DOJ Seal U.S. Department of Justice Office of Community Oriented Policing Services COPS Inter-Agency Response to Domestic Violence in a Medium Size City Erin Lane, M.P.M Joan Lucera, M.S Rachel Boba, Ph.D June 2003 Office of Community Oriented Policing Services Community Policing/Domestic Violence Test Site Project Office of Community Oriented Policing Services Community Policing/Domestic Violence Test Site Project: Inter-Agency Response to Domestic Violence in a Medium Sized City Erin Lane, M.P.M. Joan Lucera, M.S. Rachel Boba, Ph.D. POLICE FOUNDATION Prepared for the U.S. Department of Justice Office of Community Oriented Policing Services Award #98DVWXK015 June 2002 Table of Contents Pages List of Tables .....................................................................................ii-iii List of Figures ....................................................................................... iv List of Maps.............................................................................................v Chapter 1 Introduction...........................................................................................1 2 Methodology............................................................................................4 3 Literature Review: Evolution of Domestic Violence as a Policy Issue, and History of Domestic Violence Inter-Agency Response .................................................11 4 History of the Domestic Violence Coordinating Committee: 1987-2000 ...................................28 5 The Current State of the Domestic Violence Coordinating Committee: 2000-2001 ..............................................................................................57 Chapter 5 Appendix: Recommendations for Police Department Responses to Personal Violence between Intimates 6 Police Use of Dual Arrest in Intimate Assault Cases ..................................................85 Chapter 6 Appendix A: Data Cleaning Table Chapter 6 Appendix B: Table of Data Used for Analysis in Columns Chapter 6 Appendix C: Table of Data Used for Analysis in Rows 7 Mapping Reported Intimate Assaults...................................................................139 List of Tables Page Chapter 2 Table 1: Interviewees’ Agency Affiliations and Domestic Violence Related Job Functions................................................................................................6 Chapter 4 Table 1: Changes in Agency Policies and Procedures, c. 1991 ............................................41 Chapter 5 Table 1: Domestic Violence Service Delivery Community ...................................................62 Table 2: Number of Reported Working Relationships by Agency..............................................72 Chapter 6 Table 1: Demographic Comparison between the Population of the City and the Population of Reported Intimate Assaults ................................................................96 Table 2: Distribution of Police Outcomes ................................................................97 Table 3: Victim-Offender Relationship by Race of Offender ...............................................101 Table 4: Relationship, Offender’s Presence, and Arrest by Race...........................................102 Table 5: Offender’s Prior Intimate Violence Arrests......................................................104 Table 6: Offender Substance Use .........................................................................106 Table 7: Weapon Use......................................................................................107 Table 8: Offense Alleged by Victim.......................................................................107 Table 9: Evidence Collected .............................................................................109 Table 10: Dependent Variable: Dual Arrest ...............................................................113 Table 11: Dependent Variable: Emergency Protective Order Obtained........................................114 Table 12: Dependent Variable: Evidence Collected.........................................................116 Chapter 7 Table 1: Correlation Between Number of Reported Incidents and Percent of Black and White in Block Group ................................................................................164 List of Figures Page Chapter 4 Figure 1: Characterization of the History of the Domestic Violence Coordinating Committee (DVCC): 1987-2001...............................................................................35 Figure 2: Laycock and Webb’s Typology of Early Partnership Pitfalls.......................................51 Figure 3: Inter-Agency ‘Tipping Point’ Factors ...........................................................53 Chapter 5 Figure 1: The Domestic Violence Service Community.........................................................64 Figure 2: Police-Community Domestic Violence Working Relationships........................................71 Chapter 6 Figure 1: Police Outcome by Race of Offender..............................................................99 List of Maps Page Chapter 7 Map 1: Patrol Divisions and their Characteristics ........................................................148 Maps 2 and 3: Kernel Density of Reported Simple and Aggravated Intimate Assault per Square Mile...................................................................................151 Map 4: Frequency of Reported Intimate Assault by Block Group .............................................154 Map 5: Rate of Reported Intimate Assault by Block Group...................................................156 Map 6: Reported Intimate Assault and Population Density by Block Group....................................159 Maps 5 and 7: Rate of Reported Intimate Assault and Median Rent by Block Group ....................................................................................................161 Maps 5 and 8: Rate of Reported Intimate Assault and Racial Homogeneity by Block Group...............................................................................................163 Maps 9 and 10: Proportion of Reported Offending to Population by Race.....................................167 Map 11: Kernel Density of Reported Intimate Assaults with Police Stations and Medical Facilities........................................................................................170 Chapter 1: Introduction Inter-agency work is a key feature of many crime prevention and community safety efforts and part of the foundation of modern policing. No longer is crime control viewed as solely the purview of the police, but rather as an outcome for which many agencies share responsibility. As acceptance of this idea has increased, inter-agency collaborative efforts have multiplied. Nowhere have these coordinated community crime control efforts been more fully recognized and advanced than in efforts to address domestic violence. The present study examines one city’s efforts to reduce domestic violence through the coordinated work of the city police department and a wide range of criminal justice, social services, and community agencies. The city (here, and throughout this report, anonymous) is in the mid-Atlantic region of the country, has approximately 200,000 residents, and a police department with approximately 700 police officers. This research entailed study of an interagency domestic violence coalition, the Domestic Violence Coordinating Committee (DVCC), as well as two separate exploratory analyses of the city police department’s domestic violence data. After presenting the methodology and literature review, the findings of this study are presented in two parts. The first part pertains to the history and current state of the DVCC and includes Chapters 4 and 5. Chapter 4 is a description of the inception of DVCC and its evolution up until the year 2000 when the field research for this project began. Chapter 5 is a description of the DVCC at the time of the field research, a description of the domestic violence service community of which the DVCC is a part, a discussion of the perspectives about the police department expressed by members of the domestic violence service community, and a list of recommendations for improving police response to domestic violence developed by the Police Foundation. The second part is comprised of Chapters 6 and 7. Chapter 6 is an analysis of the practice known as dual arrest, where police officers arrest both parties in an intimate assault. This exploratory analysis includes comparisons among cases resulting in issuance of a warrant, single arrest, and dual arrest as well as logistic regression analysis testing the relationship between case characteristics and dual arrest. Chapter 7 explores the utility of spatially analyzing reported intimate assault data by using several different methods to map these crimes and compare geographic patterns to neighborhood demographic characteristics. Because this report covers a wide range of topics within the broad area of domestic violence, it is written so that each chapter ties in with the others, but contains a distinct qualitative or quantitative analysis that can be read separately. For this reason, each chapter includes its own conclusions, references, and in some cases summaries of relevant literature. Therefore, except for a general summary of literature on inter-agency domestic violence work, there are no overall summaries of literature, references, or conclusions. The concern of this project is violence committed against one member of a current or past intimate couple by the other, which we refer to as intimate violence. “Domestic violence” is a broader term used to describe intimate violence as well as violence between non-intimate familial or household members. In the chapters summarizing literature and describing the DVCC, we use the term “domestic violence,” because this is a term conventionally used by inter agency approaches; but regardless of the term used, throughout this report, our subject is intimate violence. Chapter 2: Methodology Overview of Methodologies This study was conducted between February 2000 and June 2001, and involved collection and analysis of qualitative and quantitative data. Part one of this report presents the qualitative findings about the Domestic Violence Coordinating Committee (DVCC) and the domestic violence service delivery community. The following section describes the methodology used to gather these data and construct these chapters. Part two of this report contains two separate quantitative analyses. The methodology of each is included in its respective chapter. The data collection process for constructing a thorough view of the DVCC and of the domestic violence service community took place between February 2000 and June 2001, and involved the collection of historical information about the period between 1986, a date slightly before the birth of the DVCC, to early 2001. Qualitative Methods Qualitative data were gathered using four methods. The first was reviewing literature on interagency domestic violence response. The second was observational field research, namely attendance at DVCC meetings, ride-alongs with police patrols, a limited amount of observation of the domestic violence court docket, and two ‘focus group’ interviews, each with four police patrol officers and supervisors. The third was examination of historical documentation relating to inter-agency work in the city, such as meeting minutes, press releases, and newspaper articles. The fourth, and primary method, was conducting in-depth interviews with officials from agencies across the city involved in domestic violence service delivery (see Table 1 for list of interviewees). Although the majority of the field research was completed between February 2000 and June 2001, the research began in August 1999 when Police Foundation researchers began attending the monthly DVCC meetings. Attendance continued until June 2001. Attending meetings served many benefits. It allowed researchers to gain knowledge about the DVCC’s background and current role in the domestic violence service delivery community, facilitated development of an interviewee list, and allowed researchers to develop a rapport with many DVCC members. In-depth interviews to capture the history of events and institutions have a rich pedigree, and have been the chosen method for a range of studies of the police and police work (e.g. Brookman, Noaks, & Wincup, 1999; Huggins, 1998; Fletcher, 1991; Fletcher, 1995; Websdale, 1992; Weinberger,1995; Wells & Triplett, 1992). Because the current study examines major events and occurrences relating to the DVCC, rather than the minutiae of individual actions and conversations, there is good reason to believe that the events portrayed here are reliable. Table 1. Interviewees’ Agency Affiliations and Domestic Violence Related Job Functions Interviewee’s Agency 1.State Dept. of Health 2.Domestic Violence Interventions 3.Hospital Emergency Room 4.Campus Sexual Assault Program 5.Battered Womens’ Shelter (BWS) 6.Magistrate’s Office 7.Office of State’s Attorney 8.Victim Witness 9.Legal Aid 10.Catholic Charities 11.Department of Social Services (DSS) 12.Family Violence Prevention Program/Department of Social Services (FVPP) 13.Family Violence Prevention Program/Department of Social Services (FVPP) 14.Family Violence Prevention Program/Department of Social Services (FVPP) 15.Domestic Violence Responders/Department of Social Services 16.Campus police from local university 17.Local agency that conducts research and training in criminal justice topics. 18.City Police Department 19.City Police Department 20.City Police Department 21.City Police Department 22.City Police Department 23.Neighboring city’s police department 24.City government 25.County domestic violence resource center 26.Additional personnel Interviewee’s Title, or Domestic Violence Related Job Function 1. Works with localities to form and monitor fatality review teams. 2. Founder and director of organization that provides counseling for batterers and monitors batterers’ compliance with court ordered treatment. 3. Nurse: provides emergency medical services to victims of domestic violence. 4. Assists students who have been sexually assaulted and promotes sexual assault awareness and education to the campus community. 5. Directs battered women’s shelter. 6. Directs agency that issues warrants and emergency protective orders. 7. Attorney: prosecutes domestic violence cases. 8. Supervises unit that provides court information and accompaniment to victims. 9. Attorney: provides legal services to domestic violence victims on civil matters. 10. Facilitates batterer groups for organization that provides a variety of service, including counseling services to victims and perpetrators, and emergency food and clothing. 11. Supervises all DSS direct service units. 12. Provides counseling and a variety of direct services to domestic violence victims. 13. Places and monitors batterers referred by court to batterer intervention programs and other programs as necessary. 14. Supervises unit that counsels and provides direct service to domestic violence victims. 15. Supervisors DSS unit that responds to domestic violence calls in tandem with police to assist domestic violence victims obtain information about and services from Social Services. 16. Supervises the department’s response to domestic violence calls on and in vicinity of campus. 17. Provides training in domestic violence to officers locally and throughout the state. 18. Chief. 19. Assistant Chief. 20. Detective: supervises Youth and Family Crimes Unit. 21. Detective located in Youth and Family Crimes Unit. 22. Sergeant acting unofficially as domestic violence point person. 23. Chief: provided historical information about the inception and early period of the DVCC. 24. Mayor. 25. Interviewee provided historical information about the inception and early period of the DVCC. 26. Interviewee was original DVCC coordinator and provides historical information about the inception and early period of the DVCC. The synthesized history presented here was formed from multiple interviews covering the same topics and events, allowing for any clear contradictions to be identified and resolved. There were no such contradictions with respect to any major historical events, although as would be expected, interviewees expressed differing interpretations of and reactions to these events. This is normal in this type of research and there is every reason to believe that what follows is a reliable portrayal of events. Twenty-four in person and two phone interviews were conducted. As the interviews were semistructured in format, they varied in length depending upon the particular respondent (i.e., some interviewees had more to say than others). Most interviews took at least one hour, the median was approximately two hours, and the longest interview lasted almost four hours. As with most research of this nature, a significant amount of time was spent scheduling interviews, followingup and rescheduling cancelled interviews, and completing partial interviews when it was impossible to complete the interview in one session. With one exception, all persons on the interview list were successfully interviewed. Between its inception in 1987 and this study, the DVCC’s membership had been reconstituted many times; thus, interviewing only present members would not provide the information necessary to produce a complete review of the DVCC’s complicated history. Since the study sought to capture many perspectives of the coalition and its history, present members from a variety of agencies were interviewed, as were past DVCC members who were no longer city residents. The latter were identified as part of a snowball sampling technique in which individuals were added to the interviewee list if they were frequently mentioned by other interviewees as being instrumental in past phases of the DVCC. 1 Several individuals were interviewed whose agencies were conspicuously absent from the DVCC even though their agency had a clear role in domestic violence service delivery. 1 Snowballing is a nonprobability sampling technique typically used when the sampling frame is unknown and members can be difficult to locate. One interviewee provides information that leads to another, so that, like a snowball rolling down a hill, the sample gathers size and momentum as the research progresses (Babbie, 1998). A central aim of the present study being to identify the role, activities and working relationships of the police, eight of the 26 interviewees were from local police agencies. This included officers from the city police department as well as officers from the campus police department of a local university. Information about current practices and activities of the police was gleaned from interviewees, ride-alongs in patrol cars, frequent contact with police personnel, and two focus groups, with four personnel each. The focus groups were conducted in two of the four police precincts. One focus group included supervisors as well as officers, and one consisted only of officers. Developing and Piloting the Interview Instrument The interview instrument was developed to capture information on interviewees’ experiences with and opinions about domestic service delivery in the city, emphasizing views on the police.2 Questions pertaining to the historical evolution of the DVCC were asked to interviewees who were knowledgeable about its early period. Also captured were subjects’ views of the police response to domestic violence in the city, the police department’s role in inter-agency work (including the DVCC), and interviewees’ impressions and experiences about working with the police department. The instrument contains seven sections. These are outlined below, along with an overview of the sub-topics covered in each section: 2 We thank Jane Sadusky of the Battered Women’s Project in Minnesota, for useful comments and suggestions during the instrument design stage. 1. Interviewee’s background in domestic violence work (e.g., how long has interviewee been involved with domestic violence work, in what capacity is interviewee involved, and how does interviewee’s role fit into the city’s service delivery system). 2. Interviewee’s agency’s involvement with domestic violence collaboration in general, through the DVCC, and with the police (e.g., what role does interviewee’s agency play in domestic violence service delivery, which if any agencies partner, or should partner with interviewee’s, does, and if so, how, does interviewee’s agency partner with police, and what are positive and negative aspects of police partnerships). 3. Interviewee’s involvement with the DVCC (e.g., to what extent, if any, does interviewee participate in the DVCC, and what is the nature of involvement). 4. Interviewee’s knowledge of DVCC’s history (e.g., interviewee’s knowledge and perspective of the evolution of the DVCC, reasons for its decline, and whether interviewee feels the resurged DVCC will avoid past pitfalls). 5. Interviewee’s perceptions of the current DVCC (e.g., what does interviewee feel are the positive and negative aspects to the current DVCC, does interviewee feel DVCC is meeting victims’ needs, and does interviewee feel DVCC is inclusive of member feedback). 6. Interviewee’s knowledge of and opinions about collaborative approaches to domestic violence (e.g., what inter-agency collaborations is interviewee aware of in the city, does interviewee believe collaborative approaches to domestic violence are effective). 7. Interviewee’s views of domestic violence services in the city (e.g., how does the interviewee feel the city addresses domestic violence effectively, what improvements need to be made, and has the presumptive arrest policy had a positive or negative effect on victims). The instrument contained 67 separate questions, most of which were open ended. It was pretested with three interviewees, which led to several modifications before being fielded with the remaining interviewees. The interviews were conducted in a semi-structured manor, allowing the interviewer to emphasize the topics most relevant to each interviewee’s position and experience. This allowed the quality and quantity of the information obtained from each interview to be maximized. For instance, although derived from the same interview schedule, the interview with the city mayor had a different emphasis than that with the police detective. All but one interview was conducted by the same interviewer. Chapter 3 Review of Literature: Evolution of Domestic Violence as a Policy Issue, and History of Domestic Violence Inter-Agency Response Joan Lucera, M.S. andErin Lane, M.P.M. POLICE FOUNDATION Prepared for the U.S. Department of Justice Office of Community Oriented Policing Services Award #98DVWXK015 June 2002 Chapter 3: Literature Review Recognition of Domestic Violence as a Public Policy Concern Historically, society tolerated and condoned men’s violence against women in intimate relationships. It was not until the end of the 19th century that “chastisement” could no longer be used as an acceptable defense to a charge of assault on a woman. Three states adopted anti-wife beating laws at that time, making it an offense punishable at the whipping post. Such laws were generally in name only, as they were infrequently and unequally enforced. When enforcement did occur, the sanctions imposed were mostly deployed against those on the fringes of society; usually blacks, vagrants, and the poor (Buzawa & Buzawa, 1990; Melton, 1999). By the beginning of the 20th century, the 48 American states as well as most of the Western countries had adopted legislation that outlawed wife beating. Through the 1960s and 1970s, social mores kept the unspoken approval of violence toward women intact, and nowhere was this more evident than in the “hands off” policy of the criminal justice system (Pence, 1996; Stelans & Lurigio, 1995). Police typically were reluctant to respond to domestic violence calls, and such calls were considered a low priority. When they did respond, an arrest was not likely to occur, as the most common law enforcement response was to separate the parties temporarily, or the use of advisement to restore peace and order. Officers felt it was not their business to interfere with family matters, and most felt that doing so was not real police work. Arrest and prosecution of offenders was generally viewed as a waste of time (Buzawa & Buzawa, 1990; Bryger & Edleson, 1987; Cordner, Gaines & Kappler, 1996; and Parnas, 1971). Courts often trivialized the victim’s experience, dismissing cases and giving lighter sentences to batterers than offenders who committed violent crimes against strangers. Use of these practices meant batterers were very rarely held accountable for their actions, thus perpetuating acceptance of violence against women (Davis & Smith, 1995). The women’s movement in the 1970s brought the issue of domestic violence to the forefront as a social concern. Grassroots battered women’s programs had already begun advocating for victims of domestic violence, providing them with safety and openly criticizing the ineffectual policies of the criminal justice system. The first hotline for these victims was set up in St. Paul, Minnesota, in 1971. The very first shelter for battered women opened in Chiswick, England, in 1972, with the first American shelter opening shortly thereafter in Pasadena, California (Pence, 1996; Steinman, 1990). Activists continued to open shelters, counsel victims and batterers, and educate the public about domestic violence, all of which led to heightened public awareness of the problem and a push to criminalize battering. The pressure was on both legislators and police to find and implement more effective solutions. Providing female victims of battering with shelter and access to available legal recourse continued to be the prime intervention strategies through the late 1970s. Demands for policy changes continued, most of them directed at the police because of their role as the first responders to battering incidents. Although police departments began to train officers in mediation of family violence, these incidents still were not treated as crimes, as police rarely made arrests. Advocates called for more attention and sensitivity for victims from the police, and for a response that would hold batterers accountable for their actions (Jolin & Moose, 1997; Steinman, 1990). While many states enacted domestic violence statues, police departments were resistant to the changes. The early 1980s brought several civil liability cases against police departments for their ineffective response to domestic violence calls. Thurman et al. v. City of Torington (1984) was the first case to find the police liable for failure to provide equal protection of the law guaranteed by the 14th Amendment. At the same time, Sherman & Berk (1984) published a landmark study indicating that arresting batterers was the most effective response to domestic violence calls, as it served as a deterrent to subsequent attacks. The groundbreaking findings of the Police Foundation, Minneapolis Domestic Violence Experiment concurred with the publication of the U.S. Attorney General’s Task Force on Family Violence report (1984) urging police to treat domestic assault as a crime. Legal requirements that limited officers from making warrantless arrests for misdemeanors they did not see were abolished. In response to the mounting political pressure, and with uncharacteristic speed, many police departments in the 1980s adopted presumptive or mandatory arrest policies for misdemeanor domestic violence assaults. Feminist and battered women’s groups lauded the change in policies, and these events gained recognition for domestic violence as a public policy concern (Gelles, 1993; Steinman, 1990). The Rise of Interagency Crime Prevention Work Despite later findings of several replication studies indicating that arrest alone was not a deterrent to continued domestic violence (see Dunford, Huizinga, & Elliott, 1990; Hirschell, Hutchinson, Dean, Kelley, & Pesackis, 1990), the tide had already turned. Increasing numbers of police departments continued to institute pro-arrest policies. Even so, a vicious cycle of ineffectual procedures continued: police would not lay charges, prosecutors did not aggressively pursue cases; and judges gave lenient sentences or dropped cases altogether (Dutton & McGregor, 1991). This fragmented response motivated advocates to develop interventions to domestic violence that actively coordinated the work of law enforcement, courts and social services. These community intervention projects (CIPs) are system-level strategies designed to ensure the responses of all involved agencies remain consistent, and do not overlap. Holding offenders accountable and keeping women safe from revictimization are the ultimate goals (Edleson, 1991; Gelles, 1993; Pence, 1983; Tolman & Weisz, 1995). Coordinated responses began to form such as arrest, aggressive pursuit of prosecutions, and the inclusion of the victims in the investigation and sentencing processes. In some projects, court mandated treatment for batterers was also an important part of the coordinated law enforcement, judicial, and social services efforts (Gelles, 1993). These efforts were in sync with the emergence of the community policing concept of proactive law enforcement which began to emerge in the early 1980s. By the early 1990’s, several coordinated interventions were in place, and evaluations of these programs had begun. Steinman (1990) examined the effects of a Domestic Violence Coalition in Lincoln, Nebraska, that incorporated pro-arrest policies and the commitment of the county attorney to prosecute whenever possible, with or without the cooperation of victims. Counseling agencies were also part of the Coalition, providing support for victims and batterer intervention programs. His study found that arrest alone led to increased violence by offenders, but that arrest in coordination with other efforts was a significant deterrent. Another study examined coordinated efforts in DuPage County, Illinois (Tolman & Weisz, 1995). The DuPage, County Domestic Violence Protocol combined a tough law enforcement approach with a social work perspective. Pro-arrest policies and aggressive prosecution were key, along with victim advocates who offered support and information for the victims about legal processes, other support services available, and worked with the prosecution to encourage victims’ cooperation. In addition, mandated participation in a domestic violence treatment program was usually required for first-time misdemeanor offenders. The evaluators found that in this program, arrest had a deterrent effect for at least 18 months, and recidivism decreased progressively for offenders who completed the most protocol levels (arrest, successful prosecution, and completion of a batterer’s program). The conclusion drawn from these studies is that coordination of criminal justice efforts can be an effective deterrent to domestic violence. This movement coincided with the new community policing philosophy, which, in sharp contrast to the traditional incident-based, reactive model of policing, takes a proactive approach to crime, and focuses on problem solving through partnerships. This philosophy calls for police to work more directly with citizens to solve crime problems before they escalate (Davis & Taylor, 1997), and was a good fit with inter-agency domestic violence work. History of the Coordinated Community Response Approach: The Duluth Model The earliest example of a coordinated, comprehensive response identified through a review of relevant literature was initiated in Duluth, Minnesota in 1980. This effort, known as the Domestic Abuse Intervention Project (DAIP) coordinated the domestic violence response of nine agencies, drawing from the areas of law enforcement, criminal justice and human services. The purpose was to provide a uniform response to domestic violence calls by creating and implementing a common set of policies and procedures