Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 165
Ensuring the Quality, Credibility, and Relevance of U.S. Justice Statistics –4– State and Local Partnerships THE UNITED STATES has a significant national justice system, composed of the federal court and penal systems and numerous federal law enforcement agencies. However, the vast majority of the activity related to crime and justice occurs at the subnational level. Most crime is pinpointed geographically, and much of the response to crime is handled by police, courts, and correctional facilities at the state, county, and municipal levels. The Bureau of Justice Statistics (BJS) thus shares with many of its fellow federal statistical agencies the challenge that it is a national government agency tasked to measure phenomena that are inherently local in nature. To meet this challenge, it has been common for federal statistical agencies to forge partnerships with state and local governments. These partnerships vary in their level of formality, in their goals and objectives, and in the fiscal resources dedicated to them on both the federal and state sides. In some cases, they are as basic as establishing regional offices to make interactions with local authorities more convenient and to serve as a venue for dissemination of information; in others, the federal agency and individual state governments are essentially equally committed to the partnership, jointly funding and staffing data collection operations. Federal-state partnerships also include models where the federal agency role is principally one of coordination and compilation, directly accumulating data provided by local authorities and piecing together national files. Under its current legal authority, BJS has at least two distinguishing characteristics in terms of its work with state and local governments, relative to its peers in the federal statistical system. One is the boldness with which state and local issues are written into BJS’s legal mandate: in no uncertain terms,
OCR for page 166
Ensuring the Quality, Credibility, and Relevance of U.S. Justice Statistics BJS’s authorizing language mandates that “[BJS] shall give primary emphasis to the problems of State and local justice systems” (42 USC § 3731), and its list of legal duties is replete with reference to performing studies “at the Federal, State, and local levels” (see Box 1-2). The second is BJS’s explicit charter—inherited from the functions of the former Law Enforcement Assistance Administration and consistent with the function of BJS’s parent Office of Justice Programs (OJP)—to provide direct financial and technical assistance to local governments and agencies, rather than solely conduct data collection functions. It follows that an assessment of BJS’s programs and functions must pay particular attention to the agency’s interactions with state and local governments, evaluating the effectiveness of these partnerships and contemplating the role of BJS’s grant programs for local authorities. In this chapter, we discuss the centerpiece of BJS’s State Justice Statistics (SJS) program—BJS’s network of state Statistical Analysis Centers (SACs; Section 4–A)—and directly compare BJS’s work with the states to the models of federal-state cooperation in other parts of the federal statistical system. We then turn to BJS’s principal grant program, the National Criminal History Improvement Program (NCHIP; Section 4–B). This chapter on state and local partnerships is also the most logical place to explore in depth federal and state roles in the compilation of one of the longest-standing statistical series in the criminal justice system—albeit not one administered by BJS. For decades, state and local police departments have supplied crime count data to the U.S. Federal Bureau of Investigation (FBI) as part of the Uniform Crime Reporting (UCR) program. As part of the panel’s charge to examine BJS’s relationship to other data-gathering entities in the U.S. Department of Justice, we discuss the current and future state of the UCR and BJS’s role relative to that of the FBI in this series; this discussion is in Section 4–C. 4–A STATE JUSTICE STATISTICS: THE STATISTICAL ANALYSIS CENTER NETWORK BJS’s network of state-based SACs actually predates the creation of BJS in its current form. The same section of the 1968 Omnibus Crime Control and Safe Streets Act that authorized the new Law Enforcement Assistance Administration (LEAA) to collect statistical information also directed the LEAA to (P.L. 90-351 § 515(c)): cooperate with and render technical assistance to States, units of general local government, combinations of such States or units, or other public and private agencies, organizations, or institutions in matters relating to law enforcement.
