arrest (see Table 3-2). There were statistically significant differences in type of relationship between 1998 and 2001. Defendants in DV cases in 2001 were less likely to be married, and more likely to be in “other” family relationships (e.g., parent-child), than defendants in 1998. Unfortunately, data on defendant-victim relationship was frequently missing. This indicates that the existence and/or the nature of the relationship was unknown to the police at the time of the arrest, and only became known to the District Attorney’s office at a later time. Because data was missing more frequently in 2001 (34% missing) than in 1998 (28% missing), it is not clear whether the statistically significant differences observed for the other categories are real differences between the 1998 and 2001 datasets, or merely differences in the rate of reporting. Finally, because we did not have information about the age of the victim, we were unable to clearly identify cases of child abuse or elder abuse using the NYPD data.13 TABLE 3-2 DEFENDANT-VICTIM RELATIONSHIP BY YEAR FOR DOMESTIC VIOLENCE CASES IN MANHATTAN Crimes Against Persons and Property Subsample DEFENDANT-VICTIM RELATIONSHIP *** Boyfriend-Girlfriend Married or Common-law Spouse Other family relationship1 Missing Total, all cases (N of cases)

3rd Quarter 1998

1st Quarter 2001

16%

100%2 (990)

100% (1,249)

Includes parent-child, grandparent-grandchild, sibling and other family relationships. 2 Percentages do not sum to 100% due to rounding error. ***

Differences between 1st Quarter 2001 and 3rd Quarter 1998 were statistically significant at p < .001.

For example, when the defendant is identified as the parent of the victim, we do not know if the child was a minor under the age of 18, or an adult. Similarly, when the defendant is identified as the son or daughter of the victim, we do not know if the victim was elderly or not.

Defendant’s Demographic Characteristics Three demographic variables were included in our analyses: gender, ethnicity and age. None of the demographic differences between 1998 and 2001 were statistically significant. About one sixth of the defendants in DV cases were women; this percentage did not change from 1998 to 2001 (see Table 3-3). The distribution of defendants in DV cases by ethnicity changed only slightly. In both years, there were slightly more Black defendants than Hispanic defendants. The age distribution of DV offenders was virtually the same in 1998 and 2001. TABLE 3-3 DEFENDANT’S DEMOGRAPHIC CHARACTERISTICS BY YEAR FOR DOMESTIC VIOLENCE CASES IN MANHATTAN Crimes Against Persons and Property Subsample DEFENDANT’S DEMOGRAPHIC CHARACTERISTICS GENDER Male Female Total, all cases (N of cases) ETHNICITY Black White Hispanic Other, non-Hispanic Total, all cases (N of cases) AGE CATEGORY Age 16-20 Age 21-29 Age 30-39 Age 40 and older Total, all cases (N of cases) 1

3rd Quarter 1998

1st Quarter 2001

83% 17

82% 18

100% (990)

100% (1,249)

45% 10 41 5

44% 11 41 4

100%1 (990)

100% (1,249)

10% 28 35 27

11% 27 35 27

100% (990)

100% (1,249)

Percentages do not sum to 100% due to rounding error.

Defendant’s Criminal History There were few differences between 1998 and 2001 in the prior criminal record of defendants in DV cases (Table 3-4). Nearly three fifths of defendants in DV cases in both years had a prior adult arrest at the time of their arrest. These similarities in prior arrest record were also reflected in prior convictions. Defendants in DV cases in 1998 averaged 1.10 prior misdemeanor convictions and 0.41 prior felony convictions, compared to 1.39 prior misdemeanor convictions and 0.44 prior felony convictions for defendants in 2001.14 Taken together, these findings indicate that defendants in DV cases in Manhattan had similar criminal histories, on average, in 1998 and 2001.

TABLE 3-4 DEFENDANT’S CRIMINAL HISTORY BY YEAR FOR DOMESTIC VIOLENCE CASES IN MANHATTAN Crimes Against Persons and Property Subsample DEFENDANT’S CRIMINAL HISTORY PRIOR ARREST HISTORY First arrest Has prior arrests Total, all cases (N of cases) NUMBER OF PRIOR CONVICTIONS Mean number of prior Misdemeanor convictions Mean number of prior felony convictions (N of cases)

3rd Quarter 1998

1st Quarter 2001

41% 59

100% (990)

100% (1,249)

1.10

1.39

.41

.44

(990)

(1,249)

Information about the number of prior misdemeanor convictions and number of prior felony convictions is based on the CJA measures of these variables. However, when CJA data were missing, information from DCJS was substituted (see description of datasets in Section II).

32 Arrest and Arraignment Charge Characteristics In addition to the arraignment charge Penal Law article described above, we examined several additional charge characteristics. The first charge variable considered is the number of arrest charges. (CJAâ&#x20AC;&#x2122;s database includes information on up to 4 arrest charges, whereas we only have information about the top (i.e., most severe) arraignment charge.) The average number of arrest charges was about 1.5 in both years (see Table 3-5). TABLE 3-5 ARREST AND ARRAIGNMENT CHARGE CHARACTERISTICS BY YEAR FOR DOMESTIC VIOLENCE CASES IN MANHATTAN Crimes Against Persons and Property Subsample ARREST AND ARRAIGNMENT CHARGE CHARACTERISTICS MEAN NUMBER OF ARREST CHARGES (N of cases) SEVERITY OF ARRAIGNMENT CHARGE *** Violation Misdemeanor Felony Total, all cases (N of cases) CHANGE IN CHARGE SEVERITY FROM ARREST TO ARRAIGNMENT Charge severity reduced No change in severity Charge severity increased Total, all cases (N of cases) 1

3rd Quarter 1998

1st Quarter 2001

1.58

1.55

(990)

(1,249)

0% 74 26

1% 82 17

100% (990)

100% (1,249)

17%

15%

100%1 (990)

100%1 (1,249)

Percentages do not sum to 100% due to rounding error.

***

Differences between 1st Quarter 2001 and 3rd Quarter 1998 were statistically significant at p < .001.

33 To measure the seriousness of the charges against the defendant, we categorized each case by severity of the top arraignment charge: whether the case was charged as a violation, misdemeanor or felony. Differences between 1998 and 2001 in severity of top arraignment charge were statistically significant. The percentage of cases charged as misdemeanors was higher in 2001 (82%) than in 1998 (74%) (Table 3-5). This reflected an increase in the volume of cases charged as misdemeanors (about 300 more misdemeanor cases in 2001 than in 1998). This may suggest that the establishment of the DV court in 2000 encouraged victims, police and/or prosecutors to go forward with more misdemeanor cases. Fewer cases were charged as felonies in 2001 than in 1998. The percentage of cases charged as felonies dropped by 9 percentage points. Although we are examining only those cases ultimately disposed as misdemeanors or violations in Criminal Court, very few DV cases were sustained as felonies and disposed in Supreme Court. Only about 4% of the DV cases in Manhattan in 1998 and 2001 were disposed in Supreme Court (data not shown). To predict case outcomes, particularly the likelihood of conviction, it is important to know the strength of the evidence in the case. Unfortunately, such data are rarely available in analyses of case processing and were not available for this study. We were, however, able to identify two indirect proxy measures of the strength of evidence. The number of arrest charges (discussed above) may be an indirect measure of strength of evidence. To evaluate the strength of the evidence presented by the police to the DA, we also examined the change in charge severity between the top arrest charge and the top Criminal Court arraignment charge. Charge severity was measured on a scale ranging from 1 (violation) to 7 (B felony). If the severity of the top arraignment charge was lower than the severity of the top arrest charge (e.g., a D felony charge at arraignment vs. a C felony charge at arrest, a B misdemeanor charge at arraignment vs. an A misdemeanor charge at arraignment, etc.), then we characterized the case as having had a reduction in charge severity. If, on the other hand, the severity of the top arraignment charge was higher than the severity of the top arrest charge (e.g., an E felony charge at arraignment vs. an A misdemeanor charge at arrest, a B felony charge at arraignment vs. C felony charge at arrest, etc.), we characterized the case as having had an increase in charge severity. There were few differences between 1998 and 2001 in the rate of charge reduction or charge increase in Manhattan DV cases. Approximately one sixth of all cases had a charge reduction between arrest and arraignment, and about 5% had a charge increase (see Table 3-5). Case Processing Characteristics Next, we consider several case processing variables, beginning with the CJA release recommendation (see Table 3-6). CJA interviews about 96% of the defendants held for arraignment in New York City. Using information from the interviews, CJA assesses the strength of the defendantâ&#x20AC;&#x2122;s New York area community ties and provides a release recommendation to the court at the time of arraignment. Since defendants with strong community ties may be more likely than those with weak ties to gain a favorable

34 TABLE 3-6 CASE PROCESSING CHARACTERISTICS BY YEAR FOR DOMESTIC VIOLENCE CASES IN MANHATTAN Crimes Against Persons and Property Subsample CASE PROCESSING CHARACTERISTICS RELEASE RECOMMENDATION ** Recommended or qualified recommendation No recommendation: Weak NYC area ties Bench Warrant attached to NYSID Other or missing Total, all cases (N of cases) CASE DISPOSED AT ARRAIGNMENT Case not disposed at Arraignment Case disposed at Arraignment Total, all cases (N of cases) DEFENDANT EVER RELEASED2 Never released Released Total, all cases (N of cases)

3rd Quarter 1998

1st Quarter 2001

58%

59%

100%1 (990)

100% (1,249)

99%

100% (990)

100% (1,249)

10% 90

100% (977)

100% (1,237)

Table continues on next page

35 TABLE 3-6, Continued CASE PROCESSING CHARACTERISTICS BY YEAR FOR DOMESTIC VIOLENCE CASES IN MANHATTAN Crimes Against Persons and Property Subsample CASE PROCESSING CHARACTERISTICS CHARGE SEVERITY REDUCED BETWEEN ARRAIGNMENT AND CONVICTION3 ** Charge severity not reduced Charge severity reduced Total, all cases (N of cases) SEVERITY OF CONVICTION CHARGE3 *** A Misdemeanor B Misdemeanor Violation Total, all cases (N of cases) MEAN NUMBER OF WEEKS FROM ARRAIGNMENT TO DISPOSITION *** (N of cases)

3rd Quarter 1998

1st Quarter 2001

31%

22%

100% (283)

100% (360)

47% 8 45

34% 4 61

100% (283)

100%1 (360)

(990)

(1,249)

Percentages do not sum to 100% due to rounding error. Data presented only for cases that were not disposed at arraignment. 3 Data presented only for cases that resulted in conviction. 2

Differences between 1st Quarter 2001 and 3rd Quarter 1998 were statistically significant at p < .01.

***

Differences between 1st Quarter 2001 and 3rd Quarter 1998 were statistically significant at p < .001.

36 case disposition, we included CJA’s release recommendation in the analyses. The release recommendation was coded in 4 categories: 1) recommended or qualified recommendation, 2) not recommended due to weak New York City area community ties, 3) no recommendation due to outstanding bench warrant and 4) other or missing. The “other” category includes defendants who were charged with bail jumping, whose NYSID’s were unavailable, whose interviews were incomplete or conducted for information only (i.e., charged with homicide or attempted homicide), or who were charged as juvenile offenders. The “missing” category includes defendants who were not interviewed by CJA.15 Defendants in DV cases were slightly more likely to have a bench warrant in 2001 than in 1998 (11% vs. 7%) and slightly less likely to receive no recommendation (25% vs. 30%) (See Table 3-6). These differences were statistically significant. DV cases were rarely disposed at arraignment (Table 3-6). As a general policy in Manhattan (and throughout New York City), ADA’s were not permitted to agree to dispositions at arraignment in DV cases.16 The policy is to gather more information on these cases and to keep them active as a means of preventing the defendant from committing further acts of domestic violence. Because so few DV cases were disposed at arraignment, we did not include this variable in our models predicting case outcomes. Among DV cases not disposed at arraignment (i.e., virtually all of them), about 90% of the cases led to release of the defendant prior to case disposition in Manhattan in both 1998 and 2001. For the models predicting likelihood of incarceration and length of jail sentence, we examined the effect of additional case processing variables related to the conviction. As reported in Table 3-6, when cases resulted in conviction, DV cases in 2001 were much more likely to have the severity of charges reduced between arraignment and conviction (78%) than in 1998 (69%). This difference was statistically significant. Not surprisingly, the severity of the final conviction charge was significantly lower for DV cases in 2001 (34% disposed as A Misdemeanors) than in 1998 (47% disposed as A Misdemeanors). These changes suggest that the establishment of the DV court changed the dynamics of plea bargaining. To achieve a similar conviction rate in 2001, after the DV court was established, greater concessions were made during plea bargaining. In 2001, 61% of DV cases that ended in conviction were disposed as violations, compared to 45% in 1998. Our models also included a measure of case processing time. DV cases in 1998 reached a final disposition in 18 weeks, on average, from arraignment to final 15

For further information about these categories, the CJA interview and CJA’s release recommendation, see NYC Criminal Justice Agency (2000). 16 The few DV cases that we identified as disposed at arraignment may have been disposed at that time due to error (e.g., failure to realize the case was a DV case) or to special circumstances involved in the case. It is also possible that the court mistakenly identified these cases as DV cases (i.e., gave the cases a DV hearing type at arraignment), when they were not in fact DV cases.