that not only limited the discretion of individuals handling the cases, but also held them to a minimum standard. This response held the offenders accountable and offered them educational and support systems designed to change their behavior. Stopping violence against women was the ultimate goal, with victim safety being the focus. The DAIP emphasized that “even if society did not create battering, it must nonetheless accept the responsibility of stopping it” (Pence, 1983, p. 252). It was the first community intervention project to assign responsibility for ending the violence to the community, instead of the victims (Pence, 1983; Pence & Shepard, 1999). Before the actual creation of the DAIP, organizers met with administrators from police, city, prosecutors, mental health, and other agencies to ask them to join a unique interagency effort, and to determine the level of commitment on the part of administrators to that effort. Each agency was asked, and agreed, to help reduce repeat domestic violence assaults by developing policies and procedures and working together as an interagency coalition. Project organizers proposed to coordinate the city’s domestic violence response from the initial call to police through the end of the court’s involvement in each case. Evaluation of the coordinated response and problem-solving guidance would also be the responsibility of the DAIP. Once the agreement was secured, the three DAIP staff members and local shelter staff worked to construct an accurate picture of how the system was supposed to work, how it was actually working, and to identify necessary changes in agency procedures to make it work effectively (Pence, 1983; 1996). Again organizers met with the agency officials to negotiate the basics of each departments’ policy agreement. Once all participating agencies agreed to the protocol, a press conference was held to announce the new approach, which went into effect immediately. Operation of the Duluth Model The intervention process starts when police respond to a domestic violence call. Typically, the officers establish probable cause to arrest the batterer, and take him to the jail for booking. Officials at the jail then contact the battered women’s shelter, tell them about the arrest and charge, and give them relevant information for contacting the victim. A volunteer from the shelter then visits the victim at home to make a safety plan, explain the court process, and to help the victim decide whether to file for an order of protection. The volunteer advocate also talks with the woman about appropriate information to give to the judge before the next day’s arraignment. The victim is then encouraged to attend weekly educational groups for battered women. Meanwhile, a volunteer from the DAIP goes to the jail to talk to the offender about educational groups that can help him address his use of violence (Pence, 1996). At the arraignment (depending on whether the defendant pleads not guilty or guilty) either a pretrial release interview or a pre-sentence investigation is ordered, both of which involve the victim communicating her concerns to the pre-trail release agent or the probation officer. In most cases, the judge will base the sentence on the probation officer’s recommendations. First time offenders usually receive a stayed 30- or 60- day jail sentence, and are ordered to attend the DAIP 29-week educational program. Once on probation, the batterer has five days to make an appointment with the DAIP for assessment (Pence, 1996). If an order of protection is handled through the civil court, the abuser is normally not allowed to enter the residence for a full year and is ordered to cooperate with the DAIP educational program. Once ordered to participate in the educational program, offenders go through intake with DAIP staff for evaluation, and attend one orientation and 27 education sessions. If the offender violates the order for protection or fails to attend the educational groups, his probation may be revoked upon request from project staff. Some, or the entire original jail sentence may then be imposed, and the offender will again be ordered to complete the program upon his release from jail (Pence, 1996). In addition to the educational program for batterers, the project also offers advocacy and support to the victims. The DAIP holds orientation sessions for women whose partners are in the program, gives out information about what to do if an abuser reoffends, and informs victims of the availability of the shelter. From its inception, the DAIP has worked to coordinate the ongoing implementation of the project, promote institutional change, and continually monitor the city’s criminal justice response. Given the wide spread reluctance of the criminal justice system to get involved with domestic violence when this program began, it is understandable how the DAIP gained recognition and respect for their success in negotiating an agreement between local criminal justice agencies to coordinate their intervention efforts. Coordinated Community Responses Today Duluth set a precedent for coordination of a multi-agency response to domestic violence. This community intervention model was the first to use a detailed, complex plan made up of several different components organized to perform designated functions working toward a common goal. The coordinated responses that developed after the implementation of the “Duluth model” were not so inclusive at first. However, change did occur, albeit slowly, and coordinated responses have continued to grow. As the criminal justice system has expanded and improved upon its handling of domestic violence, awareness of the complexity of the problem and frustration with continually fragmented solutions has increased as well. Out of this awareness and frustration, more sophisticated coordinated community responses have emerged. Modern collaborative community efforts typically involve an organizing body, such as a committee, task force, or non-profit organization, working with a wide base of agencies and services in an effort to improve the multi-system response to domestic violence (Clark, Burt, Schulte, & Maguire, 1996; Little, Malefyt, & Walker, 1998). Health care providers, religious leaders, child welfare agencies, mental health care providers, substance abuse services, school officials, and business leaders have all, on some level, recognized the social and moral costs of family violence in their communities. Many have also recognized the importance of their role in the prevention and intervention of family violence, and in reshaping the response to and attitude toward domestic violence. The National Council of Juvenile and Family Court Judges (1998) recommended that communities have a coordinating mechanism that brings together agencies and community members in an effort to “close gaps in services, to coordinate multiple interventions, and to develop interagency agreements and protocols for providing basic services to families experiencing both child maltreatment and domestic violence.” In 1994, the American Medical Association endorsed recommendations from the National Conference of Family and Violence: Health and Justice conference workgroups that stressed the importance of “interdisciplinary collaboration, empowerment of victims, perpetrator accountability, violence prevention, and the strengthening of families and communities” (Witwer & Crawfort, 1995, p.17). This commitment on the part of government and professionals from many human and criminal justice service agencies, along with increased public awareness, lends support to improved collaborative responses to domestic violence. Types and Approaches of Coordinated Community Responses While coordinated community responses share a basic goal of ending violence against women within their respective communities, not all of these efforts share the same definition or approach to reaching their goals. What they do share is the same commitment for change, which necessarily requires each collaborative effort to develop commonly held beliefs on violence against women, awareness and understanding of each others’ roles; and a plan for improving the system-wide domestic violence response (Little, et al. 1996). A coordinated community response can take one of several approaches: • Community Partnering – Using this model, a domestic violence or victim services program identifies a strategic plan of action and partners with selected community agencies and organizations to work on each task identified in the plan. Work groups are formed, using the expertise of the members to accomplish the specific task assigned to the group. Through each step of the plan, the work is collaborative, with the domestic violence program overseeing the groups’ efforts. The decentralized nature of this approach is attractive to the community and professional members who are not decision-makers in their organization or community. Because membership is not limited to executive staff, members may be more likely to volunteer instead of being drafted, which can foster team building within the work groups. The work groups are focused on problem solving as opposed to being a public forum, which may minimize public posturing and turf issues. This approach lends itself to management by grassroots organizations, as it does not require a formal infrastructure. • Community Intervention – These are private sector programs that focus on increasing criminal justice system accountability to battered women. The program staff develops and monitors the coordinated effort to promote victim safety, hold offenders accountable for their use of violence, and increase the public awareness and intolerance of domestic violence. Intervention programs provide direct services to batterers through outreach, education, and treatment groups throughout the assailant’s experience with the criminal justice system. Many of these programs use the curriculum developed by the Domestic Abuse Intervention Project in Duluth. Intervention programs can evaluate the effectiveness of the system and make necessary modifications, keep open the lines of communication between each component, and develop as well as implement system wide protocols and policies. • Task Forces or Coordinating Councils – Task forces focus on coordination of all components of the criminal justice system to improve systemic practice to end violence against women. The task force will conduct a needs assessment of the community and produce a report on the most effective practices and identify gaps in the system. Each component will then work on creating protocols and compatible guidelines for practice within the system. Task forces or councils also function as forums for agencies and citizens to work on other collaborative projects, including cross-agency training, education, and data collection. Improving communication between agencies and coordination of council participants are important functions of the task force. • Training and Technical Assistance Projects – These projects are designed to inform and improve justice and human services practice through training curricula and audiovisual materials. Training and assistance is available for every aspect of these systems as well as for clergy, educators, health care providers, civic groups, public housing staff, and more. • Community Organizing – Community activists or domestic violence programs utilize these initiatives to invite the general public to actively work together to end domestic violence. Their objectives are to include people from all walks of life to be active participants in this work, articulate the message that each citizen has a personal responsibility to end the violence, and to change society’s beliefs and awareness about the causes of violence against women as well as what can be done about it. This type of coordinated effort has been most successful in engaging communities of color and other marginalized populations in working together to end domestic violence (Hart, 1995; Little, Malefyt, &Walker, 1998; Pence, 1983). Although there is no single ideal of collaboration, recent literature (Little, et al. 1998; Pence & Sheppard, 1999) has identified eight components or activities necessary for a successful coordinated response: 1. Creation of a shared philosophical framework on violence toward women; 2. Establishing “best practice” policies and procedures for intervention agencies; 3. Coordination of interagency communication and networking; 4. Tracking and monitoring of cases to ensure accountability; 5. Providing supportive community resources and services for victims; 6. Providing sanctions and rehabilitative services for batterers; 7. Protecting children who have been exposed to domestic violence; 8. Evaluation of the coordinated response from the victims’ points of view. Membership and Leadership The composition of the formal coordinating body may vary depending on the goals and focus of the coordinated response as well as the resources and population of each community. Ideally, participants should represent a diverse cross-section of criminal justice and social service agencies, as well as other segments of the community that may be affected by the group’s activities (Mattessich & Monsey, 1992). Members may include law enforcement, local battered women’s shelters, sexual assault services, mental health services, prosecution, court services, local government leaders, child protective services, batterer intervention programs, religious leaders, business leaders, health care providers, and members of the press (Burt, Harrell, Newmark, Aron, & Jacobs, 1997). Some coordinating committees choose to keep the focus narrow and membership limited. For example, the Baltimore, Maryland Domestic Violence Coordinating Committee focuses primarily on criminal justice issues and activities, with few representatives from other community agencies. Membership is comprised of two social service agencies, the city’s only domestic violence shelter, a sexual assault center, and senior staff from criminal justice agencies and judges. In contrast, the San Diego Domestic Violence Council has representatives from over 20 agencies and individuals, from law enforcement to the military (Clark, et al. 1996). The level of staff that participates in the group is another important component of the structure. Although many groups have primarily high-level staff, others consist of line workers and community members who form at a grass-roots level. An important consideration is whether an agency representative is able to make decisions or influence policy within their organization, which may allow for quick problem resolution. If not, having the backing and support of their agency is critical for successful collaboration within the system. Leadership is another important aspect to coordinated community responses. In a study of six coordinated community responses, Clark et al. (1996) found that leadership had a definite affect on the amount and ease of changes made by the respective groups. In the name of effective coordinated responses to domestic violence, agencies that typically have not worked together and frequently hold different views on the issue are now required to change their approach and do so cooperatively. This is not an easy task, but capable leadership can be the key to effective partnerships. Strong, consistent, and reliable leadership can motivate members to seek solutions. Once a dynamic leader is in place, stability of that leadership is also important for continuity of the coordinated effort. Conclusion This paper provides an historical overview of the evolution of domestic violence first into a public policy and then into a law enforcement issue. We have briefly reviewed the transformation of domestic violence from being considered a private matter to one deserving of public condemnation and punishment. We have also reviewed the confluence of this evolution with that of inter-agency crime prevention work and community policing, and discussed the convergence of these three streams into inter-agency domestic violence collaboration. In this condensed overview, we review the many types of inter-agency domestic violence collaboration that have developed since the initiation of the Duluth Model in 1980. This discussion sets the stage for part one of this report, which is a thorough examination of one city’s inter-agency response to domestic violence. References Brygger, M.P., and Edleson, J.L. (1987). Notes From Practice. The Domestic Abuse Project. Journal of Interpersonal Violence, vol.2, no. 3, 324-336. Buzawa, E.S. and Buzawa, C.G. (1990). Domestic Violence: The Criminal Justice Response Newbury Park, CA: Sage Publications, Inc. Clark, S.J., Burt, M.B, Schulte, M., and Maguire, K. (1996). Coordinated Community Responses to Domestic Violence in Six Communities: Beyond the Justice System. Urban Institute, Washington, D.C. Cordner, G., Gaines, L.K., and Kappeler, V.E. (1996). Police Operations: Analysis and Evaluation. Cincinnati, OH: Anderson Gelles, R.J. (1993). Constraints Against Family Violence: How Well Do They Work? American Behavioral Scientist, v.36, no. 5, 575-586. Hart, B.J. (1995). Coordinated Community Approaches to Domestic Violence. Paper presented at Violence Against Women Research Strategic Planning Workshop, sponsored by the National Institute of Justice in cooperation with the Department of Health and Human Services, Washington, D.C. Little, K., Malefyt, M.B., and Walker, A.H. (1998). Assessing the Justice System Response to Violence Against Women. A Tool for Communities to Develop Coordinated Responses. Violence Against Women Office, Washington, D.C. Pence, E. (1983). The Duluth Domestic Abuse Intervention Project. Hamline Law Review, 6, 247-275. Shepard, M.F. and Pence, E.L. (1999). Coordinating Community Responses to Domestic Violence: Lessons from Duluth and Beyond. Thousand Oaks, CA: Sage Publications, Inc. Sherman, L.W. and Berk, R. A. (1984). The Minneapolis Domestic Violence Experiment. Police Foundation Reports, April. Steinman, M. (1990). Lowering Recidivism Among Men Who Batter Women. Journal of Police Science and Administration, v. 17, no. 2, 124-132. Syers, M. and Edleson, J.L. (1992). The Combined Effects of Coordinated Criminal Justice Intervention in Woman Abuse. Journal of Interpersonal Violence, v. 7 no. 4, December 1992, 490-501. Tolman, R.M. and Weisz, A. (1995). Coordinated Community Intervention for Domestic Violence: The Effects of Arrest and Prosecution on Recidivism of Woman Abuse Perpetrators. Crime and Delinquency, v. 41, no.4, 481-495. Chapter 4 History of the Domestic Violence Coordinating Committee: 1987-2000 Erin Lane, M.P.M., Graham Farrell, Ph.D., andJoan Lucera, M.S. POLICE FOUNDATION Prepared for the U.S. Department of Justice Office of Community Oriented Policing Services Award #98DVWXK015 June 2002 Chapter 4: History of the Domestic Violence Coordinating Committee: 1987-2000 Introduction This chapter describes the emergence of an entity dedicated to improving inter-agency coordination in domestic violence service delivery, the Domestic Violence Coordinating Committee (DVCC), then charts its work and activity level from its formation in 1987 to the time this research was conducted in 2000. In addition to an overview of the history of the DVCC, this chapter includes a discussion of the factors found pertinent for success and failure of DVCC, and a concluding section where these themes are drawn into an analytical framework. Special attention is given to the role of the police department and its relationships over time with the DVCC and with the individual agencies in the domestic violence service community. The aim is to identify and present local lessons with broader applicability to the policing of domestic violence. The target audience is police chiefs as well as officials from government and community agencies involved in inter-agency domestic violence collaboration. The central lessons are summarized as follows: 1. Leadership: Dynamic and committed leadership by one key individual added enormous strength and momentum to the inter-agency effort from inception until the DVCC was well established. 2. Membership: (1) Early and steadfast police involvement was critical to the productivity of the effort. When this support was withdrawn and ties between the police and the DVCC lessened, the DVCC’s effectiveness was substantially compromised. (2) A wide representation of agencies was important to achieving systemic improvement to domestic violence response. (3) The DVCC was most successful when members held sufficient rank within their respective agencies to make decisions and mobilize resources. The DVCC became ineffectual when members were relatively low level. 3. Structure: Initially the DVCC benefited from an informal structure that allowed the group’s enthusiasm to flourish productively. Over time the structure solidified, but because the group was located advantageously where it was autonomous yet supported by a prestigious city office it continued to thrive. When it was relocated to a low-profile service provision agency, it declined. To some readers, it may appear that these conclusions are self-evident. But this does not mean that stating them and exploring the circumstances in which they emerged is unimportant. Although perhaps self-evident, the DVCC still fell victim to these pit-falls. Further, the apparent obviousness of these lessons does not make them less significant. These ideas were critical to the inter-agency effort discussed here and are probably relevant to similar efforts elsewhere. Until such apparently simple issues are overcome, it is unlikely that agencies will have the opportunity to address the more nuanced issues impeding the efficiency of interagency collaborations. In other words, the basics are important and need to be addressed first. Background Domestic violence is a crime that has gained considerable attention in the last three decades. Prior to the 1970s, domestic violence was barely a public concern. The victims’, women’s, and civil rights movements could be viewed as components of an overall social movement, through which domestic violence became a major concern for public policy. Domestic violence emerged as an important part of policing as a result of the increased recognition of wife abuse as a public policy problem in the late 1960s and early 1970s. Advocates began calling for the police to be more attentive and sensitive to victims needs and to respond in a way that would hold batterers accountable for their actions (Hirschel & Hutchison, 1992; Jolin & Moose, 1997; Steinman, 1990). At this time, mutli-agency work toward solutions for specific crime problems such as domestic violence was a fairly new concept. In 1973, the U.S. National Advisory Commission on Criminal Justice Standards and Goals endorsed multiagency work in their report entitled “A National Strategy to Reduce Crime,” which presented recommendations to police, courts, corrections, and community crime prevention on implementing new crime-reducing strategies within each area. In the chapter entitled “Police,” a specific section is devoted to recommendations on coordination with other criminal justice agencies, which reads in part, No element of the criminal justice system completely discharges its responsibility simply by achieving its own immediate objectives. The police, the prosecutor, the courts, and probation, parole, and corrections agencies must cooperate with each other if the system is to operate effectively. This requires an effort on the part of each element to communicate with the other elements (National Advisory Commission on Criminal Justice Standards and Goals, 1973, p. 89). The earliest example of a coordinated, comprehensive response to domestic violence was initiated in Duluth, Minnesota in 1980. This effort, known as the Domestic Abuse Intervention Project (DAIP) coordinated the domestic violence response of nine law enforcement, criminal justice, and human services agencies. The purpose was to provide a uniform response to domestic violence calls by creating and implementing a common set of policies and procedures that not only limited the discretion of individuals handling the cases, but also held them to a minimum standard (Jolin & Moose, 1997; Pence, 1983; Shepard & Pence, 1999; Steinman, 1990). Limiting individual discretion in this way was considered an important component of ensuring that services were provided consistently. Following the implementation of the “Duluth Model,” coordinated responses began to form that included arrest, aggressive pursuit of prosecutions, and the inclusion of the victims in the investigation and sentencing processes. According to one researcher, this type of collaborative effort provided police with an opportunity to implement their most effective response to domestic violence in the history of this country (Gelles, 1993). Active participation of the police is an integral part of any effective multi-agency response to domestic violence. This is especially true in terms of the ability of the police to provide information and expert advice in relation to a crime problem. In some areas, local police were instrumental in the initial creation of a more coordinated response (Liddle and Gelsthorpe, 1994b). By the early 1990s, several coordinated interventions were in place around the country, in which police departments were key partners. These themes were especially relevant to the development and success of the subject of this chapter, the Domestic Violence Coordinating Committee. Findings The Emergence of the DVCC In the late 1980s, a captain in the homicide unit of the city police department was considering how police resources could be most effectively used to reduce the city’s high homicide rate, which was a top priority of the police department. In reviewing data, he noted that domestic violence homicides were often preceded by multiple police visits to the location, and thus, concluded that earlier intervention into domestic violence cases provided an opportunity to prevent these homicides. Serendipitously, his interest coincided with the inception of an interagency domestic violence initiative, the DVCC, which sought to enhance the coordination of agencies responding to domestic violence. The DVCC was the crystallization of the city’s concern about the violent crime rate, which had evolved into a focus on domestic violence. This focus gave rise to the DVCC in 1987 when the city created a new domestic violence coordinator position to supervise two domestic violence counselors based at court; undertake domestic violence public awareness, education, and training efforts; and to develop a ‘coordinating council’ relating to domestic violence issues in the city. It is this coordinating council that became the DVCC. In 1988 the newly hired coordinator convened a meeting of 24 local agencies that were involved in domestic violence service delivery in the city to discuss locally available resources. These individuals (including the above mentioned homicide captain) began meeting regularly, and ultimately formed a coalition known as the DVCC. The DVCC’s membership was broad and meetings were well attended. In addition to private attorneys, private clinicians, interested citizens, several domestic violence victims, and at least one judge, membership included representatives from: • Police department • Magistrate’s office • Prosecutor’s office • Department of social services • Agency that ran the battered women’s shelter (BWS) • Planned Parenthood • Department of public schools • Victim/witness program • Hospital • Mental health department • Substance abuse counseling agency • Public housing department From the start the DVCC had a strong presence in local domestic violence work because most of the participants were policy makers in their respective agencies, as well as personally motivated to address domestic violence. The fact that coordination of the DVCC was part of the paid position of a city employee was also an important factor in its early stability. The original DVCC mission, as stated by one of the original members, was to “harness the efforts of community groups and agencies to identify and reduce the incidence of domestic violence,” and their strategy was to develop resources and to identify and address gaps in service. Another long-time DVCC member, who now runs a private batterer counseling organization, reported, “the point of the DVCC was to create a bumper so that anyone who entered the system at any point will receive services.” Establishing a continuity of service was a related goal of the DVCC. One of the initial areas of agreement between DVCC members was the need to improve service to victims by integrating the criminal justice/law enforcement response. Thus, the police department’s role in and cooperation with the group was identified as critical at the outset. The city police department was particularly active from the beginning, but as we shall see, over time the police role declined substantially. Between 1988 and 1989, the inter-agency group took shape officially as the DVCC, initiating several domestic violence education projects as it did so. After its initial development as an informal group, the DVCC obtained official status as a non-profit organization on November 10, 1988. History of the DVCC 1987-2000 There has been significant variation over time in the number of active committee members, the vitality, nature of activities undertaken, and the extent of the DVCC’s influence on the local community. A graphical characterization of the DVCC’s history is shown as Figure 1, where the distinct phases of its history are described in language intended to capture the essence of these periods, such as ‘heyday’ and ‘rockbottom.’ Figure 1: Characterization of the History of the Domestic Violence Coordinating Committee (DVCC): 1987-2001 Line Graph Showing Characterization At its peak, the coalition claimed a plethora of active members working as an influential force in the community, while at its low-point, the coalition was all but dormant. Throughout, the DVCC has been a community body comprised of an array of local agencies that provide service to domestic violence victims and offenders. Because the lessons from the DVCC’s times of strength and weakness may usefully inform the work of agencies and similar coalitions elsewhere, we present a chronology of the life of the DVCC, and explicate the factors associated with its productive as well as its dormant periods. A. 1987-1988: Background to, Birth of, and ‘Gearing Up’ of the DVCC Strength through Informality Describing the early stages of the DVCC, the founder and first coordinator commented about membership and organizational strength: I think the reason it was as strong as it was, for as long as it was, is because it was a very inclusive organization – we had no power and no money, so we didn’t have to be exclusive. We did not have to set up an artificial hierarchy with a board of directors and so on. We could invite the world, and have everybody who showed up be equal. It was truly a democratic model for how to run a community-based organization. It involved everyone who could possibly be identified as a stake -holder – organizationally and individually. We made a really consistent attempt to have constituent involvement, and we did have battered women themselves come in and out of the organization. The first police representative on the DVCC was a captain when he first became involved with the newly forming DVCC, and remained an active DVCC member until his retirement as assistant chief six years later. His comments on the initial successfulness of the DVCC echoed those of the first coordinator. He suggested that the DVCC’s productivity was perhaps because there was very little in-between ideas and action, as the group did not have any internal or external bureaucratic hoops to jump through prior to taking action. Early and consistent police involvement In describing the strength and effectiveness of the DVCC, several interviewees explicitly attributed much of the success to the strong presence of the Police Department. This support lent political legitimacy as well as practical value to the DVCC’s efforts. The homicide captain, later assistant chief, was the man widely recognized as responsible for initial police interest because he was involved from the beginning and ultimately served for years on the DVCC’s board of directors. His initial motivation for involvement, as noted by himself and others, was a professional desire to reduce domestic violence homicides and a belief that working collaboratively with domestic violence service agencies was an effective means to that end. Many interviewees opined that he had the combination of rank, personal zeal, and professional commitment to produce a powerful leverage of police department resources. One significant contribution was enlisting the support of the police chief. This now-retired police official referred to the police department as one of the original “movers and shakers” in the DVCC. Although reducing the homicide rate was the impetus for the police department’s involvement with the DVCC, the department’s interest grew over time into a general commitment to improve domestic violence service delivery. The strong role of the police throughout the early and most active years of the DVCC was exemplified by donations of meeting room space, funds for printing material, and collection and sharing of crime data. In addition, police management exerted pressure on officers to take domestic violence seriously and make arrests. For example, supervisors reportedly reviewed officers’ domestic violence reports and followed-up on cases where arrests were not made. The close relationship between the Police Department and the DVCC was manifested in several ways notably that the DVCC provided training for the police department, at several points in its history. First, in 1988, the DVCC provided in-service domestic violence training to police department officers, and the department encouraged participation by offering officers continuing education credits to attend. The DVCC again trained city police officers between 1990 and 1991 after a presumptive arrest law went into affect in the state. During this two-year span, the DVCC provided in-service training on the new law for all city Police officers (at that time there were about 600). The training was an intense half-day session that included in-person testimony by domestic violence victims. B. 1989-1994: The DVCC’s ‘Heyday’ The DVCC became more secure in 1990 when the city manager’s office obtained a grant to fund an employee dedicated to violence prevention. The woman hired to, among other duties, develop a domestic violence coordinated council (what became the DVCC) accepted the position and continued her role on the DVCC, melding responsibility for coordinating the group into her new position. Several interviewees attributed the DVCC’s success in becoming central to local domestic violence service delivery to her skilled leadership. In addition, the new position within the city manager’s office, provided status, security, and funding to the DVCC. From 1989 to 1994 the DVCC was extremely active and visible in the community. Ongoing activities included producing domestic violence education materials to insert into utility bills, producing and distributing domestic violence resource directories, successfully launching a new, coordinated approach to domestic violence, training police officers, YWCA volunteers and clergy, producing a video for training police recruits in handling domestic violence, cosponsoring a lobby day with a local non-profit domestic violence group, publicizing domestic violence awareness month, and hosting “mini-conferences.” In addition, members represented the DVCC at various anti-domestic violence educational/publicity events and appearances around the city. Mini conferences Each year the DVCC put together a series of half-day conferences on a range of domestic violence-related topics. Each mini-conference included speakers and workshops around a specific domestic violence-related topic. Conference fees were minimal, and conferences were always well attended, usually by 25 to 30 people. At least one conference had almost 70 attendees. Some conferences aimed to enhance participants’ understanding of the effects and implications of domestic violence upon different populations. Other conferences aimed to assist professionals meet the needs of domestic violence victims and batterers, by offering training in recognizing victimization and information about services available in the city. Domestic Violence Initiative (DVI) For the first part of the DVCC’s existence, members focused on domestic violence education and advocacy. As they collaborated together in new ways, questions about domestic violence services and response emerged, such as: What accounted for the disparity between the high number of “domestic trouble” calls to which the police department reported responding each day and the low number of similar cases heard at juvenile and domestic court each day?; What guidelines should magistrates use to determine release times and bond levels?; Why did warrant and protective order decisions depend on which magistrate was petitioned?; Are batterer counseling programs effective at reducing recidivism, and what was the best way to measure results? Once the DVCC was “geared up,” it was prepared to tackle these questions, as it strove to fulfill its original purpose: enhancing inter-agency, domestic violence collaboration. With many issues and shortfalls to efficient integrated domestic violence response identified, the DVCC set about effecting the improvements they believed were needed. This took the shape of a new effort called the Domestic Violence Initiative (DVI). The DVI was launched in 1990 by DVCC member and non-member agencies, with the DVCC at the helm. The effort was comprised of three main activities: • Ensuring that each facet of the law enforcement/court system including police, magistrates, prosecutor’s office, court intake, judges, and service providers, had written domestic violence polices and procedures in place. • Training for personnel employed in each branch of the criminal justice system. • Development of a feedback system for evaluating the effectiveness of the policies and procedures. Many, if not most, of the key city agencies took part in the DVI, and in doing so, made important changes in their domestic violence policies and procedures. These changes are summarized in Table 1. Table 1: Changes in Agency Policies and Procedures, c. 1991 Agency -Changes in Policy or Procedure Police Department -Produced cards to give to domestic violence victims that included police officer information, and other information on legal and social services assistance. -Instituted reporting form to be used for tracking domestic violence and ensuring that officers took appropriate action. -Began feeding information on domestic violence incidents to the battered womens’ shelter for follow-up. -Attempted to institute presumptive arrest policy, but could not because other agencies did not have similar policy. -Launched academy and in-service training for officers in domestic violence and victimology. Magistrate’s Office -Policy was amended so that arrest warrants were issued for all acts of violence. Prosecutor’s Office -Institutionalized training for all attorneys in victimology and in assisting victims with the court process. -Policy changed to prosecute all misdemeanor domestic violence cases and not to drop cases at complainants’ request. -Policy changed to refrain from prosecuting complainants for refusal to testify. -Procedure changed to meet with complainants before hearings to help them understand process and hear their concerns. Juvenile and Domestic Court Services -Training for intake workers implemented. -Policy changed to consolidate cases involving the same individual into one hearing, where possible. -Complainants seeking protective orders are prioritized over other cases. Battered Womens’ Shelter -Began contacting domestic violence victims to offer services and assistance, using information provided by police Men’s Abuse Control Program -Changed batterer counseling programs to require a fee instead of being free, since experiences suggest that perpetrators take groups more seriously when they have to pay. Reviewing progress of the Domestic Violence Initiative In order for the DVI to be effective, the DVCC believed that regular and systematic reviews and documentation of its progress were important, and the requirements to do this were explicitly established at the outset. Members of the DVCC assessed and documented the work of the initiative in several ways. First, in 1992, they compiled the details of the initiative into a bound written volume. The volume included background information about the DVCC and detailed descriptions of each participating agency’s role in the initiative. In 1993 a review of the DVI was held where officials of participating agencies convened to monitor, assess, and summarize progress so that participants could see who had done what and with what results. The review was attended by influential representatives from key agencies, such as the police homicide captain, the chief of police, several judges, the chief magistrate, and a variety of local service agencies. Members convened and reviewed areas where community services had increased, and agency representatives discussed changes in their agencies. Police representatives discussed the filling out of reports, judges talked about their annual training sessions, and magistrates reported policy changes. It is unclear for how long these changes endured. As will be seen in the next section, the DVCC’s vitality dropped off precipitously in the few years after the first DVI review. In sum, the DVCC began with the appointment of a committed and able leader who had the ability to organize members as well as to effectively handle public relations. Once funding was secured from the city manager’s office the DVCC had financial security and political clout. Many local agencies became involved and continued their involvement. In particular, the police played a prominent role. The DVCC sponsored a range of educational activities and became a central figure in the domestic violence service delivery community. At the pinnacle of the DVCC’s strength, it launched the DVI, which was a multi-agency effort to improve city-wide policies and responses to domestic violence. In sum, the formation and success of the DVCC was accomplished by a small group of committed actors working at the grassroots level, whose efforts were supported by many city agencies. C. 1994 – Turning Point Around 1994-1995 the DVCC’s momentum and membership began to wane. There are several competing, but not mutually excusive, explanations about what broke the DVCC’s stride and then perpetuated its decline. The contending explanations are: (1) leadership problems, (2) departure of several key members and internal competition, and (3) a bureaucratic change within the city that detached the DVCC from the purview and status of the city manager’s office. Together it is likely that these factors came together to produce a ‘turning point’ (c.f. Gladwell, 2001) resulting in the rapid downturn of the DVCC and inter-agency domestic violence work in the city. Leadership problems In 1992, the original chairperson left the city for personal reasons, and responsibility for coordinating the DVCC shifted across a series of members elected by the DVCC membership. This was a major set-back to the DVCC because all at once it lost two of its most important assets: a dedicated, well-connected, and beloved leader, and its explicit connection to the city manager’s office, which had provided funding and status. From this time on, the DVCC did not have anyone whose paid job description included coordinating the DVCC. Further, the DVCC’s status was marginalized after it was no longer under the protective wing of the city manager’s office. The original DVCC chairperson was well connected within the city and commonly recognized as a dedicated and charismatic leader with ample political savvy to build an effective DVCC underpinned with strong community and government support. She was also the only person to hold this position as part of paid employment. It may be in large part due to an unavoidable comparison with her that none of the three subsequent chairpersons escaped criticism from interviewees. Each of them and their role is described in turn below. The second DVCC coordinator was purported to be a good leader, but overwhelmed by the work. This coordinator lasted one year. The third was an active and longtime DVCC member who was said by interviewees to have proved herself again and again, both as hardworking and dedicated to the DVCC, and to the cause of addressing domestic violence. Unlike the first chairperson, neither the second nor third was mandated to undertake the work as part of their paid employment. Hence, in retrospect, it was almost inevitable that the time, energy, and commitment they could offer were at a lower level. Nonetheless, this coordinator worked diligently for the DVCC both internally within the group and externally by representing the DVCC at various media events. Ostensibly, the group remained strong throughout her term, though its ties to the city were becoming more tenuous. Several interviewees reported that the fourth coordinator was also an extremely diligent and driven worker who was passionate about the cause; however, those interviewed felt that her strident and caustic manor alienated many would be allies and committee members. At a time when the DVCC needed a well-connected leader to mitigate the damage of losing the high-status parent of the city manager’s office, they were led by a woman who focused on community outreach and education. With the DVCC drifting and loosing momentum, interviewees reported that the DVCC was “an unproductive way to spend your time.” This may explain why many agencies, the police included, began to withdraw from activities and meetings. It did not take long until DVCC attendance and membership hit an all time low of four or five members in 1997-1998. Key members move on and there is internal competition among members Concurrent with the departure of the first coordinator, several additional central members also moved on for various reasons. One of these was the previously discussed police official, who moved to become a police chief in another city. He was the primary connection with the police department since the outset of the DVCC, and his absence created a void that was never filled. The director of BWS and the chief magistrate also terminated their involvement with the DVCC. Losing these members who were knowledgeable about, interested in, and committed to the DVCC, was a set-back; however, this could have been mitigated if their replacements were of comparable stature. Unfortunately for the DVCC, this did not occur. A police sergeant replaced the assistant chief. BWS did not designate a definite replacement representative, but instead sent staff to the DVCC meetings depending on who was available. The chief magistrate ceased his involvement, apparently due to conflicts in the DVCC pertaining to questions about whether his agency was responding adequately to domestic violence. Purportedly, he was frequently taken to task in DVCC meetings for the unresponsiveness and inconsistency of the magistrates in issuing protective orders and warrants. As a consequence of what he felt were personal attacks, he terminated his involvement. While these developments were in and of themselves harmful to the DVCC, they were indicative of a larger problem – a lack of commitment on the part of agencies. A police official familiar with the police department’s history with the DVCC noted that, “for any group to be successful, there needs to be buy-in from the department heads. I think that’s what the DVCC was lacking.” and other interviewees also referred to “lack of ownership” and “lack of support” as severe problems. Another membership issue, mentioned several times in interviews, was personality conflicts and competition among members. One member reported: Lack of ownership was the beginning of the downfall. Then the differences in philosophy and ideas about the purpose of the group continued that. Personality conflicts did not help either. Several interviewees commented specifically about the problem of individual agendas, one referring to the deteriorating situation as a “push – my – own – agenda game.” Determining which of these dynamics instigated or played the largest role in the decline of the DVCC is difficult. As often with such events, the specific sequence of events, or direct consequences of any one event, remained unclear even after many in-depth interviews. It is hypothesized that many of these factors were inter-related, and the decline due to a combination and confluence of them. Bureaucratic changes set DVCC adrift Initially the DVCC was supported by a funded coordinator position, and a safe location within a city agency. Two years after its inception, it gained status and legitimacy when responsibility for coordinating the committee was transferred to a new and advantageous location within the city government, the city manager’s office. In 1992, when the original DVCC coordinator resigned her position, the strength of the relationship between the DVCC and the city was diminished, because her position was not filled, and thus the DVCC no longer had a paid coordinator or an explicit connection to the city manager’s office. This office still supported the DVCC, however, by paying for mailings and was still informally connected to the DVCC via an active committee member who was employed therein. Time gradually eroded this tenuous connection until late 1995 when a new city manager reorganized portions of the city’s bureaucratic structure, and in doing so, severed the last functional connection between the DVCC and the city. At this point, there was no longer anyone in the city manager’s office who was affiliated with the DVCC, much less paid to coordinate it. When interviewed for this study, an original member and excoordinator stated that the DVCC “could not continue when separated from the city manager’s office.” Police department involvement after the turning point After the exit of the assistant chief, police department involvement in the DVCC declined. The sergeant tasked to represent the police department on the DVCC worked at the training academy, and, for the duration of her time on the DVCC, the committee continued to hold its meetings there. When this sergeant was transferred within the police department in 1997, the DVCC no longer met in police facilities, and police attendance at DVCC meetings ceased. The reduced police involvement with the DVCC also corresponds to the tenure of a new police chief, appointed in 1995. When interviewed, the chief stated that he did not believe that police involvement with the committee was productive and expressed skepticism about police involvement in addressing domestic violence. Several other interviewees stated that the subsequent priorities of the police department did not appear to include domestic violence. D. 1996-1998 – ‘Rock Bottom’ and ‘The Wilderness Years’ By 1996, DVCC membership had dwindled to four or five members. Some members had unrelated personal reasons for moving on. Others, however, simply felt that the group was no longer productive. One agency official noted that “the DVCC was basically a non-entity.” While the DVCC was largely absent from the city’s anti-domestic violence community, a new initiative emerged, which we term, the “Family Violence Working Group” (FVWG). The police department was a leader of the FVWG, which had one purpose: to determine the most effective use of domestic violence funds received by the police department from the U.S. Department of Justice Office of Community Oriented Policing Services and to spend the funds accordingly. A police sergeant in charge of obtaining and administering grants led the group. Other participants included representatives from Social Services, the BWS, the Prosecutors Office, the police department, and the State Supreme Court. Ultimately, the FVWG decided to produce several domestic violence education videos. One was to be used for training police recruits in the proper handling of domestic violence calls, and one was to help domestic violence victims understand the court process. Two public service announcements were also produced, one that warned against committing domestic violence and one that provided hotline numbers to victims. As soon as the project was completed, in 1999, the FVWG disbanded. In describing the impetus to launch the FVWG, one social services official stated: “Because we knew the DVCC was not capable of making any decisions, the public agencies at that time got together and started the [FVWG].” The scope of the FVWG was quite limited and the FVWG was never intended to lead to sustained inter-agency cooperation. Still, its existence indicates that inter-agency domestic violence work continued in the city, on some level. E. 1999-2001 – Resurgence During 1999, the DVCC began a resurgence. The energy and dedication of several individuals with institutional memory of the past success and value of the DVCC lay behind this attempt. The revitalization was initiated by a supervisor at Department of Social Services with responsibility for domestic violence programming. Realizing that there was no longer a viable coordinated response, she tasked a domestic violence worker with leadership experience to “get the DVCC going again.” Subsequently, meetings were held almost monthly, and interest reemerged. Almost two years after the process began, the DVCC was on the verge of again being an active part of the domestic violence service delivery system. While this development is promising, several interviewees expressed concerns about the newly emergent DVCC. When asked what made the DVCC weak, one interviewee stated: “We [the DVCC] don’t belong to anybody. There is no supporting agency to fund the group,” while another opined of the “lack of backing from the powers that be.” However, at the time of this writing, a wide range of agencies were participating in DVCC meetings, including the police department, along with individuals who had been committed to the DVCC throughout its history. One interviewee reported, “At this point the DVCC is 150 times better than it was.” Discussion and Conclusions While the scope of the study does not allow us to identify members’ specific operational decisions and actions that affected the fortunes of the DVCC, it is possible to discern several distinct themes. These are that particular aspects to leadership, membership, and group structure were critical to making and breaking the DVCC. Most likely a combination of these factors converged to initially, make the DVCC successful, and later, after certain key changes, to bring about its decline. Initially the DVCC benefited from a skilled leader who was able to focus the energy of diverse individuals, into a productive and respected organization. Though the group had many dedicated members committed to the cause of addressing domestic violence, upon this leader’s retirement, none were truly able to fill her shoes, and the void dragged down the forward momentum of the DVCC. Membership was also an important ingredient in the DVCC’s success. For the first several years of the DVCC’s existence a broad range of agencies were represented in its membership, including law enforcement and other criminal justice agencies. While this was the case the DVCC was productive and was even able to launch an ambitious multi-agency effort to improve response to domestic violence, which involved many agencies engaging in additional training, updating policies, and working together toward a more seamless service delivery system. The police in particular were very active throughout these early stages. When a few key members, representing key agencies left the DVCC, and their agencies were either no longer represented or were represented by less committed and or low-status individuals then the momentum was further undermined. The final factor identified as important to the DVCC is its structure. Initially, the DVCC capitalized on its informal structure by allowing the energy of the original members to generate and act on ideas. But this does not mean that maintaining this structure was necessary for the DVCC’s continued success. In fact, the DVCC continued to thrive after the structure became more formalized, now benefiting from new advantages, such as a prominent location within the city’s government where it was autonomous and supported. It was only when it was moved to a disadvantageous, non autonomous and non supported location within a service delivery unit when structure worked against it and further momentum was diffused. Implications for Policing In their review of inter-agency crime prevention efforts work, Webb and Laycock (2000) present telling summaries of the problems and pitfalls that inter-agency groups often encounter. Their insightfulness into ‘partnership pitfalls’ makes them worth revisiting in the present context, and their (vaguely satirical) typology is shown as Figure 2. Figure 2: Laycock and Webb’s Typology of Early Partnership Pitfalls • Who’s in charge here? • It’s not our job, it’s yours • Your priorities are not mine • We’d like to help but it would reduce our profits • We’ve got no money, can we have some of yours? • I’m in from the local voluntary group – what can I do? The Laycock and Webb (2000) analysis was based upon many years of effort implementing local inter-agency initiatives, and their discussion goes further than the typology shown here. However, the relevance of the ‘pitfalls’ typology to the present discussion is evident: The DVCC in our case study thrived when it overcame the pitfalls, and declined when they gained prominence. Many of the specific aspects highlighted by the Laycock and Webb typology were evident in the present case study: Competing agendas and personalities, funding issues, leadership issues, individual abilities, and political clout, were among the key aspects that dictated the rise and fall of the inter-agency domestic violence committee. In the present study, it seemed that several of the issues were inter-related: as funding ended, leadership shifted leading to problems in the agenda and work of the DVCC. Yet the picture was far more complex than to simply conclude that funding is everything. It is important, but it is not everything. Rather, efforts need to be made to encourage the convergence of factors that ensure such an inter-agency effort becomes, and remains, productive, producing a tipping point. This should typically mean ensuring that high-level support is gained in key agencies, particularly the police and social services. Where that support is absent, it should be sought via different channels. A reluctant police chief might be nudged into action via different approaches: perhaps simply some basic lobbying through letters and calls from prominent players in other agencies. Figure 3: Inter-Agency ‘Tipping Point’ Factors A. Key Precursors to Development and Growth: • Ability: leadership, coordination, knowledge, connections, dedication • Motivation and ability to motivate others • Other key motivated personnel • Financing • Political clout • Police involvement These positive factors were the precursors of a positive cycle of reinforcement that meant the DVCC went from strength to strength. B. Key Precursors to Decline and Decay: • Loss of leadership without equivalent replacement. • Loss of financing • Loss of political clout (the link to city manager’s office) • Loss of police contact person without equivalent replacement • New police chief with ‘old school’ views on domestic violence These negative factors were the precursors of a vicious cycle of decline and decay that resulted in loss of direction of the DVCC, and ultimately in several years without credibility or much activity. Stronger means could incorporate applications to the city manager’s or mayor’s office, who typically have influence upon the police department. What are the implications of this study for police departments and anti-domestic violence policing efforts across the country? Figure 3 summarizes some of the findings from the present study as key ‘tipping point’ issues that induced the ebb and flow, respectively, of anti-domestic violence efforts in the city.3 The summary factors shown in Figure 3 are taken from our case study of a single city. It shows the issues relating to the production of both the positive and negative ‘tipping points’ in relation to the inter-agency domestic violence work and policing. These issues almost certainly have a broader relevance to inter-agency domestic violence-related work and police work across the nation, as well, perhaps, to more general inter-agency crime control and policing efforts. Police departments can clearly benefit from the knowledge and expertise of other agencies in the community. At this point we have returned to the opening sentence of this report – inter-agency cooperation is a foundation of modern policing. However, we add to this statement the rider that the police should make efforts to ensure the convergence of positive tipping point factors, and discourage the convergence of negative tipping point factors. Inter-agency cooperation in policing needs to be tailored to the local context, individuals, and agencies for it to be productive. These are the significant broader implications of the present study for American policing. 