OCR for page 167
Ensuring the Quality, Credibility, and Relevance of U.S. Justice Statistics The LEAA’s National Criminal Justice Information and Statistics Service started the Comprehensive Data Systems (CDS) program in 1972, providing the earliest funds for establishment of SACs. As summarized by the Justice Research and Statistics Association (2008): The CDS guidelines established six objectives for the SACs: provide objective analysis of criminal justice data, including data collected by operating agencies; generate statistical reports on crime and the processing of criminal offenders in support of planning agencies; coordinate technical assistance in support of the CDS program in the state; collect, analyze, and disseminate management and administrative statistics on the criminal justice resources expended in the state; promote the orderly development of criminal justice information and statistical systems in the state; and provide uniform data on criminal justice processes for the preparation of national statistical reports. Ten states established SACs or designated existing agencies in SACs in 1972; by 1976, the SAC network had grown to 34 states and a nonprofit association—the Criminal Justice Statistics Association—developed to support and coordinate SAC activities. In 1991, the association renamed itself the Justice Research and Statistics Association (JRSA). BJS assumed responsibility for the SAC program in 1979 as LEAA was phased out (to be replaced by OJP) and it has retained this role since, as reflected in the authorizing legislation. The initial focus of SACs was to coordinate state-level data collection, act as a statistical clearinghouse, and assist the federal government with justice statistics series through contributions of state data or statistics. SAC involvement in statistical collections has historically focused and remains on state-level needs, although either direct participation in BJS programs (e.g., contributing data) or indirect participation (facilitating participation by other state agencies) has remained central to the program. As of 2008, all states and several U.S. territories had a designated SAC, though their forms and functions vary; Box 4-1 describes the types of structures that exist in the current SAC network. The funding program for SACs was reformulated from a clearinghouse orientation to a research-oriented program in 1996 under the leadership of then-BJS Director Jan Chaiken. The SJS program has since provided BJS an avenue for fostering data collection and analysis in areas consistent with BJS and Justice Department priorities. This approach has been particularly useful in collecting data on emerging local or regional issues, which often are of greater concern in specific parts of the country and states before becoming
OCR for page 168
Ensuring the Quality, Credibility, and Relevance of U.S. Justice Statistics Box 4-1 State Statistical Analysis Center Network The Justice Research and Statistics Association (JRSA) defines state Statistical Analysis Centers (SACs) as state-level units or agencies “that use operational, management, and research information from all components of the criminal justice system to conduct objective analyses of statewide and systemwide policy issues” (http://www.jrsa.org/sac/aboutsacs.html). Individual SACs receive financial support through the Bureau of Justice Statistics’ (BJS’s) State Justice Statistics Program, as does JRSA; some SACs receive additional funding from their state governments. JRSA is a nonprofit organization of SAC directors that provides a central staff and coordination effort for SAC activities; it hosts an annual research conference (with BJS funding), publishes the journal Justice Research and Policy, and maintains databases of SAC activities. SACs vary in their organizational structure and standing with respect to the state government, falling into a few basic categories: Independent State Justice Information Center: For instance, the Illinois SAC (created in 1977) is the research and analysis unit of the Illinois Criminal Justice Information Authority (ICJIA). The ICJIA was created as an independent state agency by executive order in 1982, inheriting functions from a former Illinois Law Enforcement Commission. Other states following similar models as of 2008 are Alabama and Arkansas. Unit of State Crime Commission or Planning Commission: For example, the Kansas SAC is a unit of the state sentencing commission and Montana’s is part of the state Board on Crime Control; both Georgia and the District of Columbia have SACs affiliated with the Criminal Justice Coordinating Council in those jurisdictions. Other states following this model include Arizona, Indiana, Louisiana, Maryland, Nebraska, Oregon, Pennsylvania, Rhode Island, and Utah. Unit of State Law Enforcement Agency or Justice Department: For this most common organizational structure, examples include California (branch of the state justice department), Missouri (part of the state highway patrol), and Tennessee (unit of the Tennessee Bureau of Investigation). Echoing BJS’s administrative position in the U.S. Department of Justice, both the Minnesota and South Carolina SACs are part of an Office of Justice Programs in those states’ public safety departments; similarly, Idaho’s SAC is housed in the Office of Planning, Grants, and Research of the Idaho State Police. Other states following this basic model include Colorado, Florida, Hawaii, Kentucky, Massachusetts, New Hampshire, New Jersey, New York, North Carolina, North Dakota, Ohio, South Dakota, Virginia, and West Virginia. Unit of Other State Agency or Entity: Examples include the SACs in Delaware (part of state Office of Management and Budget), Iowa (Department of Human Rights), Oklahoma (Legislative Service Bureau), Washington (Office of Financial Management), and Wisconsin (Office of Justice Assistance). The Texas SAC was reestablished, after some absence, by executive order in 2007, and is housed in the Office of the Governor.
OCR for page 169
