MENU TITLE: Domestic and Sexual Violence Data. Collection Series: NIJ Research Report Published: July 1996 51 pages 100,944 bytes ------------------------------------------------ Domestic and Sexual Violence Data Collection A Report to Congress under the Violence Against Women Act by the Justice Research and Statistics Association NCJ 161405 A Joint Publication of the National Institute of Justice and the Bureau of Justice Statistics U.S. Department of Justice Office of Justice Programs National Institute of Justice Jeremy Travis, J.D. Director Bureau of Justice Statistics Jan Chaiken, Ph.D. Director Bernard Auchter Project Monitor National Institute of Justice ------------------------------------------------ The studies reported here were mandated by Sections 40292 and 40509 of Title IV, the Violence Against Women Act, of the Violent Crime Control and Law Enforcement Act of 1994. The enabling legislation is as follows: SUBTITLE B--SAFE HOMES FOR WOMEN, DATA AND RESEARCH SEC. 40292. STATE DATA BASES. (a) IN GENERAL.--The Attorney General shall study and report to the States and to Congress on how the States may collect centralized data bases on the incidence of sexual and domestic violence offenses within a State. (b) CONSULTATION.--In conducting its study, the Attorney General shall consult persons expert in the collection of criminal justice data, State statistical administrators, law enforcement personnel, and nonprofit nongovernmental agencies that provide direct services to victims of domestic violence. The final report shall set forth the views of the persons consulted on the recommendations. (c) REPORT.--The Attorney General shall ensure that no later than 1 year after the date of enactment of this Act, the study required under subsection (a) is completed and a report describing the findings made is submitted to the Committees on the Judiciary of the Senate and the House of Representatives. SUBTITLE E--VIOLENCE AGAINST WOMEN ACT IMPROVEMENTS SEC. 40509. REPORT ON RECORDKEEPING RELATING TO DOMESTIC VIOLENCE. Not later than 1 year after the date of enactment of this Act, the Attorney General shall complete a study of, and shall submit to Congress a report and recommendations on problems of recordkeeping of criminal complaints involving domestic violence. The study and report shall examine-- (1) the efforts that have been made by the Department of Justice, including the Federal Bureau of Investigation, to collect statistics on domestic violence; and (2) the feasibility of requiring that the relationship between an offender and victim be reported in Federal records of crimes of aggravated assault, rape, and other violent crimes. The National Institute of Justice was charged with carrying out the study of domestic and sexual violence data collection by the States mandated by Section 40292. The Bureau of Justice Statistics (BJS) was given the responsibility of responding to Section 40509. Because national data collections originate at the State level, BJS joined the Federal study to the State study. ------------------------------------------ ------------------------------------------ This report was prepared by the Justice Research and Statistics Association (JRSA) under the direction of Joan C. Weiss, executive director. The primary author of this report is James Zepp, director of the JRSA National Computer Center. JRSA appreciates the invaluable contributions and support provided by the representatives of the National Institute of Justice, the Bureau of Justice Statistics, and the Federal Bureau of Investigation as well as the other U.S. Department of Justice officials who participated in this effort. JRSA also is indebted to the members of the Panel to Assess State and Federal Data on Domestic Violence and Sexual Assault for their commitment to addressing the issues of domestic and sexual violence and to this project, and to the directors of the State Statistical Analysis Centers around the country who took the time to respond to the survey on State data collection efforts. This project was supported under grant number 95- IJ-CX-0010 to the Justice Research and Statistics Association by the National Institute of Justice, Office of Justice Programs, U.S. Department of Justice. Opinions or points of view expressed in this document are those of the authors and do not necessarily reflect the official position of the U.S. Department of Justice. ----------------------------------------- TABLE OF CONTENTS Foreword Executive Summary - Key Findings - About This Report Purpose and Background Findings and Recommendations - Increasing or Improving State and Local Reporting - Recognizing the Need for Multiple Indicators To Fully Capture the Extent of the Problem - Identifying Existing Commonalities in Data Definitions and Reporting Methods Across States Data Collection Issues - Multidimensional Aspects of Domestic and Sexual Violence Offenses ----Offense type--domestic and sexual violence ----Service domains--criminal justice, health, and social services ----Criminal justice domains--law enforcement, prosecution, courts (criminal and civil), and corrections ----Intergovernmental--Federal, State, and local governmental coordination ----Multijurisdictional--multiple State and local agencies ----Definitional aspects of domestic violence - Alternative Approaches to Data Collection ----Single focus (domestic or sexual violence) law enforcement reporting systems ----Uniform crime reporting ----Crime victimization surveys ----Other criminal justice domains ----Multiservice domain data links ----Special studies - Concerns About Reporting Alternatives ----Incomplete coverage of all cases ----Possible duplication across data systems ----Incomplete reporting ----Definitional issues - Summary Appendixes Appendix A: List of Project Participants Appendix B: State Survey Form* Appendix C: List of Respondents to JRSA Survey of States and Territories Appendix D: State Survey Results* Appendix E: FBI NIBRS Status Report as of 06/02/95** Appendix F: A Note on Terminology for Crime Statistics *Appendixes B and D are not included in this ASCII text file and three tables are not available in this ASCII file. To view this information, access the Acrobat file available online or order the print copy of this publication by calling the National Criminal Justice Reference Service (NCJRS) at 800-851-3420. **Appendix E is also not included in this ASCII text file. Its information is being updated on a regular basis. For updates, contact the Bureau of Justice Statistics, U.S. Department of Justice, Washington, DC 20531, or access the Web site at http://www.nibrs.search.org. You can also obtain a print copy of this publication by calling NCJRS at 800-851-3420. --------------------------------------------------- Foreword The Violence Against Women Act provides for a fundamental change in our criminal justice system's response to violent crimes committed against women. This legislation reflects the recognition that violence against women is a crime with far- reaching, harmful consequences for families, children, and society. It recognizes the seriousness of the problem of violence against women and offers direction for a new, collaborative approach to preventing violence and protecting victims. As we seize the opportunity to make a difference in the lives of women and children victimized by violence, we want to be sure to proceed on the basis of knowledge. We need sound data to guide our policymaking. The Congress recognized this need by calling for a study to learn how the States could centralize data collection on the incidence of sexual and domestic violence offenses and to examine statistical recordkeeping at the Federal level for domestic violence-related criminal complaints. This report, supported by the National Institute of Justice and the Bureau of Justice Statistics, represents an important first step in analyzing what needs to be done to ensure reliable estimates of the extent and nature of violent crimes committed against women. The study found that the Federal Government and the majority of States currently are collecting some statistics annually on these crimes: 35 States collect data on domestic violence, 30 gather statistics on sexual violence. Some of these existing State programs can serve as working models for States that currently do not collect data on these crimes. For the future, the study identifies a need to grapple with two issues. One is the variation across States--in definitions, in types of victims included in reporting requirements, and in other elements--that make it difficult to compare or aggregate data at a national level. A second is the need to include not only law enforcement statistics but data from other parts of the criminal justice system. The study emphasizes the need for collaborative data collection within the criminal justice system and from other key sources such as health care providers, employers, and schools--in order to develop a more detailed picture of domestic and sexual violence. This study itself is evidence of the importance of collaboration: in this case, between researchers and policymakers at the Federal, State, and local levels. Research on violence against women is gaining unprecedented momentum. We look forward to continued partnership in using the tools of research and statistics to improve the way we deal with violence against women and treat women victims of violent crime. Bonnie J. Campbell Director Violence Against Women Office U.S. Department of Justice Jeremy Travis Director National Institute of Justice U.S. Department of Justice Jan M. Chaiken Director Bureau of Justice Statistics U.S. Department of Justice --------------------------------------------------- Executive Summary The objectives of the Project to Assess State and Federal Data on Domestic Violence and Sexual Assault were twofold: to identify ways that States could centralize the collection of information on the incidence of domestic and sexual violence offenses and to examine the problems of statistical recordkeeping at the Federal level for domestic violence-related criminal complaints. To meet the first objective--State data collection- -the researchers focused on the current status of reporting in the States, which could serve as a starting point for future data collection improvement efforts. The project concentrated primarily on law enforcement-based indicators of domestic and sexual violence offenses. Because of the complex nature of domestic violence as a community problem and the legislation's citation of centralized State data bases, consideration was given to information sources outside of the criminal justice system. A panel of experts representing backgrounds in criminal justice statistics, law enforcement, and victim services as well as other related concerns met twice to provide comments and suggestions to the project staff. Information from States and Territories on their data collection programs and issues was sought through a survey, to which 47 responded, and followup interviews. Information on Federal statistical recordkeeping was obtained from a wide range of Federal, State, and local agencies involved in domestic violence. The project primarily focused on the efforts of the U.S. Department of Justice--specifically the Federal Bureau of Investigation (FBI) and the Bureau of Justice Statistics (BJS)--to collect domestic violence incident data. The two major programs reviewed were the FBI's transition from the Uniform Crime Reporting (UCR) system to the National Incident-Based Reporting System (NIBRS) and the recent BJS modifications to the National Crime Victimization Survey (NCVS). Key Findings The most significant finding was that the Federal Government and a majority of States (35 for domestic violence; 30 for sexual violence) are currently collecting some statistics annually on these offenses. On closer inspection, however, there was a wide variation in how each State defines these offenses, determines what is counted, and measures or reports incidents. The variability also applies to the types of victims included in reports. Since some States have adopted family violence