3 The ‘tipping point’ concept is adapted from Gladwell (2000) who used it to refer to a convergence of factors, rather than a single factor, producing the momentum for the sharp reduction in the New York City crime rate during the 1990’s. References Babbie, E. (1998). The Practice of Social Research, 8th Edition. Belmont, CA: Wadsworth Publishing. Brookman, F., Noaks, L., & Wincup, E. (1999). Qualitative Research in Criminology. Aldershot, Hampshire: Ashgate Publishers. Fletcher, C. (1995). Breaking & Entering: Women Cops Talk About Life In The Ultimate Men’s Club. New York, NY: HarperCollins. Fletcher, C. (1991). Pure Cop: Cop Talk From The Street To The Specialized Units: Bomb Squad, Arson, Hostage Negotiation, Prostitution, Major Accidents, Crime Scene. New York, NY: Villard Books. Gelles, R.J. (1993). Constraints against family violence: How well do they work? American Behavioral Scientist, 36 (5), 575-586. Gladwell, M. (2000). The Tipping Point: How Little Things Can Make a Big Difference. New York, NY: Little, Brown and Company. Hirschel, J.D., & Hutchison, I.W. (1992). Female Spouse Abuse and the Police Response: the Charlotte, North Carolina Experiment. The Journal of Law and Criminology, 83 (1), 73-119. Huggins, M.K. (1998). Political policing: the United States and Latin America. Durham, NC: Duke University Press Jolin, A., & Moose, C.A. (1997). Evaluating a domestic violence program in a community policing environment: research implementation issues. Crime and Delinquency, 43 (3), 279 298. Liddle, M.A. & Gelsthorpe, L.B. (1994b). Crime Prevention and inter-agency cooperation. Crime Prevention Unit Series, Paper 53. London, UK: Home Office Police Department. Pence, E. (1983). The Duluth Domestic Abuse Intervention Project. Hamline Law Review, 6, 247-275. A National Strategy to Reduce Crime. (1973). US National Advisory Commission on Criminal Justice Standards and Goals. Washington, DC: US Government Printing Office. Shepard, M.F & Pence, E.L. (1999). Coordinating Community Responses to Domestic Violence: Lessons from Duluth and Beyond. Thousand Oaks, CA: Sage Publications. Steinman, M. (1990). Lowering Recidivism Among Men Who Batter Women. Journal of Police Science and Administration, 17(2), 124-132. Websdale, N. (1992). Female suffrage, male violence, and law enforcement in Lane County, Oregon 1853 to 1960: an ascending analysis of power. Social Justice, 19, (3), 82-106. Weinberger, B. (1995). The Best Police In The World: An Oral History Of English Policing. Hants, UK: Scholar Press. Wells, T. & Triplett, W., (1992). Drug Wars: An Oral History From The Trenches. New York, NY: William Morrow. Chapter 5 The Current State of the Domestic Violence Coordinating Committee: 2000-2001 Erin Lane, M.P.M. and Joan Lucera, M.S. POLICE FOUNDATION Prepared for the U.S. Department of Justice Office of Community Oriented Policing Services Award #98DVWXK015 June 2002 Chapter 5: The Current State of the Domestic Violence Coordinating Committee: 2000-2001 Introduction Now that the groundwork has been laid in the previous chapter of this report for understanding the evolution of the Domestic Violence Coordinating Committee (DVCC) through its first 13 years of existence, the view is broadened to the DVCC within the context of the domestic violence service community. This description is based on the community of agencies involved in domestic violence service delivery in the city in 2000 and 2001 and includes examination of the working relationships among the agencies in the system, between these agencies and the DVCC, and in particular, the role of the police department and its centrality to the DVCC. It is the role of the police as well as the domestic violence service delivery community’s perception of the police that is the focus of this chapter. This will be addressed by (1) describing the DVCC and its role as the explicit inter-agency domestic violence collaboration, (2) describing the DVCC member and non member agencies that comprise the domestic violence service delivery system, (3) discussing the role of the police department and its officers in response to domestic violence, and (4) presenting perspectives of the police by others in the domestic violence service community. Overall, these findings indicate that while the police are central to responding to domestic violence, not until well into the field work for this study have they been involved with the DVCC, and they generally have not been active in interagency efforts to improve domestic violence response. Findings Domestic Violence Coordinating Committee Since 1986 the Domestic Violence Coordinating Committee (DVCC) has been the primary locus of multi-agency, inter-disciplinary domestic violence service in the city. The DVCC is comprised of representatives from city government, non-profit organizations, and religious agencies working toward a more comprehensive, integrated, and seamless response to domestic violence. The DVCC, as outlined in the previous chapter, has a long history existing since 1987 as a coalition working to improve the domestic violence service delivery system, as well as to educate the public about domestic violence. During this period, the strength of the DVCC has waxed and waned. For the first years of its existence the DVCC was strong, vibrant, and an influential player in the city. Then, for myriad of reasons, its membership and activity level declined substantially. When active, the DVCC has served several important functions, most notably being a locus for cross-agency networking and inter-agency, multi disciplinary work, and a sponsor of domestic violence education and lobbying efforts. In 2000 and 2001, the study period for this research, the DVCC is reenergizing and is poised to again be a productive force for domestic violence advocacy and education. With membership growing, an organizational structure being put into place, and plans for future activities being discussed and formulated, the DVCC is becoming an active entity again. The DVCC’s mission statement states, "We are a group of professionals and concerned citizens formed to serve the citizens of [name of city] in the following ways: • To encourage the most effective means of community action in identifying and reducing the incidence of domestic violence • To enhance public awareness of domestic violence and to educate professional [sic] and the general public of the aspect of domestic violence • To provide a forum for networking and information sharing concerning issues of domestic violence" Three standing sub-committees called executive, education and membership, as well as ad hoc sub-committees that are formed when necessary pursue these goals. The six-person executive committee steers the DVCC and handles administrative tasks such as conducting meetings, and taking, writing, and distributing minutes. In conjunction with the education committee, several sub-committees have recently formed to work on four projects, some of which are ongoing: (1) producing an updated service directory, (2) assembling data on domestic violence in the city, (3) establishing a recognition program to identify and commend individuals who have handled a specific domestic violence case especially well, and (4) organizing conferences on domestic violence topics. The membership committee is concerned with building the DVCC’s roster and particularly with recruiting participants from hitherto unrepresented agencies. An elected chairperson creates agendas and presides over meetings. The DVCC holds meetings once a month at the juvenile and domestic courthouse. Meetings are always an information sharing and networking opportunity. At each meeting, members announce a variety of domestic violence related activities and trainings, oftentimes discussing legislation soon to be voted on by the state legislature. Members also share literature and report on the activities of coordinating committees from neighboring counties. Many DVCC meetings include guest speakers, usually from a local agency providing information about their agency’s domestic violence related activities and role in the domestic violence service delivery system. At each meeting, time is set aside for discussing specific domestic violence cases. This provides the opportunity for agency representatives, who may have contact with the same case, to gain information about aspects of the case of which they may not be aware. Membership and Core Members Originally, coordinating the DVCC was included in the job description of a city employee who left that position in 1992. Within a year of her departure the position was eliminated, and now the DVCC is entirely comprised of professionals who volunteer to serve on the DVCC in addition to their regular duties. DVCC members include representatives from city and state agencies, community non-profit organizations, and religious organizations that work in the area of domestic violence service delivery. While there are about 60 individuals who claim some cursory involvement with the DVCC, meeting attendance is usually between ten and 15 people. Committee membership cannot be precisely defined because participation varies, but there is a core group that forms the backbone of the DVCC and an array of individuals and agencies with varying levels of commitment and affiliation. Table 1 lists agencies comprising the domestic violence service community, showing the members of the DVCC in gray shading. Agencies shaded dark gray are those represented on the executive committee of the DVCC and agencies shaded light gray are members. We defined agencies as members if they were represented at at least one meeting between September 1999 and January 2001. We have identified 25 agencies that fit this criterion. Table 1: Domestic Violence Service Delivery Community Agency Agency’s Function Pertaining to Domestic Violence (members have *) State Dept. of Health* Works with localities to form and monitor fatality review teams. (Chairperson) Domestic Violence* Provides counseling for batterers and monitors batterers’ compliance with Interventions court ordered treatment. (Vice Chair) Hospital Emergency Room* Provides emergency medical services to victims of domestic violence. (Secretary) Department of Social* Provides counseling and a variety of direct services to domestic violence victims. Services at court (Treasurer) University sexual assault* Assists students who have been sexually assaulted and promotes service sexual assault awareness and education to the campus community. (Member at Large) BWS * Battered women’s shelter: provides safe emergency and temporary (Member at Large) housing to domestic violence victims. Magistrate’s Office Issues warrants and emergency protective orders. Prosecutor’s Office Prosecutes domestic violence cases. Victim Witness Provides court accompaniment and information to victims. Legal Aid Provides legal services to domestic violence victims on civil matters. Catholic Charities Provides a variety of service, including counseling services to victims and perpetrators, and emergency food and clothing. Family Violence Places and monitors batterers referred by court in Batterer Prevention Program at Intervention and other programs as necessary. Social Services Catholic Arch Diocese Addresses the issue of domestic violence in religious communities. Domestic Violence Task Force Local law enforcement Respond to domestic violence calls on and in vicinity of campus; agency assist students who have been sexually assaulted or stalked. City Police Department First responders to domestic violence calls. Domestic Violence Unit of Social Services that responds to domestic violence calls in Responders tandem with police to assist domestic violence victims obtain information about and services from Social Services. Criminal Justice Conducts research, implements and monitors victim witness Information and Services programs around the state, and provides police training. Center Community Corrections Provide Probation and Parole officers for offenders, as well as Batterer Intervention Groups. Mental Health Authority Government program providing mental health counseling and services for victims of domestic violence. United Way Charitable organization providing funds and services to various groups serving victims. Public Defender’s Office Provide free legal services for perpetrators at trial Sheriff’s Department Deliver warrants and summons to court, provide courtroom security Domestic Violence Non profit domestic violence advocacy and education group that Lobbying and Education provides technical assistance to community responses to domestic violence. Safe Harbor Shelter A shelter for battered women that accepts city residents but is located in a neighboring county Juvenile and domestic Court Handles cases involving juveniles and domestic violence The city police department is not shown to be a member of the DVCC because we did not observe their representatives attend any meetings during this time. It should be noted, however, that in August 1999 police representatives did attend to show a victim assistance video they had been involved in creating; additionally, the police department was becoming involved as our fieldwork was concluding in January 2001. Figure 1 provides a visual view of the domestic violence service community by displaying the component agencies visually in relation to their involvement with the DVCC. Figure 1: The Domestic Violence Service Community Image of Chart Showing These Groups in Relation to One Another The Domestic Violence Service Community DVCC Public Defender State Department of Health Community Corrections Mental health Counseling FVPP/DSS Domestic Violence Responders Police Department Catholic Charities Victim Witness Sheriff’s Department J&D Court Criminal Justice Services United Way Safe harbor Shelter Magistrate BWS Shelter City Prosecutor’s Office Legal Aid Domestic Violence Interventions Hospital University Sexual Assault Services FVPP/DSS at court Catholic Arch Diocese DV Task Force DV lobbying and education University Police In Figure 1 above, the extent to which an agency is involved in the DVCC is reflected in the proximity of its icon to the DVCC icon. The six agencies positioned on the circle closest to the DVCC icon are those represented on the executive committee. The agencies on the outer circle are those with involvement in the DVCC, defined as agencies that were observed attending at least one meeting over the study period. Agencies outside these circles have (or at the time when we began observation, had) no official involvement in the DVCC. Certain agency icons are shaded indicating that an interview with at least one representative was completed for this study. In the interview, subjects were asked with which other agencies if any they worked on domestic violence. In response to this question the following three agencies were mentioned far more than any others: the Police Department, the agency that runs the battered womens’ shelter (BWS) and Department of Social Services (DSS). These are shown in bold. The next section describes the function and role of the main elements in the domestic violence service delivery system. Battered Women’s Shelter (BWS) The BWS operates the only shelter in the city available to battered women, offering them temporary stays up to 30 days if space is available at the time of the request. There are also two shelters outside the city boundaries that accept city residents. Agencies that work with domestic violence victims – especially those whose role begins immediately after incidents such as law enforcement, emergency room personnel, and Domestic Violence Responders (described below) – rely on the BWS as a resource for emergency shelter for their clients. Agencies that enter the picture after the incident also work with the BWS to discuss the status of soon-to-be-discharged victims, or to access other services on behalf of their clients. Additional services offered by the BWS include two 24-hour hotlines where victims can receive information and referrals, a food bank, counseling, court advocacy, training in getting and keeping jobs, access to a medical clinic, and legal advice. The BWS is represented on the DVCC by various staff members who are interested and available for meetings. Batterer Counseling Programs: Office of Community Corrections, Domestic Violence Interventions, Catholic Charities, and Behavioral Health Authority In the study city, all cases involving misdemeanor domestic violence are heard at the juvenile and domestic court. Defendants, especially first time offenders, are often allowed to have their cases dismissed if within two years they complete a batterer counseling program managed by the Family Violence Prevention Program. This condition can also be a component of sentences. Offenders required to cooperate with FVPP must report to the FVPP Social Worker Specialist based at court for an assessment of their counseling needs. Based on the assessment results, they are placed in one of the many batterer treatment programs available. Once placed in a program, a FVPP case manager monitors each offender’s compliance with the court order. Several agencies and individuals run batterer groups, which focus on either domestic violence, anger management, or substance abuse. Domestic Violence Interventions, Behavioral Health Authority, the Office of Community Corrections, and Catholic Charities all offer groups, as do several licensed individuals. Many of these agencies are represented on the DVCC including Domestic Violence Interventions and Catholic Charities. Most groups charge attendees $10 or $15 a session; however, groups offered by the Office of Community Corrections (OCC) are free allowing indigent batterers to fulfill their court-imposed obligation. Groups are held all over the city and at various times of day and day of week, giving batterers every opportunity to find a group. Department of Social Services (DSS) The Department of Social Services (DSS) is a large agency that provides many direct services including all types of counseling and assistance with obtaining food stamps and other legal and financial aid. DSS houses two units that focus on domestic and family violence, the Family Violence Prevention Program (FVPP), and the Domestic Violence Responders Unit. Several staff persons from the Family Violence Prevention program participate in the DVCC including the unit supervisor and the two court workers. The majority of the FVPP – approximately eight workers and a supervisor – is based at the main Social Services building in the heart of downtown. Workers at this location assist domestic violence victims who have been referred to the unit from court, police, or other units at social services. FVPP workers can assist victims by procuring counseling, legal assistance, moving expenses, emergency food supplies, and many other forms of financial and service aid. In some cases, the FVPP works with victims only once, and in some cases they establish ongoing relationships. As mentioned above, case managers in this unit also monitor the compliance of batterers who have been court ordered to receive treatment. Two additional FVPP workers are based at juvenile and domestic court. One worker counsels juvenile offenders and their families. The other administers the batterer counseling and monitoring function of the FVPP, which involves assessing the counseling needs of batterers (e.g., anger management, substance abuse, or domestic violence counseling) placing batterers in appropriate programs, and alerting the court of successful completion or non-compliance. The Domestic Violence Responders unit is a DSS immediate response team headquartered at downtown Social Services with offices in the police precinct buildings. Domestic Violence Responders are social workers who work out of the precinct buildings during evening and nighttime hours and who respond to human services needs. Their primary, but not exclusive, role is responding to domestic violence scenes reported to police. When called, the Domestic Violence Responders work with police officers to provide the best possible service and assistance to victims by offering victims a range of information and services depending on each victim’s needs. Domestic Violence Responders are able to offer victims vouchers for emergency hotel stays, food, and baby supplies, in addition to crisis counseling and comprehensive information about available services. The Domestic Violence Responders also serve as a link between the victim and the FVPP where victims can obtain more in depth and ongoing assistance. City Police Department Domestic Violence Related Activities: Training, Response, and Investigation The city Police Department is the primary law enforcement agency responding to domestic violence in the city. To prepare its officers for this role, the department provides training in the proper handling of domestic violence calls as part of the regular academy recruit training. This training takes up one day of the 29-week academy and includes: presentations by local domestic violence services agencies on what each offers to victims and how officers can mobilize these resources; proper procedure for responding to domestic violence calls; and training on the laws relating to domestic violence in the state. From 1997 through 1998, federal funds were utilized to train officers on new laws mandating a pro-arrest policy for domestic violence incidents. This training focused on identifying the primary physical aggressor and covered changes in the laws pertaining to Emergency Protective Orders. This was the last time a department-wide in-service domestic violence training has been conducted. Currently police supervisors or representatives from the Domestic Violence Responder program occasionally conduct roll-call trainings for patrol officers, addressing procedural issues pertaining to the handling of domestic violence calls. In addition, during 1997 and 1998, the police department received a COPS office grant in a coordinated effort with the FVPP, the prosecutor’s office, and the BWS to develop a domestic violence training video for police officers, a video explaining the court process for victims, and two public service announcements. This training video is still in use at the police recruit training academy, and the public service announcements receive occasional television air-time, though funds are not available to run the announcements regularly. Although several agencies own copies of the video geared toward steering victims through the court process, such as the BWS, Victim Witness, and the Police Department, no one could say definitively how often the video is actually shown to victims. To further assist victims, as discussed above, the police department collaborates with the local Department of Social Services on the Domestic Violence Responder program. This involves training officers on the functions of the Domestic Violence Responders and encouraging officers to call them out in relevant situations. The department also provides Domestic Violence Responders office space in several of the patrol division headquarters. When the Domestic Violence Responders do not respond with police, officers give victims referrals to a variety of resources in the community. Domestic violence incident investigation is officially handled by the Police Department’s Youth and Family Crime Unit. This unit is charged with reviewing domestic violence reports and following up in particularly serious cases, or in cases where officers may not have taken the appropriate action, (e.g., if the case indicated an arrest should have been made and it was not). A supervisor and five detectives staff the unit whose purview includes investigating crimes committed by juveniles as well as those classified as domestic violence. Because juvenile offenses are prioritized above domestic offenses, however, and because the capacity of this unit is limited, domestic violence incidents are rarely investigated. Any follow-up with victims is most often handled by the police officers called to the scene who do so of their own accord. Recognizing the importance of thorough investigation and follow up with victims, several interviewees within the police department stated that a full-time domestic violence investigative unit would be immensely beneficial for all involved. The Youth and Family Crime Unit is able to review domestic violence cases by examining the domestic and family violence reports that are sent through the office on the way to archiving. These forms can be used to alert the unit supervisor to cases warranting further follow-up, but as discussed above, it is unclear how many or how extensively cases are investigated. The police department currently maintains no system for tracking domestic violence cases, offenders, or victims outside of its normal record keeping procedures. Domestic Violence Working Relationships within the Service Community Being the first responders to domestic violence incidents, police officers interact with medical, hospital, social service, and shelter personnel, as well as workers at the Emergency Protective Order issuing authority. During case follow-up, police provide and share information with prosecutors. Clearly, the police are integral to domestic violence response and case handling from beginning to end and, in doing so, must interact with other agencies. Yet, paradoxically, they are not involved with the DVCC, which, as noted earlier, is the locus of efforts to improve inter-agency response to domestic violence. Figure 2 displays the working relationships between the police department and the other agencies in the domestic violence network as reported in interviews. Figure 2: Police-Community Domestic Violence Working Relationships Graph Showing Police-Community DV Working Relationships One important measure of the strength of any multi-agency effort is the number of working relationships maintained by its members with other agencies in the network. The responses to our interview question: "What other agencies does your agency work with on domestic violence?" provide a proxy for identifying the agencies within the service delivery system and for assessing the inter-connectedness of those agencies represented in our interviewee sample. The more frequently agencies are mentioned in unprompted, self-reported working relationships, the more likely they are at the center of the service delivery web. Indeed, we found that several agencies received markedly more mentions than others. The tally of these responses is displayed in Table 2. Table 2: Number of Reported Domestic Violence Working Relationships by Agency Agency Mentions Battered Women’s Shelter 10 City Police Department 8 FVPP (DSS counseling and services for victims) 8 City Prosecutor’s Office 4 Domestic Violence Lobbying and Education 4 (Others) (1, 2 or 3) Based on the number of mentions, the city Police Department, the Department of Social Services, and the Battered Women’s Shelter, appear to be the central agencies in the service delivery community. Perceptions of the Police Department No police representatives were observed participating in the DVCC until the latter part of the period of observation and data collection for this study, and in his interview, the chief expressed lack of interest in police participation in the DVCC on the grounds that it was not a constructive use of officer time. He was doubtful that such organizations are truly collaborative when it comes to the police, remarking: Usually collaboration means, ‘what else can we find for you to do as police officers?’ All the people who sit around the table and talk about solving problems close up shop at 5:00, 5:30, but the police are open when the domestic violence incidents are going on. Everyone else is closed. It would be a true collaboration if the people with all the ideas would come out at 3:00 am and see if their ideas work for the police department like they think." In the DVCC’s heyday, however, the city police department’s view of collaborative efforts was quite the opposite. In fact, it was a police captain interested in improving the police department’s handing of domestic violence who was most responsible for the close working relationship between the department and the DVCC. At that time, the police department realized that something needed to be done to curb the escalating rate of domestic violence-related homicides in the city. With the approval and support of the previous police chief, the DVCC planned and implemented training for all city police officers on new domestic violence laws and developed a family violence form to be completed and submitted for all domestic violence calls for purposes of uniformity and tracking. This commitment on the part of the police department was frequently cited by former DVCC members and others as one reason for the DVCC’s success in its formative years, and conversely why without police department support the DVCC faltered in later years. Perceived commitment of the police department to domestic violence Interviewees repeatedly expressed the belief that the police department’s absence in the DVCC hampered its effectiveness as well as coordinated domestic violence service delivery in general. To the question, “How would you rate the level of collaboration around domestic violence within the city,” one interviewee responded, “Everyone but the police is doing a good job.” When asked about the extent of collaboration between her agency and the police department another interviewee answered “not at all.” Many interviewees expressed frustration that there is not a police department representative who regularly attends meetings or participates in any way on the DVCC. One interviewee commented, “On all the other domestic violence council meetings I have attended (in two surrounding counties) there is always a police officer there discussing recent cases, and how they could have been prevented, or what they feel the needs are in the community.” Other interviewees related a general sense of disconnection between the police department and their agency and felt that this would be mitigated if someone from the department participated in the DVCC. Many interviewees reported having a much better working relationship with the police department in the past. Lack of inter-agency communication and absence of police representation on the DVCC were the main collaboration-related concerns expressed by non-police interviewees. Many of these interviewees expressed the opinion that inter-agency communication in general is lacking with the police department. Several commented that this lack of communication perpetuates continued misunderstandings among agencies about their respective roles and purviews. An interviewee from a local agency talked about the intersection between her work and that of the police department. She articulated one issue about working with the police in this way, “There is a lack of trust from the police about our project, and people fear what will make them look bad. Police don’t want to participate [in this project] – they just don’t trust the process, and they are afraid they are going to be blamed. This could be improved if the police would give us the information we ask for more consistently.” When asked about collaboration one non-police department interviewee mentioned not getting statistics from the police anymore and expressed a general sense of disconnectedness in this way: I would like to get some statistics from the PD, on 911 calls, arrest rates, and conviction rates from the courts. It would help to know if there are changes in police policies and protocols on handling DV cases. Years ago, the DVCC got all of those reports from the police. We had access to forms that police filled out when they went out on a DV call, so we could screen the sheets and see what arrests were made, and if an arrest was not made, we could question that. I don’t know who takes care of those forms now, because they were bounced around from agency to agency. That was when the PD representative was very involved and gave us whatever we needed. We don’t get anything any more...I’m not sure what the police are doing now. I don’t know if they are giving out those cards to victims, if they are filling out those reports, if the reports go anywhere, if there is any follow-up. When the police do go out and write a report is anyone screening them? Is anyone gathering any information from those reports to write a comprehensive report? I don’t have a clue how well they are doing their job, or what they are doing. I do know that we haven’t done enough to protect women, and the laws are not utilized the way they were meant to be.” This interviewee went on to say, “a true collaboration really exists only among three city agencies – the Police Department, Department of Social Services, and the court. If they are the only ones collaborating, is that true collaboration? Where are the checks and balances with that system, and who makes sure it works? Collaboration exists, but it is questionable as to how well it works.” Lack of involvement with the DVCC in particular and lack of inter-agency collaboration in general is one of several observations leading to the opinions of many interviewees that the police department’s commitment to domestic violence is not what it should be. Interviewees reported that, “domestic violence is not a priority for the police.” A representative from another local law enforcement agency noted that city police department representatives never attend domestic violence training or other relevant events. This source states: “If you are going to do this work, it is imperative that you attend some of these functions, to make contacts, and be able to use those to offer resources to your victims.” For some non-police department interviewees, the fact that the police department does not have a dedicated domestic violence unit formed their impression that the department’s commitment to domestic violence is lacking. Interviewees mentioned three main functions that such a unit would serve: (1) expand the department’s capacity to investigate and thoroughly handle domestic violence incidents, (2) provide better services to victims, and (3) provide criminal justice agencies, service delivery agencies, and individuals with a point of contact within the department to direct questions and concerns related to domestic violence as well as to police response in general. While the Youth and Family Crime Unit previously discussed is meant to serve at least some of these roles, in