Ensuring the Quality, Credibility, and Relevance of U.S. Justice Statistics Affiliation with Academic Department or Institute: In some states, the SAC is directly affiliated with a university or university-affiliated research institute, and the SAC director or staff may be faculty members. The states organized in this manner (with their host institutions) are Alaska (University of Alaska Anchorage), Connecticut (Central Connecticut State University), Maine (University of Southern Maine, in partnership with state Department of Corrections), Michigan (Michigan State University), Mississippi (University of Southern Mississippi), Nevada (University of Nevada, Las Vegas), New Mexico (University of New Mexico), Vermont (the nonprofit Norwich Studies and Analysis Institute, affiliated with Norwich University), and Wyoming (University of Wyoming). SACs also operate in the Northern Mariana Islands and Puerto Rico. a national issue. States continue to assist BJS as a liaison for data collection efforts, enhancing state and local analytical efforts and analyses which demonstrate the utility of various systems (e.g., National Incident-Based Reporting System, criminal history records etc.). Data and policy priorities of BJS are reflected in the substantive areas under which SACs may apply for SJS program support. The “themes” in the annual SJS solicitation include issues related to BJS initiatives in an array that also allows states to focus on problems of more immediate state, local or regional concern. Themes from the 2008 SJS program for SAC solicitation are enumerated in Box 4-2. 4–A.1 State Partnerships in the Federal Statistical System In examining BJS’s partnerships, it is useful to consider some exemplars from other statistical agencies; we describe some of these arrangements in the following list. In doing so, we emphasize that this is a selective list meant to describe a range of approaches, rather than a complete canvass of federal-state partnerships, and that no assessment of the quality of the data produced by the systems is implied. Unless otherwise indicated, cost information in the following list is from the fiscal year 2008 edition of the U.S. Office of Management and Budget (OMB) publication Statistical Programs of the United States Government (U.S. Office of Management and Budget, 2007): The Behavioral Risk Factor Surveillance System (BRFSS) is a collaborative data collection system maintained by the Behavioral Surveillance Branch of the Centers for Disease Control and Prevention (CDC). We described the BRFSS in our first report (National Research Council, 2008b:Box 4-1 and Section 4–B) as the basis for one possible design alternative for the National Crime Victimization Survey (NCVS). As a federal-state partnership, the BRFSS follows a contracting model:
OCR for page 170
Ensuring the Quality, Credibility, and Relevance of U.S. Justice Statistics Box 4-2 State Justice Statistics Program Themes, 2008 Deaths in Police Custody Reporting—Obtaining statewide data on deaths occurring in the process of arrest or in pursuit of arrest. [The Bureau of Justice Statistics (BJS)] continues to request assistance from State [Statistical Analysis Centers (SACs)] to obtain specified data on these deaths and report them quarterly to BJS. [Applicants] wishing to address this theme may utilize SJS to establish a long term reporting process, rather than a one time study. Prison rape and victimization in confinement facilities—improving quality of administrative data involving criminal acts within adult and juvenile facilities. [BJS] continues to encourage SACs to examine the quality of their State administrative records and where feasible provide recommendations for the improvement of the quality and accuracy of these data. Criminal justice system crisis planning. The SAC may wish to pursue research or data collection to support criminal justice system planning for dealing with major crises, disorders, or other catastrophic incidents. [Examples include] prisoner relocation and/or alternative housing needs [and] backup records systems in the courts or other entities. Increased Web access to data. SJS funds could be used by the SAC for Internet infrastructure development, enhancements, and linkages, including building a World Wide Web site, computer support, and preparing reports for dissemination via the Internet. Performance measurement. SJS funds could be used by the SAC to help States develop and improve performance measures and the tools available to agencies to assess progress in addressing public safety and administration of justice goals. Analyses utilizing a State’s criminal history records. BJS encourages SACs to utilize the State’s criminal history records for research purposes. In particular, the SAC may wish to seek SJS funds to support studies of: Patterns of criminal behavior such as sex offending, stalking, or domestic violence; Arrests, prosecutions and convictions for firearms-related offenses; Prisoner and/or probationer recidivism, including rates of rearrest, reconviction, and return to correctional custody; The implementation and/or impact of programs such as drug courts, prisoner reentry initiatives, or specialized probation programs; or The implementation and/or impact of a State’s criminal history record improvement activities. At most one topic may be proposed in this thematic area. Funds may be requested to establish the technical capacity to conduct criminal history records-based research. The application must either state that the applicant is also the State’s administrator of [National Criminal History Improvement Program (NCHIP)] funds or include a letter or memorandum of endorsement from the State agency administering NCHIP funds. Statewide crime victimization surveys. Analysis of the uses of new or emerging biometric technologies to improve the administration of criminal justice. SJS funds may be used by the SAC to support research which describes and examines the uses of new or emerging biometric technologies (DNA evidence collection/analysis, facial recognition, etc.) to improve the administration of criminal justice in a State. Research using incident-based crime data that are compatible with the National Incident-Based Reporting System (NIBRS). [SJS] funds under this theme may be used to examine the utility of linking NIBRS incident reports to a State’s criminal history records for research purposes.
OCR for page 171
Ensuring the Quality, Credibility, and Relevance of U.S. Justice Statistics Data collection and/or research examining a special topical area: Minority overrepresentation in the criminal or juvenile justice systems.… Civil justice. SJS funds may be used by the SAC in developing estimates of the number and characteristics of tort, contract, and real property cases and the dispositions of those cases for both adjudicated and settled civil matters. The longer term objective might be to estimate change over time within the State in the nature of case issues, judgments, and awards and to evaluate the impact of civil justice reforms such as capping punitive awards or medical malpractice mediation boards. Cybercrime. SJS funds could be used by the SAC to examine the magnitude and consequences of computer crime and identity theft and fraud. Human trafficking. Justice issues in Indian Country. Criminal activity in U.S. border areas. Violent crime in schools. The impact of substance abuse on State and/or local criminal justice and public health systems. Family violence and/or stalking. Evaluation of prisoner reentry initiatives and programs. Other theme or topic identified by the SAC. SJS funds may be used by the SAC to support research examining another theme or topic provided the application is accompanied by persuasive documentation and justification that the subject is a top priority for the State’s Governor or criminal justice policy officials. SOURCE: Excerpts from