as opposed to domestic violence statutes, their statistics may include child victims along with adults. In addition, some State statutes apply regardless of the gender of the victims and the offenders, while others are not as inclusive of all possible relationships and living situations. This results in State statistics that may not be comparable or suitable for aggregation at the national level for estimates of prevalence and severity. The reasons for these disparities include differences in State criminal codes, the characteristics of existing State information systems, and the relative attention to the concerns of domestic and sexual violence given by individual States. Issues about accuracy and completeness of coverage also were raised about the FBI's and BJS's data sets and collection methodologies: o The FBI's summary UCR system, which is currently the main source of national criminal offense data, does not provide the detailed information needed to document the full extent of domestic violence- related events known to law enforcement agencies. o NIBRS, which is intended to replace the UCR system, would provide much needed detailed data for domestic violence-related offenses. The progress of law enforcement agencies toward complying with the NIBRS data standard, however, has been slower than originally anticipated when the standard was developed in the late 1980s. Further, while the NIBRS standard includes a major data element related to domestic violence (i.e., victim-offender relationship), it is not precisely tailored for measuring domestic violence as it now stands. Relying on local incident-based reporting systems that contain relevant information for measuring domestic violence but do not fully meet the current NIBRS data standard raises considerable practical difficulties at the Federal level. o The NCVS recently has been modified by BJS as a result of discussions and studies over the past 20 years about the problem of underreporting of domestic and sexual violence incidents. The revised survey instrument and procedures were fully implemented for the data set covering the 1993 calendar year, and the first data from the revised survey became available in late 1994. The first analyses based on this information were published as this report was being written. Because of the short time that these revisions have been in effect, they should continue to be evaluated for their effectiveness. About This Report This report discusses data collection issues raised by panel members, Federal, State, and local agencies, and survey responses. It suggests further actions for improving and expanding data collection and reporting at the State and national levels. The seriousness of domestic and sexual violence crimes is increasingly recognized, and the need for better measures to inform policy and planning decisions is clearly evident. Although no data system will be able to avoid all possible sources of error, efforts are being made by the Federal Bureau of Investigation, Bureau of Justice Statistics, and State agencies to improve the reporting content of their respective statistical systems. Achieving the goal of improved reporting and more comprehensive statistics for domestic and sexual violence, however, will require cooperation and coordination of many Federal, State, and local governments and agencies. Both organizational and technical solutions are needed. --------------------------------------------------- Purpose and Background Section 40292 of Title IV--the Violence Against Women Act--of the 1994 Crime Act specifies that a study shall be conducted on "how the States may collect centralized data bases on the incidence of sexual and domestic violence offenses within a State." The National Institute of Justice (NIJ) asked the Justice Research and Statistics Association (JRSA) to undertake a study of domestic and sexual violence incident data collection by the States. Section 40509 of the Violence Against Women Act authorizes that a study be conducted on the "problems of recordkeeping of criminal complaints involving domestic violence." It specifies that this study include an examination of: o The efforts that have been made by the U.S. Department of Justice, including the Federal Bureau of Investigation, to collect statistics on domestic violence. o The feasibility of requiring that the relationship between an offender and victim be reported in Federal records of crimes of aggravated assault, rape, and other violent crimes. The Bureau of Justice Statistics (BJS) was designated by the Office of the Attorney General with the responsibility of responding to this statutory requirement. Because national data collections originate at the local level, BJS joined the Federal study to the State study. The availability of comprehensive and reliable statistical data on domestic and sexual violence is a critical imperative because decisionmakers at State and local levels are confronting questions concerning appropriate policies and effective procedures for addressing this problem, and they need more information to guide their thinking. Since most laws and policies that deal directly with domestic and sexual violence offenses are passed by the States, it is appropriate that information systems are created or enhanced at this level of government. A project panel of domestic and sexual violence and criminal justice experts generally agreed that because these problems are complex issues involving many types of offenders and behaviors, statistical data are needed that can better inform policy decisions. The mixed results reported in studies on mandatory arrest policies for domestic violence offenders were cited as indicative of the possible unknown dimensions on which more information may help to provide answers or explain inconsistent policy outcomes.1 Depending on what information is collected, these data can help address questions such as: o What are the characteristics of offenders and victims? This could help agencies direct resources toward subpopulations particularly affected by these problems. o Are there different behavior patterns evident among various offender subgroups? As policymakers and agency managers wrestle with finding appropriate strategies for dealing with domestic and sexual violence, the documentation of any behavioral differences may help to explain the relative effectiveness of various approaches. o Are there differences in these offenses across communities? Unmet needs for crime prevention and treatment services could be identified for better planning and allocation of resources. o What trends are occurring in domestic violence? Do patterns vary over time by type of offender, victim, or offense? Much of this project's efforts concentrated on how the States and other key agencies define and measure these offenses. Because several States had established incident-based crime reporting systems long before the Federal Government began its efforts to implement a national system, and because many have been involved in addressing domestic and sexual violence issues, it is appropriate to study how other States can learn from this experience and be encouraged to collect data where questions remain about the prevalence and severity of these problems. For the purposes of this project, the legislation was broadly interpreted to mean an examination of crime statistics related to domestic and sexual violence. This may include measures of crime incidents, offenses, offenders, and victims. (See appendix F, which explains various terms used for crime statistics. Care in the use of terms for indicating what is being measured is important when discussing specific data sets.) Because of the wide scope of the study's topics and the limited time and resources available to conduct it, the project could only characterize issues and methods of data collection in general terms. Subsequent investigation would be needed to identify concerns such as specific differences among individual Federal, State, and agency data elements; statutory or administrative powers; and legal definitions of domestic and sexual violence. This project gathered input from two major sources: 1) a panel of experts representing a variety of backgrounds in domestic and sexual violence research and services, criminal justice statistics, and law enforcement, and 2) a survey of the States. The purpose of these efforts was to elicit information on the current status of data reporting, concerns about the quality of the data collected, and strategies for improving reporting on these offenses. The 1994 Violence Against Women Act specified that this project seek input from experts with backgrounds in law enforcement, criminal justice statistics, and victims' services. In fulfillment of this mandate, a panel of individuals representing these diverse interests was recruited. To further broaden the perspectives reflected in the discussions and in recognition of other related research activities, additional participants from other segments of the criminal justice system and health research also were invited. Although the focus was on data collection at the State government level, it was important to consider relevant Federal data collection efforts in this subject area. States often adopt Federal data standards for compliance with funding or reporting requirements, comparability with other States, and ease of adoption since no development time is required. Since the majority of information generated on domestic and sexual violence incidents is from local agencies, and they generally have the primary role in implementing criminal justice policies, their perspective on reporting concerns was important for this study. For this reason, participation by representatives from Federal and local governments was sought. Project activities included two meetings of the Panel to Assess State and Federal Data on Domestic Violence and Sexual Assault (April 10-11 and June 29, 1995; see appendix A for a list of participants). Information relevant to Federal data collection efforts on domestic and sexual violence was presented by representatives of BJS and the Federal Bureau of Investigation's Uniform Crime Reporting Program Office. In addition, representatives from other Federal agencies contributed their expertise, including the Bureau of Justice Assistance, National Institute of Justice, the Centers for Disease Control and Prevention, and the Department of Justice's Violence Against Women Office. Because of the complementary nature of Federal, State, and local goals, comments on the Federal statistical efforts also were received from non- Federal criminal justice professionals, including representatives from the International Association of Chiefs of Police (IACP), National Center for State Courts, six State Statistical Analysis Centers (Alabama, Connecticut, Colorado, Iowa, Massachusetts, and Michigan), and three local government agencies (the Washington Metropolitan Police Department; the Alexandria, Virginia, Police Department; and the Baltimore, Maryland, Mayor's Coordinating Council on Criminal Justice). Representatives from several domestic violence victims' advocacy/service organizations--e.g., National Resource Center on Domestic Violence, Pennsylvania Coalition Against Domestic Violence, and the House of Ruth--also participated in these discussions. During the first project panel meeting on April 10-11, 1995, in Washington, D.C., the Statistical Analysis Center (SAC) directors from Alabama, Connecticut, Colorado, Massachusetts, and Michigan provided descriptions of domestic and sexual violence data collection efforts in their respective States. Representatives from domestic and sexual violence agencies discussed data collection systems to which they contribute information, concerns about existing statistics on these problems, and issues involving data sharing among service providers and criminal justice agencies. Law enforcement and other criminal justice representatives cited some of the problems and limitations that domestic violence incidents present for the classification and processing of cases. Additional approaches to estimating domestic and