37 disposition. Case processing time was significantly shorter in 2001 (13 weeks, on average) (Table 3-6). This suggests that the specialized DV court was more efficient, reducing processing time for DV cases. C. Summary and Discussion of Findings Our review of case outcomes, re-arrest rates and defendant and case characteristics revealed a number of differences between 1998 and 2001. Some of these differences were apparently related to changes brought about by the establishment of the specialized DV court in 2000. Other differences were apparently related to changes in the volume and type of DV cases. The volume of DV cases in Manhattan was 26% higher in the first quarter of 2001 than in the third quarter of 1998. While some of this increase may be due to increases in the rate at which offenses were committed, this seems unlikely. DV complaints (which often do not lead to an arrest) declined by about 9% citywide between 1998 and 2001 (NYPD 1998, 2001). Most of the increase in DV case volume was probably due to better efforts to identify and track DV arrests and to monitor re-offending by DV defendants. These efforts were made as a direct result of the establishment of the specialized DV court part. As we expected, the distribution of case dispositions in DV cases in Manhattan was the same in 1998 and 2001. The conviction rate was 29% in both years. About 55% of DV cases were dismissed in 1998 and 2001. This finding is consistent with our previous research showing that the use of specialized DV courts in Brooklyn and the Bronx in 1998 did not result in similar case dispositions in those boroughs. There was no clear distinction between Manhattan, which had no specialized DV court in 1998, and the other two boroughs, which did (Peterson 2002). These results suggest that case dispositions were not affected by the establishment of the specialized DV court. Case dispositions are affected primarily by the strength of evidence in the case. Although DV courts are designed to improve the processing of DV cases, they are not likely to affect the strength of evidence in these cases. Among comparable Non-DV cases, case dispositions did change between 1998 and 2001. There were more convictions and fewer dismissals in Non-DV cases in 2001 than in 1998. Conviction rates were considerably higher in Non-DV cases than in DV cases. About two thirds of Non-DV cases in Manhattan ended in conviction, compared to less than one third in DV cases. This finding is consistent with our previous studies of citywide and borough differences between DV and Non-DV cases (Peterson 2001, 2002). Conviction rates are lower in DV cases because victims are less likely to cooperate with the prosecution than victims in Non-DV cases. Without victim cooperation, the remaining evidence in many DV cases is often insufficient to obtain a conviction. The DA’s office in Manhattan actively prosecuted many DV cases without victim cooperation. Sometimes an “evidence-based” prosecution was possible, relying on “hearsay exceptions” (e.g., statements made by victims and recorded on 911 tapes), photographs, police testimony, medical evidence, etc. While the case was pending, ADA’s tried to develop additional evidence, to encourage the victim to cooperate, and to provide services to the victim (e.g., counseling, housing assistance). In spite of their

38 efforts, the evidence in many of these DV cases was weak, and the DA’s office was often unable to obtain a conviction. Also as we expected, sentence outcomes for convicted defendants in DV cases were affected by the introduction of the specialized DV court in Manhattan. Convicted DV defendants were slightly less likely to receive a jail sentence in 2001 (27%) than in 1998 (31%). Conditional discharges (often with a requirement that the defendant complete a batterer intervention program) were much more common in 2001 than in 1998 (67% vs. 54%). This pattern reflects the greater emphasis on the use of batterer intervention programs in the specialized DV court in 2001 than in the mixed-docket courts in 1998. Our interviews with ADA’s in Manhattan suggest that this change occurred primarily because the court, not the DA’s office, changed its approach. According to the ADA’s, the Manhattan DA’s office continued to seek jail sentences in DV cases as often in 2001, after the DV court was established, as it did in 1998. Among comparable Non-DV cases, there was virtually no change in sentence outcomes between 1998 and 2001. This suggests that the changes observed for DV cases did not reflect general trends for comparable cases. These findings are consistent with our previous analysis of cross-borough differences in 1998. The boroughs with specialized DV courts (Brooklyn and the Bronx) were less likely to impose incarceration as a sentence in DV cases than Manhattan, where DV cases were heard in mixed-docket courts in 1998. While we do not have data on the use of batterer intervention programs, our observations in the Criminal Courts and conversations with DA’s and judges support this explanation. In many DV cases in the specialized courts, DA’s and/or the court may view a conditional discharge with a program requirement as a more appropriate sentence than jail. Once assigned to a program, defendants’ cases are transferred to a Domestic Violence Compliance Part, where a judicial hearing officer (JHO) monitors their progress. The JHO receives reports on defendants’ compliance with the programs. Defendants who fail to comply can have their cases sent back to the specialized DV part for re-sentencing. (See Puffett and Gavin (2004) for a detailed description of program monitoring in a similar DV court in the Bronx.) Among those sentenced to jail, sentences were shorter, on average, in 2001 than in 1998 (48 versus 65 days, respectively). However, this difference was not statistically significant. Furthermore, our earlier research comparing boroughs with and without specialized DV courts found that DV courts did not affect sentence length. Since the number of defendants sentenced to jail is relatively small, caution should be exercised when interpreting these results. The findings on sentence length for DV cases are not strong enough to support our prediction that jail sentences would be shorter after the DV court was established. In comparable Non-DV cases, jail sentences were shorter in 2001 than in 1998. This suggests that any trend toward shorter sentences in DV cases may reflect general trends toward shorter sentences.

39 As we expected, the re-arrest rate for DV offenses was higher after the establishment of the specialized DV court in Manhattan. In the 18 months after case disposition, the re-arrest rate for new DV offenses was 12% in 1998 and 16% in 2001. The re-arrest rate increased for two reasons. First, the establishment of the specialized DV court led to greater efforts to identify DV cases to be processed in the new court. In 1998, as noted earlier, the courts were less successful at identifying DV cases accurately. Second, the re-arrest rate increased as a result of closer monitoring of DV offenders for new DV offenses. In 2001, post-disposition judicial monitoring in a compliance part increased a defendant’s likelihood of being re-arrested for a new DV offense. The compliance part received reports from batterer intervention programs about new DV offenses and violations of program conditions, some of which resulted in the defendant’s re-arrest. Because this type of judicial monitoring was not available in 1998, there were not as many re-arrests for new DV offenses. These changes associated with the specialized DV court preclude us from determining whether the establishment of the court deterred re-offending among defendants in DV cases. The characteristics of defendants in DV cases were very similar in 1998 and 2001. There were no statistically significant differences in terms of demographic characteristics or criminal history. There were fewer defendants in 2001 than in 1998 who were married or in common-law relationships. However, there also was an increase in the percentage of cases where information about the relationship was missing. These changes may simply have reflected a decline in reporting the relationship, not real changes in the types of relationships. There were several significant differences in case characteristics between 1998 and 2001. The percentage of harassment and “other” arraignment charges was higher, while the percentage of assaults was lower, in 2001 than in 1998. There was also a decline in severity of arraignment charges between 1998 and 2001. The percentage of felony arraignment charges was lower in 2001 than in 1998. The establishment of the specialized DV court in 2001 may have encouraged victims, police and/or prosecutors to go forward with more misdemeanor cases. Among cases that ended in conviction, charge reduction between arraignment and conviction was common in both 1998 and 2001. However, there were significantly more cases where charges were reduced in 2001 (78%) than in 1998 (69%). Not surprisingly, the severity of the top conviction charge was much lower in 2001 than in 1998. About 61% of DV convictions were disposed as violations in 2001, compared to only 45% in 1998. These findings suggest that ADA’s or the court were more willing to reduce charges in 2001 than in 1998. In spite of this greater flexibility in plea bargaining, the conviction rate was the same in both years. This indicates that after the specialized DV court was established, greater plea bargaining concessions were needed to maintain the same conviction rate. One final change in case processing between 1998 and 2001 is worth noting. The number of weeks between arraignment and disposition declined significantly, from 18 weeks to 13 weeks. The establishment of the specialized DV court appears to have made the processing of DV cases more efficient.

[This page intentionally left blank]

41 IV. MODELS PREDICTING LIKELIHOOD OF CONVICTION A. Comparison of Models for Third Quarter 1998 and First Quarter 2001 As noted earlier, the conviction rate for DV cases in Manhattan was about 29% in both 1998 and 2001. In this section of the report, we examine whether the factors that affected the likelihood of conviction were the same in both years. The models predicting the likelihood of conviction are similar to those used in our previous research reports. (For details on how those models were developed, see Peterson 2001, 2002). Specifically, we used logistic regression models to control for defendant and case characteristics that we believe might affect the likelihood of conviction (see Appendix A, Section 1 for a description of logistic regression analysis). We first analyzed separate models for each year. We then evaluated whether the factors predicting likelihood of conviction differed by year by testing for the statistical significance of the interaction of each of the independent variables with year (see Appendix A, Section 2 for a description of interaction tests). In 1998, the statistically significant predictors of the likelihood of conviction were whether the defendant was ever released (if the case was not disposed at arraignment), whether the defendant was arraigned on felony charges, and whether the defendant was female (all of which reduced the likelihood of conviction) and whether the DA increased the severity of charges between arrest and arraignment (which increased the likelihood of conviction). None of the other variables in the model had a statistically significant impact on the likelihood of conviction in 1998. Overall, the model explained 26% of the variation in likelihood of conviction (see Nagelkerke R2 in Table 4-1). In 2001, the statistically significant predictors of the likelihood of conviction included two of the predictors that were statistically significant in 1998: whether the defendant was female, and whether the defendant was ever released. Each of these variables reduced the likelihood of conviction in 2001 as it did in 1998. In 2001, the following variables were also statistically significant predictors of the likelihood of conviction: the number of weeks from arraignment to disposition (which decreased the likelihood of conviction), the number of arrest charges (which increased the likelihood of conviction), whether the defendant had prior arrests (which increased the likelihood of conviction) and whether the defendant was between the ages of 30 and 39 (which increased the likelihood of conviction relative to defendants who were aged 16-20). Surprisingly, defendants who had more prior felony convictions were less likely to be convicted. Overall, the model explained 19% of the variation in the likelihood of conviction. An examination of the interaction effects showed that 4 variables had significantly different effects in 1998 and in 2001 (see Table 4-1, column 5). First, the effect of being released was much weaker in 2001 than in 1998. In 1998, the odds of conviction for a defendant who was released were 20 times lower (1 / .05) than for a defendant who was never released. In 2001, the odds of conviction for a defendant who was released

42 TABLE 4-1 LOGISTIC REGRESSION MODEL PREDICTING LIKELIHOOD OF CONVICTION FOR DOMESTIC VIOLENCE CASES IN MANHATTAN 1 CRIMES AGAINST PERSONS AND PROPERTY SUBSAMPLE 3rd Quarter 1998 DV Cases 2

INDEPENDENT VARIABLES

1st Quarter 2001 DV Cases

Standardized β

Odds Ratio

Standardized β

Odds Ratio

Significance

0.05 0.09 -0.02 -0.01

1.15 1.52 0.90 0.94

-0.07 0.00 0.01 0.03

0.85 1.00 1.05 1.07

ns ns ns ns

-0.21 *

0.55

-0.32 ***

0.49

0.05 0.08 0.01

1.19 1.19 1.06

0.01 0.17 0.15

1.03 1.35 1.86

ns ns ns

-0.24 0.02 -0.08

0.57 1.04 0.83

0.22 0.33 * 0.15

1.53 1.82 1.34

+ ns ns

-0.06 0.02 0.11 -0.01

0.83 1.06 1.43 0.99

-0.18 -0.06 -0.09 -0.16

0.65 0.87 0.81 0.75

ns ns ns ns

0.05 0.00 -0.03

1.11 1.00 0.97

0.22 * 0.09 -0.21 *

1.48 1.02 0.82

ns ns ns

of Interaction

CONTROL VARIABLE ARRAIGNMENT CHARGE PENAL LAW ARTICLE:

Reference Category: Assault (PL 120) Criminal Contempt (PL 215) Harassment (PL 240) Crimes Against Children (PL 260) Other DEFENDANT'S DEMOGRAPHIC CHARACTERISTICS SEX (Female) ETHNICITY:

Reference Category: Black White Hispanic Other AGE:

Reference Category: Age 16-20 Age 21-29 Age 30-39 Age 40 and over DEFENDANT-VICTIM RELATIONSHIP:

Reference Category: Married Boyfriend-Girlfriend Common-Law Spouse Other Family Missing DEFENDANT'S CRIMINAL HISTORY ANY PRIOR ARRESTS NUMBER OF PRIOR MISDEMEANOR CONVICTIONS NUMBER OF PRIOR FELONY CONVICTIONS

Page 1 of 2

TABLE 4-1 (continued)

3rd Quarter 1998 DV Cases 2

INDEPENDENT VARIABLES

Standardized β

Odds Ratio

1st Quarter 2001 DV Cases Standardized β

Odds Ratio

Significance 3

of Interaction

ARREST AND ARRAIGNMENT CHARGE CHARACTERISTICS NUMBER OF ARREST CHARGES ARRAIGNMENT CHARGE IS A FELONY

-0.01 -0.25 **

0.99 0.54

-0.12 0.22 **

0.71 2.97

-0.11 -0.08 -0.03 -0.88 *** 0.09

0.78 0.71 0.87 0.05 1.01

0.18 * 0.08

1.22 1.21

ns ++

-0.08 0.01

0.82 1.05

ns ns

-0.13 -0.04 -0.12 -0.66 *** -0.22 *

0.79 0.90 0.61 0.16 0.98

ns ns ns ++ +

CHANGE IN CHARGE SEVERITY FROM ARREST TO ARRAIGNMENT:

Reference Category: No Change Charge Severity Reduced from Arrest to Arraignment Charge Severity Increased from Arrest to Arraignment CASE PROCESSING CHARACTERISTICS RELEASE RECOMMENDATION:

Reference Category: No Recommendation (Weak NYC Ties) Recommended or Qualified Recommendation Open Bench Warrant At Time of Arrest Missing DEFENDANT EVER RELEASED NUMBER OF WEEKS FROM ARRAIGNMENT TO DISPOSITION

Nagelkerke R2 (N of Cases)

.26 *** (990)

.19 *** (1,249)

NOTES 1

See text for a description of the datasets and the subsamples. See Appendix B for information about the measurement and coding of the variables. 3 This column indicates whether the interaction between the independent variable and year (3rd Quarter 1998 vs. 1st Quarter 2001) was statistically significant. See discussion in text and in Appendix A, Section 2. 2

* Statistically significant at p < .05 ** Statistically significant at p < .01 *** Statistically significant at p < .001 ns Interaction effect was not statistically significant at p < .05 + Interaction effect was statistically significant at p < .05 ++ Interaction effect was statistically significant at p < .01 +++ Interaction effect was statistically significant at p < .001