reality it does not. All interviewees were asked whether they thought establishing a domestic violence unit would be a positive thing, and 21 out of 23 – including four police personnel – stated unequivocally that it was a good idea. Notably, many interviewees offered that while it was a good idea, they entertained little hope that the police department would establish such a unit. One interviewee felt that funding for such positions was possible to obtain and expressed frustration that the department was not making any efforts to procure such funding. While interviewees felt that a domestic violence unit would be ideal, it was thought that it would be beneficial to have even one person in the police department dedicated to domestic violence or at least designated as a contact person. A representative from an agency that works closely with the police stated: [Establishing a domestic violence unit] would definitely be a positive thing…if not a specialized unit, at least someone who could be a liaison in the department for exactly that purpose. For instance, right now in the department there is one particular person who deals with fugitives from other states. If you have questions, that is your contact. It is not a whole unit, but everyone knows who to go to on that issue. Non-police interviewees did mention one sergeant that seems to serve as an ad-hoc contact person. This sergeant’s only domestic violence-related task is training, but his current assignment (detective in auto-theft unit) has taken him out of the “domestic violence loop.” Our interviewees were unaware of anyone who can officially or unofficially fill that void. One interviewee stated, “I don’t know who to call, where to go – no point of contact.” A representative of an agency whose work is directly linked to the police department felt that a liaison between her agency and the Police Department would improve communication. She felt that increased communication would allow her office to better address some concerns expressed by officers about obtaining warrants, as well as making the police department’s role clearer. For want of a contact within the department, one interviewee who works with batterers leaves messages on the “crime stoppers” line when she gets intelligence about, for example, a batterer in possession of a firearm. Police Department’s Response to Domestic Violence Calls for Service In addition to these concerns about the level of collaboration with city agencies and organizations (including the DVCC) and the perceived level of commitment to domestic violence by the police, interviewees also had concerns about how the police handled domestic violence calls. Although none of our questions pertained specifically to the city officers’ responses to domestic violence calls, several interviewees offered comments on this topic. A representative from a law enforcement agency whose jurisdiction overlaps the city police department’s jurisdiction mentioned that differences in the way officers in the two departments are trained to handle domestic violence presents a challenge, because in the interviewee’s opinion, less domestic violence follow-through is expected of city police officers. Apparently, confusion sometimes results when officers of both agencies respond to the same call since they have different views of what action is required, with city police officers usually opting for less action. Regarding the presumptive arrest policy in place in the city, there was unanimous support among the police and civilian interviewees alike for this policy, although enthusiasm varied. Commenting on whether or not the policy has a positive or negative effect on victims, one non police interviewee stated: “The victims feel reassured that the police do care and are listening.” Although many civilian interviewees were aware of the law, several expressed doubt about the zeal and consistency with which city police officers actually carry it out. One interviewee stated, “It depends on the officer. Some take it seriously and go by the book, while others just blow it off.” She associated inconsistent application of the policy with lack of discipline in the ranks, stating, “The officers need to be held accountable for their actions, which really speak to their respect for the chief and his authority. They ignore an executive order from him when they don’t arrest.” Other interviewees were generally pleased with their relationship with the police, although interviewees acknowledged there was room for improvement. In the eyes of one DSS worker, “There is still an invisible barrier between the police department and the DSS in general. We meet some resistance about becoming a part of an investigation because we are Social Services and not police officers. There is an attitude of ‘why do we need a social worker…’ Some feel it is redundant to have a second person present at the interview.” Several interviewees believed that the Domestic Violence Responders program has greatly enhanced the working relationship between the police and DSS. One interviewee commented, “We work well together, and they treat us like family – they call us to come to community events and police training.” Another interview reflected, “That they have opened the divisions to us is pretty incredible. The sharing of information and the level of trust is very good, and it works both ways. The division commanders are wonderful and are very receptive to our suggestions.” Finally, interviewees repeatedly expressed the opinion that the city police officers need more training in handing domestic violence cases. The following are some of the comments made by non-police interviewees regarding training: • “Get the police more training.” • “Now when I go to the police department they say, ‘oh no, not another domestic violence training. We’ve had enough training.’ They may have had the training, but they are still not doing it right. Domestic violence is not going to go away, and we have to keep pecking at it so we can teach the new blood not to develop the resistance that is dying out with the old.” Civilian interviewees continually expressed how important they felt police department participation is within the domestic violence service delivery system. For example, one interviewee stated, “We share information – often they can find out things I can’t, like whether someone has a criminal record. By working with the police, we provide better victim services by coordinating our response. We help the police understand the victims’ perspectives, and they help us with the legal issues.” A representative from one agency put it this way, “The police have greater access to the public than we do. If an officer goes with me to a home, the person is more likely to talk to me than if I went alone.” The above discussion summarizes the views of the police department’s response to domestic violence expressed through interviews by members of the domestic violence service delivery community. The opinions of non-police department interviewees were mixed. All interviewees approved of the department’s presumptive arrest policy, and several spoke of positive working relationships between themselves and police personnel in patrol districts. Others, however, expressed frustration about what they perceived as the department’s lack of commitment to domestic violence. These interviewees pointed to the absence of personnel actively involved in domestic violence case follow-up and investigation, the lack of police department collaboration and communication with other agencies involved in service delivery, and the inconsistency with which officers apply the presumptive arrest policy. Non-police department interviewees also lamented the absence of police involvement in the Domestic Violence Coordinating Committee. These interviewees believed that active police department involvement was essential for the inter-agency approach to be effective. Discussion Recommendations submitted to the Police Department Toward the end of the field-work of this project, the police department formed a domestic violence working group, called the “Intimate Partner Initiative, ” in partnership with the city prosecutor’s office and several other agencies The goal of this effort was to reduce the number of intimate partner homicides through increased timely prosecutions and increased public awareness about intimate violence and available assistance. In doing to, the initiative was to make a strong point to offenders that the city is taking intimate violence offences seriously. The Intimate Partner Initiative began meeting and drawing up a plan for implementing the changes deemed necessary to pursue these goals. A Police Foundation field researcher was invited to attend and subsequently began serving on the committee. After it became clear that the involvement and contributions of this researcher were encouraged, we developed a list of recommendations the department could make to improve intimate violence service delivery that were based on the interview data. This list is attached as an Appendix to this chapter. The recommendations fell into two categories: (1) those that could be implemented quickly and at a low cost, with the anticipation of immediate improved results, such as officers always taking photographs of injuries; and (2) more long-term solutions that may require money or at least additional manpower to implement, (e.g., the formation of a dedicated domestic violence investigative unit, or at the very least appointing a domestic violence point person within the department). Chapter 5 Appendix Recommendations for Police Department Responses To Personal Violence Between Intimates Police Foundation Logo Recommendations for Police Department Responses to Personal Violence between Intimates March 2001 The following is a summary list of recommendations that may inform the work of the Interpersonal Violence Reduction Team. It is based on domestic violence research conducted by the Police Foundation for the COPS Office of the Department of Justice. This work included interviews with police personnel in [study city] and elsewhere; interviews with officials from other community agencies; and a review of good practice from other areas. The recommendations are in two categories: (A) Simple, tangible, quick-return interventions, and (B) Longer-term measures. A. Quick-return interventions 1. Monitored use of photographic evidence. Photographic evidence of violence can significantly increase the chances of batterers’ guilty pleas at court. The PD might consider requiring photos to be taken as a follow-up, a few days after a violent incident when bruising is more visible. This should be monitored to ensure implementation. 2. Provision of information to victims. A card or leaflet listing current referral numbers could be adopted for mandatory distribution at incidents of domestic interpersonal violence. Officers would explain the resources and encourage victims to use them. The Second Responders have thousands of copies of a leaflet containing pertinent information for victims: numbers for hotlines, police precincts, social services, the YWCA, and more, which might be used or adapted for these purposes. 3. Consistent use of dedicated 911 cell phones for victims. Recent technological developments mean that victims known to be at high risk can receive a prioritized, informed and incident-specific response from the police. Procedures/protocol to allow identification of high-risk victims, use and return of cell phones, and monitoring of use and responses, would facilitate this intervention. 4. Involvement with the Domestic Violence Coordinating Committee. Routine PD participation would facilitate communication between agencies, and improve information sharing on policies, procedures, and cases. The aim would be to enhance service delivery to victims and offenders by all agencies involved. The PD would benefit from the cooperation and information from other agencies and from the kudos of consulting the community on these issues. B. Longer-term measures. 1. A dedicated Domestic Violence Unit, Detectives, and/or Point Person. Due to the size of the intimate interpersonal violence problem, larger cities often have a specialized Domestic Violence Unit, dedicated domestic violence detectives, and a liaison person to work with agencies and the community: • A dedicated Unit could perhaps incorporate much of this work. It would need to be sufficiently staffed, and managed at a senior level (perhaps Deputy Chief) to avoid marginalization of domestic violence within the department. • Specialized Detectives could respond to and monitor cases from first call through investigation and prosecution, They could provide follow-up services with victims, monitor repeat offenders, and monitor police response (consistent adherence to policies and procedure). • A domestic violence Point Person could be appointed as a liaison with and contact point for other agencies and the community. 2. Improved and consistent training for all police officers. Training should include understanding the needs and viewpoints of victims (to avoid ‘revictimization’) and familiarity with available resources (to better inform victims). All officers should understand criteria by which an arrest should be made. The department could require officers to carry a card listing these criteria, and require them to consult it at domestic violence incidents. 3. Continual Monitoring of pro-arrest policies. Monitoring and control checks is a mechanism that can improve compliance with policy and consistency of application. Creating and maintaining a database of PD 109 Forms (and checking to ensure they are completed for all domestic violence calls) would facilitate this. The compliance of individual officers and the PD as a whole can then be monitored. 4. A domestic violence database. Such a database could allow tracking of repeat callers, addresses, victims, and offenders. It could also allow cross-referencing between persons and places as well as being used to inform responding officers of the relevant history and information. Social Services have an existing database that may inform this work. There is a range of literature available on many of these subjects and others, addressing police responses to domestic violence. The Police Foundation would be happy to supply the Interpersonal Violence Reduction Team with copies of relevant articles upon request. Chapter 6 Police Use of Dual Arrest in Intimate Assault Cases Erin Lane, M.P.M. and Rachel Boba, Ph.D. POLICE FOUNDATION Prepared for the U.S. Department of Justice Office of Community Oriented Policing Services Award #98DVWXK015 June 2002 Chapter 6: Police Use of Dual Arrest in Intimate Assault Cases Introduction In this chapter, we present exploratory findings on dual arrests for intimate partner assaults in the study city, which has a primary physical aggressor, presumptive arrest law in effect. Although we did not set out to examine the practice of dual arrests in the broader study of which this analysis is a part, our interest was peaked by local service providers who repeatedly expressed concern about this police action. Intimate partner assault refers to violence between male and female couples that are married, formerly married, have a child in common, or cohabit. Dual arrest is the arrest of both members of the couple and is applied by police officers when they find probable cause to believe an assault occurred but are unable to identify a single primary aggressor, or judge both parties to be culpable. Similar to other studies that have used police data to identify case characteristics correlated with arrest, this study uses supplementary domestic violence reports filled out by police officers to identify case characteristics associated with dual arrest. As with any police data, the reports available for this analysis are only a portion of the incidents that come to the attention of the police, which are only a portion of actual intimate violence. There are two components to this analysis. The first is a comparison between characteristics of intimate assault by three police actions: warrant, single arrest and dual arrest. We found that dual arrest is the least common police action occurring in about 7% of these cases. The demographic characteristics associated with dual arrest are more similar to those of arrest cases than to those in which a warrant is issued. By several measures, dual arrest cases appear to be some degree less serious than those where a single arrest is made. The second component uses logistic regression to determine what case and demographic characteristics are predictive of dual arrest in comparison to single arrest. Warrant cases were not considered for this analysis, in order to focus the analysis on the police decision to make a dual arrest instead of a single arrest. We found that the only significant variables were those pertaining to victim and offender substance use. In cases where the victim is suspected of having taken alcohol or drugs prior to the incident, dual arrest is over three times as likely. Two additional logistic regression models were tested to see whether dual arrest was predictive of officers obtaining emergency protective orders or collecting evidence. Dual arrest was not found to be predictive of these actions. Intimate Violence And Dual Arrest According to the book Violence in Families, “Arrest for domestic violence is perhaps the beststudied intervention for family violence” (Chalk & King, 1998: 174). Violence in Families discusses many studies on the effectiveness of arrest, beginning with the Minneapolis Domestic Violence Experiment (Sherman & Berk, 1984), and continuing through the “Spouse Abuse Replication Program (SARP)” studies that replicated it (Berk, Campbell, Klap, & Western, 1992; Dunford, Huizinga, & Elliott, 1990; Hirschel & Hutchinson, 1992; Pate & Hamilton, 1992; Sherman, Schmidt, Rogan, Smith, Gartin, Cohn, Collins, & Bacich, 1992). While the SARP studies, and others have examined whether and under what circumstances arrest is effective, they have produced contradictory and inconclusive results (Garner, Fagan, & Maxwell, 1995; Melton, 1999; Chalk & King, 1998). Whether this inconclusiveness is due to non-uniformity in the applied treatments used in the experiments or because arrest is an effective deterrent for certain types of people and situations and not others remains unclear. Despite inconclusiveness on the effectiveness of arrest, states have responded to the finding that arrest is the most effective deterrent to intimate violence by implementing mandatory arrest laws, and police department policies were changed accordingly. A 1997 national summary of domestic violence laws revealed that in 23 states arrest for domestic violence incidents is mandated in some or all circumstances, and that in an additional six states arrest is officially preferred (Institute for Law and Justice, 1997). At least one police department (Concord, NH) instituted a mandatory arrest policy independent of state laws (Holmes, 1993). As encouragement of police officers to make arrests in intimate violence situations continues, the desirability of mandatory arrest policies is being questioned, in part because their effectiveness is in doubt and in part because they are thought to effectuate dual arrests, which is considered by many to be a problematic outcome (Hamberger, 1997; Holmes, 1993; Law Enforcement News, 2000; Martin, 1997). The underlying assumption of dual arrest is that assaults occur where there is probable cause to arrest both parties. On the surface, such a police response appears quite appropriate and useful, especially considering research from the 1980s using the Conflict Tactics Scale indicating women to have rates of relationship violence as high if not higher than men (Straus & Gelles, 1990). Yet some advocates argue that acts of violence should be considered in context, rather than at face value and believe dual arrest often revictimizes battered women. They claim that many women arrested in dual arrest incidents are in fact battered women who participate in violence as self-defense or in response to an ongoing pattern of violence that they did not initiate. (Martin, 1997; Hamburger, 1997; Law Enforcement News, 2000). Further analysis of the Conflict Tactics Scale data supports the advocates’ claim, showing that although men’s and women’s rates of violence are similar, women tend to suffer more frequent and serious injuries from intimate violence (Gelles &Cornell, 1990). Additional justification for this view comes from a study by Hamberger (1997) who explored women’s use of violence in intimate relationships, by studying the culpability of 52 female intimate violence arrestees. He found that (1) 51% of women reported that their partners had initiated the overall pattern of violence in their relationship, (2) 37% of women reported that their partner always initiated individual episodes of violence, (3) many of the women who acknowledged initiating violence more often than their partners noted that they had begun to use violence after many years of being victimized by their partners, and (4) 24 women in the sample reported self-defense/protection as a motivation for their violence. Hamberger concluded that 67% of the women in the sample do not appear to be husband beaters or mutual combatants. Rather, in many instances, they are, “battered women who are fighting to defend themselves from an assault by their partners” (Hamberger, 1997: 125). While the intention of mandatory arrest policies is to protect victims by ensuring that police arrest perpetrators, detractors assert that in practice, officers have interpreted them to mean that they must arrest all parties who have engaged in any violence – even in self-defense. William Holmes comments, “officers may find it easier to arrest both parties than to interpret poorly worded statutes or policies. On rare occasions, evidence may even show that the parties are mutual aggressors” (Holmes, 1993: 105). His implication is that dual arrest is not warranted in the majority of cases in which it is applied. It may also be the case that the laws are so broad that they do not allow officers discretion in considering the context of the case when making arrest decisions. The question of how to handle the murkiness of an incident where both participants used violence, but perhaps unequally, is of concern to police agencies endeavoring to effectively and fairly address intimate violence. A December 2000 article in Law Enforcement News states: Reconsidering strategies for reducing domestic violence was on the agenda in a number of police agencies last year, particularly with regard to the practice of dual arrest. Growing out of mandatory-arrest policies established in the 1980s and early 90s, dual arrests often result in the inadvertent arrest of a battered woman who was fighting back against her attacker, victim advocates claim – and law enforcement has begun to agree with that finding. Another example is New Mexico’s Violence Against Women Task Force’s recommendation which comments, “Dual arrest trivializes the seriousness of the offense. It may increase the danger to the victim because the next time it happens, she may not call the police for fear of being arrested herself” (New Mexico, Office of the Attorney General, Violence Against Women Task Force, 8/17/2001). The federal government’s concern about dual arrest is evident in solicitations from the Office of Justice Programs for “Grants to Encourage Arrest Policies and Enforcement of Protection Orders” which define eligible recipients as government entities that “demonstrate that their laws, policies, or practices and their training programs discourage dual arrest of the offender and the victim” (OJP, 2001). While concern about dual arrests is frequently expressed, very few efforts have been made to compare the characteristics of cases to which dual arrest is applied to those of other cases. Two of these, one conducted by the Connecticut Department of Public Safety and the other by an independent researcher, are based on Connecticut State data. Both projects differentiate cases resolved by dual arrest from those resolved by single arrest. The State of Connecticut Department of Public Safety (CDPS) framed their study of dual arrests in the executive summary of the report as follows: A major area of concern since the inception of the mandatory arrest requirement has been the effects of arrest of all parties involved in a family violence incident, or dual arrest. Every evaluation of the family violence law has devoted space to dual arrest and has cited it as cause for concern. None of the evaluation projects to date have included input from law enforcement as to why dual arrests are made and what impact Connecticut’s reporting mechanism has on the rate of dual arrest. (p.1). The CDPS studied the application of dual arrests by analyzing arrest data to draw a comparison between dual arrest and single arrest incident characteristics and conducted police officer interviews to discern their opinions about mandatory arrest, dual arrest, and the training received in family violence.1 The arrest data revealed that in the three years following implementation of the mandatory arrest policy, the rate of dual arrests grew disproportionately to the increase in total family violence arrests. Analysis of police incident data revealed that dual arrest cases tend to be less serious, to involve cohabiting couples, and to involve offenders who are younger than those arrested in single arrest incidents. 1 In this jurisdiction "family violence" generally was the concern, rather than intimate violence, which is the interest of the present paper. In a separate study, Margaret Martin (1997) examined dual arrest by analyzing a sample of 448 family violence cases disposed in Connecticut state criminal courts, originating from police agencies throughout the state. Her study followed implementation of the statewide mandatory arrest policy and compared characteristics of intimate violence cases resulting in single arrest to those resulting in dual arrest. Martin found that defendants in dual arrest cases were more likely to be female, white, younger, cohabitors (living with but unmarried to partner), and to have less serious family violence arrest histories, as well as less serious current charges, than single arrest defendants. Dual arrest incidents were also more likely to involve alcohol or drugs and were less likely to result in a court conviction. Interestingly, there was evidence of prior victimization among 40% of female defendants in dual arrest cases, which lends support to the notion that the women arrested in dual arrest situations are likely to be victims themselves. The possibility that intimate violence victims may be harmed by policies established to protect them underscores the importance of understanding the true effects of policies such as mandatory arrest. According to the editors of Violence in Families, the cost and consequences of arrest policies “merit consideration,” specifically in terms of “improper or unwarranted arrests” (Chalk and King, 1998: 178). Quantifying and describing the phenomenon of dual arrests in specific localities and overall is important groundwork toward this goal. An evaluation of the Queens County, New York Arrest Policies Project includes a recommended strategy for addressing dual arrest stating, “To best attack the problem, statistics on the extent and nature of dual arrests need to be developed” (Miller, 1999: 17). While there is a wealth of expressed concern about dual arrest, so far there is little quantification and examination of the practice. Some states, such as California and Minnesota, have sought to decrease the use of dual arrest by including language discouraging it in legislation (National Council of Juvenile and Family Court Judges, 2001a; National Council of Juvenile and Family Court Judges, 2001b). Others such as Washington, Alabama, Wisconsin, and New York have modified their intimate violence statutes or policies to include directives about identifying a “primary physical aggressor” (Holmes 1993; Sutton, 1999; Alabama Coalition Against Domestic Violence, 2001), and in some cases (California), a “primary dominant aggressor” (National Council of Juvenile and Family Court Judges, 2001b). These localities direct officers, through law or policy, to conduct a primary aggressor analysis if more than one party in an intimate violence dispute claims to have been victimized. Elements of primary aggressor analyses differ, but include considerations such as (Alabama Coalition Against Domestic Violence, 2001; Sutton, 1999): • Prior complaints of domestic violence, • The relative severity of injuries to each person, • The likelihood of future injury to each person, • Whether one of the persons acted in self-defense, and • Consideration of the physical strength of the parties. Our analysis seeks to contribute to the discussion on dual arrest by providing an exploratory overview of its use in a presumptive arrest jurisdiction with primary aggressor language. The Present Analysis Using a database of domestic violence police reports, we compare characteristics of cases resolved by single arrest, dual arrest, and with a warrant,2 over a two-year period. This exploration has two components. First, a descriptive case study of the three officer actions in intimate violence incidents in a jurisdiction with a mandatory arrest, primary aggressor law in effect, comparing differences between cases addressed with the three possible police actions. Second, logistic regression analysis exploring whether certain demographic, history of violence, and incident characteristics explain dual arrest, and whether officers taking certain actions in handling intimate assault cases is explained by dual arrest. 2 Police obtain warrants if the offender is not present during their investigation of the crime. In some cases, the victim obtains the warrant with or without the assistance of the officer. In our study site, violence between intimates was viewed as assault prior to July 1, 1991. That date marked the implementation of a new state code making assault against a family or household member a crime distinct from other assaults, with its own set of guidelines and penalties. The new code stated that officers may arrest without a warrant for an alleged crime against a family or household member regardless of whether the violation was committed in the officer’s presence, if the officer established probable cause to believe an offense occurred (State Code Section 19.2-81.3, A.). This gave officers permission to arrest for domestic violence misdemeanors they did not witness, removing a major restraint on their ability to address domestic violence. In 1997, the laws changed again, when a presumptive arrest law was put into place, and the term “primary physical aggressor” introduced with section B of Code Section 19.2-81.3: “A lawenforcement officer having probable cause to believe that a violation…has occurred shall arrest and take into custody the person he has probable cause to believe, based on the totality of the circumstances, was the primary physical aggressor unless there are special circumstances which would dictate a course of action other than an arrest” [italics added]. Following these changes, officers were trained to identify primary physical aggressors and were provided guidelines for seeking emergency protective orders. Methodology and Data This analysis draws from domestic violence incident reports, referred to here as (DVRs), which officers are required to complete for all domestic violence incidents. The DVRs provide more detailed information than the standard criminal incident report, which officers fill out in addition to the DVR if a crime has been committed. Information on the DVR includes demographic, relationship, and address information on the parties involved; the location, date, and time, of the incident; police action taken, presence of children, apparent use of alcohol or drugs by either of the parties, weapon use, victim injuries, and whether protective orders were on file. In this study, we analyze intimate assaults among persons who are or were married, have a child in common, or cohabit.3 We eliminated cases that did not meet the state’s definition of domestic violence, such as violence between same-sex couples. We also eliminated cases where the type of police report submitted was other than assault or was unknown, to ensure that we were only considering incidents involving assaults, since assault is the predominate charge in intimate violence. Applying these steps to the data available to this project produced a database of 1,376 intimate violence assaults reported to the police department between February 1, 1999, and March 31, 2001. This process is laid out in Appendix A. The data used in this analysis are also summarized in columns and rows and included as Appendices B and C respectively. 