Bureau of Justice Statistics (2008f); emphases in the original. the CDC executes contracts with state health agencies, paying for them to conduct monthly telephone interviews and administer a core questionnaire. In return, the states get processed returns in terms of state-level estimates (by design, the BRFSS is an amalgam of state samples and is not meant to be a nationally-representative sample); they also have the latitude to add their own topic supplements. This form of partnership—in which the federal agency exercises strong control over content and resulting data but pays the states to provide data collection—is expensive. Of the $453.1 million estimated to be spent in fiscal year 2008 on purchasing statistical services from state and local governments, CDC’s $162.2 million (not including activities of the National Center for Health Statistics [NCHS], which is a component of CDC) is the largest single share. The Vital Statistics program of NCHS compiles information from birth and death certificates that have been collected by state health (vital registration) departments. NCHS’s fiscal year 2008 allocation for statis-
OCR for page 172
Ensuring the Quality, Credibility, and Relevance of U.S. Justice Statistics tical purchases from state governments is about $18.6 million—lower spending relative to the BRFSS because of the different relationship between the federal agency and the localities. The law directs that (42 USC § 242k(h)(1)): There shall be an annual collection of data from the records of births, deaths, marriages, and divorces in registration areas. The data shall be obtained only from and restricted to such records of the States and municipalities which [NCHS] determines possess records affording satisfactory data in necessary detail and form. … Each State or registration area shall be paid by [NCHS] the Federal share of its reasonable costs (as determined by [NCHS]) for collecting and transcribing (at the request of [NCHS] and by whatever method authorized by [NCHS]) its records for such data. Hence, the collection costs of vital statistics data are largely assumed at the state level (with some reimbursement of “the Federal share of the reasonable cost”). NCHS’s costs involve compilation and processing, as well as the promulgation of standards. This model has its difficulties, because disagreements over the costs of providing the data can put a damper on participation. In recent years, NCHS has struggled to achieve cooperation by state health departments (and the county health departments and local facilities coordinated by the states) in adopting new standard formats for birth and death certificates. Participation in supplying divorce records declined sufficiently that NCHS abandoned their collection in 1996, and the most recent comprehensive study by NCHS of marriage and divorce information from vital records is based on 1989–1990 data (see http://www.cdc.gov/nchs/mardiv.htm). The Bureau of Labor Statistics allocation for purchases of statistical services from states and localities was the second largest among statistical agencies in fiscal year 2008, estimated at $96 million. BLS operates federal-state cooperative arrangements through its regional offices; limited contracts for collection and sharing of employment data between BLS and individual states had been crafted as early as 1916, but the establishment of the regional offices in 1942 formalized those arrangements (Hines and Engen, 1992). Today, the BLS Federal-State Cooperative Programs (administered through six regional offices) encompass a number of labor market information programs, including surveys and records collections on occupational safety and health.1 1 In recent years, BLS has switched from operating eight regional offices to six: its Kansas City, Missouri, regional office was merged with the Dallas office and the operations of its Boston and New York offices were consolidated into a Boston (Northeast) office. However, New York City still retains a Regional Office for Economic Information and Analysis headed by a regional commissioner.
OCR for page 173
Ensuring the Quality, Credibility, and Relevance of U.S. Justice Statistics As an operational model, BLS’s Federal-State Cooperative Program is closer to the contract-driven BRFSS model than the vital statistics example, involving contractual agreements to collect and transfer specific data series. The Census Bureau employs numerous mechanisms to work with state and local governments, including the coordination of field activities through 12 regional offices. Two of its major partnerships are of particular interest because they share a similar structure with the BLS model. Between 1967 and 1973, the Census Bureau formalized loose arrangements with state agencies to create the Federal-State Cooperative Program for Population Estimates, under which states supply some of the raw information (vital statistics data on births and deaths, estimates of prison population, and other records) needed to update decennial census information and generate intercensal population estimates. By 1979, a parallel Federal-State Cooperative Programs for Population Projections was forged to build collaboration with the states on the production of population forecasts. As an agency, the National Agricultural Statistical Service (NASS) of the U.S. Department of Agriculture (USDA) dates back to the 1961 formation of a Statistical Reporting Service, but its roots—and partnership with states—are more extensive. Wisconsin was the first state to enter into a memorandum of understanding (MOU) for data gathering and sharing with USDA in 1917; over the years, similar MOUs were executed with state agriculture departments, land grant universities, and other agricultural entities, and all states currently have an MOU on file with NASS (Dantzler, 2008). The defining characteristic of NASS’s state partnerships is the high degree to which labor and other resources are shared, by means of a third party. In 1972, NASS established an agreement with the National Association of State Departments of Agriculture (NASDA) under which NASDA bears the principal costs of data collection, including salaries and travel expenses of about 3,700 field interviewers. Field work is coordinated through NASS’s 46 field offices,2 which are staffed by a mix of federal/NASS employees (675 total) and state “cooperator” employees (151 total). Because of the intermediary role of NASDA, OMB tabulations indicate that NASS purchases no statistical services directly from state and local governments, but rather from a private-sector entity ($29 million, out of NASS’s $167.7 million total estimated budget). BJS’s state partnerships do not correspond neatly with any of these organizational models. The NCHIP and related grant programs provide rela- 2 All states are covered by the NASS-state partnerships; however, the New England states are coordinated through a single regional office in Concord, New Hampshire.