Page 2 of 2

44 were only 6 times lower (1 / .16). Second, the number of weeks from arraignment to disposition had a statistically significant negative effect on the likelihood of conviction in 2001, but not in 1998. Although the conviction rate was the same in both 1998 and 2001, recall that case-processing time was 5 weeks shorter in 2001 than in 1998 (13 weeks in 2001, vs. 18 weeks in 1998). This suggests that the establishment of the DV court routinized the processing of DV cases so that cases that were likely to end in a conviction could be disposed of earlier. Third, defendants who were arraigned on felony charges were less likely than other defendants to be convicted in 1998, but not in 2001. This indicates that cases arraigned on felony charges were much weaker than expected in 1998, and often failed to result in convictions. As reported in Table 3-5, only 17% of DV cases in Manhattan were charged as felonies in 2001, compared to 26% in 1998. In 2001, the DAâ&#x20AC;&#x2122;s office was apparently more selective in arraigning defendants on felony charges. As a result, defendants arraigned on felony charges were no less likely to be convicted than defendants arraigned on lesser charges in 2001. Finally, defendants aged 21-29 were more likely to be convicted than those aged 16-20 in 2001. In 1998, defendants aged 21-29 were less likely to be convicted than those aged 16-20. No explanation for this difference is apparent, so we do not discuss it further. Another interesting difference between 1998 and 2001 is that the 2001 model explained less of the variation in the likelihood of conviction than the 1998 model (19% vs. 26%). This decline in the explanatory power of the models occurred despite the fact that there were only a few statistically significant changes in defendant and case characteristics between 1998 and 2001. Taken together, these results suggest that the establishment of the DV court led to more individualized attention to each DV case in 2001. In 1998, DV cases were apparently viewed as more similar to each other, and the factors affecting the likelihood of conviction had larger effects. In 2001, the defendant and case characteristics examined here had less influence. Unmeasured or unique characteristics of each case played a greater role in 2001. In spite of the differences between 1998 and 2001 noted above, it is important to recognize that there were also many similarities in the models for each year. Type of arraignment charge, ethnicity, defendant-victim relationship and the CJA release recommendation had no influence on the likelihood of conviction in either year. B. Summary and Discussion of Findings Although the conviction rate for DV cases in Manhattan was the same in 1998 and 2001, there were differences between the two years in the factors that influenced the likelihood of conviction. The models predicting likelihood of conviction revealed two interesting differences between 1998 and 2001. First, cases that took longer to reach a disposition were less likely to end in conviction in 2001, but not in 1998. In 2001, when most cases were heard in the specialized DV court, cases that would end in conviction were identified and disposed earlier. Why? In the specialized DV court, it may be easier for all parties to agree on an appropriate sentence if the defendant pleads guilty. This may make it easier to reach a

45 plea bargain. Also, as noted in Section III above, it appears that the court and/or the DA’s office were willing to make greater concessions in plea bargaining in 2001 than in 1998. About 61% of the convictions in 2001 were disposed with a violation as the top charge, compared to only 45% in 1998. Defendants were more willing to plead guilty to violations since such convictions can often be removed from their criminal records after one year. In 1998, when all DV cases were heard in mixed-docket courts, it may have been more difficult to agree on a sentence or to identify an appropriate batterer intervention program or drug or alcohol treatment program. Furthermore, the courts and the DA’s office were apparently less willing to reduce charges in exchange for a guilty plea in 1998. As a result, it took longer, on average, to obtain convictions in 1998 than in 2001. Finally, the Manhattan DA’s office increased its efforts to try DV cases as quickly as possible to avoid evidence problems that might develop if the cases continued for a long time. These case-processing differences between 1998 and 2001 may help to explain why case-processing time was 5 weeks shorter in 2001 than in 1998. Second, the severity of arraignment charge affected the likelihood of conviction in 1998 but not in 2001. Defendants arraigned on felony charges in 1998 were less likely to be convicted than defendants arraigned on lesser charges. The evidence in these felony cases was often not sufficient to sustain a conviction. In 2001, the DA’s office was less likely to arraign defendants on felony charges in DV cases. This reflected both an increase in the number of misdemeanor cases and a decline in the number of felony cases (data not shown). As a result of this greater selectivity, defendants arraigned on felony charges in 2001 were no less likely to be convicted than defendants arraigned on lesser charges. Taken together, these differences between 1998 and 2001 suggest that the establishment of the specialized DV court affected the process by which dispositions were reached. There were more DV cases charged as misdemeanors in 2001, and there was more flexibility in plea bargaining. Case processing time was shorter in 2001, in part because the court was more efficient at reaching convictions. The greater emphasis on placing defendants in batterer intervention programs as well as the routinization of information about programs may have contributed to the increased efficiency.

[This page intentionally left blank]

47 V. MODELS PREDICTING LIKELIHOOD OF INCARCERATION We next turn our attention to models predicting the likelihood of incarceration. As discussed in our overview of case outcomes, among cases that lead to a conviction, the percentage of defendants in DV cases in Manhattan who were sentenced to jail was lower in 2001 than in 1998 (27% vs. 31%). Our analytic strategy parallels the one we used to examine the likelihood of conviction in the previous section. We begin by examining separate models for 1998 and 2001. We then examine whether the predictors of likelihood of incarceration were different in 1998 and 2001. Since our dependent variable was measured in two categories (sentenced to jail vs. not sentenced to jail), we used logistic regression models (see Appendix A, section 1 for a description of this statistical technique). Because our analyses were based only on those cases that ended in a conviction, we were able to use two additional case processing variables as predictors in our models. First, we included a measure of whether the charge severity was reduced between the arraignment charge and the conviction charge. We expected that charge reduction would be associated with a lower likelihood of incarceration. Charge reduction is more likely in cases where the evidence is weaker, and defendants may be induced to plead guilty in return for a promise of a non-incarcerative sentence. Second, we also included the severity of the conviction charge in the modelâ&#x20AC;&#x201D;we expected cases disposed as violations or as B misdemeanors to be less likely to result in a jail sentence than cases disposed as A misdemeanors. Finally, our models predicting the likelihood of incarceration included a correction for selection bias to improve the accuracy of the statistical results (see Appendix A, section 3A for a discussion of selection bias). A. Comparison of Models for Third Quarter 1998 and First Quarter 2001 The logistic regression models predicting likelihood of incarceration for 1998 and 2001 are presented in Table 5-1. In 1998, the strongest predictors of incarceration in the model (as measured by standardized beta) were whether the case was disposed as a violation, whether the defendant was ever released (both of which reduced the likelihood of incarceration), whether the defendant had a prior adult arrest record, the number of arrest charges and whether the defendantâ&#x20AC;&#x2122;s release recommendation was missing (all of which increased the likelihood of incarceration). Overall, the model explained 66% of the variation in the likelihood of incarceration (see Nagelkerke R2 in Table 5-1). This indicates that the model was relatively successful in identifying factors that affected the likelihood of an incarcerative sentence for convicted defendants. In 2001, as in 1998, cases disposed as violations were less likely to result in incarceration. In both years, convicted defendants who were ever released were less likely to receive jail sentences than those who were never released. Surprisingly, the number of arrest charges had a statistically significant negative effect on the likelihood of incarceration in 2001. Two variables affected the likelihood of incarceration in 2001

TABLE 5-1 LOGISTIC REGRESSION MODEL PREDICTING LIKELIHOOD OF INCARCERATION FOR CONVICTED DOMESTIC VIOLENCE DEFENDANTS IN MANHATTAN 1 CRIMES AGAINST PERSONS AND PROPERTY SUBSAMPLE 3rd Quarter 1998 DV Cases INDEPENDENT VARIABLES

1st Quarter 2001 DV Cases

Standardized β

Odds Ratio

Standardized β

Odds Ratio

Significance

-0.38

0.00

0.10

4.29

-0.08 0.06 0.04 -0.04

0.50 2.18 2.37 0.64

0.06 -0.08 -0.25 -0.01

1.79 0.38 0.00 0.91

ns ns ns ns

-0.03

0.72

-0.19 *

0.13

-0.01 0.12 -0.05

0.86 2.34 0.45

0.01 -0.12 -0.01

1.07 0.44 0.79

ns + ns

-0.02 0.02 0.06

0.87 1.13 1.58

-0.12 -0.05 -0.10

0.40 0.70 0.47

ns ns ns

-0.01 0.02 -0.07 0.05

0.92 1.15 0.51 1.46

-0.02 0.14 0.04 -0.01

0.80 3.09 1.45 0.95

ns ns ns ns

9.58 1.08 1.02

0.14 0.26 ** -0.07

2.80 1.21 0.79

ns ns ns

of Interaction

CONTROL VARIABLE SELECTION BIAS CORRECTION: LIKELIHOOD OF CONVICTION ARRAIGNMENT CHARGE PENAL LAW ARTICLE:

Reference Category: Assault (PL 120) Criminal Contempt (PL 215 Harassment (PL 240) Crimes Against Children (PL 260 Other DEFENDANT'S DEMOGRAPHIC CHARACTERISTICS SEX (Female) ETHNICITY:

Reference Category: Black White Hispanic Other AGE:

Reference Category: Age 16-20 Age 21-29 Age 30-39 Age 40 and over DEFENDANT-VICTIM RELATIONSHIP:

0.31 * 0.09 0.01

Page 1 of 2

TABLE 5-1 (continued)

3rd Quarter 1998 DV Cases INDEPENDENT VARIABLES

Standardized β

1st Quarter 2001 DV Cases

Odds Ratio

Standardized β

Odds Ratio

Significance

0.21 * -0.01

2.34 0.90

-0.16 * -0.09

0.55 0.47

+++ ns

0.10 0.12

2.75 4.13

-0.12 -0.02

0.31 0.76

+ ns

-0.05 0.04 0.26 ** -0.49 **

0.70 1.62 40.48 0.03

-0.11 0.01 0.12 -0.25 *

0.48 1.09 6.95 0.16

ns ns ns ns

-0.07

0.59

-0.08

0.53

-0.51 ** 0.02 -0.08

0.03 1.32 0.98

-0.29 * -0.06 0.15

0.14 0.39 1.05

ns ns ns

of Interaction

ARREST AND ARRAIGNMENT CHARGE CHARACTERISTICS NUMBER OF ARREST CHARGES ARRAIGNMENT CHARGE IS A FELONY CHANGE IN CHARGE SEVERITY FROM ARREST TO ARRAIGNMENT:

Reference Category: No Change Charge Severity Reduced from Arrest to Arraignmen Charge Severity Increased from Arrest to Arraignmen CASE PROCESSING CHARACTERISTICS RELEASE RECOMMENDATION:

Reference Category: No Recommendation (Weak NYC Ties) Recommended or Qualified Recommendatio Open Bench Warrant At Time of Arres Missing DEFENDANT EVER RELEASED CHARGE SEVERITY REDUCED BETWEEN ARRAIGNMENT AND DISPOSITION SEVERITY OF DISPOSITION CHARGE:

Reference Category: Case Disposed as an A Misdemeanor Case Disposed as a Violation Case Disposed as a B Misdemeano NUMBER OF WEEKS FROM ARRAIGNMENT TO CONVICTION

Nagelkerke R2 (N of Cases)

.66 *** (283)

.66 *** (360)

NOTES 1

Page 2 of 2

50 but not in 1998. The number of prior misdemeanor convictions increased the likelihood of incarceration. Also, convicted female defendants were less likely to receive jail sentences than male defendants. Overall, the 2001 model explained 66% of the variation in the likelihood of incarceration, the same as in 1998. The results for the statistically significant interactions (column 5 of Table 5-1) show that three variables had significantly different effects in 1998 and 2001. First, the number of arrest charges increased the likelihood of incarceration in 1998. Each additional arrest charge more than doubled the odds of incarceration (odds ratio = 2.34) in 1998. In 2001, each additional arrest charge reduced the odds of incarceration by nearly half (odds ratio = .55). We had expected the number of arrest charges to increase the odds of incarceration, since cases with more arrest charges may have involved more serious incidents of domestic violence. The reduction in the odds of incarceration that we found in 2001 is therefore somewhat surprising. Second, the effect of having charge severity reduced between arrest and arraignment also differed between 1998 and 2001. Reductions in charge severity increased the odds of incarceration in 1998, but decreased the odds of incarceration in 2001. Finally, Hispanic defendants were more likely than Black defendants to be incarcerated in 1998, and less likely than Black defendants to be incarcerated in 2001. There also were several similarities in the models for 1998 and 2001: arraignment charge penal law article, age, defendant-victim relationship and number of weeks from arraignment to conviction had no impact on the likelihood of incarceration in either year, after taking other case and defendant characteristics into account. Furthermore, the models were equally effective in explaining the likelihood of incarceration: each explained two thirds of the variation in the likelihood of incarceration. B. Summary and Discussion of Findings We found several differences between 1998 and 2001 in the factors that influenced the likelihood of incarceration. In each case, the effect of a variable shifted from positive to negative between 1998 and 2001. First, number of arrest charges increased the odds of incarceration in 1998 but decreased them in 2001. Since we believe that number of arrest charges reflects the strength of evidence in the case, it is surprising that this variable reduced the odds of incarceration in 2001. We expected that jail sentences would be more likely in cases with more arrest charges, as we found in 1998. The difference between 1998 and 2001 may reflect the emphasis of the specialized DV court on mandating defendants to batterer intervention programs with a conditional discharge in 2001. In cases with stronger evidence (i.e., more arrest charges) the court may have focused primarily on obtaining a conviction with a program requirement in 2001. Second, reductions in charge severity between arrest and arraignment increased the odds of incarceration in 1998, but reduced them in 2001. Since a reduction in charge severity between arrest and arraignment may indicate that the evidence in the

51 case is weak, it is surprising that this variable increased the odds of incarceration in 1998. We expected that jail sentences would be less likely in cases where charge severity was reduced, as we found in 2001. This difference between 1998 and 2001 is probably not attributable to the operation of the specialized DV court in 2001. Finally, Hispanic defendants were more likely than Black defendants to be incarcerated in 1998 and less likely than Black defendants to be incarcerated in 2001. We had no reason to expect differences in the likelihood of incarceration by ethnicity and the difference we found may be due to random error. For these reasons, we do not discuss this finding in further detail.