3 Cohabitors refer here to unmarried couples who live together. Findings In order to establish a context for the intimate assault data, we compared the DVR data to 2000 Census data, where possible, for the study city. The comparison includes demographic data such as race and age of individuals as well as type of couple. See Table 1. Table 1: Demographic Comparison between the Population of City and the Population of Reported Intimate Assaults City-wide** Domestic violence victims Domestic violence offenders Race Black 57.2% 83.1% 75.7% White 38.3% 14.0% 8.5% Other 4.5% 2.0% 2.8% Unknown 0 0.9% 12.9% Relationship Spouse 27.1% 29.2% 27.9% Female householder 20.4% 20.4%* N/A Median Age 33.9 29.0 31.0* This is the percentage of female victims whose relationship to the offender was Child-in-Common." If these victims lived with the offender, they would be listed as cohabitors. It is possible that some of these victims live with parents or others rather than heading a household, thus this should be considered only an estimate. **Source: 2000 Census data from the Census Website These comparisons revealed: • As a share of the population in the city, blacks are over represented as both victims and offenders of intimate violence. • Based on the closest estimate that can be made, the percentage of spouses and female- headed household in the database of intimate assaults is similar to that of city residents. • The median age of domestic violence victims is four years less than for the city population, and the median age of offenders is two years less. Focusing on the population of victims and offenders that comprise this analysis, it is notable how unevenly distributed offenders and victims are by race. Blacks comprise the great majority of offenders. Less than 10% of the offenders are white, and only 2.8% are neither black nor white (other). While, in the interest of describing the complete dataset, we present comparisons between all three race groups, the uneven distribution – especially the small number of other race offenders – prevent us from drawing anything but speculative conclusions based on race. Three possible police responses to intimate assaults are distinguished in the data: warrant, single arrest, and dual arrest, and there are specific conditions under which each outcome generally is most likely to occur. Officers obtain warrants when they find probable cause to believe that one person has assaulted another, but the person they have identified as the primary physical aggressor is not present at the scene. When the identified primary physical aggressor is present in these circumstances, officers make single arrests. Officers make dual arrests when they have probable cause to believe an assault occurred and are unable to determine who is the primary aggressor. Table 2 displays the distribution of police action across these three outcomes, in these data. Dual arrest is the least common outcome, occurring in only 7.3% of reported intimate assaults. Table 2. Distribution of Police Outcomes Frequency Percent Warrant 581 42.2 Arrest 694 50.4 Dual arrest 101 7.3 Total 1376 99.9* *Percentages do not add to 100 due to rounding. Part 1: Comparisons Across Three Police Outcomes The first part of this analysis seeks to describe cases to which dual arrest is applied and to identify and explore any systematic differences between dual arrest cases and those addressed through single arrest or warrant. To explore case variation by outcome, we present comparisons in five categories: (1) demographics, (2) history of violence, (3) case characteristics, (4) incident characteristics, and (5) police action. These results are displayed and discussed below. A. Demographics The majority of cases in this database involve cohabiting black couples, where the offender is male and the victim female. Ninety-four percent of cases involve male offenders and female victims. The average age of victims is 31 and the average age of offenders is 33 – a two-year difference that is consistent across all police outcome categories. While the average age of offenders in dual and single arrest categories is the same, offenders in the warrant category are on average two years younger. This difference is also true for victims. Since many of the demographic variables are best described and understood in relation to one another or to other variables, the key findings are synthesized and discussed below. Tables displaying all of the frequencies and percentages by columns and rows are included in Appendices 2 and 3. Of the offenders whose race is known, 86.3 % of the offenders are black, 10.5 % are white, and 3.2 % are of another race (“other”). When examined by police action, an interesting difference emerges. While single arrest was the most common outcome for cases involving black and white offenders, among black offenders, the percent disposed by single arrest and by warrant is almost equal, whereas among white offenders, the percentage of cases disposed by arrest is much higher than the percentage where a warrant was issued. Similarly, the percentage of whites to whom dual arrest is applied is greater than that of blacks and other race offenders. Police outcomes for other race offenders are more similar to cases involving black offenders than to white offenders, but again, since there were only 41 offenders of other races, 21 who received a warrant, 19 who were arrested, and one who was dual arrested, generalizations about this group are tentative. Below are three pie charts that display the police outcome by race of offender. Figure : 1 Police Outcome by Race of Offender Black: N= 1,101 Warrant 44% Arrest 49% Dual Arrest 7% White: N= 134 Warrant 24% Arrest 66% Dual Arrest 10% Other Race: N= 41 Warrant 51% Arrest 46% Dual Arrest 3% Since warrant issuance is an outcome applied when the offender is not present during police service, these race differences by arrest (single and dual) and warrant may indicate that there is a relationship between offender race and offender presence, perhaps through the type of relationship with the victim. For example, couples in certain relationships more often live together, such as spouses versus former spouses. Offenders living with the victim may be more likely to be present during police service, compared to offenders who have a home elsewhere. In a study of police action in 189 domestic violence calls, Lynette Feder found that offenders not present during police service were more likely to be unmarried (Feder, 1996). She also found that police arrested 44% of offenders who were present when they arrived, and only 8% who were not. We believe it is reasonable to assume that this is the case in the analysis. Below, we explore the relationship between these variables. Four, not necessarily mutually exclusive, victim-offender relationships exist in these data, which, in a hierarchy from most to least formal, are spouse, former spouse, cohabitants, and child-in common. Clearly, it is possible for any of the couples to have a child in common, not just the “child-in-common” group. Cases that fit more than one category were placed in the most formal category into which they fit. For example, if a couple has a child together and lives together, they would be classified as cohabitors. The largest relationship category in these data is cohabitors, who account for 50% of all cases. The next largest relationship category is married couples (29% of all cases), then child-in- common (20%), and the smallest category is former spouses, who accounted for less than 1% of cases. Although the largest share of offenders in all race categories are cohabitors, a greater percentage of whites and other race offenders are married than are blacks, while blacks have a greater percentage of child-in-common relationships. Table 3 displays these results. Table 3:Victim-Offender Relationship by Race of Offender Black White Other Total % N % N % N % N Spouse 25.6 282 41.0 55 41.5 17 27.7 354 Former Spouse .4 4 1.5 2 4.9 2 .6 8 Cohabitors 52.4 577 50.0 67 51.2 21 52.1 665 Child in Common 21.6 38 7.5 10 2.4 1 19.5 249 To explore the intersection of race, type of relationship, offender presence during police service, and arrest, we assume that spouses and cohabitors live together and that former spouses and couples designated as child-in-common do not. Under this assumption, approximately 78% of black offenders, 91% of white offenders, and 93% of other race offenders live together. The offender was present in 51% of all cases. Differences among the race groups are apparent in that of cases with black offenders, the offender was present 50% of the time, white offenders, 62%, and other race, 45 %. Table 4 displays the results of these variables by race, relationship, whether victim and offender live together, and arrest. Table 4: Relationship, Offender’s Presence, and Arrest by Race Relationship Offender Offender Offender lives with present during arrested (single victim police service or dual) Black Spouse 25.6% 78.0% 50.2% 56.0% Cohabitor 52.4% Former Spouse .4% Child in Common 21.6% White Spouse 41.0% 91.0% 61.8% 76.0% Cohabitor 50.0% Former Spouse 1.5% Child in Common 7.5% Other Spouse 41.5% 92.7% 45.2% 49.0% Cohabitor 51.2% Former Spouse 4.9% Child in Common 2.4% A lower proportion of black offenders live with the victim, are present during police service, and are arrested, than are whites. Other race offenders are similar to whites in terms of living with the victim, but are not in terms of being present for police service and being arrested. Thus, the relationship between these variables appears to work similarly for blacks and whites, but not for offenders of other race. These findings should be considered cautiously, however, because of the unequal racial distribution of the data, in particular the small number of offenders of other races. On the surface, there does not seem to be a direct relationship between race and officers’ arrest decisions; the more important variable seems to be presence during police service, which is directly related to arrest across all three race groups. In sum, we found average age of single and dual arrest victims and single and dual arrest offenders to be more similar to each other than to victims and offenders respectively in cases where a warrant was issued. In warrant cases, victims and offenders are on average two years younger than their respective single and dual arrest counterparts. Police outcome differences by race are interesting. A greater percentage of whites than blacks or other race offenders are single or dual arrested. This may be because a greater percentage of whites are present during police service. Racial differences by arrest were also true for dual arrest, while warrant cases again, had a different offender make-up. The distribution of single and dual arrest offenders across types of relationships also tended to be more similar than the relationship distribution of warrant cases. Based on all the of these demographic measures the characteristics of single and dual arrest are more similar to each other than to cases where a warrant is issued. Regarding dual arrest cases specifically, cohabitors have the highest proportion of dual arrests, and in terms of race, whites have the highest proportion of dual arrests. Again, readers are cautioned that the small number of dual arrest cases as well as those involving offenders of “other” races greatly limit the conclusions that can be made. B. History of violence In differentiating the types of cases more often ending in dual arrest, one question is whether dual arrest is generally applied more often in cases where there is a history of violence. This may suggest whether dual arrest is being applied to cases where both parties are using violence equally, or, whether it is being applied to parties who acted in self-defense. If there is a history of violence in the relationship, then dual arrest may be more likely a result of self-defense by one of the parties. Variables such as whether protective orders were in place (either emergency, preliminary, or permanent), or whether the offender has been arrested for domestic violence in the past are indications of this – at least indications of past incidents that have come to the attention of police. Given that so much intimate violence is not reported, there are likely many other cases in this analysis whose histories of violence are not reflected in these variables. Even if the data were considered complete, since the results of this analysis are ambiguous, no conclusions are drawn. Of the three outcome categories, dual arrest cases have the lowest percentage of protective orders in place (4.0%), which could indicate less serious history of violence. The percentages of warrant and arrest cases where protective orders were in place are almost identical (6.4% and 6.8% respectively). As displayed in Table 5, most cases had no prior arrests. The range among those that did have prior arrests is 1-15. Table 5: Offender’s Prior Intimate Violence Arrests Warrant Single Arrest Dual Arrest Total % N % N % N % N None 77.9 441 69.9 485 73.0 73 73.5 999 One 14.1 80 18.3 127 15.0 15 16.3 222 Two 4.8 27 8.1 56 9.0 9 6.8 92 Three or more 3.2 18 3.7 26 3.0 3 3.5 47 In terms of prior arrests, dual arrest cases appear more similar to single arrest cases than to warrant cases. The average number of prior arrests among dual arrest cases is .46 and the average among single arrest cases is .52. Warrant cases have an average of .34 prior arrests – the lowest of the three. However, an Anova test on the difference between these means showed them not to be significantly different. C. Incident characteristics To identify the types of incidents in which officers are most likely to arrest both parties, the following topics were explored: whether an aggravated or simple assault charge was filed (indicating seriousness of incident), substance use by the victim, the offender, both, neither; weapon use; what offense the victim alleged; whether the aggressor was present; whether children witnessed the incident; and whether medical treatment was administered. Aggravated or simple assault Officers can classify assaults as simple or aggravated, depending on the seriousness. Officers classified 7.3% of single arrest cases and 6.2% of warrant cases as the more serious type of assault, aggravated. Far fewer (3.0%) dual arrest cases were classified as aggravated. Substance use The DVR contains several questions pertaining to victim and offender substance use. It asks officers to record whether there is evidence of alcohol and or drug use for each member of the party. Since officers recorded very little drug use, suspected drug use was combined with suspected alcohol use into a substance use variable for the victim and for the offender. A third variable was created indicating whether both parties were suspected of using substances. Table 6 displays these results. For all categories, officers reported that 17% of victims appeared to have used substances prior to the incident and 34% of offenders. At least one party appeared to have used substances in 39% of all cases. Both the victim and the offender appeared to have used substances in 12.5% of all cases. In 61%, neither party appeared to have used substances. It is likely that offender substance use for the warrant outcome is less accurate than the others, because in many of these cases the offender was not at the scene. Officers may have reported offender substance use based on statements made by the victims, who might be disinclined to report drug use. Table 6: Offender Substance Use Warrant Single Arrest Dual Arrest Total % N % N % N % N Victim 9.0 52 20.9 145 38.6 39 17.2 236 Offender 26.2 152 41.4 287 35.6 36 34.5 475 Both 6.0 35 15.9 110 26.7 27 12.5 172 Neither 70.9 412 53.6 372 52.5 53 61.0 837 It is notable that of the three outcome categories, dual arrest has the greatest percentage of cases where both offender and victim appeared to have used substances prior to the incident. The relationship between dual arrest and suspected substance use by the victim will be explored later in the logistic regression section. Weapon use Officers recorded weapon use in several categories on the DVR: gun, knife, hands and feet, and other. Other includes a range of objects such as “vehicles,” telephones, hammers, belts, bats, and even a baby monitor handset. Since hands and feet were used in almost every case, this variable was not considered. Overall, a weapon – other than hands and feet – was only used in 12.5% of all cases. Dual arrest was the police outcome category with the highest rate of weapon use, followed by single arrest, then warrant (see Table 7). Table 7: Weapon Use Variable Warrant Single Arrest Dual Arrest Total % N % N % N % N Gun .9% 5 2.0% 14 1.0% 1 1.5% 20 Knife 2.8% 16 3.5% 24 6.9% 7 3.4% 47 Other 8.3% 48 7.8% 54 8.9% 9 8.1% 111 Any 11.7% 68 12.8% 89 14.9% 15 12.5% 172 Alleged offense Simple and aggravated assault were the official charges filed by the police, but on the DVR, officers also record what offense(s) was alleged by the victim. Physical assault was by the predominant alleged offense, claimed in 97.3% of cases. In the remaining 2.7% of cases, victims alleged that another offense had been committed by the offender, such as sexual assault, property crime, psychological abuse, or offense against a child. In some cases victims claimed that more than one type of offense had been committed; thus the alleged offenses total more than 100% across the outcome categories. These results are displayed in Table 8 on the following page. Because there are so few cases where an offense other than physical has been alleged, no conclusions about the alleged offense and police outcome are drawn. Table 8: Offense Alleged by Victim Variable Warrant Single Arrest Dual Arrest Total % N % N % N % N Physical assault 96.6 561 97.8 679 98.0 99 97.3 1339 Sexual assault .2 1 0 0 0 0 .1 1 Property crime 1.7 10 1.3 9 1.0 1 1.5 20 Psychological 1.4 8 2.2 15 3.0 3 1.9 26 Offense against child .2 1 .4 3 0 0 .3 4 Other offense .9 5 1.7 12 0 0 1.2 17 Children witnessed incident Overall, children witnessed 42.8% of the incidents examined here. This percentage is virtually the same across police outcome categories. Medical treatment received Police reported that medical treatment was administered at the scene or at a hospital after the incident in 15.6% of all cases. In an additional 6%, victims reported to officers that they planned to see their personal physician. Cases where warrants were issued had the highest rate of receiving medical treatment (17%) closely followed by cases where arrest was made (14.8%). A marginally lower rate of medical treatment received (12.8%) was observed among the dual arrest cases compared to arrest cases, which could indicate that these cases were generally less serious. D. Police action The type of discretionally action officers take in intimate violence cases is a useful gauge for assessing officers’ perceptions of the crime. Variables in this database that capture officer discretionary actions are whether they collected evidence and whether they obtained an emergency protective order (EPO) for the victim. Overall, some type of evidence, such as statements, photographs, 911 tapes, or hospital records was collected in 16.3% of cases. Dual arrest cases had the lowest rate of evidence collection (see Table 9). Table 9: Evidence Collected Variable Warrant Single Arrest Dual Arrest Total % N % N % N % N Statements 6.0 35 6.3 44 4.0 4 6.0 83 Photographs 7.9 46 9.2 64 7.9 8 8.6 118 911 Tapes 0 0 .1 1 0 0 .1 1 Hospital records 1.2 7 .4 3 0 0 .7 10 Other .02 2 .06 8 .1 2 .1 12 Any 15.5 90 17.3 120 13.9 14 16.3 224 Officers obtained EPOs in 25.7% of cases overall. While they obtained them in close to the same percentage of warrant (27.4%) and single arrest (25.9%) cases, they did so at the lowest level for dual arrest cases (13.9%). It is not surprising that dual arrest cases have the lowest level of EPOs, in fact is surprising that there are any in this group. It indicates that in 13.9% of dual arrest cases, the officer was concerned for the safety of one party, even though that party was perceived to have committed a violent crime for which they were arrested. The previous analysis seeks to differentiate characteristics of intimate assault handled by police through warrants, arrest, and dual arrest. Based on the data, interpretations are offered as to the seriousness of dual arrest cases compared to those where an arrest was made or a warrant issued. These interpretations are highly speculative and limited, because they are based on just a few variables which may not be the most relevant ones, and because of the uneven distribution of cases in certain of the variable categories. In terms of comparing demographic, history of violence, and incident characteristics by police outcome, these data indicate that the demographic conditions under which dual arrests are made resemble those of arrest more closely than those of warrant. In terms of case seriousness, dual arrest cases appear to be less serious than those resulting in arrest or warrant, in that dual arrest cases have the lowest percentage classified as aggravated assault. Based on history of violence, the results are ambiguous, but dual arrest cases appear to have a less serious history than those of warrants, and single arrest. In a slightly higher percentage of dual arrest incidents, a weapon was used, compared to the other two outcomes. Part 2: Analysis of Dual Arrests Through Logistic Regression Dual arrests occur when the police believe both parties participated in the violence and when they are unable to identify the primary physical aggressor. As noted earlier, dual arrest was the outcome in 101 of the cases in the sample of 1,376 cases, which is 7.3 %. While it was the least common specified outcome, dual arrests nonetheless accounted for 13.6 % of all arrests in the sample, since each incident produced two arrests. Now that we have presented descriptive data comparing case characteristics of warrant, single arrest, and dual arrest, we focus our analysis further by comparing characteristics of single arrest cases to those of dual arrest, since these outcomes occur under similar conditions – in both cases the offender is present during police service. This allows for the analysis to be focused on police decision making. To examine the influence of demographic, history of violence, and incident characteristics, on the police decision to make a dual arrest, and since dual arrest events are relatively rare, we use logistic regression to examine characteristics associated with dual arrest. We used the backward likelihood method as this is the stepwise method recommended by Andy Field in his SPSS statistics textbook (Field, 2000). Stepwise was used to allow us an additional way to compare the importance of individual variables. Using this method, we ran three models with three different dependent variables: dual arrest, whether police collected evidence, and whether Emergency Protective Orders were obtained. In the first model we tested whether any of the demographic and case characteristic variables explained dual arrest, and in the latter two models we tested whether dual arrest explained whether police obtained evidence then whether the police obtained an Emergency Protective Order. In each case the models were run with each set of independent variables (demographic, history of violence, and incident characteristics) separately, and then together as a full model. This allowed us to see whether dual arrest is a predictor of either police action. Results of these regression models are displayed and discussed below. For each of the three models, we report the Nagelkerke R2, the overall Chi2 , degrees of freedom, -2 Log Likelihood, goodness of fit, and the percent correct overall. Two demographic variables are included: relationship of couple, and race of offender. The relationship variable is dummy coded, using spouse as the reference category, so statistics for each of the other categories are reported in comparison to the spouse category. For race, we used black as the reference category because the vast majority of offenders were black; thus, white and other race is reported in comparison to blacks. Overall, we found that models did not result in strong goodness of fit scores or high R2s; however, in each model there were a few interesting findings worthy of discussion. Model 1: Using Demographic, Case, and Incident Characteristics to Predict Dual Arrest In this model, dual arrest is the dichotomous dependent variable, coded as one if there was a dual arrest, and zero if there was a single arrest. A range of explanatory variables is used to describe demographic, history of violence, and incident characteristics, so that any variables associated with dual arrest can be identified. Regressions for each of these sets of variables, as well as for the complete model, are presented below. Neither demographic variables, nor history of violence variables are predictors of whether the police make a dual arrest, and none of the variables in either group is individually significant. Clearly, most of the predictive power of these variables is in the incident characteristics category. The model for incident characteristics explains only, 5.9% of the variance in dual arrest, but two variables pertaining to suspected substance use are statistically significant. Dual arrest was 3.3 times as likely in cases where the victim is suspected of using substances (p<.01), and about half as likely when the offender is suspected of using substances (p<.05). The full model is a better fit but the only significant variables remain those pertaining to victim and offender suspected substance use. These results are displayed in Table 10 on the following page. Table 10: Dependent variable: Dual Arrest Demographics Case Characteristics Incident Characteristics Full model N= 738 N= 795 N= 756 N= 704 R2= .006 R2= .003 R2= .059 R2= .073 Chi2= 4.6 Chi2= 1.48 Chi2= 23.75 Chi2= 28 B Wald sig. Odds B Wald sig. Odds B Wald sig. Odds B Wald sig. Odds Demographic characteristics Former Spouse -3.9 .09 .75 .02 -3.4 .07 .78 .03 Cohabitants .18 .50 .48 1.2 .22 .67 .41 1.2 Child in Common -.38 .81 .37 .68 -.18 .16 .68 .83 White -.04 .02 .89 .98 -.17 .24 .62 .83 Other -1.1 1.1 .30 .34 -.83 .62 .43 .44 History of violence Prior arrests -.55 1.1 .29 .57 -.20 .02 .90 .99 Protective order in place -.04 .16 .69 .96 -.30 .30 .58 .74 Incident characteristics Victim substance use 1.12 19.9 .00** 3.3 1.2 18.4 .00** 3.3 Offender substance use -.64 5.9 .02* .53 -.59 4.6 .03* .55 Any weapon used .21 .42 .52 1.2 .31 .86 .35 1.4 Medical attention given .02 .00 .96 1.0 -.29 .50 .48 .75 Aggravated assault filed -1.2 3.2 .07 .31 -.95 2.0 .15 .38 Degrees of freedom 5 2 5 12 -2 Log likelihood 565.96 603.8 543.97 510.11 Goodness of fit 734.22 794.90 746.23 686.70 Percent Correct total 86.9 87.3 87.5 87.2 Notes: ** P<.01 * P<.05 Model 2: Using Emergency Protective Order as the Dependent Variable Using demographic, history of violence, and case incident, variables to predict whether police obtained an Emergency Protective Order produced a relatively weak model, although, of the three full models presented here, it has the highest R2 (17%). Since this analysis focuses on dual arrest, we present only the full model rather than also including the partial models in which dual arrest is not included (see Table 11 below). Table 11: Dependent Variable: Emergency Protective Order Obtained N= 704 R2= .17 Chi2= 87.19 B Wald sig. Odds Demographic characteristics Former Spouse -3.9 .28 .59 .02 Cohabitants -.37 3.1 .08 .69 Child in Common -.56 2.7 .09 .57 White .32 1.4 .23 1.4 Other -1.3 2.7 .09 .25 History of violence Prior arrests .06 .73 .39 1.1 Protective order in place 2.6 47.8 .00** 12.6 Incident characteristics Victim substance use -.37 2.2 .14 .69 Offender substance use .42 4.1 .04* 1.5 Any weapon used -.14 .22 .64 .86 Medical attention given -.15 .20 .65 .86 Aggravated assault filed .04 .01 .92 1.0 Dual Arrest -.49 2.3 .13 .61 Degrees of freedom 13 -2 Log likelihood 693.39 Goodness of fit 703.77 Percent correct of total 79.26 Note: * P<.01 ** P<.05 The single strongest predictor of having an EPO obtained is whether a protective order was already in place. This increased the likelihood of obtaining an EPO 12.6 times (P<.01). Whether the offender was suspected of using substances was also significant (P<.05), and increased the likelihood 1.5 times. Notably, several demographic variables were almost significant. In comparison to cases involving spouses, all others were less likely to have EPOs obtained. We were especially interested in whether dual arrest was a predictor of officers obtaining EPOs. This analysis indicates that EPOs are less likely to be obtained in dual arrest situations, but the difference is not significant. Overall, officers were 12.6 times more likely to obtain EPOs when protective orders were already in place and were about 1.5 times more likely to do so if the offender was suspected of using substances. Model 3: Using Whether Officers Collected Any Evidence as the Dependent Variable When regressed against whether officers collected any evidence, these same demographic, history of violence, and incident variables produce an R2of 8%, indicating a relatively weak model. The regression results are displayed in Table 12 on the next page. The strongest predictor of officers collecting evidence is whether medical attention was administered, which increased the likelihood of officers collecting evidence 2.2 times (P<.01). Having protective orders in place also doubles the odds of officers collecting evidence. Interestingly, the other history of violence variable, whether the offender had been arrested before for domestic violence, is almost significant, but in the opposite direction, indicating that officers may be less likely to collect evidence in cases where they are aware of prior intimate violence arrests. As with EPOs, dual arrest was inversely related to whether police collected any evidence, but was not significant. Table 12: Dependent Variable: Evidence Collected N= 704 R2= .08 Chi2= 38.22 B Wald sig. Odds Demographic characteristics Former Spouse 1.4 1.2 .27 4.2 Cohabitants -.16 .46 .49 .85 Child in Common .52 2.4 .12 1.7 White .45 2.4 .12 1.6 Other .45 .61 .43 1.6 History of violence Prior arrests -.25 3.5 .06 .78 Protective order in place .76 4.4 .04* 2.1 Incident characteristics Victim substance use .34 1.6 .20 1.4 Offender substance use -.04 .03 .87 .96 Any weapon used .28 .93 .33 1.3 Medical attention given .78 6.9 .00** 2.2 Aggravated assault filed .41 1.1 .29 1.5 Dual arrest -.18 .28 .59 .84 Degrees of freedom 13 -2 Log likelihood 604.68 Goodness of fit 707.22 Percent correct of total 83.38 Note: * P<.01 ** P<.05 Discussion Since presumptive and mandatory arrest laws have taken hold, dual arrests have become a more common outcome in domestic violence incidents. This has raised concern among advocates as well as police who want to ensure that, to the extent possible, police action protects victims by inhibiting revictimization. Many have expressed concern, however, that dual arrests are being applied to victims who acted in self-defense rather than to co-combatants. While answering this