sexual violence incidents were described by a researcher from the Centers for Disease Control and Prevention. Staff from the Bureau of Justice Statistics and the Federal Bureau of Investigation (FBI) provided information on current Federal data collection systems and recent efforts to improve these data sets. The panel members reviewed a draft survey instrument developed by JRSA project staff to obtain information from the States and Territories on existing domestic and sexual violence data reporting systems and on problems and solutions they have experienced. Based on suggestions offered during the panel meeting and on a field test conducted in four States, the survey instrument was revised. In May the final version (see appendix B) was mailed to each State's or Territory's SAC director; this official was considered most likely to have experience with the broad range of data sources on which information was sought. If a State did not have a SAC, the respective State Uniform Crime Reporting Program Office received the survey as the next most appropriate agency. A total of 47 responded--42 by the initial June 5, 1995, deadline, and 5 after reminder phone calls were made and additional copies of the survey were sent to nonrespondents. (See appendix C for a listing of responding jurisdictions. A summary of the survey responses is included in appendix D.) The panel reconvened on June 29, 1995, in Washington, D.C., to review the State survey results. In addition, Bonnie Campbell, the newly appointed director of the Violence Against Women Office, U.S. Department of Justice, spoke to the panel on the need for reliable crime data on domestic and sexual violence. She indicated that as national policies are being decided on these issues, more data would help to inform these discussions. Additional comments were made by representatives from the National Institute of Justice, Bureau of Justice Statistics, Bureau of Justice Assistance, National Academy of Sciences, and the National Center for State Courts. The FBI representative also briefed the panel on the FBI Criminal Information Systems Advisory Committee's recent decision to review that agency's position on the National Incident-Based Reporting System. The comments and suggestions from panel members were used in preparation of this project report. --------------------------------------------------- Findings and Recommendations The major finding is that many States are already collecting or are implementing systems to collect data on domestic and sexual violence offenses. According to the State survey results, 35 of 47 responding States and Territories collect domestic violence statistics annually, and 30 respondents gather sexual violence statistics. But there is wide variation among States that have systems in place or nearing completion with regard to what information is collected and how it is gathered. The variability reflects differences in how States have approached these two issues and the existing structures for collecting general crime incident data. For example, some States have passed specific domestic or family violence statutes that clearly define this as an offense and may even have statewide reporting requirements. Other States have not designated domestic violence as a separate offense but have instituted reporting systems for cases that can be characterized as such. If a State already had an incident-based crime reporting system, then it may have simply added or derived domestic violence crime statistics from this. Lacking this capability, other States have had to create domestic violence-specific reporting systems. Since law enforcement agencies are the primary point of first contact with the criminal justice system and are responsible for enforcing protection orders, their data are a major source for estimating the overall prevalence and severity of domestic and sexual violence problems known to the criminal justice system. Every State and Territory provides Uniform Crime Reporting (UCR) data; there are about 16,000 agencies contributing information to this statistical data set. Discussions at both panel meetings focused on the National Incident- Based Reporting System, one of the major developments in law enforcement statistics in the past 10 years. (See "Incident-based data" under "Uniform Crime Reporting," which describes the issues raised about this data system's future.) Information from prosecutors, courts, and corrections can provide important insights into the processing and subsequent dispositions of cases. However, comprehensive statewide statistical data bases from these sources are less frequently available than are law enforcement data sets. This may be due to the fact that most of these systems originally were created for operational information needs, such as case or agency management, and not necessarily for the production of statistical reports. Unless these capabilities were included in their initial system design or subsequently added, data on the characteristics of victims, offenders, offenses, and other related information may not be readily available. Although most States collect data on domestic and sexual violence, the State survey indicates that few draw on noncriminal justice information sources. As noted in the previous discussion of criminal justice data bases, unless specific statistical reporting capabilities have been built into these information systems, data may not be easily obtainable for research and policy analysis purposes. Despite data collection difficulties cited in the project panel discussions and the State survey responses, the importance of having reliable and comprehensive indicators for domestic violence was repeatedly highlighted. Data are needed for basic research on the problems of domestic and sexual violence, for resolving debates over proposed legislation and budgets, for guiding administrative or managerial decisions over program policies and resources, and for informing the public about the seriousness and extent of these problems in their communities. The following suggestions for future work on improving and expanding domestic and sexual violence data collection emerged from the panel discussions and survey responses. Increasing or Improving State and Local Reporting Although the majority of States are collecting some form of information on domestic and sexual violence offenses, there is still a significant number that do not. In addition, there are substantial disparities in the types and quantity of data collected. If the availability of reliable indicators for domestic and sexual violence is considered an important goal, then several strategies can be employed to improve and expand this reporting effort. To encourage States to improve or establish a data reporting program for domestic and/or sexual violence offenses, various educational activities could be provided to increase State officials' awareness of successful efforts by other States. Through publications or meetings, exemplary programs could be highlighted as working models for other States to adopt. An important factor in maintaining ongoing data collection programs is showing that information gathered is used for analyses or is worthwhile for decisionmaking. Such feedback can be a strong motivating influence to the staff who are generating data. By providing examples of analyses and applications based on these data, policymakers may be more willing to allocate the authority and resources necessary to establish or continue collecting information on these concerns. These may be analytic approaches to policy alternatives, crime analysis systems for police, or planning tools for agencies to direct staff and other resources. Recognizing the Need for Multiple Indicators To Fully Capture the Extent of the Problem It was evident from this study that domestic and sexual violence offenders are not single populations; there are a number of subgroups that have distinct behaviors and motivations. Although the reporting of aggregate statistical figures can provide some rough estimates of prevalence, important differences may be lost that could be significant for policy and planning decisions. Therefore, consideration should be given to identifying indicators that could inform these discussions. Special studies could help to provide detailed information beyond general statistical data sets. This includes methods for estimating offender or victim characteristics not captured in general statistical reports and examining individual histories for patterns of recidivism among offender subpopulations. Identifying Existing Commonalities in Data Definitions and Reporting Methods Across States A more indepth analysis of State reporting systems could be conducted to identify existing commonalities in definitions, forms, and practices across States. Although this study received materials from many of the States in response to the survey, not all States responded or provided all of the referenced items. The constraints of this project did not allow for a systematic comparison of individual data elements, procedures, and definitions across jurisdictions and agencies, but a review was done of the available materials that looked for examples and general trends among the respondents. A State-by-State analysis could serve as a good starting point for any national reporting standards or help in compiling multi-State data on domestic and sexual violence incidents. By disseminating this analysis to the States, some may be persuaded to modify their current practices to increase data comparability across States. --------------------------------------------------- Data Collection Issues The following pages present issues and concerns raised during the two panel meetings, during discussions with individuals involved with domestic and sexual violence problems or statistical information systems, and from the responses to the State survey. Any data collection efforts by the States will have to address many of these concerns in order to implement or expand their systems. Multidimensional Aspects of Domestic and Sexual Violence Offenses The many aspects related to domestic and sexual violence include possible data sources, point-of- case involvement or intervention, and definitional criteria that can determine what information may be collected and some of the resulting limitations. Recognizing these multiple dimensions is important to ensure that the resulting data sets respond to policy and other informational needs. Because both domestic and sexual violence are complex issues involving many social and psychological forces, reliance on a single statistical indicator does not present a complete picture of either problem. Offense type--domestic and sexual violence The legislation mandated that this project study data collection alternatives for both domestic and sexual violence offenses. These are separate crimes that usually have their own respective statutory definitions and reporting mechanisms. Although the concept of sexual violence has been expanded over time in many States to include a broader range of victimizations, its acceptance as a major criminal offense was well established. Recognition of domestic violence as a serious criminal offense is a more recent phenomenon. One indication of the difference between these crimes is that forcible rape was included in the State data compiled in the Federal Uniform Crime Reporting (UCR) system at its inception in 1930, while estimates of domestic violence-related incidents can only be inferred from UCR statistics for murder and manslaughter. Victim-offender relationship information is not collected for other crime categories, so no domestic violence estimates can be derived from these counts. The data standards published in 1988 for the National Incident-Based Reporting System (NIBRS), which is intended to replace the UCR system, broaden the definition of forcible rape and increase the sexual violence-related offense categories, but they still do not list domestic violence as an offense category or provide any other means for explicitly identifying these incidents.2 It is possible to infer domestic