[This page intentionally left blank]

53 VI. MODELS PREDICTING LENGTH OF JAIL SENTENCE The final case outcome we examined in this study was the length of jail sentence. Here, the analyses were restricted to those cases where the defendant was convicted and sentenced to jail (including both “time served” and definite sentences). As reported in Section III, the average length of jail sentence in DV cases in Manhattan was 65 days in 1998 and 48 days in 2001. In this section of the report, we develop statistical models predicting the length of the jail sentence. We continue to follow the analytic strategy used in the analyses of conviction and incarceration. The models we present include an additional correction for selection bias that was not used previously: a correction for the probability of incarceration (see Appendix A, Section 3B for a discussion). Our models predicting length of jail sentence also controlled for the severity of the conviction charge. This is important because severity of conviction charge places statutory limits on the length of the jail sentence that may be imposed (New York State Penal Law §70.15). A defendant can be sentenced to a maximum term of 15 days for a violation, 90 days for a B misdemeanor and 365 days for an A misdemeanor. A. Comparison of Models for Third Quarter 1998 and First Quarter 2001 We began by examining the predictors of sentence length separately for 1998 and 2001 (Table 6-1). Because the dependent variable, number of days sentenced to jail, is a continuous variable, we used linear regression analysis (see Appendix A, Section 4 for a description of this statistical technique.) In 1998, as expected, the statutory limits associated with the severity of the conviction charge were strong predictors of the length of jail sentence. Defendants convicted of violations or B misdemeanors received shorter sentences than those convicted of A misdemeanors. Other strong predictors of length of jail sentence, as measured by the standardized beta, were the selection bias correction for likelihood of incarceration and whether the defendant was recommended for release (which had a negative effect on length of jail sentence), and whether the defendant had a prior adult arrest record, the number of prior misdemeanor convictions, whether charges were reduced between arrest and arraignment, whether the defendant and victim had a nonintimate (“other”) family relationship, whether the defendant-victim relationship was missing, and whether the defendant was age 30-39 (all of which increased the length of the jail sentence). Overall, the model explained 64% of the variation in length of the jail sentence (see R2 statistic in Table 6-1). In 2001, there were fewer statistically significant predictors of sentence length. Two of the statistically significant predictors in 1998 were also statistically significant in 2001: whether the defendant and victim had a non-intimate family relationship, and the number of prior misdemeanor convictions. In 2001, Hispanic defendants received longer sentences than Black defendants and women received shorter sentences than men. Defendants who were not recommended for release because of an outstanding

TABLE 6-1 REGRESSION MODEL PREDICTING LENGTH OF SENTENCE FOR CONVICTED DOMESTIC VIOLENCE DEFENDANTS SENTENCED TO JAIL IN MANHATTAN 1 CRIMES AGAINST PERSONS AND PROPERTY SUBSAMPLE 3rd Quarter 1998 DV Cases INDEPENDENT VARIABLES

Standardized β

1st Quarter 2001 DV Cases Standardized β

Significance of Interaction3

CONTROL VARIABLES SELECTION BIAS CORRECTION: LIKELIHOOD OF CONVICTION SELECTION BIAS CORRECTION: LIKELIHOOD OF INCARCERATION

-0.20 -0.59 *

-70.54 -169.97

-0.58 0.45

-171.16 98.06

ns ++

-0.07 -0.05 0.01 -0.21

-9.71 -11.44 1.87 -43.23

0.01 0.05 4 --0.17

1.03 12.85 4 --26.07

ns ns ns +

-0.18

-44.00

-0.20 *

-64.75

-0.01 0.04 0.03

-1.43 4.54 10.41

-0.08 0.21 * -0.03

-20.59 26.10 -13.80

ns ns ns

0.28 0.43 * 0.37

43.10 53.72 50.28

0.09 0.23 0.14

13.17 25.70 18.30

ns ns ns

-0.03 -0.14 0.31 * 0.28 *

-4.64 -21.82 61.62 36.90

0.10 0.13 0.22 * 0.11

15.92 15.93 40.14 13.69

ns ns ns ns

0.41 ** 0.26 * -0.04

88.14 2.62 -2.42

0.15 0.24 * -0.02

35.07 2.17 -1.05

ns ns ns

ARRAIGNMENT CHARGE PENAL LAW ARTICLE:

Reference Category: Assault (PL 120) Criminal Contempt (PL 215 Harassment (PL 240) Crimes Against Children (PL 260) Other DEFENDANT'S DEMOGRAPHIC CHARACTERISTICS SEX (Female) ETHNICITY:

Reference Category: Black White Hispanic Other AGE:

Reference Category: Age 16-20 Age 21-29 Age 30-39 Age 40 and over DEFENDANT-VICTIM RELATIONSHIP:

Page 1 of 2

TABLE 6-1 (continued)

3rd Quarter 1998 DV Cases INDEPENDENT VARIABLES

Standardized β

1st Quarter 2001 DV Cases Standardized β

Significance of Interaction3

6.93 -12.25

0.07 0.11

5.13 15.43

ns ns

0.25 * 0.10

41.82 15.53

0.15 -0.02

24.19 -5.10

ns ns

-0.45 ** -0.18 0.03 -0.42

-56.92 -33.66 5.56 -58.21

-0.15 -0.29 * 0.01 -0.37

-17.82 -36.16 1.68 -44.27

ns ns ns ns

47.49

0.51

57.80

-159.83 -66.67 1.30

-0.08 -0.16 0.13

-10.98 -35.06 0.85

+ ns ns

ARREST AND ARRAIGNMENT CHARGE CHARACTERISTICS NUMBER OF ARREST CHARGES ARRAIGNMENT CHARGE IS A FELONY CHANGE IN CHARGE SEVERITY FROM ARREST TO ARRAIGNMENT:

Reference Category: No Change Charge Severity Reduced from Arrest to Arraignment Charge Severity Increased from Arrest to Arraignment

0.10 -0.10

CASE PROCESSING CHARACTERISTICS RELEASE RECOMMENDATION:

Reference Category: No Recommendation (Weak NYC Ties) Recommended or Qualified Recommendation Open Bench Warrant At Time of Arrest Missing DEFENDANT EVER RELEASED CHARGE SEVERITY REDUCED BETWEEN ARRAIGNMENT AND DISPOSITION SEVERITY OF DISPOSITION CHARGE:

Reference Category: Case Disposed as an A Misdemeanor Case Disposed as a Violation Case Disposed as a B Misdemeanor NUMBER OF WEEKS FROM ARRAIGNMENT TO CONVICTION

Nagelkerke R (N of Cases)

0.39

-0.54 *** -0.37 * 0.19

.64 *** (87)

.63 *** (98)

NOTES 1

See text for a description of the datasets and the subsample See Appendix B for information about the measurement and coding of the variable 3 This column indicates whether the interaction between the independent variable and year (3rd Quarter 1998 vs. 1st Quarter 2001) was statistically significant. See discussion in text and in Appendix A, Section 4 This variable was removed from the analysis because there were no cases of Crimes Against Children among defendants sentenced to jail in 2001. 2

Page 2 of 2

56 bench warrant received shorter sentences. Overall, the model explained 63% of the variation in length of jail sentence, about the same as the 1998 model. There were three statistically significant interactions in Table 6-1. Cases disposed as violations received much shorter sentences in 1998 than cases disposed as A misdemeanors. Since the length of sentences is statutorily limited by the severity of the conviction charge (15 days for a violation, 90 days for a B misdemeanor, and 1 year for an A misdemeanor), this finding is not surprising. However, in 2001, the impact of severity of disposition charge was much weaker, and not statistically significant. An examination of sentence length by severity of disposition (data not shown) reveals that the length of sentences for A misdemeanors was considerably shorter in 2001 than in 1998 (56 vs. 72 days). As a result, the differences in sentence length by severity of disposition were smaller in 2001, accounting for the weaker effect of severity of disposition on sentence length. The other statistically significant differences were found in two control variables, and neither difference appears to reveal any insight about the operation of the DV court. The selection bias correction for the likelihood of incarceration had a negative effect on length of jail sentences in 1998, and a positive effect in 2001. Those arraigned on “other” charges (e.g., criminal mischief, robbery, burglary, sex offenses) received shorter sentences in 1998 and longer sentences in 2001. We do not discuss these differences further. A comparison of the 1998 and 2001 models shows that several variables had no impact on length of jail sentence in either year: number of arrest charges, whether the arraignment charge was a felony, whether the defendant was ever released, whether charge severity was reduced between arraignment and disposition, and case processing time. B. Summary and Discussion of Findings We found one interesting difference between 1998 and 2001 in the factors that influenced the length of jail sentence. Reflecting the statutory limits on sentences, defendants who were sentenced to jail for violations received shorter sentences than those sentenced to jail for A misdemeanors in 1998. In 2001, there was no statistically significant difference between the lengths of jail sentences for those convicted of violations and those convicted of A misdemeanors. This difference between 1998 and 2001 is primarily the result of shorter sentences for A misdemeanors in 2001 than in 1998. We believe this finding reflects the shift of the specialized DV court away from the use of longer jail sentences as a sanction in DV cases. It may also reflect the greater efficiency of the specialized DV court. Defendants who received “time served” sentences for convictions on A misdemeanors served shorter sentences in 2001 than in 1998. Since case-processing time was shorter in 2001 than 1998, the shorter sentences are attributable in part to the quicker processing of DV cases in the specialized DV court.

57 VII. MODELS PREDICTING LIKELIHOOD OF RE-ARREST We now turn our attention to models predicting the likelihood of re-arrest for DV offenses. As discussed in Section III above, the re-arrest rate for DV offenses was 16% in 2001, compared to 12% in 1998. The increase in the re-arrest rate is most likely due to improved identification of DV cases after the establishment of the specialized DV court in Manhattan in June 2000. In this section of the report, we focus on the factors that predict the likelihood of re-arrest. We begin by examining separate models for DV cases in each year. Since our dependent variable was measured in two categories (rearrested for a DV offense within 18 months following case disposition vs. not re-arrested for a DV offense within 18 months following case disposition), we used logistic regression models (see Section 1 of Appendix A for a description of this statistical technique). Our analyses of the likelihood of re-arrest use many of the independent variables used in the models of conviction, incarceration and sentence length. However, there are two key differences. First, we include several variables to measure the impact of case outcomes on the likelihood of re-arrest. Specifically, we categorize each case as to whether it ended in dismissal, an ACD, a conviction without a jail sentence, or a conviction with a jail sentence. (As noted earlier, most defendants who were convicted without a jail sentence received a conditional discharge). In previous research, analyzing citywide data for the third quarter of 1998, we found that case outcomes had no effect on re-arrest rates once other defendant and case characteristics were taken into account (Peterson 2003a). However, it is possible that case outcomes affected rearrest rates in Manhattan in 1998 and/or 2001 and we will test for this possibility in our models. Second, instead of using CJAâ&#x20AC;&#x2122;s release recommendation as a predictor of rearrest, we use the individual â&#x20AC;&#x153;community tiesâ&#x20AC;? measures on which the recommendation is based. Our previous research has found that defendants with weak community ties are more likely to be re-arrested for a DV offense (Peterson 2003a). A. Comparison of Models for Third Quarter 1998 and First Quarter 2001 Our logistic regression model assesses the factors that affected the likelihood that a defendant was re-arrested for a DV offense. As described earlier, our dependent variable is whether or not the defendant was re-arrested at least once for a DV offense in the 18 months following case disposition. The logistic regression models predicting likelihood of re-arrest in 1998 and 2001 are presented in Table 7-1. In 1998, there were only three statistically significant predictors of re-arrest. First, defendants who were convicted and sentenced to jail were less likely to be rearrested than those whose cases were dismissed. Defendants who received other case outcomes (ACD, conviction without a jail sentence) were no less likely to be re-arrested than those whose cases were dismissed. This pattern of findings suggests that in Manhattan in 1998, jail sentences may have deterred defendants from committing future

TABLE 7-1 LOGISTIC REGRESSION MODEL PREDICTING LIKELIHOOD OF RE-ARREST FOR A DOMESTIC VIOLENCE OFFENSE: DOMESTIC VIOLENCE OFFENDERS IN MANHATTAN 1 CRIMES AGAINST PERSONS AND PROPERTY SUBSAMPLE 3rd Quarter 1998 DV Cases INDEPENDENT VARIABLES

CASE OUTCOME Reference Category: Dismissed Adjourned in Contemplation of Dismissa Convicted, no Jail Sentence Convicted, with Jail Sentence

Standardized β

1st Quarter 2001 DV Cases

Odds Ratio

Standardized β

Odds Ratio

Significance

-0.03 -0.07 -0.29 *

0.92 0.84 0.36

-0.05 0.08 0.20 *

0.89 1.18 1.91

ns ns ++

0.18 0.03 0.02 0.04

1.64 1.11 1.13 1.15

0.03 -0.24 0.02 0.15

1.09 0.50 1.14 1.48

ns ns ns ns

-0.44 **

0.32

-0.27 *

0.55

-0.06 -0.03 -0.36

0.81 0.94 0.20

-0.15 0.01 -0.01

0.66 1.02 0.95

ns ns ns

0.00 0.04 -0.07

1.01 1.08 0.86

0.01 -0.17 -0.11

1.02 0.74 0.81

ns ns ns

0.20 0.23 0.01 0.17

1.70 1.71 1.03 1.44

2.02 1.84 1.33 1.28

ns ns ns ns

of Interaction

CONTROL VARIABLE ARRAIGNMENT CHARGE PENAL LAW ARTICLE:

Reference Category: Assault (PL 120) Criminal Contempt (PL 215 Harassment (PL 240) Crimes Against Children (PL 260 Other DEFENDANT'S DEMOGRAPHIC CHARACTERISTICS SEX (Female) ETHNICITY:

Reference Category: Black White Hispanic Other AGE:

Reference Category: Age 16-20 Age 21-29 Age 30-39 Age 40 and over DEFENDANT-VICTIM RELATIONSHIP:

Reference Category: Married Boyfriend-Girlfriend Common-Law Spouse Other Family Missing

0.30 * 0.28 * 0.12 0.14

Page 1 of 2

59 TABLE 7-1 (continued)

3rd Quarter 1998 DV Cases INDEPENDENT VARIABLES

1st Quarter 2001 DV Cases

Standardized β

Odds Ratio

Standardized β

Odds Ratio

0.15 0.01 0.09

1.34 1.00 1.10

0.25 * 0.16 -0.06

1.55 1.04 0.95

ns ns ns

0.27 **

2.61

3.69

0.08

1.28

-0.10

0.71

0.13 -0.05 0.15 0.17 -0.11 -0.21

1.29 0.90 1.33 1.42 0.78 0.38

-0.05 0.06 0.13 0.01 0.21 * -0.05

0.91 1.13 1.25 1.02 1.52 0.83

ns ns ns ns ns ns

-0.14 0.06

0.84 1.13

-0.15 0.12

0.86 1.31

ns ns

-0.01 0.04

0.97 1.20

0.11 0.07

1.31 1.38

ns ns

-0.10 -0.15

0.74 0.99

0.07 -0.27 *

1.23 0.97

ns ns

Significance 3

of Interaction

DEFENDANT'S CRIMINAL HISTORY ANY PRIOR ARRESTS NUMBER OF PRIOR MISDEMEANOR CONVICTIONS NUMBER OF PRIOR FELONY CONVICTIONS ANY ARRESTS FOR A DV OFFENSE PRIOR TO CASE DISPOSITION ANY ARRESTS FOR A NON-DV OFFENSE PRIOR TO CASE DISPOSITION