question is beyond the scope of this study, as the data do not allow for such specific conclusions to be drawn, it is possible to speculate based on comparisons between cases with different police outcomes. This first set of comparisons is between all three police outcomes and is descriptive. Second, logistic regression is used to compare cases to which dual arrest and single arrest are applied, to ascertain which variables are the best predictors of dual arrest. The logistic regression analysis excludes cases where a warrant was issued instead of an arrest being made to focus on the police decision between dual and single arrest. Dual arrest cases appear in general to be relatively less serious than those where a single arrest is made in that, compared to single arrest and warrant cases, they are less often classified as aggravated assault, involve medical treatment, or have protective orders in place; however, they have about the same average number of prior arrests. Since advocates worry that dual arrest is applied to particularly serious cases where one party if driven to violence in self-defense, and these data indicate dual arrest cases to generally be less serious, our data do not support the advocates’ concerns. However, we must once again caution against drawing definitive conclusion from these data. The logistic regression indicates that overall, the variables included do not fit and are not found to be strong predictors of dual arrest. The two variables that are significant are suspected substance use by the victim and by the offender. Suspected victim substance use tripled the odds of dual arrest. Since dual arrest is applied when officers cannot determine who is most to blame, this may indicate that victim intoxication increases the chance that officers view both parties as sharing substantial responsibility. While it is understandable that substance use by both parties might indicate mutual guilt to officers, these data do not explain officers’ decision making. We do not know whether the officers decided upon dual arrest due to true ambiguity, or whether the fact that both parties appeared to be using substances was interpreted as mutual responsibility. The authors feel that the possibility that officers interpret intoxication to mean complicity in violence is important to investigate, as it could have serious negative ramifications for victims. To explore the relationship between dual arrest and police action, two additional logistic regression models were run with dual arrest as a predictor variable (1) of whether emergency protective orders were obtained, and (2) of whether the police collected evidence. These police officer actions are somewhat discretionary, and thus may indicate how serious officers’ perceive these incidents to be. Dual arrest was not found to be a significant predictor of getting an EPO, though this model was the strongest of the three. The strongest predictor for obtaining EPOs, was whether other protective orders were in place at the time, which increased the likelihood 12.6 times. This makes sense, because if a police officer is informed that protective orders are in place, he or she would be expected to consider the case more serious. Interestingly, an offenders’ record of past intimate violence arrests known to officers is not found to be a predictor of whether officers obtain EPOs. One would think that these cases would appear more serious and thus merit EPOs. In addition, dual arrest was not a predictor of collecting evidence. Whether medical attention was administered was the strongest predictor, which more than doubled the chances of officers obtaining evidence. This outcome is not surprising. In cases where medical attention is administered, either at the scene, or after the victim is transferred from the site by ambulance, clearing the call takes much longer, giving the officer more time to collect evidence. Additionally, cases where medical attention is given may be the cases most likely to go to court where evidence will be needed by the prosecutors. The fact that the single most common form of evidence collected is photographic also may explain this outcome. Even though photographs of property damage or disarray can also be evidence, officers may be more inclined to take pictures of visible injuries than property damage, and visible injuries are probably more common in victims to whom medical attention is administered. One of the most interesting findings in this model is that measures of history of violence are contradictory. One would expect officers to take incidents more seriously and thus to collect evidence if there is a history of violence. This seems to be the case when protective orders are in place, where the likelihood of evidence collection more than doubles. However, in cases where the offender has prior arrests for domestic violence, the likelihood of officers collecting evidence decreases (odds ratio .78). One possible interpretation for this is that in cases where officers are aware that there are past offenses, officers may feel that the couple is resistant to outside assistance and that additional efforts may be ineffectual. Conclusion Of vital concern to law enforcement, victim advocates, and the community at large is ensuring that current responses to intimate violence do not “revictimize the victim.” The question has been raised that dual arrest may do just that, by arresting individuals, who, though they participated in violence, did so in self-defense and are more accurately classified as victims than offenders. Ascertaining to what extent, if any, this is the case is important in evaluating law enforcements’ response to intimate violence and specifically, its success at assisting and protecting victims. Distinguishing the types of cases in which dual arrest is most likely to be applied by officers may not answer this question directly or decisively, but we believe it has some utility for shedding light on it. This is the approach we have employed in the present analysis. Here we have sought to describe characteristics of dual arrest in comparison with those resulting in other police outcomes. We have also conducted analysis to determine whether certain case characteristics are predictive of dual arrest. Based on our data, cases to which dual arrest is applied appear to be somewhat less serious. We feel that much more needs to be done in this area to replicate as well as expand on our findings. Ideally, further analysis would use a dataset that includes more dual arrest cases to allow for greater sensitivity in detecting patterns among dual arrest cases. We also believe it would be useful to compare a dual arrest analysis based on a primary aggressor jurisdiction, such as this one, to one based on a non primary aggressor jurisdiction. In the absence of further research on the dual arrest outcome, dual arrests will continue to be made, yet it will be unknown whether intended law enforcement goals are being achieved, or whether victims are being “revictimized” by the system that has been sent to help. Continuing when so much is unknown is risking missing opportunities to help victims as well as increasing the chance that victims will become alienated from what may be the best protection available to them. References Alabama Coalition Against Domestic Violence. (2001). News: New bill names domestic violence as a separate crime. Retrieved August 17, 2001, from http://www.acadv.org/news.html Berk, R.A., Campbell, A., Klap, R., & Western, B. (1992). A Bayesian analysis of the Colorado Springs spouse abuse experiment. Journal of Criminal Law and Criminology, 83 (1), 170-200. Chalk, R., & King, P.A. (Eds). (1998). Violence in families: Assessing prevention and treatment programs. Washington, DC: National Academy Press. Domestic Violence Law Search a. (2001, November 29). National Council of Juvenile and Family Court Judges. Retrieved November 29, 2001, from http://www.dvlawsearch.com/database/search/show.php?id=1604 Domestic Violence Law Search b. (2001, November 29). National Council of Juvenile and Family Court Judges. Retrieved November 29, 2001, from http://www.dvlawsearch.com/database/search/show.php?id=323 Dunford, F.W., Huizinga, D., & Elliott D.S. (1990). The role of arrest in domestic assault: The Omaha experiment. Criminology, 28, 183-206. Feder, L. (1996). Police handling of domestic violence calls: The importance of offender’s presence in the arrest decision. Journal of Criminal Justice, 24, 481-490. Field, A. (2000). Discovering Statistics: Using SPSS for Windows. Thousand Oaks, CA: Sage Publications. Garner, J., Fagan, J., & Maxwell, C. (1995). Published findings from the Spouse Assault Replication Program: A critical review. Journal of Quantitative Criminology, 11 (1), (3-27). Gelles, R.J., & Cornell, C.P. (1990). Intimate violence in families, 2nd Edition. Newbury Park, CA: Sage. Grants to Encourage Arrest Policies and Enforcement of Protection Orders. (2001, August 17). Office of Justice Programs, Department of Justice. Retrieved August 17, 2001, from http://www.cfda.gov/public/viewprog.asp?progid=539. Hamberger, K. L. (1997). Female offenders in domestic violence: A look at actions in their context. Spouse/Partner Maltreatment, 117-129. Hirschel, J.D., & Hutchinson, I.W. (1992). Female spouse abuse and the police response: The Charlotte, North Carolina experiment. Journal of Law and Criminology, 83(1): 73-119. Holmes, W. M. (1993). Police arrests for domestic violence. American Journal of Police, 12 (4), 101-125. Institute for Law and Justice. (1997, June 30). Police warrantless arrest authority: Domestic violence and other violations. Retrieved November 27, 2001 from http://www.ilj.org/dv/exhibits/exhibit4.html. Martin, M. E. (1997). Double your trouble: Dual arrest in family violence. Journal of Family Violence, 12, (2), 139-157. Melton, H. (1999). Police response to domestic violence. Journal of Offender Rehabilitation, 29 (1/2), 1-21. Miller, N. (1999). Process evaluation of the Queens County arrest policies project. Retrieved April 8, 2002 from http://www.ilj.org/dv/CaseStudies/queens.pdf New Mexico, Office of the Attorney General, Violence Against Women Task Force: Final report. (2001, August 17). Retrieved August 17, 2001, from http://www.ago.state.nm.us/Publications/Violence/violence_against_women_task_fo.html Pate, A.M., & Hamilton, E.E. (1992). Formal and Informal deterrents to domestic violence: the Date County spouse abuse experiment. American Sociological Review 57: 691-697. Sherman, L.W., & Berk, R.A. (1984). The specific deterrent effects of arrest for domestic violence. American Sociological Review, 49(2), 261-272. Sherman, L.W., Schmidt, J.D., Rogan, D.P., Smith, D.A., Gartin, P.R., Cohn, E.G., Collins D.J., & Bacich, A.R. (1992). The variable effects of arrest on criminal careers: The Milwaukee domestic violence experiment. Journal of Criminal Law and Criminology, 83 (1), 137-169. State of Connecticut Department of Public Safety. (1991). Study of family violence incidents which result in arrest of both parties. Meridian, Ct.: State of Connecticut Department of Public Safety. Strauss, M.A. & Gelles, R.J. (1990). Physical violence in American families: risk factors and adaptations to violence in 8,145 families. New Brunswick, NJ: Transaction. Sutton, E. (1999). Assaulted, then arrested: How laws to protect battered women are landing them in jail. The Village Voice. Retrieved August 17, 2001 from http://www.villagevoice.com/features/9923/sutton.shtml. “The Right Mix on Domestic Abuse.” (2000,December 15-31).Law Enforcement News, XXVl (545). Retrieved August 17, 2001, from http://www.lib.jjay.cuny.edu/len/2000/12.31/looking.html Chapter 6 Appendix A Data Cleaning Table Reason for Deletion Number of records deleted Entered in database twice. 110 The relationship between the offender and victim was non-intimate. 112 The relationship was boyfriend/girlfriend – not included in VA code. 17 Relationship was noted by officer as “unknown”.* 8 The case accounted for two reports because it was a dual arrest.** 14 Police outcome listed as “No arrest.” 182 Police outcome listed as “unknown” or blank. 99 Victim and offender same sex. 8 Sex of offender could not be established. 64 Type of report not indicated. 73 Type of report other than assault (couldn't be recoded). 13 Number of cases remaining for analysis 1,376 *We deleted cases where the officer recorded the relationship as unknown to ensure that we were only considering cases between intimate or past intimate couples. **When dual arrests are made, each arrest is assigned its own incident number. Most officers report dual arrests by recording both incident numbers on the same PD 109, since there is a space designated for that purpose. Some officers report dual arrests by completing two forms, one for each offender. For consistency, we removed from the database the second report, leaving us with exactly one record per incident. Chapter 6 Appendix B Table of Data Used for Analysis in Columns Percent of Victims and Offenders in Various Demographic Categories by Outcome Variable Warrant Single Arrest Dual Arrest Total % N % N % N % N Sex of Offender Male 96.6% 561 93.2% 647 N/A N/A 93.8% 1291 Female 3.4% 20 6.8% 47 N/A N/A 6.2% 85 Race of Offender Black 90.1% 485 83.2% 534 85.4% 82 86.3% 1101 White 5.9% 32 13.9% 89 13.5% 13 10.5% 134 Other 3.9% 21 3.0% 19 1.0% 1 3.2% 41 Average Age Average victim age 29.2 (559) 32.4 (673) 32.8 (99) 31 (1331) Average offender age 30.6 (521) 34.7 (645) 34.6 (95) 33 (1261) Relationship Spouse 27.0% 157 31.1% 216 28.7% 29 29.2% 402 Former spouse 1.0% 6 .6% 4 0 0 .7% 10 Cohabitants 41.1% 239 54.9% 381 62.4% 63 49.6% 683 Child in Common 30.8% 179 13.4% 93 8.9% 9 20.4% 281 History of Violence Variable Warrant Single Arrest Dual Arrest Total % N % N % N % N Protective order in place 6.4% 37 6.8% 47 4.0% 4 6.4% 88 Offender's prior DV arrests None 77.9% 441 69.9% 485 73.0% 73 73.5% 999 One 14.1% 80 18.3% 127 15.0% 15 16.3% 222 Two 4.8% 27 8.1% 56 9.0% 9 6.8% 92 Three or more 3.2% 18 3.7% 26 3.0% 3 3.5% 47 Chapter 6 Appendix C Table of Data Used for Analysis in Rows Incident Characteristics Variable Warrant Single Arrest Dual Arrest Total % N % N % N % N Substance Use Victim 9.0% 52 20.9% 145 38.6% 39 17.2% 236 Offender 26.2% 152 41.4% 287 35.6% 36 34.5% 475 Both 6.0% 35 15.9% 110 26.7% 27 12.5% 172 Weapon Use Gun .9% 5 2.0% 14 1.0% 1 1.5% 20 Knife 2.8% 16 3.5% 24 6.9% 7 3.4% 47 Other 8.3% 48 7.8% 54 8.9% 9 8.1% 111 Any 11.7% 68 12.8% 89 14.9% 15 12.5% 172 Alleged Offense Physical assault 96.6% 561 97.8% 679 98.0% 99 97.3% 1339 Sexual assault .2% 1 0 0 0 0 .1% 1 Property crime 1.7% 10 1.3% 9 1.0% 1 1.5% 20 Psychological 1.4% 8 2.2% 15 3.0% 3 1.9% 26 Offense against child .2% 1 .4% 3 0 0 .3% 4 Other offense .9% 5 1.7% 12 0 0 1.2% 17 Additional Characteristics Aggressor present 7.1% 41 82.4% 572 90.1% 91 51.2% 704 Children witnessed incident 44.8% 260 41.1% 285 43.6% 44 42.8% 589 Medical treatment given 17.0% 92 14.8% 98 12.8% 12 15.6% 202 Police Action Variable Warrant Single Arrest Dual Arrest Total % N % N % N % N Type of Report Aggravated assault 6.2% 36 7.3% 51 3.0% 3 6.5% 90 Simple assault 93.8% 545 92.7% 643 97% 98 93.5% 1286 Evidence Collected Statements 6.0% 35 6.3% 44 4.0% 4 6.0% 83 Photographs 7.9% 46 9.2% 64 7.9% 8 8.6% 118 911 Tapes 0 0 .1% 1 0 0 .1% 1 Hospital records 1.2% 7 .4% 3 0 0 .7% 10 Other .02 2 .06% 8 .14% 2 .05% 12 Any 15.5% 90 17.3% 120 13.9% 14 16.3% 224 Obtained EPO 27.4% 159 25.9% 180 13.9% 14 25.7% 353 Victim and Offender Demographic Categories by Outcome Variable Warrant Single Arrest Dual Arrest % N % N % N Sex of Offender Male 43.5% 561 50.1% 647 6.4% 83 Female 23.55% 20 55.3% 47 21.2% 18 Total 42.2% 581 50.4% 694 7.3% 101 Race of Offender Black 44.1% 485 48.5% 534 7.4% 82 White 23.9% 32 66.4% 89 9.7% 13 Other 51.2% 21 46.3% 19 2.4% 1 Total 42.2% 538 50.3% 642 7.5% 96 Relationship Spouse 39.1% 157 53.7% 216 7.2% 29 Former spouse 60% 6 40% 4 0 0 Cohabitants 35% 239 55.8% 381 9.2% 63 Child in Common 63.7% 179 33.1% 93 3.2% 9 Total 42.2% 581 50.4% 694 7.3% 101 History of Violence Variable Warrant Single Arrest Dual Arrest % N % N % N Protective order in place 42.0% 37 53.4% 47 4.5% 4 Total 42.2% 581 50.4% 694 7.3% 101 None 44.1% 441 48.5% 485 7.3% 73 One 36.0% 80 57.2% 127 6.8% 15 Two 29.3% 27 60.9% 56 9.8% 9 Three or more 38.3% 18 55.3% 26 6.4% 3 Total 41.6% 566 51.0% 694 7.4% 100 Incident Characteristics Variable Warrant Single Arrest Dual Arrest % N % N % N Substance Use Victim 22.0% 52 61.4% 145 16.5% 39 Offender 32.0% 152 60.4% 287 7.6% 36 Both 20.3% 35 64% 110 15.7% 27 Weapon Use Gun 25.0% 5 70.0% 14 5.0% 1 Knife 34% 16 51.1% 24 14.9% 7 Other 43.2% 48 48.6% 54 8.1% 9 Any 39.5% 68 51.7% 89 8.7% 15 Alleged Offense Physical assault 41.9% 561 50.7% 679 7.4% 99 Sexual assault 100% 1 0 0 0 0 Property crime 50.0% 10 45.0% 9 5.0% 1 Psychological 30.8% 8 57.7% 15 11.5% 3 Offense against child 25.0% 1 75.0% 3 0 0 Other offense 29.4% 5 70.6% 12 0 0 Aggressor present 5.8% 41 81..3% 572 12.9% 91 Children witnessed incident 44.1% 260 48.4% 285 7.5% 44 Medical treatment given 45.5% 92 48.5% 98 5.9% 12 Police Action Warrant Single Arrest Dual Arrest % N % N % N Type of Report Aggravated assault 40.0% 36 56.7% 51 3.3% 3 Simple assault 42.4% 545 50.0% 643 7.6% 98 Total 42.2% 581 50.4% 694 7.3% 101 Evidence Collected Statements 42.2% 35 53.0% 44 4.8% 4 Photographs 39.0% 46 54.2% 64 6.8% 8 911 Tapes 0 0 100% 1 0 0 Hospital records 70.0% 7 30.0% 3 0 0 Any 40.2% 90 53.6% 120 6.3% 14 Obtained EPO 45.0% 159 51.0% 180 4.0% 14 Dependent variable: Dual Arrest Demographics Case Characteristics Incident Characteristics Full model N= 738 N= 795 N= 756 N= 704 R2= .006 R2= .003 R2= .059 R2= .073 Chi2= 4.6 Chi2= 1.48 Chi2= 23.75 Chi2= 28 B Wald sig. Odds B Wald sig. Odds B Wald sig. Odds B Wald sig. Odds Demographic characteristics Former Spouse -3.9 .09 .75 .02 -3.4 .07 .78 .03 Cohabitants .18 .50 .48 1.2 .22 .67 .41 1.2 Child in Common -.38 .81 .37 .68 -.18 .16 .68 .83 White -.04 .02 .89 .98 -.17 .24 .62 .83 Other -1.1 1.1 .30 .34 -.83 .62 .43 .44 History of violence Prior arrests -.55 1.1 .29 .57 -.20 .02 .90 .99 Protective order in place -.04 .16 .69 .96 -.30 .30 .58 .74 Incident characteristics Victim substance use 1.12 19.9 .00 3.3 1.2 18.4 .00** 3.3 Offender substance use -.64 5.9 .02 .53 -.59 4.6 .03* .55 Any weapon used .21 .42 .52 1.2 .31 .86 .35 1.4 Medical attention given .02 .00 .96 1.0 -.29 .50 .48 .75 Aggravated assault filed -1.2 3.2 .07 .31 -.95 2.0 .15 .38 Degrees of freedom 5 2 5 12 -2 Log likelihood 565.96 603.8 543.97 510.11 Goodness of fit 734.22 794.90 746.23 686.70 Percent Correct total 86.9 87.3 87.5 87.2 Notes: ** P<.01 * P<.05 Dependent Variable: Emergency Protective Order Obtained N= 704 R2= .17 Chi2= 87.19 B Wald R Odds Demographic characteristics Former Spouse -3.9 .28 .59 .02 Cohabitants -.37 3.1 .08 .69 Child in Common -.56 2.7 .09 .57 White .32 1.4 .23 1.4 Other -1.3 2.7 .09 .25 History of violence Prior arrests .06 .73 .39 1.1 Protective order in place 2.6 47.8 .00** 12.6 Incident characteristics Victim substance use -.37 2.2 .14 .69 Offender substance use .42 4.1 .04* 1.5 Any weapon used -.14 .22 .64 .86 Medical attention given -.15 .20 .65 .86 Aggravated assault filed .04 .01 .92 1.0 Dual Arrest -.49 2.3 .13 .61 Degrees of freedom 13 -2 Log likelihood 693.39 Goodness of fit 703.77 Percent correct of total 79.26 Note: * P<.01 ** P<.05 Dependent Variable: Evidence Collection N= 704 R2= .08 Chi2= 38.22 B Wald R Odds Demographic characteristics Former Spouse 1.4 1.2 .27 4.2 Cohabitants -.16 .46 .49 .85 Child in Common .52 2.4 .12 1.7 White .45 2.4 .12 1.6 Other .45 .61 .43 1.6 History of violence Prior arrests -.25 3.5 .06 .78 Protective order in place .76 4.4 .04 2.1 Incident characteristics Victim substance use .34 1.6 .20 1.4 Offender substance use -.04 .03 .87 .96 Any weapon used .28 .93 .33 1.3 Medical attention given .78 6.9 .00 2.2 Aggravated assault filed .41 1.1 .29 1.5 Dual arrest -.18 .28 .59 .84 Degrees of freedom 13 -2 Log likelihood 604.68 Goodness of fit 707.22 Percent correct of total 83.38 Note: * P<.01 ** P<.05 Chapter 7 Mapping Reported Intimate Assault Rachel Boba, Ph.D., Laura Wyckoff, M.A., and Erin Lane, M.P.M. POLICE FOUNDATION Prepared for the U.S. Department of Justice Office of Community Oriented Policing Services Award #98DVWXK015 June 2002 Chapter 7: Mapping Reported Intimate Assaults Introduction Since the advent of desktop-based geographic information systems (GIS) software, the application of spatial analysis to understanding and responding to a variety of crime and public safety problems has expanded. Researchers as well as practitioners have found GIS an invaluable tool for examining crime. For example, researchers have used GIS to study the spatial dimensions of various types of crime (Radcliff & McCullagh, 1998; Rogers & Delores, 1996), and police managers have adopted GIS as a means for understanding and responding to crime in their jurisdictions (LaVigne & Wartell, 1998; Taxman & McEwen, 1994; Weisburd & McEwen, 1997). GIS has, however, rarely been used as a tool to examine intimate violence. The paucity of spatial applications to the study of intimate violence is easy to understand. Researchers and practitioners have been disinclined to spatially analyze intimate violence because it is not considered a crime affected by place and environmental characteristics such as street lighting, type of road, or even neighborhood crime levels. Additionally, most types of intimate violence data are not well suited for mapping. There are many sources of intimate violence data, including victimization surveys, hospitals, shelters, and police departments, and each type has shortcomings. Victimization data are ideal for examining the spatial nature of intimate violence because they are the most comprehensive and least biased, providing a snapshot of intimate violence across all demographic categories. Yet, victimization data do not typically include location information, which is necessary for spatial examination. Another source, hospital records of admissions are biased toward more serious incidents because victims without serious physical injuries are unlikely to go to a hospital. Additionally, these records may be incomplete because medical staff may treat a patient for injuries without being aware that the injuries resulted from intimate violence. Even when records are complete, confidentiality issues often restrict the availability of these data to researchers. Still another data source is domestic violence shelters. However, since for a variety of reasons most victims do not go to shelters, records of shelter clients reflect only a small amount of actual intimate violence. Confidentiality concerns also make obtaining data from helters very difficult. Lastly, since police data include only incidents that are reported to the police, they are also subject to bias, but they do have two important advantages for spatial analysis. First, they are relatively accessible in that strict confidentiality does not apply, and second, they include location information that makes spatial examination possible. In the exploratory analysis presented here, we examine intimate assault reported to the police using spatial analysis methods that are frequently applied to other types of crime to determine if these methods are also useful for analyzing intimate assault. We examine intimate assault by frequency, type (simple or aggravated), and in relation to several demographic variables commonly considered with crime data such as population, population density, race, and median rent, and in relation to locations where victims could seek assistance. It is important to keep in mind that the incidents comprising this analysis represent intimate assaults reported to police, which is a subset of all intimate violence. In addition to the general biases of reported intimate assault data, the data used in this study are subject to several additional biases. One important characteristic of intimate violence is its repetitiveness, in that violence reoccurs in violent relationships. Unfortunately, it is not possible to analyze repeat addresses in these data because reliable apartment numbers are not present. Finally, the time range of the data used here is March 1999 to February 2001. It would be appropriate to use 2000 census data for this analysis, but block group level variables from the 2000 census were not available at the time of this examination. The Present Analysis This is an exploratory spatial analysis of intimate assaults, both simple and aggravated, reported to the city police department during a two-year period, using 1990 census data at the block group level provided by Environmental Systems Research Institute (ESRI). Census block group boundaries are used as proxies for neighborhoods, allowing for examination of reported intimate assault, by neighborhood characteristics such as population, population density, socio-economic level, and race. Research Questions The following are research questions that guide this exploratory analysis. 1.What is the spatial distribution of reported intimate assaults in this city? First, we examine reported simple and aggravated intimate assault separately, using the kernel density method. This method uses point data to determine the relative density of intimate assaults in the city (see ESRI, 1996). Because the demographic data used in the subsequent analyses are aggregated to the block group level, the reported intimate assaults are also examined through thematic shading of the block groups by number of assaults. 2.What is the relationship between reported intimate assault and demographic variables commonly examined in criminal justice research such as population, population density, socio- economic level (median rent), and race? A.What is the spatial distribution of reported intimate assaults by population? Using the block group as the unit of analysis, this analysis shows the rate of intimate assaults by population to determine if there are any areas of the city that have relatively higher rates of reported intimate assault. Population is obtained through the census data. B. What is the spatial distribution of reported intimate assaults by population density? A question often asked anecdotally about reported intimate assaults is whether areas with high population density have relatively higher numbers of reported intimate assault. The assumption is that people who live in closer proximity to others are more likely to hear intimate disputes and call the police – “thinner walls, more calls.” We examine the relationship between population density and number of reported intimate assaults. To determine areas as relatively high population density and high levels of reported intimate assaults or the reverse, which also follows the same argument, (i.e., lower density, lower number of reported intimate assaults), we break down the distribution of block groups into quartiles. That is, 25% of the distribution of block groups with the highest and lowest rates of population density are selected as are the 25% with the highest and lowest numbers of reported intimate assaults. The remaining block groups (that fall into these categories for both variables) then are coded into one of the following categories: 1. High population density/high frequency of reported intimate assaults 2. Low population density/low frequency of reported intimate assaults 3. High population density/low frequency of reported intimate assaults 4. Low population density/high frequency of reported intimate assaults These block groups are then mapped to determine their spatial distribution, comparing groups #1 and #2 to groups #3 and #4. Population density is obtained through the census data. C. How does the spatial distribution of the rate of reported intimate assaults by population compare to the median rent? Using the block group as the unit of analysis, a map depicting rate of reported intimate assault by population is compared to one depicting median rent by block group. Median rent is obtained through the census data. D.How does the spatial distribution of the rate of reported intimate assault compare to racial homogeneity of the population? Again, using the block group as the unit of analysis, a map depicting the rate of reported intimate assault by population is visually compared to one depicting racial homogeneity. Racial homogeneity is defined as a block group’s population being at least 80% one race and is obtained through the census data. E.What is the spatial distribution of block groups in which blacks or whites are overrepresented as offenders of reported intimate assault? Another notion in criminology is that minorities are overrepresented in phenomena such as arrests, convictions, and the prison population. Is this also true with intimate violence? All things being equal, we would expect the proportion of offenders of a given race to be equal to that races’ proportion of the population. This analysis compares the percent of black and white offenders residing in a block group with the percent of black and white residents of that block group to determine whether each race is under, over, or equally represented in the population of reported offenders. The maps shade block groups according to a ratio of these percentages to determine any spatial pattern. 