violence incidents from NIBRS information by using the victim-offender relationship data. However, this could be challenged as including incidents that are not necessarily part of a violent relationship. A further distinction between sexual and domestic violence incidents is their defining characteristics. Sexual violence is a crime that is generally determined by specifically prohibited sexually related acts taken by a perpetrator against another person. On the other hand, domestic violence-related incidents not only require a criminal act but also must occur between individuals with an existing or prior close relationship. Because any violent act may be considered domestic violence related if a legally accepted relationship exists between the offender and victim--and the definitions used vary from jurisdiction to jurisdiction--it is extremely difficult to collect consistent and reliable statistics for this offense. Furthermore, as with hate crimes, a distinguishing characteristic for domestic violence incidents may be the offender's intent in committing an act such as intimidation of the victim. In this instance, a relatively minor offense like vandalism of property may have a much greater significance when considered in the context of a violent domestic relationship. Consequently, the approaches for collecting data on domestic violence are different from those for sexual violence. In general, domestic violence requires more data elements and greater detail to derive statistical estimates and to differentiate the severity or nature of cases. While personnel generating incident reports need training to ensure accurate reporting for both offenses, domestic violence appears to present more difficulties for staff in the field to properly identify these cases. Other concerns that can affect accurate data reporting will be discussed in later sections. Service domains--criminal justice, health, and social services A truly comprehensive data collection effort for domestic and sexual violence incidents would involve multiple service domains. Because of the strong psychological and social impact that domestic and sexual violence can have on victims, awareness and treatment of cases often may not originate in or involve the criminal justice system. Data from health and social service providers can be used to supplement prevalence estimates based on criminal justice sources. Interest in integrating or sharing information across service domains to promote coordination of efforts has led the Bureau of Justice Assistance to fund three demonstration projects (in Baltimore, Maryland; Santa Clara, California; and the Commonwealth of Virginia). The experience from the Baltimore domestic violence demonstration project indicated that definitional differences as well as the varying responsibilities across agencies can create significant problems for data collection and integration. Concerns about client confidentiality also may prevent health and social service providers from sharing information on specific individuals with the criminal justice system. Even with individual case information, it may be impossible or extremely expensive to link data across agencies and service domains in order to avoid duplicate counting of incidents or persons. Further, the disparate sources of data and uses for which they are intended may argue against their close integration. A practical concern is that few States indicated in their survey responses that there were noncriminal justice data bases available for analyses. For this and the previously cited reasons, the active integration of criminal justice, health, and social services information systems may not be feasible or desirable. Among the alternatives raised in the panel discussions were special studies that could incorporate information across service domains and may provide some insights into the overall prevalence and handling of domestic violence cases. Limited tracking of cross-service domain interactions also may provide some estimates of this phenomenon without requiring an elaborate coordination of separate information systems. Examples of this are the New York State Domestic Incident Report form, which includes information on whether an officer made a referral to a protective or victim services agency, and the Wisconsin Domestic Abuse Report form, which tracks a case through sentencing and includes various forms of counseling/treatment as sentencing options and information on whether medical treatment was required. Since many States have mandatory referral laws for health and social service providers, report forms used for these referrals could provide some information on cases originating from noncriminal justice sources. Criminal justice domains--law enforcement, prosecution, courts (criminal and civil), and corrections Within the criminal justice system, data collection is complicated by the division of responsibilities across many independent entities--law enforcement agencies, prosecutors, courts, and corrections. Although some local jurisdictions have established integrated criminal justice information systems, most criminal justice data are fragmented along operational boundaries. The continued difficulties in obtaining complete and accurate criminal history records were cited as one indicator of the inability to track individuals as they move through the criminal justice system or recidivate for subsequent crimes. Improving this situation requires surmounting technical, procedural, and organizational obstacles across many agencies. Although this can be done, it is a lengthy, resource-intensive process that requires coordination and cooperation at all levels of the participating organizations. Factors that work against achieving this goal and other forms of information systems integration are changing governmental priorities for information, short-term budget horizons, frequent personnel changes, the absence of or conflicting data standards across information sources, and rapidly shifting technologies. Any major data collection effort should consider these issues as part of its planning. Intergovernmental--Federal, State, and local governmental coordination Even with a State-centered data collection effort, both the Federal and local levels of government need to be considered. Federal systems often influence the ways State agencies design or implement their data systems. As previously noted, the Federal Government has several major statistical data and records systems (UCR/NIBRS, NCIC 2000 the Project, National Crime Victimization Survey, and National Criminal History Improvement Program) that provide data standards, funding, and national data-sharing capabilities. Although many States have developed their own information systems independent of any Federal effort, when feasible, planning for future data systems should try to maximize their usefulness and potential for piggybacking onto other reporting systems to avoid repetitive data collection or incompatible information systems. In addition to the obstacles cited in "Criminal justice domains-- law enforcement, prosecution, courts (criminal and civil), and corrections," there can be wide variations in offense classifications and procedures from State to State because each State sets its own criminal code. This situation often requires a translation of State-generated information, such as offense codes, into a national coding scheme. The result may sometimes be a grouping of otherwise dissimilar cases. For example, in Louisiana an assault may be simply a verbal threat, while other States require physical contact or injury by an assailant. Consequently, Louisiana may appear to have a greater problem with assaults than other States when national compilations of State crime data are reviewed. Although automated criminal records systems can be programmed to perform many of these report translations, some cases may require human judgment to interpret the circumstances. In these situations, State agencies frequently find themselves between Federal data standards and the applicable State and local practices. Even though States generally determine criminal statutes, health and social service regulations, and many public services, it is local agencies that often implement or enforce these policies and programs. Consequently, local personnel are usually responsible for collecting information from victims and offenders and must process the reports for submission to State agencies. Given these circumstances, most statewide data collection programs will involve substantial intergovernmental coordination and cooperation among a large number of agencies. In the area of crime data, some of the larger States must collect and process data from 500 to nearly 1,000 different agencies. Gathering complete and accurate data across all of these entities in a timely manner can be a challenging task that requires careful planning, adequate resources (particularly training of local staff in reporting procedures), and good communications at all levels of participation. Multijurisdictional--multiple State and local agencies Because of victim and offender mobility, agencies increasingly must be able to share information across State and local boundaries. There are several Federal and regional efforts under way to either provide mechanisms to do this or to encourage the development of these systems. Achieving this goal will take time, however, and many technical and organizational obstacles will have to be overcome. One situation where this has been identified as a serious problem is with court protection orders because officials outside of the originating jurisdictions generally do not have ready access to the information required for enforcement. In addition, policies and standards for issuing court protection orders can vary from area to area, making enforcement across jurisdictional boundaries a complex issue for local authorities. Definitional aspects of domestic violence Because the recognition of domestic violence as a serious criminal offense is a relatively recent phenomena for many communities, understanding of the problem continues to evolve. Hence, the legal definitions and procedures for this offense can vary widely from jurisdiction to jurisdiction. The following discussion of two aspects of this problem explains some of the differences in how domestic violence incidents are treated. Domestic abuse as a continuum of violent actions. Domestic violence often does not consist of a single incident; it is instead a continual state of victimization.3 Therefore, recognizing and counting domestic violence incidents can be somewhat different from other types of crimes, which are generally limited in time and do not often involve repeated offender victimization of the same person. In the National Crime Victimization Survey, this circumstance is acknowledged by the use of a "series crime incident" designation for situations in which the respondent cannot identify details for discrete victimization events, and more than five victimizations occurred in the previous 6 months. Because intimidation also may be part of the behavior exhibited by abusers, offenses that otherwise would not be considered violent may be characterized as part of an overall behavior pattern. This further complicates the recording of criminal events since it involves judging an offender's motivation. With this concept of domestic violence as a continuum of behaviors rather than a discrete event, it is understandable why domestic violence statistics may include many different criminal actions that vary from relatively minor offenses to extremely serious ones. This situation particularly applies as the definition of domestic violence is made more inclusive of the range of abusive behaviors. Some of these may or may not be part of the governing statutes for any given State, which may make comparing data across States exceedingly difficult. It also becomes a measurement issue because thresholds along this continuum must be established to determine whether and how any given event or action is counted. These decisions have obvious implications for resulting statistical