0.44 ***

DEFENDANT'S COMMUNITY TIES UNEMPLOYED AT CURRENT ADDRESS 1 YEAR OR LESS LIVES WITH SOMEONE DOES NOT EXPECT ANYONE AT ARRAIGNMENT HAS NO TELEPHONE LIVES OUTSIDE NYC AREA

ARREST AND ARRAIGNMENT CHARGE CHARACTERISTICS NUMBER OF ARREST CHARGES ARRAIGNMENT CHARGE IS A FELONY CHANGE IN CHARGE SEVERITY FROM ARREST TO ARRAIGNMENT:

Reference Category: No Change Charge Severity Reduced from Arrest to Arraignmen Charge Severity Increased from Arrest to Arraignmen CASE PROCESSING CHARACTERISTICS DEFENDANT EVER RELEASED NUMBER OF WEEKS FROM ARRAIGNMENT TO DISPOSITION

Nagelkerke R2 (N of Cases)

.12 *** (955)

.14 *** (1,188)

NOTES 1

Page 2 of 2

60 DV offenses. However, in previous research, including our own analyses of citywide data in 1998, jail has not been found to deter re-arrests (Peterson 2003a). Second, women were less likely than men to be re-arrested. This result is not surprising, since women are much less likely than men to be arrested for DV offenses (Peterson 2003a). Finally, defendants who were arrested for a DV offense prior to the disposition of their case in the third quarter of 1998 were more likely to be re-arrested after the case was disposed. These defendants are likely to be chronic offenders whose future behavior is not easily deterred by criminal justice interventions. In 2001, as in 1998, women were less likely to be re-arrested for a DV offense. Also, defendants who were re-arrested for a DV offense while the original case was pending were more likely to be re-arrested for a DV offense after the original case was disposed. Several variables were statistically significant in 2001 but not in 1998. The odds of re-arrest were twice as high for defendants whose victim was a girlfriend or boyfriend as for defendants whose victim was their spouse. Similarly, the odds of rearrest were twice as high for defendants whose victim was a common-law spouse as for defendants whose victim was their spouse. Because we do not have data on the identity of the victim, we cannot determine whether the victim in the re-arrest was the same boyfriend, girlfriend or common-law spouse, or a different one. In 2001, defendants who had prior adult arrests were more likely to be re-arrested than those who did not. Also, defendants who did not have a telephone were more likely to be rearrested. Finally, defendants whose cases took longer from arraignment to disposition were less likely to be re-arrested. Each additional week between arraignment and disposition reduced the odds of re-arrest by 3% (see Table 7-1). Although there were several variables that had statistically significant effects in 2001, but not in 1998, statistical tests of the difference in their effects (in the significance of interaction column) show that the effects of most defendant and case characteristics on likelihood of re-arrest were the same in 1998 and 2001. However, there was one statistically significant difference between the two years. Interestingly, while defendants sentenced to jail in 1998 were less likely to be re-arrested than those whose cases were dismissed, defendants sentenced to jail in 2001 were more likely to be re-arrested. About 9% of 78 defendants sentenced to jail in 1998 were re-arrested, compared to 32% of 90 defendants in 2001 (data not shown). One possible explanation for this finding is that jail sentences meted out in mixed-docket courts in 1998 were more effective at deterring re-arrest than jail sentences meted out in the specialized court in 2001. An alternative explanation is that the difference, although statistically significant, is based on relatively small samples and may be an aberration. A third explanation is that the rate of identification of DV re-arrests was much less accurate in 1998 than in 2001. The rate of re-arrest for any offense (DV or Non-DV) for defendants sentenced to jail was slightly higher in 1998 than in 2001 (58% vs. 53%, data not shown). This pattern suggests that the third explanation is the most credible. If jail sentences had a deterrent effect in 1998, they presumably would have deterred re-arrests for Non-DV as well as DV offenses.

61 Many variables in the model had no statistically significant effect on the rate of re-arrest in both 1998 and 2001. Arraignment charge, ethnicity, age, all but one of the community ties variables and all the arrest and arraignment charge characteristics had no detectable impact on the re-arrest rate. As a result, the models explained relatively little of the variation in likelihood of re-arrest: 12% in 1998 and 14% in 2001 (see Nagelkerke R2 in Table 7-1). Stated another way, 86% to 88% of the variation in likelihood of re-arrest cannot be accounted for by the variables included in our models. This indicates that the models are relatively ineffective in identifying the factors that influence the likelihood of re-arrest for a DV offense. While we have discussed the influence of many variables that had a statistically significant impact on re-arrest, the combined effects of these variables were relatively small. Why is it so difficult to predict recidivism among DV offenders? First, re-arrest for a DV offense is relatively rare, occurring for only 12% to 16% of DV offenders in these datasets. It may be more difficult to develop a strong predictive model for a rare event. Second, the measures of recidivism may be unreliable, and subject to significant measurement error. It is also possible that the predictors of re-arrest included in the models are poorly measured. Measurement error reduces the statistical power of predictive models. More refined measures of criminal history, community ties or the defendant-victim relationship might improve the predictive power of the model. Third, factors not currently included in the model may play a larger role in predicting recidivism. For example, psychological problems (e.g., borderline personality disorder, anti-social personality), being physically, sexually or emotionally abused as a child, or other factors may influence the likelihood of recidivism. Finally, random events or unique situational factors may trigger new DV offenses. B. Summary and Discussion of Findings There are several noteworthy findings in our models predicting the likelihood of re-arrest. First, we found one statistically significant difference between 1998 and 2001 in the factors that influenced the likelihood of re-arrest. The odds of re-arrest for defendants sentenced to jail were lower than for defendants whose cases were dismissed in 1998. In 2001, defendants sentenced to jail had higher odds of re-arrest than those whose cases were dismissed. Although we considered drawing the conclusion that jail sentences had a deterrent effect in 1998 but not in 2001, this seems unlikely. The most likely explanation is that the courts improved their efforts at identifying DV re-arrests in 2001. Specifically, greater efforts may have been made to identify DV cases at arraignment. Also, the courts monitored DV offenders more closely in 2001 than in 1998. Second, very few variables were strong predictors of the likelihood of re-arrest. Women were less likely to be re-arrested than men in both years. Defendants who were re-arrested for a DV offense while their original case was still pending were more likely to be re-arrested after the original case was disposed. Overall, our statistical models were relatively unsuccessful at predicting the likelihood that a DV offender would be re-arrested for a DV offense. We accounted for 12% of the variation in

62 likelihood of re-arrest in 1998 and 14% in 2001. These findings appear to be typical of studies of recidivism among DV offenders. Considerable work remains to be done to measure and identify the factors that predict re-arrest for DV offenses. Third, in 2001, defendants whose cases took longer to reach a disposition were less likely to be re-arrested after disposition. Case processing time also reduced the likelihood of re-arrest in 1998, although the effect was not statistically significant then. This suggests that the closer monitoring of DV cases in the specialized DV court may have had a stronger deterrent effect. Fourth, our study found two statistically significant effects of the type of relationship between the defendant and the victim on the likelihood of re-arrest. Defendants who were in a boyfriend-girlfriend relationship or were common-law spouses were more likely to be re-arrested for a DV offense than married defendants in 2001. These findings are tentative, since the defendant-victim relationship was listed as â&#x20AC;&#x153;missingâ&#x20AC;? for 34% of the defendants in this sample. More complete data would provide a better test of the impact of defendant-victim relationship. Furthermore, we have no data on whether the relationship was a current or former relationship at the time of the arrest.17 Finally, none of the community ties variables had a statistically significant impact on the likelihood of re-arrest in 1998, and only one community ties variable (whether the defendant had a telephone) was statistically significant in 2001. This pattern of findings is at first glance surprising, given that many of these variables had statistically significant effects in our earlier analysis of citywide data in 1998 (Peterson 2003a). However, the general failure to find statistically significant effects in the Manhattan samples is probably due to the smaller sample sizes (955 and 1,188 in the Manhattan datasets in 1998 and 2001, compared to approximately 6,500 in the citywide dataset in 1998).

A previous study comparing those who were current vs. former intimate partners at the time of arrest found no impact of defendant-victim relationship on the likelihood of re-arrest (Davis et al. 1998). However, this study was limited by its small sample size.

63 VIII. CONCLUSION A. Major Findings This study was designed to address two major questions about the processing of DV cases: 1) What are the effects of the use of a specialized domestic violence court on case outcomes in DV cases? 2) What is the effect of the use of a specialized domestic violence court on the re-arrest rate for DV offenders? To address these questions, we used the Third Quarter 1998 Dataset and the First Quarter 2001 dataset. Each dataset included CJA data for a three-month cohort of arrests. We analyzed data for 990 DV cases disposed in Criminal Court in Manhattan in the third quarter of 1998 and 1,249 DV cases in the first quarter of 2001. In 1998, Manhattan did not have a specialized DV Criminal Court. In 2001, a specialized DV Criminal Court had been operating in Manhattan since June 2000. There are four major findings from our study. First, the specialized DV Criminal Court in Manhattan did not change case dispositions. This finding confirmed our expectations. The distribution of case dispositions was the same before and after the specialized DV court was established. The conviction rate was 29% in both periods; about 56% of DV cases were dismissed, and about 15% of DV cases were adjourned in contemplation of dismissal. These findings were consistent with our previous study comparing boroughs with and without specialized DV courts. In that study, we found no consistent difference in case dispositions between boroughs that used a specialized DV court in 1998 and Manhattan, which did not (Peterson 2002). Our finding that the specialized DV Criminal Court did not change the distribution of case dispositions in Manhattan is also consistent with the findings of Newmark et al.â&#x20AC;&#x2122;s (2001) study of the Brooklyn DV Supreme Court and Angeneâ&#x20AC;&#x2122;s (2000) study of the specialized DV court in San Diego. However, the conviction rates in Brooklyn and San Diego were already very high (over 90%) even before the specialized DV courts were established, so there was little opportunity for conviction rates to increase. Furthermore, our results are inconsistent with the findings from four studies of specialized DV courts (Eckberg and Podkopacz 2002, Goldkamp et al. 1996, Miller 1999 and Davis et al. 2001), all of which reported an increase in the conviction rate. Other changes that took place simultaneously with the establishment of these specialized DV courts may account for the change in case dispositions. Second, the specialized DV Criminal Court in Manhattan did change sentence outcomes. This finding also confirmed our expectations. After the specialized DV court was established, about two thirds of convicted defendants were sentenced to

64 conditional discharges (often accompanied by a requirement that the defendant complete a batterer intervention program or drug or alcohol treatment program). Prior to the establishment of the DV court, just over half of convicted defendants in DV cases were sentenced to conditional discharges. Jail sentences were used slightly less often after the DV court was established. About 27% of convicted defendants in DV cases were sentenced to jail in 2001, compared to 31% in 1998. The findings on sentence outcomes were consistent with our previous research comparing boroughs with and without specialized DV courts. We found that convicted DV defendants were slightly less likely to be sentenced to jail in boroughs that used specialized DV courts in 1998, than in Manhattan, which did not have a specialized DV court in 1998 (Peterson 2002). Our findings were also consistent with the findings of the Angene (2000) and Davis et al. (2001) studies. However, in each of those studies the percentage of convicted defendants sentenced to jail declined by half, a much sharper decline than we found in Manhattan. Jail sentences for DV offenders were much more common in San Diego (61%) and Milwaukee (75%) before the DV courts were established than they were in Manhattan (31%). Since the incarceration rate was initially much lower in Manhattan, the potential for sharply reducing the incarceration rate was much more limited than it was in San Diego or Milwaukee. Third, we did not find much support for our prediction that the specialized DV Criminal Court in Manhattan would reduce the average length of jail sentences. Average jail sentences did decline, from 65 days to 48 days. However, these results are based on small samples of defendants (87 and 98 cases, respectively), and this difference may have been due to chance alone. Furthermore, our previous research comparing boroughs with and without specialized DV courts found that the use of specialized DV courts did not affect sentence length (Peterson 2002). Only one other study examined the impact of a specialized DV court on the length of jail sentences. Unlike our study, Angene (2000) found that jail sentences in San Diego increased after the DV court was established. As noted above, San Diego imposed jail sentences in three fifths of DV cases before the specialized DV court was established. It appears that most of these sentences were relatively short. After the DV court was established in San Diego, jail sentences were used much less frequently, but were of longer duration when they were imposed. Finally, the re-arrest rate for DV offenses within 18 months of case disposition was higher after the establishment of the specialized DV Criminal Court in Manhattan. This confirms our expectations, not because we expected the DV court to be less effective at deterring DV offenders, but because we expected the DV court to lead to better efforts to identify DV arrests and to monitor DV offenders more closely. Our previous research comparing boroughs with and without specialized DV courts is consistent with this interpretation. We found that re-arrest rates were higher in boroughs that used specialized DV courts in 1998, than in Manhattan, which did not (Peterson 2003a). Our findings are also similar to those of Newmark et al. (2001), who found that re-arrest rates increased after the specialized DV Supreme Court was established in Brooklyn. They attributed the increase to closer monitoring of DV offenders. Other studies (Angene 2000, Davis et al. 2001, Eckberg and Podkopacz