OCR for page 174
Ensuring the Quality, Credibility, and Relevance of U.S. Justice Statistics tively unfettered funding to the states, not the more formal data collection contracts executed by the other agencies. In its SAC network, BJS’s system bears some similarities to the NASS arrangement, including the presence of a third-party coordinator (NASDA for NASS and JRSA for the SACs), but is neither as formal nor intensive as the state agricultural arrangements. In large part, the more-limited role of the SACs is dictated by basic logic and the breadth of BJS’s scope: in the criminal justice arena, there is no ideal, single, state-level point of contact with which a strong data collection arrangement can be brokered, because of the differences in state justice organizations. That is, state police or public safety departments may be distinct from state corrections departments, which in turn are distinct from state court systems, which are distinct from state victims’ offices and other related agencies, all of which are distinct from their local or large-city equivalents. As noted in our first report (National Research Council, 2008b:60), BJS has provided technical support to state and local agencies as well as brokering partnerships to disseminate and collect data. This is particularly true of its development of software tools for local data collection. BJS developed, and made freely available to state and local agencies, software for conducting victimization surveys, using the NCVS (including the detailed Incident Report) as a template. The software has since been relabeled Justice Survey Software and is administered by SureCode Technologies, and remains available at http://www.bjsjss.org; templates have been added for victimization surveys conducted in individual states, as well as for BJS’s National Survey of Prosecutors, Police-Public Contact Survey supplement, and State Court Processing Statistics inventory. More recently, this survey-building software has been ported to the Web as “BJScvs” and made available to state and local agencies through the website http://www.bjscvs.org (Justice Research and Statistics Association, 2006b:5). 4–A.2 Assessment Some of the basic benefits of a strong and active partnership between BJS and its state SACs can be listed in brief: SACs are familiar with state, local, and often regional justice issues and can provide context for federal initiatives. Federal-state cooperation promotes consistency in definitions and concepts across data collections, which in turn has the benefit of facilitating more effective comparisons between jurisdictions and agencies. With partnerships, BJS is also in a position to provide guidance to states and localities on common standards for data quality, measurement, and analysis. The response to crime is predominantly local and state in nature, hence
OCR for page 175
Ensuring the Quality, Credibility, and Relevance of U.S. Justice Statistics the need for familiarity with justice systems, data, and agencies at this level. The SACs are often able to facilitate access to key agencies, data systems, and collection mechanisms that benefit federal statistical system efforts. States benefit from a strong state-federal partnership in several key ways: BJS is able to provide technical assistance that would not otherwise be available to states, or that might be cost-prohibitive for any one jurisdiction to obtain. Although the financial benefit of the SAC program has not been large, states have been able to leverage BJS assistance (including NCHIP) for system, data quality, and analytical enhancements that might otherwise not be available from state resources alone. A major benefit for BJS and the nation is improvement in justice data systems and a national perspective on crime and justice. Policy development in the states often requires benchmarking and an understanding of crime on the national scene; BJS is uniquely situated to provide this perspective. BJS benefits from a strong relationship with the SACs through the inventiveness of research performed at the state level and through the states’ direct contact with issues of local interest; feedback from SAC partners and successes with state-level activities can inform the development of national-level data collections. The capacity of the state partnerships to assist BJS to more nimbly meet new data needs was clearly illustrated by the data collection efforts set up to comply with the Deaths in Custody Reporting Act of 2000 (P.L. 106-297). As we also discuss in Box 3-3 and Section 3–B.3, the act tasked BJS with quarterly data collection on incidents involving the death of persons while in criminal justice system physical custody. BJS was able to use its ongoing relationship with state Departments of Correction to collect data on deaths occurring while in correctional custody. However, a data collection system for deaths occurring in law enforcement or other justice system custody proved more problematic, given significant variation in the ability to identify and capture data on incidents even within states. SACs assisted in developing data collection systems initially and in some cases continue to assist BJS in preparing reports for the Deaths in Custody project mandated by Congress; it remains one of the suggested program themes in the 2008 program solicitation. In the panel’s assessment, the BJS-state SAC network stands as a relatively low-cost activity on BJS’s part with great dividends in terms of outreach and feedback, as well as dissemination of data and products to state policy makers. Consistent with other recommendations we make in Chapter 5 on
OCR for page 198
Ensuring the Quality, Credibility, and Relevance of U.S. Justice Statistics generating particularly interesting or useful findings, could spur greater interest and participation in the series. As Faggiani (2007:2) summarizes: The implementation of NIBRS by local law enforcement agencies is an evolving but slow moving process and this has had an impact on its use for research. The early NIBRS data releases (1996 through 1998) were mostly limited to small and medium-sized law enforcement agencies representing primarily rural states. For example, the 1996 data covered only nine cities with populations in excess of 100,000 and no cities with populations over 250,000. Researchers examining the utility of NIBRS for scientific research began to raise serious questions about the overall representativeness of this supposed national data system. For the purpose of studying the occurrence of crime in the United States, a healthy UCR program and, particularly, a full-fledged NIBRS are both critical data systems. A fully featured NIBRS with high participation has the potential to shed light on some dynamics of law enforcement operations that are not visible in current data (and not even envisioned within the tight management and administrative focus of the LEMAS series). One such example is the potential for NIBRS to provide detail on police clearance rates, which are currently reported as a gross indicator of departmental success. However, the aggregate rates, minus the type of contextual information on incidents and the extent of police contact with victims and offenders, mask a great deal of potentially useful information. On an explanatory basis, Addington (2007a) used the incident and police clearance date recorded in NIBRS data to test (and confirm) the conventional wisdom that those murder cases that are cleared by police tend to be cleared early and that there is a major drop in the clearance rate after more than a week has passed since the homicide. With fuller data resources, and study of different crime types, NIBRS could provide a useful platform to study the factors that influence the successful clearance of crimes by police. Finding 4.3: A full-fledged NIBRS would be a source of basic information on police responses to public complaints (911 calls), including whether or not a case is “cleared” by police through an arrest. Having concluded that the UCR and NIBRS programs have their merits—and that the nation benefits from having multiple data systems (UCR/NIBRS and the NCVS) to measure the incidence and circumstances of crime from different perspective—the question that remains is whether they should be managed by separate parts of the Justice Department. Put more bluntly, the question is whether it makes sense for BJS, the principal statistical agency in the Justice Department, to lack authority for what is arguably the most prominent statistical data series produced by the department, and whether it would be preferable for BJS to “take over” UCR operations from the FBI.