figures and observations. A range of relationships from spouses to intimate acquaintances and nonrelatives. Some States are inclusive in their definition of domestic violence as far as the range of victim-offender relationships allows, while others require either a spousal relationship or a child in common for a criminal act to be considered domestic violence.4 As legal definitions of domestic violence expand to include more informal relationships such as current or ex-boyfriends/girlfriends, roommates, and cohabitants, the reporting systems must be modified to reflect these broader perspectives and to differentiate the characteristics of these victim/offender subgroups within the general population. Otherwise, important details may be lost in working with aggregate statistics, which could blur the diversity among domestic violence victims and offenders. Alternative Approaches to Data Collection In both the panel discussions and State survey responses, it was evident that the States have adopted a wide variety of approaches to collecting data on domestic and sexual violence incidents.5 This is largely explained by differences in statutory legal definitions and procedures, individual agency authority, and existing data and reporting systems. For example, the State of Connecticut uses a separate form to collect family violence offense data since it does not have an operational general crime incident-based reporting system. This data collection program is mandated by a statute that also specifies many of the data items to be gathered. The State intends to incorporate this information into NIBRS, which it is currently implementing. Single focus (domestic or sexual violence) law enforcement reporting systems Information specific to domestic and sexual violence offenses often has not been included in general criminal justice reporting systems. Because some State statutes call for reporting on these concerns, one approach used by several States has been the development of a special report form for either domestic violence or sexual violence incidents.6 In some instances this is a supplement to a State's Uniform Crime Reporting form (Kansas, Michigan, and Puerto Rico), while in others it is a completely separate reporting system (Connecticut, Massachusetts, New York, and Wisconsin). As States move from the aggregate data reporting system (UCR) to incident-based, NIBRS-type systems, many States operating separate reporting systems for domestic and sexual violence offenses (e.g., Connecticut) are planning to merge them into the general data collection program. One advantage of a single-focus approach to data collection is that it allows gathering of offense- specific information that may be more difficult to include in a general form intended for all possible crimes. The trade-off is that specialized forms may be more difficult to implement on a statewide basis and may be considered an additional reporting burden. Uniform crime reporting Since 1930 the Uniform Crime Reporting system has been the major national statistical series for crimes known to law enforcement agencies. All States contribute data to this program in some manner. This system, however, only reports aggregate totals for crimes, which limits any analyses to indicating general trends in crime and not much more. In the 1980s the FBI conducted studies that led to the decision to replace the summary UCR system with NIBRS. The following discussion highlights some of the issues related to UCR and NIBRS and the implementation status of these data systems. Aggregate crime data. The summary UCR program provides only aggregate or total data on eight major crimes and on arrestee age, sex, and racial/ethnic characteristics for all offenses. Known as the Part I Offenses, the eight offense categories are: o Criminal homicide (murder and nonnegligent manslaughter). o Forcible rape (rape by force and attempts to commit forcible rape). o Robbery (theft using a firearm, knife or cutting instrument, other dangerous weapon, or strong-arm method to force or threaten a victim). o Aggravated assault (an unlawful attack on a victim using a firearm; knife or cutting instrument; other dangerous weapon; or hands, fists, and feet). o Burglary (includes forcible entry; unlawful, nonforcible entry; and attempted forcible entry of a structure to commit a felony or a theft). o Larceny theft (the unlawful taking, carrying, leading, or riding away of property from the possession or constructive possession of another person, except for a motor vehicle). o Motor vehicle theft (theft or attempted theft of an auto, truck, bus, or other vehicle that is self- propelled and runs on land surface and not on rails). o Arson (willful or malicious burning of or attempt to burn a structure, vehicle, or other property, with or without intent to defraud).7 About 16,000 law enforcement agencies participate in the FBI's UCR system, and almost every State has a functioning UCR Program Office. With its long history and stability, UCR statistics are the most frequently cited data for national, State, and local crime trends in this country. Unfortunately, the weaknesses of the UCR are generally acknowledged, and its value for policy and planning decisions is extremely limited. At this time the majority of States still produce only UCR data reports. Although most States are working toward implementing a statewide incident- based reporting system, only a few have fully operational programs. Even among States with operational systems, the data reported may reflect only a small proportion of the jurisdictions or crimes in that State.8 Although it would be technically feasible to add domestic violence to the offenses reported as aggregate statistics, there potentially would be some confusion with other offenses. Since domestic violence can be any criminal act between persons having a legally recognized relationship for this purpose, in an aggregate statistical reporting system that includes domestic violence as a category, it would be impossible to tell how many domestic violence incidents were murders, how many were assaults, or even how many were acts of intimidation that did not involve physical harm. Conversely, in an aggregate reporting system, the proportion of assaults that were related to domestic violence incidents could not be determined because only a single figure is given for each offense category. Recognizing the dual nature of domestic violence, incidents could be counted for both domestic violence and other appropriate offense categories. This practice, however, would result in possibly double-counting crime incidents with no way to clearly determine the extent of overlap between domestic violence and other offense categories. More complete estimates of sexual violence from aggregate data systems could be derived by expanding the number of sexual violence offenses to include more than just forcible rape. These statistics would have the same limitations as other aggregate information. However, because of the complexity of domestic and sexual violence as interpersonal events and as crimes, aggregate statistics would be of minimal value for policy and planning decisions. One solution to the problem of lack of detail in aggregate statistics is exemplified by a study conducted by the Centers for Disease Control and Prevention (CDC), which reviewed individual nonfatal incident reports from the Atlanta Police Department to identify how many involved family or intimate assaults.9 Even though the department does not specifically record family or domestic violence as an offense, CDC staff determined from a sample of cases reported in 1984 which ones could be classified as such. Based on the proportion of family and domestic violence incidents found in this sample, estimates by known offense category could then be calculated for the total reported crimes. Assuming that the rates found in this study are transferable to other jurisdictions and time periods, this approach could be used to estimate domestic violence incidents occurring in other areas. Similar studies could be conducted to assess the validity of these estimators for other jurisdictions and time periods. Incident-based data. With local agencies increasingly automating their records systems, more will have the ability to generate electronic data files that can be used for a variety of reports and analyses, and they will be able to develop greater capabilities to conduct incident-based analyses. Although each agency may be able to electronically access individual records, the information identified may not be consistent across agencies because of a lack of State data standards, differences in software used, or local operational characteristics. Therefore, although automation may improve the likelihood that States can obtain data from local agencies, it will not guarantee ready input for centralized data bases. In 1988 the FBI published its NIBRS data standards. This system is the most prominent national effort to have States collect crime incident data and represents the most likely source of reasonably comparable State statistics on reported domestic and sexual violence incidents. The NIBRS data standard is a substantial departure from the summary UCR system. Instead of aggregate data for 8 crimes, NIBRS specifies 52 data elements (one more data element was subsequently added for identifying hate crimes) to be reported for every crime incident known to law enforcement.10 Reporting NIBRS information is even more complex than it appears from the previous statement because NIBRS data elements are divided into six groups or segments--Administrative Data, Offense, Property, Victim, Offender, and Arrestee. Except for Administrative Data, all of the segments may need to be addressed for the same incident depending on the circumstances. For example, a criminal kills a security guard while robbing a bank. In reporting this incident, two Offense Segments would be entered--one for murder and one for robbery. Two Victim Segments also would be submitted--one for the murdered guard and one for the bank robbed. Depending on what actually was taken or damaged by the robber, one or more Property Segments also may be required. The UCR system used a hierarchy rule, which allowed for the reporting of only the most serious offense committed during an incident; and it had much more limited, if any, related information about the crimes being committed. Because most automated information systems must allow for the largest possible data record, NIBRS data files can be quite large relative to the amount of data being reported. NIBRS also increases the number of crime categories about which offense and arrestee data are collected from 8 to 22. Group A includes the following crimes: - Arson - Gambling offenses - Assault offenses - Homicide offenses - Bribery - Kidnaping/abduction - Burglary/breaking and entering - Larceny/theft offenses - Motor vehicle theft - Counterfeiting/forgery - Pornography/obscene material - Destruction/damage/vandalism of property - Prostitution offenses - Robbery - Drug/narcotic offenses - Sex offenses, forcible - Embezzlement - Sex offenses, nonforcible - Extortion/blackmail - Stolen property offenses - Fraud offenses - Weapon law violations Arrestee data only are collected for 11 additional offenses, which are referred to as Group B offenses: - Bad checks - Liquor law violations - Curfew/loitering/vagrancy violations - Peeping Tom - Runaway - Disorderly conduct - Drunkenness - Family offenses, nonviolent - Trespass of real property - All other offenses Although NIBRS does not include a specific domestic violence offense code, it is possible to estimate domestic violence incidents from NIBRS data because victim-offender relationships are reported for crimes against persons and robbery. Using this data element, domestic violence estimates can be calculated for any offense category. The absence of a national definition of domestic violence causes irregularities in the inclusion/exclusion of more informal relationships such as current or ex-boyfriends/girlfriends, roommates, and cohabitants. Some States define domestic violence