65 2002, and Gover et al. 2003) found that specialized DV courts reduced recidivism. However these studies are not directly comparable to ours. The Angene (2000) and Davis et al. (2001) studies did not use re-arrests to measure recidivism. Angene (2000) used any police contact for a DV incident as a measure of recidivism, while Davis et al. (2001) used information obtained from interviews with victims. Eckberg and Podkopacz (2002) did use re-arrests to measure recidivism, but limited their measure to re-arrests for a new domestic assault. Our recidivism measure includes any DV re-arrest, not just re-arrests for assault. Gover et al.’s (2003) study evaluated a rural jurisdiction, where arrest volume was relatively low and DV offenders were easier to identify and monitor both before and after the court was established. Because the establishment of the specialized DV court in Manhattan improved the identification of DV arrests and the monitoring of DV offenders, our data on rearrests do not enable us to determine whether the specialized court had a deterrent effect on DV offenders. The finding that the re-arrest rate actually increased suggests that the specialized DV court was accomplishing many of its other goals: increasing defendant accountability, court monitoring and victim safety. Although we would like to address the question of deterrence, our study does not permit us to do so. B. Discussion Significant changes in the prosecution and adjudication of DV cases have been introduced in the U.S. over the last two decades. Specialized domestic violence courts were established in many jurisdictions. These courts seek to increase defendant accountability, promote victim safety and coordinate the activities of criminal justice agencies that respond to domestic violence. While these courts have been designed to improve the criminal justice system’s response to domestic violence, little is known about their impact. The current study has sought to provide new information about the impact of specialized DV courts. Specifically, we have compared the processing of DV cases before and after the establishment of a specialized DV Criminal Court in Manhattan. Although specialized DV courts have multiple goals, our study has focused primarily on the court’s impact on one of those goals: increasing defendant accountability. Has the specialized DV Criminal Court accomplished its goal of increasing defendant accountability? How? The findings from our study suggest that the specialized DV court increased defendant accountability through three mechanisms: processing DV cases more quickly, improving identification of DV cases, and monitoring defendants more closely for new DV offenses. We found evidence of these changes throughout our analyses. 1. Faster Processing of Domestic Violence Cases The average time between arraignment and disposition declined significantly between 1998 and 2001, from 18 weeks to 13 weeks. There was no change in the percentage of cases disposed at arraignment—about 99% of DV cases in both years were continued beyond arraignment. However, we found that in 2001, cases that would

66 end in a conviction were identified and disposed of earlier than in 1998. We also found that among defendants who were given a jail sentence of time served, sentences were shorter in 2001 than in 1998, suggesting that these cases were processed more quickly. How did the specialized DV court reduce case processing time? One way may be through greater flexibility in plea bargaining after the specialized DV court was established. Among cases that ended in conviction, 78% had a reduction in charge severity between arraignment and conviction in 2001, compared to 69% in 1998. As a result, the severity of the top conviction charge was considerably lower in 2001 than in 1998. About 61% of convictions in 2001 had a violation as the top charge, compared to only 45% of convictions in 1998. (A violation is less severe than a misdemeanor.) This may explain why cases ending in conviction were disposed of more quickly in 2001 than in 1998. Defendants were more willing to plead guilty if the conviction charge was a violation, and the sentence did not include jail time. We found further evidence of greater flexibility in plea bargaining in our analyses of the likelihood of jail sentences. Surprisingly, we found that in 2001 convictions in which the defendant had a larger number of arrest charges were less likely to end in a jail sentence than those where the defendant had a smaller number of arrest charges. Since a larger number of arrest charges is an indication of a potentially stronger case, we expected that the ADA or the court would have greater leverage to hold out for a jail sentence. While this occurred in 1998, when DV cases were heard in mixed-docket courts, it did not occur in the specialized DV court in 2001. In 2001, the court focused primarily on using that leverage to obtain a conviction with a conditional discharge and a program requirement. This change between 1998 and 2001 suggests that the specialized court changed the goals of plea bargaining in DV cases. (As we noted in Section III-C, the DAâ&#x20AC;&#x2122;s office in Manhattan has indicated that it did not change its approach on seeking jail sentences in DV cases, i.e., it continued to seek jail sentences as often after the specialized DV court was established as it did before.) In stronger cases, jail sentences were less likely and conditional discharges with program requirements were more likely in the specialized DV court. A second way that the court reduced case-processing time may be through greater routinization in the processing of DV cases in the specialized DV court. One of the rationales for specialized DV courts is that the court becomes familiar with the special features of DV cases. In a specialized DV court, it may be easier for ADAâ&#x20AC;&#x2122;s, defense attorneys and judges to reach agreement on the strength of the evidence in the case and on an appropriate sentence for the defendant. Defendants may be more willing to agree to sentences requiring program completion rather than jail time. It is also easier to identify appropriate batterer intervention programs or drug or alcohol treatment programs in a specialized DV court, since these programs are routinely used in large numbers of cases. Finally, the specialized DV court may also have encouraged the DAâ&#x20AC;&#x2122;s office to improve its charging decisions. In 1998, cases that were arraigned on felony charges and disposed of in Criminal Court were less likely to end in conviction than Criminal

67 Court cases arraigned on lesser charges. This suggests that the evidence in these felony cases was often not sufficient to sustain a conviction. In 2001, the DA’s office arraigned fewer cases on felony charges, and the conviction rate in these cases was the same as the conviction rate in cases arraigned on lesser charges. This suggests that the DA’s office was more selective in charging cases appropriately. As a result, more of the cases charged as felonies and disposed in Criminal Court ended in a conviction in 2001. 2. Improved Identification of Domestic Violence Cases The improved identification of DV cases became evident as we examined the volume of DV Cases in 1998 and 2001. In the third quarter of 1998, before the specialized DV Criminal Court was established, the courts identified 990 DV cases in Manhattan involving crimes against persons and property. In the first quarter of 2001, after the specialized DV court was established, the courts identified 1,249 DV cases. Since domestic violence complaints declined by about 9% citywide between 1998 and 2001 (NYPD 1998, 2001), we do not believe this increase in DV cases reflected an increase in the incidence of domestic violence offenses in Manhattan. Instead, the increase in DV cases in Manhattan between 1998 and 2001 probably reflects the increased accuracy of the courts in identifying DV cases. The improvements in identifying DV cases were achieved in three ways. First, better efforts were made to identify DV cases at arraignment by gathering more information about the relationship between the defendant and the victim. After the DV court was established, it became more important to accurately identify DV cases. Improvements in record-keeping insured that post-arraignment appearances would be scheduled in the specialized DV court. Second, better efforts were made after arraignment to identify DV cases that were not properly identified at arraignment. Third, better efforts were made to monitor defendants who had previously been arrested for a DV offense. This improved monitoring was accomplished, for example, by having batterer intervention programs report to the DV compliance part about new offenses. Puffett and Gavin’s (2004) study of a similar DV part in the Bronx suggests that judicial monitoring increases a defendant’s likelihood of being re-arrested for a new DV offense. The improvements in identifying DV cases changed the distribution of DV cases, primarily because the volume of misdemeanor cases increased. We found that in 2001 there were about 300 more DV cases arraigned on misdemeanor charges than in 1998. Since the specialized DV Criminal Court was specifically designed to focus more attention on these cases, our data suggest that the court succeeded in identifying these cases and scheduling them for appearances in the specialized DV court. The improved identification of DV cases also affected the results of our analyses of re-arrests before and after the specialized DV court was established. As we expected, the re-arrest rate was higher after the DV court was established. We do not believe that this reflects a higher rate of re-offending; instead it reflects the cumulative results of the improvements made to identify DV cases.

68 3. Improved Monitoring of Domestic Violence Defendants The specialized DV court increased the monitoring of DV defendants by requiring more of them to complete batterer intervention programs and/or drug or alcohol treatment programs. To accomplish this, convicted defendants were often given a sentence of a conditional discharge with program completion as one of the conditions. Although we do not have information about enrollment in the programs, we do know that the use of conditional discharge sentences for convicted DV defendants increased significantly. In 1998, before the specialized DV court was established, just over half of the sentences in DV cases were conditional discharges. In 2001, after the specialized DV court was established, over two thirds of the sentences in DV cases were conditional discharges. The use of jail sentences declined slightly. The conditional discharge sentence allowed the court to monitor the defendant for up to a year after case disposition. Compliance with the requirements of batterer intervention programs and drug and alcohol treatment programs was monitored in a DV compliance part. These programs reported to the DV compliance part about defendantsâ&#x20AC;&#x2122; progress, as well as about violations of program conditions and re-offending. While our study did not have access to data on program compliance, Puffett and Gavin (2004) provide compliance data for a similar DV compliance part in the Bronx. They found that only half of defendants successfully completed their programs. The completion rate varied by the type of program mandate. Defendants mandated to a batterer intervention program only were more likely to successfully complete the program (58%) than those mandated to substance abuse treatment only (40%) or those mandated to both types of programs (33%). Defendants who failed to complete their programs were usually sent to jail, and were much more likely to be re-arrested after their release. If results in Manhattan are similar to those in the Bronx, judicial monitoring is insuring that the court responds to defendants who fail to complete their programs. In our Manhattan data, we found that in 2001, but not in 1998, the postdisposition re-arrest rate for new DV offenses was lower for defendants whose cases took longer to process. Each additional week between arraignment and disposition decreased the odds of re-arrest. This finding suggests that the longer the defendant was monitored through appearances in the specialized DV court, the less likely the defendant was to be re-arrested for a new DV offense after the case was disposed. Since we did not find a similar effect in 1998, we speculate that monitoring through court appearances was more effective when the defendants were appearing in the specialized DV court in 2001. This finding is particularly noteworthy because the average case-processing time was 5 weeks shorter in 2001 than in 1998. In spite of this decline in the time available to monitor defendants prior to disposition, monitoring by the specialized DV court apparently had a longer-lasting effect on the postdisposition re-arrest rate. In 1998, when defendants were appearing for longer periods of time in all-purpose courts with mixed dockets, monitoring through court appearances did not appear to deter re-offending.

69 C. Conclusions Has the specialized DV Criminal Court in Manhattan been successful? Mazur and Aldrich (2002) suggest that success is difficult to define for domestic violence courts. One possible measure of success is whether the specialized DV court deterred future violence by DV offenders. However, unlike other problem-solving courts, such as drug courts, DV courts do not seek to “rehabilitate” defendants (Casey and Rottman 2003, Mazur and Aldrich 2002). As we noted in Section I of the report, batterer intervention programs are used primarily to monitor defendants—evidence is mixed on whether the programs deter future violence. In our study, it was impossible to determine if the establishment of the specialized DV court in Manhattan deterred future violence by the defendants. We measured recidivism by determining whether the defendant was re-arrested for a new DV offense within 18 months of case disposition. Because the establishment of the court improved the identification of DV cases we found that the re-arrest rate was higher after the court was established. Other possible measures of success would focus on the victims. Were victims satisfied with the outcome of the case? Were they re-victimized by the defendant? Did the court refer them to services, such as emergency housing or legal services? Very little is known about the impact of the specialized DV courts on victims, either in Manhattan or in other jurisdictions. Furthermore, long-term studies based on victim interviews are generally not feasible, due to the difficulties of finding and interviewing victims (Mazur and Aldrich 2002). Given the difficulties encountered in measuring success in terms of deterrence or victim outcomes, how can we draw conclusions about the impact of the specialized DV Criminal Court in Manhattan? Alternative measures of success could be based on one of the intermediate goals of the court: improving defendant accountability. While deterring future violence and promoting victim safety may be the ultimate goals, one of the ways the specialized DV court seeks to meet these goals is by holding defendants accountable. Judged by this standard, it seems clear that the specialized DV court in Manhattan has been a success. As noted above, we found evidence that the specialized court was doing more monitoring of DV defendants. We also found that defendants who appeared in court over a longer period of time were less likely to be rearrested after their case was disposed. These findings suggest that the specialized DV court was able to hold defendants accountable through court monitoring, and the monitoring had a deterrent effect. What are the implications of our research for future efforts to use specialized DV courts? For the foreseeable future, the specialized misdemeanor DV courts will continue to process most of the misdemeanor domestic violence cases throughout New York City. All the boroughs in New York City now use specialized misdemeanor DV courts similar to Manhattan’s specialized DV Criminal Court. Although all the boroughs will also soon have Integrated Domestic Violence (IDV) courts, the IDV courts will only handle cases for defendants who have at least one additional concurrent felony DV or Family Court case involving the same family. Since most cases involve

70 defendants whose only case is a misdemeanor DV case, the majority of misdemeanor DV cases will continue to be processed in the specialized misdemeanor DV courts in each borough. Given these plans, the key policy question in New York City is not whether specialized misdemeanor DV courts should be retained, but how they can be improved. A recent report addresses this question and outlines the key principles for insuring effective implementation of specialized DV courts (Mazur and Aldrich 2002). We do not offer specific recommendations for improvements in our report. Instead, we conclude by identifying questions for future research on the operation of the specialized misdemeanor DV courts in New York City as well as in other jurisdictions throughout the United States. First, more research is needed to determine whether improved monitoring of offenders by a specialized DV court can increase victim safety. Although the Manhattan specialized DV court increased defendant accountability, we were not able to assess accurately whether defendants were deterred from committing future DV offenses. Additional research, perhaps interviewing victims about re-offending by the defendant, is needed to address this question. Second, more research is needed to improve the effectiveness of batterer intervention programs. Specialized DV courts rely heavily on batterer intervention programs, yet many of these programs have not been evaluated. Are some types of programs more effective than others? Are certain types of defendants more likely to complete these programs successfully? Third, more research is needed on victim outcomes. Although comprehensive long-term studies may be difficult to conduct, other types of studies may provide useful information. Small-scale focus groups or interviews with victims would be able to address questions about victim satisfaction with the court system. Did the victim feel s/he was provided clear information about how the case would proceed and about available victim services? Was s/he satisfied with the disposition of the case? Was s/he satisfied with information received about the defendantâ&#x20AC;&#x2122;s participation in a batterer intervention program or drug or alcohol treatment program, if any? Finally, research is needed on the views of court actors. Do judges, ADAâ&#x20AC;&#x2122;s, defense attorneys, DV resource coordinators, and other staff members feel the court is achieving its goals? What features of the courtâ&#x20AC;&#x2122;s processing of DV cases do they believe have the greatest impact? What suggestions do they have for improvements? Specialized DV courts represent a major effort to change how the judicial system responds to domestic violence cases. Our study has provided some answers about the impact of the specialized DV Criminal Court in Manhattan. However, research on the success of specialized DV courts is still in its early stages, and much more remains to be learned. A more comprehensive assessment is needed to determine whether specialized DV courts are more appropriate for DV cases than mixed-docket courts.