3.What is the spatial relationship between reported intimate assault and locations of victim assistance? Localities seeking to ensure that victims are adequately served and protected would value knowing where intimate assault occurs in relation to locations where these victims could s eek assistance. While reported intimate assaults cannot be thought to accurately describe the actual intimate violence problem, they are a starting point. In this part of the analysis, we present a kernel density map showing reported intimate assault in relation to police stations and medical facilities. The reported intimate assault data come from the police department, and the city’s GIS coordinator provided the hospital, clinic, and police station point data. Methodology This analysis draws data from the city’s GIS coordinator, the city police department, as well as by ESRI. The police data come from domestic violence incident reports (DVRs), which officers are required to complete for every domestic violence substantiated calls. They provide demographic, relationship, and address data on the parties involved; the location, date, and time, of the incident; police action taken, presence of children, apparent use of alcohol or drugs by either of the parties, weapon use, victim injuries, and whether protective orders were on file. This analysis includes only those DVRs in which an aggravated or simple intimate assault has occurred between persons who are or were married, have a child in common, currently cohabit, or have cohabited within the previous 12 months. The city’s GIS coordinator provided a layer with addresses of hospitals and clinics, a layer with addresses of the four police stations, and a layer of the police patrol division boundaries. The resulting database contains 1,350 intimate assaults reported between March 1, 1999, and February 28, 2001. The street addresses of the incident were geocoded to the city’s street centerline files obtained through ESRI®, resulting in a geocoding rate of 94% (1223). Eighty- two (6%) of the addresses did not geocode. This is attributed to slightly out-of-date street files and, in some cases, to incomplete address information recorded by officers on the DVR form. Study City The study city is in the mid-Atlantic region of the country, covers approximately 62 square miles, and has a police department of approximately 700 sworn officers. A river runs through the city horizontally, and forms the northern border of the largest of the four patrol divisions Division D, and the other three, A, B, and C, which are north of the river. Between 1990 and 2000 the population decreased from 203,000 to 200,000. During this time the percent of whites decreased from 43% to 38%, the percent of blacks increased from 55% to 57%, the percent of Hispanics increased from 1% to 3%, and the other and mixed race category increased from 1% to 2% (ESRI; City Website). Map 1 on the next page displays the boundaries of the four patrol divisions and provides demographic context to each one. As noted earlier, we use the 1990 data for this analysis. Division A is the smallest of the four divisions, covering two square miles, and is the center city business district. This division has the lowest population per square mile of the four police patrol divisions (1,700 persons per square mile). Twenty-four percent of the population is white, 71% is black, and a small number of Hispanics, Asians, and others make up the remaining 5% (Police Department’s Web site). Three percent of the city’s reported intimate assault occurs in this division. Map 1: Patrol Divisions and their Characteristics Image of Map Patrol Division A 2 square miles 1,700 persons per square mile 24% white 71% black 5% other 3% of city's reported intimate assaults Patrol Division B 5 square miles 6,134 persons per square mile 9% white 90% black 1% other 27% of city's reported intimate assaults Patrol Division C 19 square miles 4,463 persons per square mile 54% white 45% black 1% other 21% of city's reported intimate assaults Patrol Division D 36 square miles 2,318 persons per square mile 46% white 53% black 1% other 49% of city's reported intimate assaults Within its small, five square miles, Division B has several public housing complexes that are well known to the police. Division B has the highest population density of the four divisions (6,134 persons per square mile), and is 90% black, 9% white, and 1% other. Twenty-seven percent of the city’s reported intimate violence occurs in this division. Division C is the most diverse of the four patrol divisions with three distinct areas. The northern band of this division is similar to Division B, with a good deal of low income, multi-family housing. In the southern section, the area closest to the center houses a university, and thus, many students. The wealthiest section of the city is in the southwest portion of this division. Division C covers 19 square miles and has a population density of 4,463 persons per square mile. This division accounts for 21% of the reported intimate assaults, is 54% white, 45% black, and 1% other. It is the only division where blacks are not the majority of residents. Division D is the largest of the patrol districts, covering 36 square miles, but has a relatively low population density of 2,318 persons per square mile. Blacks comprise 53% of the population, whites 46%, and 1% is made up of other races. Forty-nine percent of the city’s reported intimate violence occurs in this division. Analysis This analysis explores the spatial distribution of intimate assault in three parts. The first part uses two methods to depict the spatial pattern of citywide reported intimate assault. Second is a spatial examination of reported intimate assault in relation to population, population density, median rent and race. The third part is a visual examination of the spatial patters of intimate assault in relation to the locations of police stations and medical facilities – institutions where intimate assault victims could seek assistance. Our analysis includes two types of maps, kernel density and graduated area. The kernel density maps were created using a cell size of 50 feet and a search radius of one mile. The graduated area maps are thematically shaded by census block group. Six block groups were excluded from consideration in the thematically shaded maps. Five were excluded because they had no recorded population in the 1990 census, making rates impossible to compute. The sixth was excluded because its reported intimate assault rate was far outside the distribution due to an extremely small population and several reported intimate assaults. Part I: Overall Spatial Patterns of Intimate Assault “Hotspots’’ of Reported Intimate Assault: Maps 2 and 3 In order to (1) identify concentrations of reported intimate assault across block group boundaries, and (2) to compare the geographic concentrations of simple intimate assault to aggravated intimate assault, we produced kernel density maps of simple (Map 2) and aggravated (Map 3) reported intimate assault. The maps on the next page were created with a cell size of 50 feet and a search radius of one mile. Unlike graduated area maps, which consider the map sections (in this case block groups) separately, kernel density maps facilitate citywide comparison by indicating relative density of activity irrespective of block group boundaries. Map of Kernel Density of Reported Simple and Aggravated Intimate Assault per Square Mile Assaults can be charged as simple or aggravated depending on the seriousness. Aggravated assaults are more serious, and account for only 6.6% of all intimate assault charges in the city. To compare the geographic distribution of simple and aggravated assaults, Maps 2 and 3 are shown together. It is important to note that while the density patterns on these two maps can be compared their respective magnitudes cannot. Because there are so many more simple assaults than aggravated assaults, it takes many more simple assaults in one area to register as a high- density area. As shown in the legends, the “hottest’’ areas for simple assault have between 93.464-97.358 incidents per square mile while the “hottest’’ areas for aggravated assault have just 8.409-8.759 incidents – a vast difference. The concentrations of reported intimate assaults across block groups are readily apparent in Map 2, which, though it only shows simple assaults, is a suitable proxy for all intimate assaults since simple assaults are the vast majority of assaults. This map shows (1) that the highest concentration of reported intimate assaults occur in Division B, specifically in the northwest section, (2) that there are several areas of lesser concentration in the east of Division D, and (3) that there are few reported intimate assaults in the western areas of Divisions C and D. These comparisons indicate that the highest concentrations of reported intimate assaults occur in the northwest area of Division B, with some lesser concentrations in Division D, and that simple and aggravated assaults cluster in similar geographic patterns in the city. Since these maps show the geographic pattern of simple and aggravated assault to be nearly identical, for the remainder of the paper they are considered together as “intimate assaults.’’ Comparing the pattern of simple assaults (Map 2) and aggravated assaults (Map 3) reveals little differences in the clustering of these two types of assault. At first glance some of the aggravated assault clusters in Division D appear tighter, but this is because there are so few aggravated assaults, and just one or two assaults may appear as a cluster. Frequency of Reported Intimate Assault by Block Group: Map 4 To provide a baseline for the analyses by block group demographic characteristics, we created a graduated area map displaying frequency of reported intimate assault by block group, which is displayed on the next page. Block groups are shaded in blue with darker hues indicating higher frequencies of reported intimate assault. This method reveals that many block groups had no reported intimate assault over the two-year period. Many of these block groups are located in the western area of Division C, with a few scattered elsewhere in the city. Comparing this map to Map 2 confirms that the greatest concentrations of reported intimate assault are located in the northwest area of Division B and that several less intense concentrations are located in Division D. This map also identifies the specific blocks within these areas that have the highest frequencies of intimate assault. Using the maps together can provide a double check to the conclusions drawn based on only one. Map of Frequency of Reported Intimate Assault by Block Group Part 2: Relationship between Reported Intimate Assault and Demographic Variables Rate of Reported Intimate Assault by Block Group: Map 5 To examine the geographic distribution of intimate assault while controlling for population, we thematically mapped the rate of reported intimate assault per 100,000 persons by block group. In Map 5 on the next page, block groups are shaded in purple by rate of intimate assault, with darker hues indicating higher rates. We used the natural breaks method so that outliers as well as clusters of block groups with similar rates could be easily identified. The median rate is 546 reported intimate assaults per 100,000 persons for the two years of data. The mean is 706, and the standard deviation is 715. The high standard deviation indicates that there is a wide range of rates across block groups and that outliers are present. In terms of the overall distribution of intimate assault, Map 4 showing frequency, and Map 5 showing rate are very similar, in that the western areas of Divisions C and D have strikingly low frequencies of intimate assault and rates per population, and block groups in Division B and in some areas of Division D have the highest frequency and rates per population. Within smaller areas, however, the maps show differences. For example, along the northern border of Division B, the frequency map (Map 4) indicates the highest concentration to be in the northwest corner of Division B, whereas the rate map (Map 5) indicates the highest concentration to be in the northeast corner. This is also the case along the eastern border of Division D where Map of Rate of Reported Intimate Assault by Block Group Map 4 shows the three block groups to have high frequencies of reported intimate assault, while Map 5 shows these block groups to be in different rate categories, reinforcing the importance of considering rates when comparing block groups. Reported Intimate Assault and Population Density by Block Group: Map 6 Another demographic characteristic that may be related to reported intimate assault is population density. It is plausible that intimate assault would be reported more often in high density areas where neighbors cannot help but be aware of each others’ activities, and thus may more often call the police because of disturbances, than in areas where neighbors have more privacy. To e xplore the premise that population density is positively related to reported intimate assault, we created a map showing block groups with low intimate assault and low population density (shaded dark red), low intimate assault and high population density (shaded light red), high reported intimate assault and high population density (shaded in dark blue), and high intimate assault and low population density (shaded light blue). As discussed earlier, we achieved these classifications by ranking block groups in order of population density and then by frequency of reported intimate assault, and then grouping them into quartiles. Those block groups in the lowest intimate assault quartile that are also in the lowest population density quartile are classified as low/low and shaded dark red. Those block groups in the lowest intimate assault quartile that are in the highest population density quartile were classified as low/high and shaded light red. Those block groups in the highest intimate assault quartile that are also in the highest population density quartile are classified as high/high and shaded dark blue. Finally, those block groups in the highest intimate assault quartile that are in the lowest population density quartile are classified as high/low and shaded light blue. Block g roups that did not fall into any of these four categories are not considered in this examination and are not shaded. The more block groups in the high/high and low/low categories the more evidence for our assumption that intimate assault is positively related to population density. In Map 6 (on the next page) there are a total of 24 block groups that were either dark blue (7) high/high or dark red (17) low/low. All but one of the dark blue block groups is located in the area with the greatest concentrations and highest rates of reported intimate assault – the northern areas of Divisions B and C. Most of the dark red block groups are concentrated in the western areas of the city where the lowest frequencies and rates of domestic violence were observed. Several of these block groups are also in the central, commercial area of the city, Division A. The placement of the dark blue and red block groups fits our expectations about patterns of reported intimate assault in relation to population density because it comports with what we observed in the previous maps. Conversely, there were 23 block groups that are either light red (12) low/high or light blue (11) high/low. The light red block groups are mainly in two areas, Division C near the university and in Division B on the periphery of the highest concentration of reported intimate assault. All but one of the light blue block groups is in Division D, where most of these light blue block groups lie on the edge. The single light blue block group not in Division D is in Division C, near the concentration of reported intimate assault. Map of Reported Intimate Assault and Population Density by Block Group Since only 47 out of 190, or 25% of the block groups fall into any of these categories and the breakdown is nearly identical between high/high and low/low and the block groups that are mixed (low/high and high/low), no conclusions can be drawn. The assumption that more intimate assaults will be reported in more densely populated areas can neither be supported or denied. Rate of Reported Intimate Assault and Median Rent by Block Group: Maps 5 and 7 To explore the relationship between rate of reported intimate assault and socio-economic level (here measured as median rent), we compare Map 5 (rate of reported intimate assault) to a thematically shaded map showing median rent by block group, Map 7, displayed on the next page. In Map 7, the color green is used to show the median rent category of each block group, with darker shades of green indicating higher median rent categories. The citywide median rent is $429, and the mean and standard deviation are $426 and $158 respectively, indicating that median rents are normally distributed with 68% falling between $268 and $584, and that there are not extreme outliers. By examining these maps side by side, we found that rate of intimate assault and median rent appeared to be inversely related to each other. Map 5 indicates that block groups with higher rates of intimate assault tend to be in the eastern area of side of the city, where Map 7 indicates block groups with lower median rents tend to be and vice versa. To test this apparent relationship statistically, we ran a correlation on median rent and rate of reported intimate assaults by block group which resulted in a -0.35 Pearson R which was significant at the .01 Map of Rate of Reported Intimate Assault and Median Rent by Block Group level (N=190). The test indicates that median rent is inversely and significantly related to rate block group rate of intimate violence. In other words, as median rent decreases, the rate of reported intimate assault increases. Rate of Reported Intimate Assault and Racial Homogeneity by Block Group: Map 5 and 8 As mentioned in the introduction to this report, race is a demographic variable that is frequently considered in criminal justice research. There are many ways to examine race in relation to crime problems, and we employ two. On the next page we compare Map 5, rate of reported intimate assault by block group (discussed earlier), to Map 8, which shows racial homogeneity by block group. We define racial homogeneity as a block group’s population being at least 80% one single race. Map 8 contains three racial homogeneity categories. If the population of a block group is at least 80% white it is shaded pink, block groups where blacks make up at least 80% of the population are shaded purple, and block groups where neither race comprises 80% of the population are shaded gray. Examination of Map 8 reveals that there are distinct portions of the city where the 80% black block groups are clustered (predominately in the northeast in Divisions B and C), and where the 80% white block groups are clustered (predominately in the western areas of Divisions C and D). Racially heterogeneous block groups are scattered around most of the city, mainly in Division D. Generally, rates of reported intimate assault tend to be lower in the block groups that are at least 80% white. Many of the block groups with the highest rates of reported intimate assault are those that are at least 80% black. The area where the more heterogeneous block groups are, in Map of Rate of Reported Intimate Assault and Racial Homogeneity by Block Group Division D, is also where the rates of reported intimate assault appear to vary the most. A correlation was conducted (see Table 1) to see whether number of reported intimate assaults in block groups was correlated with the percent of block group race. The table indicates that the number of incidents in the block group is positively and significantly related to the percent black in the block group and significantly inversely related to the percent white in the block group. Table 1: Correlation Between Number of Reported Incidents and Percent of Black and White in Block Group % Black % White Intimate assaults Pearson Correlation 0.33 -0.34 Sig. (2-tailed) 0.00** 0.00** N 190.00 190.00 ** Correlation is significant at the 0.01 level (2-tailed). Proportion of Reported Offending to Population by Race: Maps 9 and 10 To examine race and intimate assault in another way, we explored whether offending by whites and blacks was proportionate to their share of the population in each block group. We created maps showing percent of total offending by each race in relation to percent population of each race by block group. That is, all things being equal we would expect the percent of offenders by race to match the racial make up of the block group. For example, if whites make up 50% of the population in a block group, we would expect them to make up 50% of the reported offenders. But if they make up 75% of the reported offenders, because 75%/50% is 1.5, whites are one and a half times as likely to be offenders than would be expected given their share of the population. In other words, they would be overrepresented. Calculations comparing proportion of observed to expected offenders were made for whites (Map 9) and for blacks (Map 10) for each block group and are illustrated on page 28. Maps 9 and 10 are thematically shaded by the ratio of observed versus expected offending by block group. Dark gray shading indicates that offenders of the race displayed are equally represented as offenders in that block group, and light gray shading indicates that offenders of the race displayed are underrepresented. The equal representation category captures block groups where the expected value is within two tenths of one. In other words, a ratio of observed to expected anywhere between .91 and 1.1 was considered to be equal representation. Color shading indicates that offenders of the race displayed are overrepresented as offenders in the block group. Block groups may be put into the “No intimate assault” category either because there were no intimate assaults reported in that block group, or because there are no residents of the race displayed in that block group. Taking the entire city into account, we found that blacks are 1.34 more likely to be reported offenders than expected given their share of the population. Whites, on the other hand, are .25 as likely, meaning that they are underrepresented as reported offenders compared to their proportion of the population. Specifically, they were 75% less likely to be reported offenders. Examination of the maps indicates that there are more block groups throughout the city in which black offenders are overrepresented (53), than there are block groups where white offenders are overrepresented (13). In examining Map 9, we found that the relatively few block groups where whites are overrepresented tend also to be block groups where the population is at least 80% black – where whites are the minority. In many areas where whites are at least 80% of the population, they are underrepresented as offenders. Map 10, displaying the ratio by block group of observed black offending to expected black offending based on the population, shows that block groups where blacks are overrepresented are distributed throughout the city in each of the three types of block groups. In block groups that are at least 80% black, blacks tend to be equally represented rather than overrepresented. This is not surprising since it would be more difficult to achieve overrepresention of a race when that race comprises such a substantial majority of the population. It is interesting that both blacks and whites seem more likely to be overrepresented as offenders in block groups where they are the minority. Map of Proportion of Reported Offending to Population by Race Part 3: Reported Intimate Violence in Relation to Service Providers Medical Facilities and Police Stations in Relation to Reported Intimate Assault: Map 11. To show the geographic clustering of reported intimate assault in relation to locations where intimate assault victims could seek assistance, we created a kernel density map of all reported intimate assaults and overlaid it with the locations of hospitals, clinics, and the headquarters of the four police divisions. We also include a cutout of the central city and “hotspot’’ (high density area of reported intimate assaults) area to make it easier to see all of the locations shown, and for closer examination of the sources of assistance near the “hotspot.’’ Such a comparison can be useful for localities wishing to assess how well the community’s needs are being served, by showing where service providers are in relation to crime problems. Since these data are reported intimate assaults rather than actual intimate assaults, this comparison is an imperfect measure of how well needs are being met, but it still offers an idea of how close possible sources of assistance are to concentrations of known intimate assault. Although analyzing reported crime should not be an end to assessingcommunity service needs, it can be a useful beginning. After all, reported crime tells you where at least part of the problem is, and the known part of the problem still needs attention. Map 11 on page 31 is not meant to be a definitive assessment of victim needs in relation to sources of assistance for the reason above, and, because it includes only a selection of possible sources of victim assistance. In this map, hospitals and police stations are self-explanatory, and we assume to be locations victims could access on their own or to which they might be transported. Clinics are a different matter, since there are many different types. We do not have enough descriptive information to differentiate between them, and thus cannot distinguish between clinics that would be appropriate sources of assistance to victims and those that would not. Additionally, there are probably many additional service agencies in the city to which victims could appeal for assistance. This map is meant merely as a general comparison between spatial patterns of reported intimate assault and a selection of possible victim assistance locations. Based on the locations that are included, overall it appears that there are many sources of victim assistance, including one of the four police stations, in and around the main hotspot area in Division B. Also, there seem to be many more hospitals and clinics throughout Divisions A, B, and C, and relatively few in Division D. Of the eight hospitals in the city, seven are in the northern part of the city in Divisions A, B, and C. Only one is south of the river in Division D, where, incidentally, 49% of the reported intimate assaults are generated. Map of Kernel Density of Reported Intimate Assaults with Police Stations and Medical Facilities Discussion This exploratory examination of the geographic patterns of reported intimate assault has sought to examine three questions: 1. What is the spatial distribution of reported intimate assaults in this city? 2. What is the spatial relationship between reported intimate assault and population, population density, median rent and race? 3. What is the spatial relationship between reported intimate assault and locations where victims could seek assistance? We addressed the first question by creating kernel density maps showing the spatial distribution overall and with graduated area maps showing the spatial distribution using frequency of reported intimate assault by census block group. We found that simple and aggravated assault had virtually the same pattern even though the magnitude was much different. The block group analysis indicated a similar pattern as the kernel density maps which served as a double check of our findings. In general these maps revealed the highest areas of reported intimate assault to be the northern area of Division B, and the adjacent area of Division C and the lowest levels of reported intimate assault to be the western areas of Divisions C and D adjacent to the river. To assess the relationship between reported intimate assault and several demographic variables commonly used in criminal justice research, we first created a graduatedarea map showing the rate of reported intimate assault by block group population. This map identifies roughly the same spatial patterns of reported intimate assault as the map showing frequency of reported intimate assaults. We then explored population density by displaying block groups with the highest and lowest population densities and highest and lowest frequencies of reported intimate assault to test the theory that areas of high population density generate greater reporting of intimate assault. We did not find strong evidence in support or against our assumption that population density was positively related to reported intimate assault. To examine median rent and race, we created three comparisons. First, we compared rate of reported intimate assault by block group to median rent by block group, and found a visually apparent inverse relationship between these variables. Areas where block groups had high median rents tended to be areas where rates of reported intimate assault were low. Conversely, areas where block group median rents tended to be low tended to have high rates of reported intimate assault. This relationship was confirmed statistically through a correction which found the variables to be inversely and significantly related. We then compared rate of reported intimate assault by block group to block group racial homogeneity. In the study city, block groups with populations of over 80% black tended to be located in areas of the city with high block group reported intimate assault rates and block groups with populations of over 80% white tended to be in areas with low reported intimate assault rates. This relationship was also confirmed using a correlation between frequency of reported intimate assault and percent black or white in the block group. The lower the percentage of whites in a block group, the fewer reported intimate assaults, and the higher the percentage of blacks in a block group, the more reported intimate assaults. We also examined race by mapping the ratio of each races’ expected offending to its observed offending by block group. We found that there were many more block groups where blacks were overrepresented as offenders of reported assault than where whiteswere reported offenders. Interestingly, many of the block groups where whites were overrepresented as offenders are those comprised of at least 80% black. The converse is also true for blacks; thus it seems that offenders are more likely to be overepresented in block groups where their race is the minority (20% or below). Overall, in this city, our results suggest that blacks are overrepresented as offenders of reported intimate assault and whites are underrepresented. The last section of this paper explores the relationship between reported intimate assault and locations of institutions victims could go for assistance. We found that many clinics, several hospitals, as well as one police station are located in and around the densest cluster of reported intimate assault, in the northern part of city, in Division B. On the other hand, only three clinics, and one hospital – of the eight in the city – are located in Division D. We caution against drawing conclusions about how well victims are served because our analysis uses reported intimate assault rather than actual intimate assault. This type of analysis could be very useful if victimization or hospital data were used, as these data would provide a more accurate view of victims needs which could then be compared to available resources. This paper is meant to explore the utility of using GIS to examine intimate violence by testing several possible applications of spatial analysis to intimate assault. While the reported intimate assault data analyzed here do not describe the geographic patterns of actual intimate violence, they do describe the geographic and demographic patterns of intimate offending known to police, thus offering a means for understanding the basis for police officers’ perceptions of intimate violence. In this case, it is easy to see how officers could get an inaccurate view of intimate violence in relation to socio-economic level and race, since the victims and offenders they encounter fall into particular categories. The majority of the offenders in these reported data are black and living in lower income areas (unfortunately two related characteristics). This does not mean, however, that blacks, or those less wealthy have a greater propensity for intimate assault. These observed preponderances may simply be functions of the fact that these individuals tend to live in higher population density areas where intimate violence is more likely to be reported to police. Our efforts to examine this issue were inconclusive, but we believe it to be an important subject to explore, especially so that individuals with certain characteristics are not unfairly maligned. This analysis employs several methods for examining intimate assault data. While the findings of each method generally reinforce each other, there are also small differences. We believe that descriptions of reported intimate assault based on areas of agreement among the maps are more reliable than those of areas where the maps indicate slightly different results. Conclusion We hope that the ideas for examining intimate violence spatially presented here will generate additional explorations of this topic. Ideally, future efforts will use hospital or victimization data to either replicate our methods or apply new ones. We believe that such efforts could prove useful in improving responses to intimate violence or at least in contributing to a better understanding of it. Given the ever-increasing importance of GIS as tool for addressing crime, the value of determining whether it can usefully be applied to addressing intimate violence is evident. If constructive uses exist, then they should be identified so that agencies responding to intimate violence can put them into practice. On the other hand, knowing that such uses do not exist creates an opportunity to direct resources for understanding and addressing intimate violence more constructively elsewhere. References Environmental Systems Research Incorporated. (1996). ArcView Spatial Analyst: Advanced spatial analysis using raster and vector data. Redlands, CA: ESRI. LaVigne, N. & Wartell, J. (Eds.). (1998). Crime mapping case studies: Success in the field. Washington, DC: Police Executive Research Forum. Taxman, F. & McEwen, T. (1994). High-tech computer mapping and low-tech community policing. Journal of Urban Technology, 2 (1), (85-103). Weisburd, D. & McEwen, T. (Eds.). (1997). Crime Mapping and Crime Prevention:Vol. 8. Crime prevention studies. Monsey, NY: Criminal Justice Press.