in terms of a wide range of victim-offender relationships, while others require either a spousal relationship or a child in common for a criminal act to be considered domestic violence.11 The exclusion of informal relationships obscures the distinctions between the characteristics of these victim/offender subgroups and the general population and minimizes the aggregate national estimate. During the April 10-11 meeting of the project panel, FBI representatives indicated that adding to the allowable codes for a specific NIBRS data element would be much easier to implement than expanding the number of data elements. For example, codes for indicating "ex-boyfriend/girlfriend," "child in common," and "shared domicile" could be introduced for the victim-offender relationship data element, which would reflect expanded definitions for domestic violence that are used by some States. Similarly, domestic violence could be added to the list of NIBRS offense codes. Adding a separate domestic violence data element (such as Michigan and Kansas have done to their incident- based reporting data standard), however, would create significant problems for Federal, State, and local governments because they would have to reprogram software and reprint manuals and forms to accommodate this change. Including victim-offender relationship data in the national statistical data standard does have one important benefit. Because the definitions for domestic and sexual violence can vary widely across States regarding which types of relationships and offenses are included, national estimates based on State-reported statistics for these crimes would not be uniformly consistent. Calculating prevalence estimates using selected offense categories and victim-offender relationship data would produce more reasonably consistent national statistics since these figures would be less affected by differences in State domestic/sexual violence definitions. Another major difference between the NIBRS and UCR systems is that the NIBRS record links original incident reports to subsequent arrestees. This feature will allow further analysis of crime characteristics related to the eventual capture of a suspect. For domestic violence incidents, this would provide some indication of how frequently arrests are made in these cases and for various related offenses or types of offenders. As long as a State can produce a data file that complies with the FBI's standards, it is permitted to expand the incident-based data elements collected from law enforcement agencies within its jurisdiction. One reason for doing so is to increase the utility of the data for resource allocation studies, crime analysis, and other applications that need information on crime patterns. Because NIBRS was originally designed as a statistical data base, it intentionally lacks certain information items that help agencies with their operational needs. For example, some States include geocodes to map crime data and offender physical characteristics, such as scars, marks, and tattoos, to identify specific criminals and to link previously unknown cases to particular suspects. The variability across States in their incident- based reporting standards has created some difficulties for automated information systems vendors and agency data processing managers. It generally increases the amount of computer software customization that is required for each agency before it is able to produce NIBRS-compliant data files. Consequently, reporting agencies have experienced difficulties either in locating records management software that complies with the NIBRS data standard as well as their own State standard or in obtaining the resources necessary to revise their existing reporting systems. Because of these and other concerns, NIBRS implementation across the States has been slow, and many State and local agencies have complained about the additional reporting burden placed on them. Since participation in UCR and NIBRS is voluntary, almost any expansion of these reporting requirements can be a significant resource issue for many agencies that are already coping with difficult budgetary choices. According to the FBI UCR Program Office, as of June 2, 1995, 8 States had been approved as capable of producing NIBRS-compliant data, and 20 States were testing for NIBRS compliance. Only 7 States do not have plans or have not indicated interest in NIBRS participation.12 The survey on State domestic and sexual violence data collection conducted by JRSA also indicated that a majority of the States (38, or 81 percent) have or plan to have an incident- based reporting system in the next 2 years.13 Although only a few States are currently producing NIBRS-compliant data, most are moving in the direction of implementing some level of incident- based reporting capability statewide. Even some of the States with no intention of meeting the FBI's data standards either had or were considering an incident-based reporting system. Consequently, there will probably be a subset of States for which crime incident data can be aggregated for national- level analyses, and there even may be certain data elements for which some additional States can provide information to further expand data coverage. Attaining the goal of a crime incident- based reporting system that is national in scope and contains uniform data elements, however, may take substantially more time than originally envisioned and may never be fully achieved without a significant commitment of Federal resources. State survey responses also revealed that States have adopted various reporting methods for collecting domestic violence data. For example, Michigan and Kansas have added a box on their crime incident report forms that officers must mark to indicate whether an incident was domestic violence related. Other States (e.g., Connecticut, New York, and Wisconsin) have separate forms for reporting domestic violence. The special domestic violence report form enables States to collect offense- specific information that may be more difficult to include in a general crime incident report form. Use of a separate form, however, does carry the physical and psychological burden of additional paperwork, which increases the likelihood that officers will fail to complete or submit a report. Finally, it should be noted that any incident-based reporting system also will be subject to the underreporting concerns that have been raised about the summary UCR system. Law enforcement agencies cannot be expected to provide data on crimes of which they are not aware. Consequently, additional sources may be consulted to estimate the degree to which the problem of underreporting affects the crime statistics gathered by law enforcement agencies--such as a major Federal statistical data base, which is discussed in the next section. Crime victimization surveys Crime victimization surveys supplement crime statistics generated by law enforcement agencies. Their purpose is to provide data about crimes that may not be reported by victims to police and to obtain detailed information that may not be collected in traditional police reports. State survey responses showed that only a few States (10, or 21 percent) conduct such surveys, and only half of these collect domestic or sexual violence data.14 Among the possible reasons for the low number of States that do these studies are the costs involved, controversies that can arise over survey questions or methods, and resistance by government agencies to burden citizens with more information requests. National Crime Victimization Survey. At the national level, the Bureau of Justice Statistics operates the National Crime Victimization Survey (NCVS). The survey itself is administered by the Bureau of the Census through a cooperative agreement. It collects information on the following completed and attempted crimes: rapes, robberies, assaults, larcenies, burglaries, and motor vehicle thefts. The survey's purpose is to gather information on unreported crimes as well as on aspects of reported crimes that may not be collected through police-generated crime statistics. The data consists of interviews from approximately 50,000 housing units and 100,000 persons. All members of a selected household are interviewed every 6 months for a period of 3 years. The first and fifth interviews are conducted in person; the remainder are done over the phone.15 To encourage cooperation in relating what may be very personal or potentially embarrassing information, interviewees are assured that their responses will be kept confidential. Interviewers receive special training to address those portions of the survey instrument requiring greater sensitivity. Past NCVS-based family violence statistics have been substantially lower than those from specific family violence surveys and studies. In order to address specific concerns about the NCVS's ability to accurately reflect the incidence of domestic and sexual violence, several revisions in the survey's questions and procedures have recently been implemented. These changes in the NCVS should bring its estimates closer to those from other information sources, but some differences will remain due to NCVS's purpose, which is to measure potential criminal behavior, as opposed to other studies' orientation toward family or domestic relations. The 1993 statistics are the first data set to fully incorporate these changes. The actual data became available in late 1994, and the first analyses of this data were published in August 1995.16 One important distinction is that the questions on the NCVS instrument are intentionally phrased to elicit information about respondents' experiences during the previous 6 months regardless of whether or not they could be considered crimes. This approach is taken to avoid underreporting due to interviewee misconceptions, which may be either culturally based or personal perceptions concerning what constitutes criminal behavior. Therefore, in addition to those crimes not reported to police for reasons of inconvenience or fear of involving authorities, these statistics include events that respondents may not consider to be crimes and would probably not have reported, even if the other barriers did not exist. From this perspective, responses to the NCVS questions "are not crime-specific; rather, they are behavior specific."17 Some may argue that this effort to be more inclusive or comprehensive in coverage may lead to an overestimation of crime because it counts events that are not truly criminal behavior. For domestic violence in particular, this can be a difficult issue since the public's and the criminal justice system's concept of this crime continues to evolve. Consequently, keeping statistical measurement methods consistent and relevant under such circumstances can be challenging. By their very nature, victimization survey-based crime estimates will be higher than law enforcement-generated crime statistics. Concerns have been raised as to how many of the discrepancies between the two sets of figures are attributable to measurement methods as opposed to the actual incidence of crimes. The Bureau of Justice Statistics has conducted and sponsored many studies to assess the possible impact of various survey methodologies and procedures on reporting results.18 Although some measurement concerns can be addressed by rephrasing questions or making changes in the administration of the survey instrument, other errors may be more difficult to control for. This fact was acknowledged in comments by BJS staff in describing the NCVS's recent revisions: This redesign project is only one step in the evolution of our thinking about issues of rape and family violence. Research, however, including the NCVS, cannot be divorced from the social context in which it is conducted. Existing attitudes and stereotypes regarding these victimizations are linked inextricably to our ability, as researchers, to quantify the problem. Unfortunately, research indicates the persistence of impediments that prevent many women from reporting these victimizations.19 Other criminal justice domains In addition to law enforcement and