71 IX. REFERENCES Aldrich, Liberty and Julie A. Domonkos. 2000. “Navigating the NYPD: A Guide to New York City Police Department Policies and Procedures for Family Court Attorneys Handling Order of Protection Cases.” Pp. 279-292 in Julie A. Domonkos and Jill Laurie Goodman, Lawyer’s Manual on Domestic Violence: Representing the Victim. New York: Supreme Court of the State of New York, Appellate Division, First Department. Angene, Lyn. 2000. Evaluation Report for the San Diego County Domestic Violence Courts. San Diego, CA: San Diego Superior Court. Berk, Richard A. 1983. “An Introduction to Sample Selection Bias in Sociological Data.” American Sociological Review 48:386-398. Berman, Greg and John Feinblatt. 2001. “Problem Solving Courts.” Law and Policy 25:125-140. Buzawa, Eve, Gerald T. Hotaling, Andrew Klein and James Byrne. 1999. Response to Domestic Violence in a Pro-Active Court Setting: Final Report. Washington, DC: National Institute of Justice. Casey, Pamela M. and David B. Rottman. 2003. Problem-Solving Courts: Models and Trends. Williamsburg, VA: National Center for State Courts. Chalk, Rosemary and Patricia A. King. 1998. “Assessing Prevention and Treatment Programs.” Pp. 158-205 in Violence in Families. Washington, DC: National Academy Press. Davis, Robert C., Barbara E. Smith and Laura B. Nickles. 1998. “The Deterrent Effect of Prosecuting Domestic Violence Misdemeanors.” Crime and Delinquency 44:434-442. Davis, Robert C., Barbara E. Smith and Caitilin R. Rabbitt. 2001. “Increasing Convictions in Domestic Violence Cases: A Field Test in Milwaukee.” The Justice System Journal 22: 62-72. Davis, Robert C. and Bruce G. Taylor. 1999. “Does Batterer Treatment Reduce Violence? A Synthesis of the Literature.” Pp. 66-93 in Lynette Feder (ed.), Women and Domestic Violence: An Interdisciplinary Approach. New York: Haworth. Eckberg, Deborah A. and Marcy R. Podkopacz. 2002. Domestic Violence Court: Case Processing Update and Recidivism Analysis. Minneapolis: Fourth Judicial District of the State of Minnesota.

72 Eckert, Mary A., and Mari Curbelo. 2000. Alternative to Incarceration Information Services: First Half Fiscal Year 2000. New York: New York City Criminal Justice Agency. Fritzler, Randal B. and Leonore Simon M. J. 2000. “Creating a Domestic Violence Court: Combat in the Trenches.” Court Review 37:28-39. Goldkamp, John S., Doris Weiland, Mark Collins and Michael White. 1996. The Role of Drug and Alcohol Abuse in Domestic Violence and Its Treatment: Dade County’s Domestic Violence Court Experiment, Final Report. Philadelphia, PA: Crime and Justice Research Institute. Gover, Angela R., John M. MacDonald and Geoffrey P. Alpert. 2003. “Combating Domestic Violence: Findings from an Evaluation of a Local Domestic Violence Court.” Criminology and Public Policy 3:109-132. Heckman, James. 1979. “Sample Selection Bias as a Specification Error.” Econometrica 47:153-161. Kaye, Judith S. 2001. The State of the Judiciary. Albany, NY: New York State Unified Court System. Kaye, Judith S. and Susan K. Knipps. 2000. “Judicial Responses to Domestic Violence: The Case for a Problem Solving Approach.” Western State University Law Review 27:1-13. Keilitz, Susan. 2000. Specialization of Domestic Violence Case Management in the Courts: A National Survey. Williamsburg, VA: National Center for State Courts. Klein, Stephen P., Patricia Ebener, Allan Abrahamse and Nora Fitzgerald. 1991. Predicting Criminal Justice Outcomes: What Matters? Washington, DC: Bureau of Justice Statistics, U.S. Department of Justice. Lewis-Beck, Michael S. 1980. Applied Regression: An Introduction. Sage Publications: Thousand Oaks, CA. MacLeod, Dag and Julia F. Weber. 2000. Domestic Violence Courts: A Descriptive Study. San Francisco, CA. Judicial Council of California-Administrative Office of the Courts. Mazur, Robyn and Liberty Aldrich. 2002. What Makes a Domestic Violence Court Work? Key Principles. New York: Center for Court Innovation. Menard, Scott. 1995. Applied Logistic Regression Analysis. Sage Publications: Thousand Oaks, CA.

73 Miethe, Terance D. 1987. â&#x20AC;&#x153;Stereotypical Conceptions and Criminal Processing: The Case of the Victim-Offender Relationship.â&#x20AC;? Justice Quarterly 4:571-593. Miller, Neal. 1999. Process Evaluation of the Queens County Arrest Policies Project. Washington, DC: National Institute of Justice. Mohr, Lawrence B. 1990. Understanding Significance Testing. Sage Publications: Thousand Oaks, CA. Newmark, Lisa, Mike Rempel, Kelly Diffily and Kamala Mallik Kane. 2001. Specialized Felony Domestic Violence Courts: Lessons on Implementation and Impacts from the Kings County Experience. Washington, DC: Urban Institute Justice Policy Center. New York City Criminal Justice Agency. 2000. Semi-Annual Report: Second Half of 1998. New York: New York City Criminal Justice Agency. NYPD. 1998. Complaints and Arrests 1998. New York: Crime Analysis Unit, New York City Police Department. NYPD. 2000. Patrol Guide. New York: New York City Police Department. NYPD. 2001. Complaints and Arrests 2001. New York: Crime Analysis Unit, New York City Police Department. Peterson, Richard R. 1989. Women, Work and Divorce. Albany, NY: State University of New York Press. Peterson, Richard R. 2001. Comparing the Processing of Domestic Violence Cases to Non-Domestic Violence Cases in New York City Criminal Courts. New York: New York City Criminal Justice Agency. Peterson, Richard R. 2002. Cross-Borough Differences in the Processing of Domestic Violence Cases in New York City Criminal Courts. New York: New York City Criminal Justice Agency. Peterson, Richard R. 2003a. The Impact of Case Processing on Re-arrests Among Domestic Violence Offenders in New York City. New York: New York City Criminal Justice Agency. Peterson, Richard R. 2003b. Combating Domestic Violence in New York City: A Study of DV Cases in the Criminal Courts. New York: New York City Criminal Justice Agency. Peterson, Richard R. 2003c. Combating Domestic Violence in New York City, 2001. New York: New York City Criminal Justice Agency.

Puffett, Nora K. and Chandra Gavin. 2004. Predictors of Program Outcome and Recidivism at the Bronx Misdemeanor Domestic Violence Court. New York: Center for Court Innovation. Siddiqi, Qudsia. 1999. Assessing Risk of Pretrial Failure to Appear in New York City: A Research Summary and Implications for Developing Release-Recommendation Schemes. New York, NY: New York City Criminal Justice Agency. SPSS, Inc. 1999. SPSS Regression Models 9.0. Chicago: SPSS, Inc. Steketee, Martha Wade, Lynn S. Levey and Susan L. Keilitz. 2000. “Implementing an Integrated Domestic Violence Court: Systemic Change in the District of Columbia.” Williamsburg, VA: National Center for State Courts. Tsai, Betsy. 2000. “The Trend Toward Specialized Domestic Violence Courts: Improvements on an Effective Innovation.” Fordham Law Review 68:1285-1325. Watson, Charlotte A. 2000. “Programs for Men Who Batter: What Have We Learned?” New York State Office for the Prevention of Domestic Violence Bulletin 12:8-9. Weis, Joseph G. 1989. “Family Violence Research Methodology and Design.” Pp. 117-162 in Lloyd Ohlin and Michael Tonry (eds.), Crime and Justice: A Review of Research, Volume 11: Family Violence. Chicago: University of Chicago.

75 APPENDIX A: STATISTICAL METHODS 1. Logistic Regression Analysis The statistical technique used to predict likelihood of conviction, likelihood of incarceration and likelihood of re-arrest is logistic regression, a technique that is used when the outcome to be explained (i.e., the dependent variable) has two categories. In this appendix, we discuss the use of logistic regression analysis to predict the likelihood of conviction. However, our discussion applies equally well to the prediction of the likelihood of incarceration, and the likelihood of re-arrest. In the analysis of the likelihood of conviction, all cases were coded on our dependent variable in one of two categories: not convicted (coded 0) and convicted (coded 1). The models we present are structured to predict the likelihood that cases are disposed as convictions. These predictions are made on the basis of information we have about a variety of defendant and case characteristics (i.e., the independent variables). Logistic regression techniques provide several ways of evaluating the effect of these independent variables on the likelihood of conviction. This report examines three statistical measures to evaluate the effect of the independent variables. First, we report the statistical significance of each independent variable. Statistical significance takes into account the size of the sample as well as the magnitude of the effect of the independent variable. Based on this information, statistical significance assesses the probability that the effect observed in the sample could have occurred by chance alone. In this report, following standard convention, significance levels of .05 or less are treated as statistically significant. In other words, when an effect has a 5% or less probability of having occurred by chance, we conclude that the independent variable is a statistically significant predictor of the likelihood of conviction. We use the more conservative “two-tail” test of statistical significance (which makes no assumption about the direction of the effect of the independent variables) rather than the “one-tail” test (which assumes that the effect of the independent variable is in one direction only, either positive or negative). One weakness of using statistical significance to measure the effect of an independent variable is that when sample sizes are large (e.g., more than several thousand cases), many independent variables will have statistically significant effects even when the magnitude of the effects is small. For example, in a very large sample, we may be able to say that having a prior adult arrest has a statistically significant effect on the likelihood of conviction, even though it increases the likelihood of conviction only from 49% to 51%. In this situation, we can say that this difference in conviction rates is unlikely to be due to chance. However, it is also clear that knowing whether or not a defendant had a prior adult arrest does not explain much of the variation in likelihood of conviction. The second statistical measure used to evaluate the effect of the independent variables is the odds ratio. The odds ratio supplements information about statistical

76 significance by evaluating the magnitude of the effect of the independent variable. Specifically, it tells us how much the odds of an outcome (e.g., conviction) change, for each one unit increase in the independent variable. If an independent variable is coded in two categories (e.g., 0 and 1), then the odds ratio tells us how the odds of the outcome change when cases are coded 1 on the independent variable (vs. cases coded 0). An odds ratio greater than one indicates an increase in the likelihood of the outcome occurring, while an odds ratio less than one indicates a decrease in the likelihood of the outcome occurring. An odds ratio of 1 indicates that the odds of an outcome occurring are not affected by the independent variable. To return to our previous example, if the odds ratio for the effect of having a prior adult arrest on the likelihood of conviction was 1.12, this would mean that in cases where the defendant had a prior adult arrest the odds of conviction are 1.12 times greater than in cases where the defendant did not have a prior adult arrest. In contrast, if we examined the impact of whether the defendant was ever released from custody while the case was pending, we might find an odds ratio less than 1. For example, if the odds ratio was .83, this would mean that in cases where the defendant was released from custody the odds of conviction are only .83 times as large as the odds when the defendant was not released from custody. To simplify interpretation of odds ratios less than 1, it is common to examine the inverse of the odds ratio (1 divided by the odds ratio). When this is done, the interpretation of the effect of the independent variable is reversed. For example, if the odds ratio for being released from custody is .83, we can take the inverse of the odds ratio, 1.20 (1 divided by .83), and say that in cases where the defendant was not released from custody, the odds of conviction were 1.20 times greater than in cases where the defendant was released. Finally, if the odds ratio was 1.00, this would mean that whether or not the defendant was released from custody had no impact on the odds of conviction. (These examples are hypothetical and do not necessarily reflect our expectations about the findings). In the analyses presented in this report, results are presented for independent variables coded in three different waysâ&#x20AC;&#x201D;categorical variables that have two categories, categorical variables that have more than two categories, and continuous variables that measure the quantity of a defendant or case characteristic (e.g., the number of prior felony convictions for the defendant). When a categorical independent variable has two categories, the odds ratio measures the change in the odds when cases are in one category vs. another (e.g., defendant had a prior adult arrest vs. did not have a prior adult arrest). When a categorical independent variable has more than two categories, one of the categories is chosen as a reference category, and the odds ratios measure the effect of being in each of the other categories vs. being in the reference category (e.g., cases in the Bronx and Manhattan are compared to cases in Brooklyn, which is used as the reference category). Finally, when the independent variable is continuous, the odds ratio measures the change in the odds associated with an increase of one unit on the scale of the independent variable (e.g., for number of prior felony convictions, the odds ratio measures the effect of having one additional felony conviction).

77 The third statistical measure used to assess the effect of the independent variables is the standardized beta coefficient (Menard, 1995). Standardized betas take into account not only the change in the likelihood of the outcome associated with a change in the independent variable, but also the distribution of the cases among the categories of the independent variable. Being in one category of an independent variable may have a large effect on the likelihood of an outcome (and therefore have a large odds ratio), but if there are relatively few cases in that category, the variable will not help to explain much of the variation in the likelihood of the outcome. For example, a case where the charge severity increased from arrest to arraignment might have a high probability of conviction, and hence a high odds ratio. However, if charge severity increased in only a small number of cases, this variable will not be able to explain much of the variation in likelihood of conviction. Standardized betas measure this overall effect of the independent variable on the dependent variable. Standardized betas vary from -1 to +1; values closer to zero indicate that the effect of the independent variable is relatively small, while values closer to +1 or -1 indicate that the effect of the independent variable is relatively strong. There are no commonly accepted absolute standards to decide whether a standardized beta is strong or weak. Consequently, we will discuss the relative strength of variables, describing some as stronger or weaker than others. In this report, we discuss results for all three of the measures described above. We use the statistical significance level to distinguish those independent variables that have a detectable18 effect on the dependent variable from those that do not. We use the odds ratio to evaluate the size of the effect of the independent variable, and we use the standardized beta to evaluate the ability of the independent variable to account for variation in the dependent variable. The models we discuss include a large number of predictors of the dependent variable. In these models, the measures of the effect of each independent variable (statistical significance, odds ratio, and standardized beta) evaluate the effect of that independent variable after controlling for the effects of all the other independent variables in the model. These effects represent the net effect of a given independent variable after the effect of all the other independent variables have been taken into account. This net effect differs from the total effect of the independent variable, which is the effect of the independent variable when it is used as the only predictor of the dependent variable. To evaluate the overall ability of all the independent variables in the logistic regression model to predict the dependent variable, we use a statistical measure called Nagelkerke R2 (SPSS, Inc., 1999). This measure varies from 0 to +1. It can be roughly interpreted as indicating what proportion of the variation in the dependent variable is explained by all the independent variables in the model (see Menard 1995 for a full discussion of the R2 statistic in logistic regression models). Low values of R2 (closer to 18

Due to sampling error, and limitations of logistic regression techniques, it is possible that some independent variables that do affect the dependent variable are found to be statistically insignificant in our particular sample of cases. See Mohr (1990) for a further discussion of these issues.