OCR for page 199
Ensuring the Quality, Credibility, and Relevance of U.S. Justice Statistics Rosenfeld (2007:830) argues that “the FBI is no longer the appropriate institutional home for the UCR program, if it ever was.” The principal argument for the transfer is that BJS is more likely to be able to provide the technical support and capability for ongoing improvement of the UCR than the FBI, given that “tracking conventional crime is not a high priority in the [FBI’s] post 9/11 focus.” The “appropriate focus and necessary human and technical resources” to best monitor locally recorded crimes reside in BJS rather than the FBI and its administrative placement is more historical artifact than organizational efficiency: “had the BJS existed 75 years ago, the responsibility to compile local crime statistics would have been placed there,” but by the time BJS was founded in 1979, the UCR was well entrenched as part of the FBI and little incentive existed to transfer authority for the UCR program (Rosenfeld, 2007:831). To be sure, Rosenfeld (2007:831) argues that transfer of the UCR program “is a necessary but not sufficient condition to upgrade the nation’s crime monitoring capabilities,” and that BJS would require additional resources to make substantial improvements in the timeliness and quality of UCR estimates. Rosenfeld (2007) further cites a recent example of both UCR and NCVS being bested by another data source—in terms of timeliness of information—as motivation for improving both benchmark measures of crime through an organizational realignment. By August 2006, “local police chiefs had been complaining for months … that violent crime was on the rise and that they lacked the resources to combat it.” However, such a shift in crime rate (reversing several years of declining trends) could not be measured by UCR: “the FBI report [of UCR results for 2006] was not released until September 2006 and covered only the period through the end of 2005.” The FBI released a preliminary report covering data from the first half of 2006 in December 2006—rapid dissemination by UCR standards, but in a sense the information was still too late. Any reading from the NCVS lagged behind the FBI’s figures, meaning that “no single source of [publicly available] systematic data” existed to refute or corroborate the chiefs’ claims. That August, the Police Executive Research Forum (PERF) convened a Violent Crime Forum of police chiefs, the result of which was a compilation of current crime data from several of the participating police departments. PERF and the chiefs used these data to describe apparent crime increases in its report A Gathering Storm: Violent Crime in America (Police Executive Research Forum, 2006)—a report that went public in October 2006. An update of that report, published in April 2007, pointedly reminded readers that PERF encourages police agencies “not to wait” for the FBI to release its UCR crime figures and to send their data directly to PERF for compilation and early release (Rosenfeld, 2007:826).