victimization data, information from prosecutors, courts, and corrections can show charging practices, conviction rates, and sentencing patterns for these offenses. Further analysis of these data sources would go somewhat beyond the original scope of this project, which was to study how data on the incidence of domestic and sexual violence offenses could be collected. Obtaining information from these additional sources would expand knowledge about case dispositions, possible treatment outcomes, and recidivism among types of offenders. Many of the data collection problems cited for police information also apply to these segments of the criminal justice system, i.e., lack of uniform data definitions and standards, incompatible computer hardware and software, and other organizational barriers. Another problem is that many of the records systems on which any statistical information is based were developed for internal processing of individuals or cases. Consequently, they may lack summary reporting capabilities. Finally, data quality also may be a concern, particularly for data items that are not required for the originating agency's information needs and, therefore, may not be thoroughly checked or verified before submission. Multiservice domain data links Although such an information system would present valuable data on victims and offenders, it would need to overcome substantial technical and organizational obstacles and probably would require greater funding resources than are currently available. Concerns such as client confidentiality also would prevent tracking individuals through multiple service systems. Special studies With the diversity of victim-offender subpopulations for both domestic and sexual violence offenses, no general data reporting system will be able to capture sufficiently detailed information for all possible analyses. Therefore, special studies that focus on specific victim- offender groups or behavior patterns may be a more reasonable solution for obtaining answers to policy questions that general statistical data sets may not be able to address. If properly planned, such special studies can supplement what is known from general crime statistical data and could be used to estimate specific characteristics that may be lost in larger data categories or groupings. Concerns About Reporting Alternatives Unfortunately, no single data reporting system will be able to find all instances of domestic or sexual violence. Each has weaknesses in definitional detail or implementation that can lead to uncertainties about possible under- or overreporting in the resulting statistics. The following sections explore some of these concerns. Incomplete coverage of all cases No data source will be able to capture all possible instances of domestic and sexual violence. Victims' refusal to report crimes to or cooperate with public officials will contribute to underreporting problems. Any failures by police, prosecutors, and service providers to recognize or accurately record domestic and sexual violence offenses also will result in underestimates of prevalence in agency- generated statistics. Surveys and victims' self- reports may suffer from errors stemming from how the survey questions are phrased or administered or from the respondents' unwillingness or inability to answer accurately. Prevalence estimates, which may be based on samples or indirectly derived from crime statistics, may be affected by these and other measurement errors. Unfortunately, because of the realities imposed by data collection resources and methods and the nature of the crimes being studied, it is not always possible to eliminate or control for all of these concerns. Consequently, any analyses of these statistics must simply acknowledge these limitations and, when available, use alternative measures that may provide additional support for these indicators. Possible duplication across data systems Individual victims and offenders may come into contact with several different criminal justice, health, and social service agencies. Within an agency, the same person also may be involved in multiple incidents or service events. Without a universal personal identifier for victims and offenders, it is extremely difficult and expensive to track individuals across records that are usually organized around cases or incidents. The task is even harder when multiple agencies are involved, since records systems may not be compatible. Given that domestic abuse and sexual violence are behaviorally motivated crimes, there is a tendency for these offenders to recidivate. This may result in the duplication of individuals in statistical figures based on agency records. Depending on the nature of the analysis, this may or may not be important. For example, research related to mandatory arrest policies require specific followup information on known abusers to identify intervention points and subsequent behaviors. Studies looking at other aspects of victim-offender characteristics may not be so dependent on the elimination of duplicate records. Incomplete reporting Failures to provide complete and accurate data plague most information-gathering efforts. Criminal offense reporting may be affected by factors such as victim noncooperation and agency failures to collect data or properly implement procedures. The level of difficulty these problems create for statewide data collection was indicated in the State survey. Because domestic and sexual violence victims can face possible reprisals by the offenders, a heavy burden of embarrassment, and other repercussions, obtaining their cooperation can be extremely difficult for law enforcement and other agencies. The act of reporting domestic violence and some sexual violence incidents may be considered by victims as a last resort or as a way to make an irreparable break in a relationship. Consequently, the victims may perceive reporting as an admission of personal failure that they cannot face or believe is avoidable. Additionally, the problem of adequate training of personnel in handling these cases often was cited in the project panel discussions and survey responses. As many jurisdictions are recognizing the seriousness of these offenses, new laws and policies are being adopted, which in turn may require time to train all relevant staff in new procedures and to fully implement them. For all the above reasons, underreporting of domestic and sexual violence can be more of a problem than for other types of offenses. Although efforts can be made to overcome some of the factors hindering accurate reporting, some barriers may never be completely surmounted given the nature of these crimes and the social and behavioral issues involved. Definitional issues With any statistical data set, it is important to be aware of what definitions are used to determine what cases or persons are being included or excluded. Both domestic and sexual violence can be viewed from different perspectives, such as the frequency of abuse, level of abuse, categories of victims, or types of offenders. Obviously, what is counted will affect the resulting statistical figures. This situation should be recognized in any discussion of the prevalence of these social problems. The panel members and the State survey responders indicated many definitions for both domestic and sexual violence. This is due in part to statutory differences across the States, but it also reflects an evolving understanding of domestic and sexual violence that is occurring in society in general, which is reflected in the policies and practices of various agencies around the country. Because domestic and sexual violence are being treated as much more serious criminal offenses than in the past, these differences can result from some jurisdictions adopting more aggressive or comprehensive approaches than others. Unfortunately, this situation makes collecting uniform and consistent data across jurisdictions much more difficult. During the second panel meeting, the representative from the Washington Metropolitan Police Department announced that a new sexual assault law had been recently passed by his government. This announcement highlighted the challenge of trying to implement statistical measures for events about which relevant criminal laws may be rapidly changing. Because identifying domestic violence crimes may involve consideration of a criminal act, the relationship between the victim and offender, and the offender's motive for committing the act, properly classifying cases can be more difficult than for other types of offenses. For example, a crime incident that would normally be considered a property crime (e.g., a burglary, motor vehicle theft, larceny, or vandalism) could be classified as a domestic violence incident if the perpetrator's intent is to harass or intimidate the victim. This may require the investigating officers to go beyond the facts initially presented for a complaint to its possible underlying circumstances. An additional complication for data reporting is that some States have not mandated a specific domestic violence offense with which to charge an offender. In these instances, the offender is charged with another offense, but his case may be flagged as a domestic violence case for reporting purposes. Other States have broader family violence statutes that include domestic violence. Summary In summary, different definitions, legal procedures, and records management systems that are used to identify domestic violence and sexual assault complicate data collection efforts. Jurisdictional and agency differences can contribute to problems that prevent the compilation of reliable and uniform national estimates for these offenses. Definitional differences between service domains such as law enforcement agencies and victim services providers, along with incomplete sharing of information, speak to the importance of using multiple sources of data to develop a comprehensive understanding of the problem. ---------------------------------------- Notes 1. Blumstein, Alfred, and Joan Petersilia, "NIJ and Its Research Program," 25 Years of Criminal Justice Research: The National Institute of Justice, Washington, D.C.: U.S. Department of Justice, National Institute of Justice, December 1994. Limited copies of this document are available from the National Criminal Justice Reference Service, 800-851-3420. Ask for NCJ 151287. Or you can view the document online at http://www.ncjrs.org. 2. Federal Bureau of Investigation, Uniform Crime Reporting: National Incident-Based Reporting System--Volume I: Data Collection Guidelines, Washington, D.C.: U.S. Department of Justice, Federal Bureau of Investigation, July 1, 1988:5-16. 3. Wiersema, B., "The Impact of Series Incidents on Estimates of Violent Victimization," paper presented at the annual meeting of the American Society of Criminology, Phoenix, 1993. 4. See appendix D, question 38. 5. See appendix D, questions 19, 20, 21, and 22. 6. See appendix D, questions 21d and 22d. 7. Federal Bureau of Investigation, Uniform Crime Reporting Handbook, Washington, D.C.: U.S. Department of Justice, Federal Bureau of Investigation, 1984:5-32. 8. See appendix D, questions 4 and 5. 9. Saltzman, Linda E., James A. Mercy et al., "Identification of Nonfatal Family and Intimate Assault Incidents in Police Data," American Journal of Public Health, 62, No. 7 (July 1992):1018-1020. 10. Federal Bureau of Investigation, Uniform Crime Reporting Handbook--NIBRS Edition, Washington, D.C.: U.S. Department of Justice, Federal Bureau of Investigation, 1992. 11. See appendix D, questions 37 and 38. 12. See appendix C. 13. See appendix C, question 1. 14. See appendix D, questions 7, 8, and 9. 15. Bachman, Ronet, and Bruce M. Taylor, "The Measurement of Family Violence and Rape by the Redesigned National Crime Victimization Survey," Justice Quarterly, 11, No. 3 (September 1994). 16. Bachman, Ronet, and Linda E. Saltzman, Violence Against Women Estimates from the Redesigned Survey, Special Report, Washington, D.C.: U.S. Department of Justice, Bureau of Justice Statistics (NCJ 154348), August 1995. 