78 0) indicate that the model as a whole is relatively weak in accounting for variation in the dependent variable. High values (closer to 1) indicate that the model as a whole is very successful in accounting for variation in the dependent variable. 2. Interaction Effects To test for differences between the models for different years, we used statistical tests called interaction tests. These tests allow us to determine whether the influence of an independent variable (e.g., a defendant or case characteristic) on the dependent variable was significantly different in one year than its effect in the other year. If an interaction test was statistically significant, we concluded that the effect of the independent variable depended on (interacted with) year. We then examined and discussed the difference in the impact of the variable in 1998 vs. 2001. 3. Correcting for Selection Bias A. Models of Incarceration Our models predicting likelihood of incarceration included an additional control variable: a correction for selection bias. Selection bias is a problem that arises in statistical analysis when the group of cases that could have ended up in the sample is restricted to a selected set of respondents (Berk 1983). When selection bias occurs, the statistical estimates of the effects of the independent variables may be biased. These estimates may overstate, or understate, the influence of an independent variable. If problems of selection bias are not addressed, the interpretation of the results may be misleading. In our analysis, the models predicting likelihood of incarceration include only cases that resulted in conviction. The variables that influence whether a case results in conviction also may influence whether the case results in incarceration. For example, number of prior misdemeanor convictions may affect both the likelihood of conviction and the likelihood of incarceration. If a model predicting the likelihood of incarceration among those convicted does not control for selection bias, the estimate of the effect of number of prior misdemeanor convictions will be overstated. Part of its effect on likelihood of incarceration will actually be due to its influence on the likelihood of conviction, i.e., on the likelihood that the case ended up in the sample of convicted cases. The remainder of its effect, if any, will be due to its influence on the likelihood of incarceration. To control for this kind of selection bias, we included in the models a control variable that measures the predicted probability of conviction. This predicted probability of conviction was created using a model similar to that presented in Table 4-1 (data not shown). To avoid statistical problems19 the predicted probability of conviction used as a

When the predicted probability of conviction is included in a model predicting the likelihood of incarceration, it is important that the predicted probability of conviction not be highly correlated with other variables in the model. This problem, known as multicollinearity, is particularly likely if the same set of independent variables is used in both the conviction and incarceration models.

79 correction for selection bias was created using a somewhat different set of independent variables than the model presented in Table 4-1. The predicted probability of conviction can theoretically vary from a low of 0.00 to a high of 1.00. Of course, the model predicting likelihood of incarceration included only those cases where the defendant actually was convicted. The predicted probability of conviction for convicted cases was skewed toward the higher end of the scale (the mean predicted probability for the DV cases in our analyses was .50 in 1998 and .51 in 2001). Nevertheless, even among cases that resulted in conviction, there was significant variation: the predicted probability of conviction ranged from .14 to .95. It is this variation that enables the predicted probability of conviction to correct for selection bias. Among convicted cases, those with a low predicted probability of conviction are more similar to those who were not convicted, while those with a high predicted probability of conviction are more representative of those actually in the sample. The influence of the predicted probability of conviction on the likelihood of incarceration measures, and therefore controls for, the influence of variables that affect both outcomes. When the predicted probability of conviction is included as a control variable in models predicting the likelihood of incarceration, the estimates of the effects of the other independent variables in the model are more accurate.20 To return to the example discussed above, the number of prior misdemeanor convictions may influence both the likelihood of conviction and the likelihood of incarceration. In our models predicting the likelihood of incarceration, the estimate of the effect of number of prior misdemeanor convictions is more accurate because the models control for the influence of number of prior misdemeanor convictions on the likelihood of conviction.21 As a result, we have greater confidence in our estimates of the effects of this and other independent variables, as well as in our interpretation of the results of the models. Although we include the predicted probability of conviction as a control variable in all the models presented in this section, we do not discuss the impact of this variable since its primary purpose is to enable us to estimate accurately, and to interpret, the effects of the other variables. B. Models of Length of Jail Sentence The analyses of length of sentence included only cases that resulted in incarceration. The variables that influence whether a case resulted in incarceration also may influence the length of the sentence. For example, number of prior misdemeanor convictions may affect both the likelihood of incarceration and the length of the For this reason, the model creating the predicted probability of conviction uses some different independent variables than those used in the analyses presented in this report. 20 See Heckman (1979) and Peterson (1989) for a more detailed discussion of selection bias and corrections for it. 21 The extent to which the estimates are more accurate depends on the ability of the model predicting the probability of conviction to explain a significant portion of the variation in likelihood of conviction. Our models were reasonably successful at explaining variation in likelihood of conviction. The models correcting for selection bias accounted for approximately 35% of the variation in likelihood of conviction.

80 sentence. The estimate of the effect of number of prior misdemeanor convictions will be overstated if it influences both the likelihood of being in the sample and the outcome being analyzed. In this circumstance, part of its effect on length of sentence will be due to its influence on the likelihood of incarceration, i.e., on the likelihood that the case ended up in the sample of cases sentenced to jail. The remainder of its effect, if any, will be due to its influence on the length of jail sentence. To correct for this kind of selection bias, we included an additional control variable in the models: the predicted probability of incarceration. This predicted probability of incarceration was created using a model similar to that presented in Table 5-1 (data not shown).22 The predicted probability of incarceration can theoretically vary from a low of 0.00 to a high of 1.00. Since the model predicting length of sentence included only those cases where the defendant was actually sentenced to jail, the predicted probability of incarceration for these cases was skewed toward the higher end of the scale (the mean predicted probability was .63 in 1998 and .70 in 2001). Nevertheless, even among cases that resulted in incarceration, there was significant variation: the predicted probability of incarceration varied from .04 to .97. When the predicted probability of incarceration is included as a control variable in models predicting the length of sentence, the estimates of the effects of the other independent variables in the model are more accurate.23 4. Linear Regression Analysis Length of jail sentence was measured in days.24 We used a statistical technique known as linear regression, which is appropriate when the dependent variable measures a quantity, such as the number of days.25 Length of jail sentence varied from 1 to 243 days, however most sentences (over 80%) were 120 days or less. Like logistic regression, linear regression results evaluate the effect of each independent variable after controlling for the effects of all the other independent variables in the model. These effects represent the net effect of the independent variable, in contrast to the total effect (i.e., when the independent variable is used as the only predictor). Also, the linear regression results report the statistical significance of

As before, to avoid statistical problems the predicted probability of incarceration used as a correction for selection bias was created using a somewhat different set of independent variables than the model presented in Table 5-1. 23 The extent to which the estimates are more accurate depends on the ability of the model predicting the probability of incarceration to explain a significant portion of the variation in likelihood of incarceration. The models correcting for selection bias accounted for approximately 47% of the variation in likelihood of incarceration. This is a sufficiently large proportion of the variation for the purpose of controlling for selection bias. 24 See discussion in Section III of this report explaining how length of jail sentence is measured. 25 In our previous analyses, the dependent variable was dichotomous (i.e., had only two categories) and we used logistic regression techniques.

81 each independent variable, just as the logistic regression results did. The interpretation of statistical significance is the same for both types of regression models. However, there are some differences between linear regression and logistic regression. The linear regression results include both a beta and a standardized beta for each predictor (independent variable) in the model. The beta measures the magnitude of the effect of the independent variable. Its purpose is similar to that of the odds ratio in logistic regression, but its interpretation is slightly different. In the linear regressions, the beta can be interpreted as the change in the number of days sentenced to jail associated with a one-unit change in the independent variable. The interpretation of the standardized beta in linear regression is similar to that in logistic regression. It measures the influence of the independent variable in explaining variation in the number of days sentenced to jail, after taking into account the effects of all the other independent variables in the model. Standardized betas vary from -1 to +1; a value close to zero indicates that the effect of the independent variable is relatively small, while a value closer to +1 or -1 indicates that the effect of the independent variable is relatively strong. To evaluate the overall ability of all of the independent variables in the linear regression model to predict the dependent variable, we use the coefficient of determination, also referred to as R2. R2 varies from 0 to +1. It indicates what proportion of variation in the dependent variable is explained by all the independent variables in the model (see Lewis-Beck 1980 for a full discussion of R2 in linear regression models). Low values of R2 indicate that the model explains only a small proportion of the variation in the dependent variable, while high values indicate that the model explains a large proportion of the variation in the dependent variable.

[This page intentionally left blank]

83 APPENDIX B: CODING OF VARIABLES FOR REGRESSION MODELS

CODING SCHEME

VARIABLES DEPENDENT VARIABLES

LIKELIHOOD OF CONVICTION LIKELIHOOD OF INCARCERATION LENGTH OF JAIL SENTENCE RE-ARREST FOR A DV OFFENSE WITHIN 18 MONTHS OF CASE DISPOSITION

Convicted = 1, Not convicted = 0 Sentenced to jail = 1, Not sentenced to jail = 0 Number of days, ranges from 1 to 365 Re-arrested = 1, Not re-arrested = 0

CASE OUTCOME (Re-Arrest Model, Table 7-1) Reference Category: Dismissed Adjourned in Contemplation of Dismissal

Dismissed: Reference Category Adjourned in Contemplation of Dismissal = 1, All other categories = 0 Convicted, no Jail Sentence = 1, All other categories = 0 Convicted, with Jail Sentence = 1, All other categories = 0

Convicted, no Jail Sentence Convicted, with Jail Sentence CONTROL VARIABLES SELECTION BIAS CORRECTIONS: LIKELIHOOD OF CONVICTION LIKELIHOOD OF INCARCERATION

Continuous, ranges from 0.00 to 1.00 Continuous, ranges from 0.00 to 1.00

ARRAIGNMENT CHARGE PENAL LAW ARTICLE: Reference Category: Assault (PL 120) Criminal Contempt (PL 215) Harassment (PL 240) Crimes Against Children (PL 260) Other

Assault: Reference Category Criminal Contempt = 1, All other categories = 0 Harassment = 1, All other categories = 0 Crimes Against Children = 1, All other categories = 0 Other = 1, All other categories = 0

DEFENDANT'S DEMOGRAPHIC CHARACTERISTICS SEX (Female)

Female = 1, Male = 0

ETHNICITY: Reference Category: Black White Hispanic Other

Black: Reference Category White = 1, All other categories = 0 Hispanic = 1, All other categories = 0 Other = 1, All other categories = 0

AGE: Reference Category: Age 16-20 Age 21-29 Age 30-39 Age 40 and over

Age 16-20: Reference Category Age 21-29 = 1, All other categories = 0 Age 30-39 = 1, All other categories = 0 Age 40 and over = 1, All other categories = 0

DEFENDANT-VICTIM RELATIONSHIP Reference Category: Married Boyfriend-Girlfriend Common-Law Marriage Other Relationship Missing

Married: Reference Category Boyfriend-Girlfriend = 1, All other categories = 0 Common-Law Marriage = 1, All other categories = 0 Other Relationship = 1, All other categories = 0 Missing = 1, All other categories = 0

Table Continues on Next Page

APPENDIX B: CODING OF VARIABLES FOR REGRESSION MODELS (continued)

VARIABLES

CODING SCHEME

DEFENDANT'S CRIMINAL HISTORY ANY PRIOR ARRESTS

Any Prior Arrests = 1, All other categories = 0

NUMBER OF PRIOR MISDEMEANOR CONVICTIONS

Number of convictions, ranges from 0 to 67

NUMBER OF PRIOR FELONY CONVICTIONS

Number of convictions, ranges from 0 to 7

ANY ARRESTS FOR A DV OFFENSE PRIOR TO CASE DISPOSITION

Any Arrests for a DV Offense Prior To Case Disposition = 1, All other categories = 0

ANY ARRESTS FOR A NON-DV OFFENSE PRIOR TO CASE DISPOSITION

Any Arrests for a Non-DV Offense Prior To Case Disposition = 1, All other categories = 0

DEFENDANT'S COMMUNITY TIES (Re-Arrest Model, Table 7-1) UNEMPLOYED

Unemployed = 1, All other categories = 0

AT CURRENT ADDRESS 1 YEAR OR LESS

At Current Address 1 Year or Less = 1, All other categories = 0

LIVES WITH SOMEONE

Lives with Someone = 1, All other categories = 0

DOES NOT EXPECT ANYONE AT ARRAIGNMENT

Does Not Expect Anyone At Arraignment = 1, All other categories = 0

HAS NO TELEPHONE

Has No Telephone = 1, All other categories = 0

LIVES OUTSIDE NYC AREA

Lives Outside NYC Area = 1, All other categories = 0

ARREST AND ARRAIGNMENT CHARGE CHARACTERISTICS NUMBER OF ARREST CHARGES

Number of charges, ranges from 1 to 4

ARRAIGNMENT CHARGE IS A FELONY

Charged as a Felony = 1, All other categories = 0

CHANGE IN CHARGE SEVERITY FROM ARREST TO ARRAIGNMENT: Reference Category: No Change Charge Severity Reduced from Arrest to Arraignment Charge Severity Increased from Arrest to Arraignment

No Change: Reference Category Charge Severity Reduced = 1, All other categories = 0 Charge Severity Increased = 1, All other categories = 0

CASE PROCESSING CHARACTERISTICS RELEASE RECOMMENDATION: Reference Category: No Recommendation (Weak NYC Ties) Recommended or Qualified Recommendation Open Bench Warrant At Time of Arrest Other or Missing DEFENDANT EVER RELEASED

Defendant ever released = 1, Defendant never released or case was disposed at arraignment = 0

CHARGE SEVERITY REDUCED BETWEEN ARRAIGNMENT AND CONVICTION

Severity reduced between arraignment and conviction = 1, Severity not reduced between arraignment and conviction = 0

SEVERITY OF CONVICTION CHARGE: Reference Category: Conviction Charge was an A Misdemeanor Conviction Charge was a Violation Conviction Charge was a B Misdemeanor

Convicted of an A Misdemeanor: Reference Category Convicted of a Violation = 1, All other categories = 0 Convicted of a B Misdemeanor = 1, All other categories = 0

NUMBER OF WEEKS FROM ARRAIGNMENT TO DISPOSITION

Number of weeks, ranges from 0 to 56

NOTE 1

No recommendation (weak NYC ties): Reference Category Recommended or qualified recommendation = 1, All other categories = 0 Open bench warrant = 1, All other categories = 0 Other or Missing = 1, All other categories = 0

See text for a description of the variables in the models.