OCR for page 200
Ensuring the Quality, Credibility, and Relevance of U.S. Justice Statistics The basic arguments for BJS acquiring authority for the UCR and NIBRS programs include, in brief: The operational transfer would solidify BJS’s position as the preeminent statistical agency within the Department of Justice and the preeminent governmental source for justice statistics, generally. The transfer would permit the FBI to sharpen its new organizational focus on antiterrorism efforts. Placing authority for the UCR and NIBRS within a true statistical agency would facilitate attention to methodological problems in the series, including adjustments for nonresponse and imputation routines. The strongest argument against BJS acquiring control of the UCR is the potential disruption of the relationships that have built up over the decades of FBI administration of the UCR, brokered through outlets such as the IACP. A great strength of BJS’s correctional data series and a key to their quality is the network of ties that BJS has built with state departments of corrections and individual facilities; likewise, BJS continues to develop ties to state court systems. Just as it is reasonable to expect that the quality of resulting data would be impaired by a sudden change in reporting structures, shifting lines of data reporting that have existed, in some cases, since the 1930s is not something that should be taken lightly. There is, moreover, a trust that may be implicit in UCR reporting relationships—law enforcement agencies providing data to a fellow law enforcement agency—that is nontrivial. It is certainly possible that, with strong endorsement and assistance from collaboratives such as the IACP and PERF, a transition from FBI to BJS could be successful in time, but the short-term impact on response rates could be significant. In our assessment, the prospect of BJS “taking over” the UCR picks unnecessary turf fights with both the police community and the FBI, both of which have historically been protective of the program. Although the organizational transfer of UCR from one Justice Department agency to another would seem to be a fairly easy task, we think that this appearance is deceptive. The suggestion severely underestimates the level of energy and expense that would be necessary to get the UCR SRS (much less a full-fledged NIBRS) to function efficiently and effectively under a new administrative parent and as a part of the statistical system. In our interim report, we noted the inherent rigidity of the UCR—that, for instance, the core set of crime types covered by the UCR has remained the same since the UCR’s creation in 1929 (save for the addition of arson as a top-tier “Part I” crime. We commended the value of the NCVS as an independent check on the UCR (and vice versa), but urged that “the utility of an UCR-independent measure of crime should not prevent consideration of [NCVS] design options that reduce lockstep similarity between the UCR and the NCVS” (National Research Council, 2008b:77). To make clear the
OCR for page 201
Ensuring the Quality, Credibility, and Relevance of U.S. Justice Statistics tacit criticism in that statement, the UCR is in some respects an antiquated and inefficient system for collecting and disseminating annual estimates of level and changes in crime reported to the police at the national level. As a census-type measure of all jurisdictions, the UCR has and will continue to have essential roles, including such purposes as allocating funding across all jurisdictions based on crime counts. However, our concern is that the products of a principal statistical agency are held to high standards (as we describe in great detail throughout Chapter 5). In particular, recast as a core statistical collection within a principal statistical agency, a BJS-led UCR would require much more intensive—and expensive—attention to issues of data quality, response, data collection instrumentation, and documentation than has previously been brought to bear on the UCR. Clearly, as we have indicated, it is in the national interest to have a high-quality UCR program. It follows that if the FBI’s strategic goals shift even more heavily toward its expanded portfolio in terrorism surveillance, and hence that attention to and resources for administration of the UCR become so scarce that the UCR and NIBRS programs will atrophy, then an administrative transfer of authority of UCR to BJS would be sensible (short-term effects on response notwithstanding). Barring these conditions, we find no compelling reason, other than the organizational neatness of consolidating statistical functions in one agency, for UCR to shift away from the FBI. Timely Records-Based Collection from Local Law Enforcement Files Rather than “take over” UCR, as the option might be bluntly described, our recommendation is that BJS explore the possibility of doing something that is different from either the UCR or NCVS and that we think is better than the UCR in some respects. Specifically, we suggest that BJS work with local law enforcement agencies to develop a system under which BJS could regularly extract records from individual departments’ own computer systems—data that many departments regularly compile on their own and some departments post on their websites—for a sample of jurisdictions. This system would shift much of the burden of response and data gathering from the local authorities (i.e., filling out UCR summary forms) to BJS (i.e., sample design, data editing, and inference). Such a system would be capable of providing more timely (if less detailed) glimpses of crime trends than is possible under UCR, NIBRS, or the NCVS. The strategy we propose is consistent with one commonly used by statistical agencies, pairing sample- and census-based methods. Among U.S. federal statistical agencies, as well as in government statistical systems around the world, it is common to conduct sample-based measurements in parallel with more exhaustive, census-type measurements of the same basic phenomena. For example, BLS interviews a large sample of employers (covering
OCR for page 202
Ensuring the Quality, Credibility, and Relevance of U.S. Justice Statistics about 390,000 worksites) each month as part of its Current Employment Statistics (CES) program. Each month, the CES data are used to produce a count of changes in the number of jobs in the country; this monthly “Employment Situation” report is a familiar and highly publicized barometer of economic conditions. The CES also supports production of employment figures for states and metropolitan areas.12 This monthly measure of national, state, and local employment conditions is complemented by the Quarterly Census of Employment and Wages (QCEW), which taps into data from the state-based unemployment insurance systems. Thus, the QCEW provides a more comprehensive census-type count in the change of jobs at the national level (with QCEW coverage estimated to include 98 percent of U.S. jobs), while yielding detailed estimates down to the county level.13 Another example is the national Vital Statistics program administered by NCHS, which we described in Section 4–A.1. Drawing from reports from state health departments and reporting authorities, the vital statistics estimates represent a census-type measure of births in the United States. However, NCHS also fields the National Survey of Family Growth (NSFG; see http://www.cdc.gov/nchs/NSFG.htm), which measures self-reported births among probability samples of females, providing independent measures of births and fertility. The NSFG is also capable of generating more detail about the pregnancies and related births and yields national estimates of marriage and divorce, topics that are no longer covered by the Vital Statistics program. The UCR and the NCVS do not fit this paired census- and sample-based measurement approach because of the much more extensive scope of the NCVS, including crimes and general incidents of victimization that are not reported to the police. There are three reasons that such paired census-based and sample-based measures are useful to policy makers and professionals. First, the sample-based measurements can be constructed to be more timely than the census-based measurements. Individual attention can be paid to each sample unit; efforts of interviewers and other agents can raise the level of quick response to the survey request. Second, by focusing survey research resources on a small number of sample units, the quality of reporting can be raised, over that expected from administrative systems. Often this higher quality is obtained through increased standardization in reporting across the various sample units. Because participation is assisted through the survey data collectors, estimates can be published more quickly, to the benefit of the country. However, and the third reason for parallel samples and censuses, 12 Because of cuts in BLS’s funding for fiscal year 2008, BLS has had to eliminate the production of some CES estimates for all metropolitan areas and completely eliminate the generation of estimates for the smallest metropolitan areas; see http://www.bls.gov/sae/msareductions.htm. 13 See http://www.bls.gov/ces/and http://www.bls.gov/sae/ for additional information on the CES program and http://www.bls.gov/cew/regarding the QCEW.