17. Bachman, Ronet, and Bruce M. Taylor, "The Measurement of Family Violence and Rape by the Redesigned National Crime Victimization Survey," Justice Quarterly, 11, No. 3 (September 1994):509-510. 18. Bureau of Justice Statistics, The National Crime Survey: Working Papers, Volume I: Current and Historical Perspectives and Volume II: Methodological Studies, Washington, D.C.: U.S. Department of Justice, Bureau of Justice Statistics, December 1981. 19. Bachman, Ronet, and Bruce M. Taylor, "The Measurement of Family Violence and Rape by the Redesigned National Crime Victimization Survey," Justice Quarterly, 11, No. 3 (September 1994):511. --------------------------------------------------- Appendix A: List of Project Participants Panel to Assess State and Federal Data on Domestic Violence and Sexual Assault Yoshio Akiyama, Ph.D. Chief Statistician Program Support Section Criminal Justice Information Services Division Federal Bureau of Investigation 10th and Pennsylvania, N.W. Washington, DC 20535 Fax: 202-324-2258 Bernard V. Auchter Program Manager National Institute of Justice U.S. Department of Justice 633 Indiana Avenue, N.W., Room 842 Washington, DC 20531 Fax: 202-307-9907 No‰l Brennan Deputy Assistant Attorney General Office of Justice Programs U.S. Department of Justice 633 Indiana Avenue, N.W., Room 1300 Washington, DC 20531 Fax: 202-514-7805 Debra Bright Project Coordinator Mayor's Coordinating Council on Criminal Justice 10 South Street, Suite 400 Baltimore, MD 21202 Fax: 410-625-2107 Timothy Bynum, Ph.D. Michigan Justice Statistics Center School of Criminal Justice Michigan State University 560 Baker Hall East Lansing, MI 48824 Fax: 517-432-1787 Bonnie J. Campbell Director Violence Against Women U.S. Department of Justice 10th and Constitution, N.W., Room 5302 Washington, DC 20531 Fax: 202-307-3911 Nancy Crowell, Research Associate Commission on Behavior and Social Science and Education National Academy of Sciences 2101 Constitution Avenue, N.W., Suite HA156H Washington, DC 20418 Fax: 202-334-3829 Kim English Director Office of Research & Statistics Colorado Division of Criminal Justice 700 Kipling Street, Suite 3000 Denver, CO 80403 Fax: 303-239-4491 Mai Fernandez Special Assistant to the Deputy Assistant Attorney General Office of Justice Programs U.S. Department of Justice 633 Indiana Avenue, N.W., Room 1300 Washington, DC 20531 Fax: 202-514-7805 John Firman Director of Research International Association of Chiefs of Police 515 North Washington Street, 4th Floor Alexandria, VA 22314 Fax: 703-836-4543 Therese Ford Director Alabama Criminal Justice Information Center 770 Washington Avenue, Suite 350 Montgomery, AL 36130 Fax: 334-242-0577 Gilford Gee Survey Statistician Programs Support Section Criminal Justice Information Services Division Federal Bureau of Investigation 10th and Pennsylvania, N.W. Washington, DC 20535 Fax: 202-324-2258 Sgt. Scott Gibson Patrol Division Alexandria Police Department 2003 Mill Road Alexandria, VA 22314 Fax: 703-838-6309 Adele Harrell, Ph.D. Director Program on Law and Behavior The Urban Institute 2100 M Street, N.W., Suite 500 Washington, DC 20037 Fax: 202-659-8985 Barbara Hart Legal Director, Pennsylvania Coalition Against Domestic Violence, and Associate Director, Battered Women's Justice Project 6400 Flank Drive, Suite 1300 Harrisburg, PA 17112 Fax: 717-545-9456 Gwen Holden Executive Vice President National Criminal Justice Association 444 North Capitol Street, N.W., Suite 618 Washington, DC 20001 Fax: 202-508-3859 William Holmes, Ph.D. Director Statistical Analysis Center Executive Office of Public Safety Division of Programs 100 Cambridge Street, Room 2100 Boston, MA 02202 Fax: 617-727-5356 Jann Jackson Associate Director House of Ruth 2201 Argonne Drive Baltimore, MD 21218 Fax: 410-889-9347 Neal Kauder Research Associate National Center for State Courts 300 Newport Avenue Williamsburg, VA 23187-8798 Fax: 804-220-0449 Susan Keilitz Senior Research Associate National Center for State Courts 300 Newport Avenue Williamsburg, VA 23187-8798 Fax: 804-220-0449 Jackie McCann Program Specialist Office for Victims of Crime U.S. Department of Justice 633 Indiana Avenue, N.W. Washington, DC 20531 Fax: 202-514-6383 Patti Dobbs Medaris Branch Chief Violence Against Women Bureau of Justice Assistance U.S. Department of Justice 633 Indiana Avenue, N.W. Washington, DC 20531 Fax: 202-307-2019 Ann Menard Director National Resource Center on Domestic Violence 6400 Flank Drive, Suite 1300 Harrisburg, PA 17112 Fax: 717-545-9456 Richard Moore Administrator Statistical Analysis Center Criminal & Juvenile Justice Planning Lucas State Office Building Des Moines, IA 50319 Fax: 515-242-6119 Lisa Doyle Moran Associate Director for Legal Affairs National Criminal Justice Association 444 North Capitol Street, N.W., Suite 618 Washington, DC 20001 Fax: 202-508-3859 Mary Morgan Deputy Assistant Attorney General Office of Policy Development U.S. Department of Justice 10th and Constitution, N.W., Room 4242 Washington, DC 20530 Fax: 202-514-5715 Ashley Oliver Presidential Management Intern Office for Victims of Crime U.S. Department of Justice 633 Indiana Avenue, N.W. Washington, DC 20531 Fax: 202-514-6383 Carol Petrie Director Planning and Management Division National Institute of Justice U.S. Department of Justice 633 Indiana Avenue, N.W., Room 854 Washington, DC 20531 Fax: 202-514-6212 Cathy Poston Senior Counsel Office of Policy Development U.S. Department of Justice 10th and Constitution, N.W., Room 4242 Washington, DC 20530 Fax: 202-514-5715 Dolly Reed Director Office of Policy and Management 80 Washington Street Hartford, CT 06106 Fax: 203-566-1589 Benjamin H. Renshaw, III Deputy Director Bureau of Justice Statistics U.S. Department of Justice 633 Indiana Avenue, N.W., Room 1144 Washington, DC 20531 Fax: 202-307-5846 Stephen Rickman Director Crime Bill Support Division Bureau of Justice Assistance U.S. Department of Justice 633 Indiana Avenue, N.W., 4th Floor Washington, DC 20531 Fax: 202-307-2019 Emanuel Ross, Jr., Ph.D. Operations Research Analyst Metropolitan Police Department 300 Indiana Avenue, N.W., Room 3142 Washington, DC 20001 Fax: 202-727-0826 Elisabeth Sachs Senior Counsel Office of Policy Development U.S. Department of Justice 10th and Constitution, N.W., Room 4242 Washington, DC 20530 Fax: 202-514-5715 Linda Saltzman, Ph.D. Senior Scientist Family and Intimate Violence Prevention Team Centers for Disease Control and Prevention Mail Stop K60, 4770 Buford Highway Atlanta, GA 30341-3724 Fax: 404-488-4349 Kathy Schwartz Administrator Violence Against Women Program U.S. Department of Justice 633 Indiana Avenue, N.W., 4th Floor Washington, DC 20531 Fax: 202-307-2019 Edwin Zedlewski, Ph.D. Director Office of Criminal Justice Research National Institute of Justice 633 Indiana Avenue, N.W. Washington, DC 20531 Fax: 202-307-6394 ------ JRSA Staff ------ Justice Research & Statistics Association 444 North Capitol Street, N.W., Suite 445 Washington, DC 20001 Fax: 202-624-5269 Joan C. Weiss Executive Director Jim Zepp Computer Center Director Karen Maline Assistant Director for Information Services --------------------------------------------------- Appendix B: State Survey Form (Not included in this ASCII text file. See Table of Contents for more information.) --------------------------------------------------- Appendix C: List of Respondents to JRSA Survey of States and Territories States and Territories Responding to JRSA Survey on Domestic and Sexual Violence Data Collection Alabama Montana Alaska Nebraska Arizona Nevada Arkansas New Hampshire California New Jersey Colorado New York Connecticut North Carolina Delaware North Dakota District of Columbia Florida Northern Mariana Islands Georgia Ohio Hawaii Oklahoma Idaho Pennsylvania Illinois Puerto Rico Indiana Rhode Island Iowa South Dakota Kansas Tennessee Louisiana Texas Maine Utah Maryland Virgin Islands Massachusetts Washington Michigan West Virginia Minnesota Wisconsin Missouri Wyoming --------------------------------------------------- Appendix D: State Survey Results (Appendix D is not included in this ASCII text file. See Table of Contents for more information.) ------------------------------------------------- Appendix E: FBI NIBRS Status Report as of 06/02/95* (Appendix E is not included in this ASCII text file.) * For updates, contact the Bureau of Justice Statistics, U.S. Department of Justice, Washington, DC 20531, or access the Web site at http://www.nibrs.search.org. ------------------------------------------------- Appendix F: A Note on Terminology for Crime Statistics A Note on Terminology for Crime Statistics In any discussion of domestic and sexual violence crime statistics, it is important to be precise in the terminology used to describe the crime data being collected and analyzed. Otherwise, what appear to be conflicts between various prevalence or severity indicators may be actually due to the use of different units of measure. A distinction should be made between the terms "incident" and "offense." According to the FBI's National Incident-Based Reporting System (NIBRS) data standards, An "incident" is defined for NIBRS reporting purposes as one or more offenses committed by the same offender, or group of offenders acting in concert, at the same time and place.1 From this definition, certain relationships are evident among the terms used, e.g., an incident is a single event during which multiple offenses or crimes may be committed and with which several victims and offenders may be associated. Consequently, a jurisdiction's statistics for offenses, victims, and offenders may be greater than its number of reported incidents. The number of offenses, victims, and offenders also may be different from one another since each is an independent phenomenon. In other words, a single offender may have multiple victims or may commit multiple offenses against a single victim, or multiple offenders may attack a single victim. A further refinement is that most crime incident data sets rarely have the ability to distinguish between first-time offenders and recidivists. This would require examining criminal history records to determine whether someone is a repeat offender when reporting every crime incident. Since the offender is often unknown when a crime is first reported, this information would have to be retroactively added when a case is cleared. Additionally, since the crime is reported at arrest, subsequent prosecutorial or judicial actions may invalidate the assignment of a crime to an individual. Tracking these decisions and updating personal records and higher level statistical data sets require more of a resource commitment than most jurisdictions are willing to make. As a result, the number of offenders reported in incident-based data sets may be greater than the actual criminal population. This would be particularly true for offenses such as domestic and sexual violence that are behaviorally motivated crimes and where offenders may have a greater tendency to recidivate. All of this means that the use of specific statistical figures in discussing prevalence or severity of domestic and sexual violence problems should be done with a full awareness of what they represent. In addition to concerns about under- or overreporting problems, there may be legitimate reasons for discrepancies across various data sets because of differences in definitions and procedures for data collection employed in each. -------------------- Note 1. Federal Bureau of Investigation, Uniform Crime Reporting: National Incident-Based Reporting System--Volume I: Data Collection Guidelines, Washington, D.C.: U.S. Department of Justice, Federal Bureau of Investigation, July 1, 1988:17. ----------------------------------------------- For more information on the National Institute of Justice, please contact: National Criminal Justice Reference Service Box 6000 Rockville, MD 20849-6000 1-800-851-3420 e-mail: askncjrs@ncjrs.org The National Institute of Justice is a component of the Office of Justice Programs, which also includes the Bureau of Justice Assistance, Bureau of Justice Statistics, Office of Juvenile Justice and Delinquency Prevention, and the Office for Victims of Crime.