OCR for page 203
Ensuring the Quality, Credibility, and Relevance of U.S. Justice Statistics those benefits come at a price: the sample sizes are generally inadequate to offer stable estimates of the prevalence of rare events or estimates of differences of small subsets of the population. Only with full census-based measurement can analysis at such levels be accomplished. In short, the sample-based methods are appealing because they can offer higher-quality responses, obtained in a timelier fashion. In contrast, census-based methods are necessary to understand small subpopulations or rare events related to the phenomenon of interest. Facilitating both of these systems makes sense, especially when the census-based system is already in place for administrative and management reasons. The panel believes that the country would be well served by a similar parallel structure applied to crimes reported to police. The new system would rely on the UCR as the census-based vehicle that would permit fine-grade comparison of areal patterns of crime, while a sample-based measurement drawn from the crime information compiled and disseminated in real time by many police departments would provide a more timely indicator of general crime trends. The NCVS, of course, would be a third component of this new structure, adding benefit through its rich contextual information and coverage of incidents not reported to police. Recommendation 4.4: To improve the timeliness of crime statistics, BJS should explore the development of a crime reporting system based on a probability sample of police administrative records. The goals of such a system would be national representativeness, high response, high data quality, timeliness and flexibility in terms of crime classification and analysis, and national statistics for the monitoring of crime trends. Such a system has two notable precursors. First, it hearkens back to the original purpose of the NIBRS program, “to take advantage of available crime data maintained in modern law enforcement records systems” (Federal Bureau of Investigation, 2004b:3). Where NIBRS has endured slow implementation because of the need to develop systems to provide rich incident detail, the hope of a timely records-based summary system would be to tap the more aggregate crime statistics used by departments to measure their own progress. The second key precursor is the one that makes such a records-based system feasible at this time, when it has not in the past. The New York City Police Department, under then-commissioner William Bratton, is credited with developing the COMPSTAT approach to measuring and planning responses to crime in 1994. Alternatively described as a contraction for “computational statistics” or “comparative statistics,” COMPSTAT combined a managerial focus emphasizing internal accountability among district and regional commanders for crime activity in their areas with a technical basis in the use of timely crime incident data to iden-
OCR for page 204
Ensuring the Quality, Credibility, and Relevance of U.S. Justice Statistics tify problems and tailor responses. The COMPSTAT program was popularly credited, at least in part, for major crime rate decreases in New York, which in turns spurred the adoption of similar programs in other cities. COMPSTAT implementation and use in a variety of sites, including detailed case studies, have been reviewed by Willis et al. (2003), Weisburd et al. (2003), and Weisburd et al. (2004). The system we suggest is akin to, but not as extensive as, the “national COMPSTAT” that has been suggested by PERF in the wake of PERF’s initial work in combining local-agency records (and hence monitoring crime-rate shifts earlier than data provided by either the UCR or the NCVS). In his introduction to PERF’s report on its second Violent Crime Summit, PERF Executive Director Chuck Wexler described the “national COMPSTAT” idea (Police Executive Research Forum, 2007:iv; emphases in original): We want to change the way that people view crime. In the past, criminologists waited several years to make conclusions about crime trends. They were cautious about drawing conclusions and waited until they could state with scientific certainty that there was a changing pattern. The problem with that approach is that by the time a crime trend has been identified, the information is so old as to make it useless, because new trends, new crime patterns, and new causes of crime have taken hold. Programs and policies that we undertake today, to respond to the crime problems of last year, are not likely to succeed. [A “national COMPSTAT” approach would use] accurate, timely information to track crime as it happens, to search for pockets of violence wherever and whenever they occur, and to react quickly. In a sense, we believe that police leaders should act more like public health epidemiologists, who don’t wait for a pandemic to overtake the nation, with hundreds or thousands of people dead, before they sound an alarm and start implementing countermeasures. We say that our proposal is akin to but not identical to a “national COMPSTAT” in that our proposal is to generate a timely summary or index value of crime, and not a direct analysis of the geographically coded incident data that can be used to inform COMPSTAT managers’ decisions to allocate personnel and resources. Valuable though such data would be, there is an inherent tension between the statistical information that an agency such as BJS can provide and the fine-grained, “tactical” assessments—informing specific interventions and used to hold line officers and commanders accountable—that are of particular interest to law enforcement agencies. Rather, our intent is to provide policy makers with crime data that provide information that is both contemporaneous (immediately of interest and relevance to consumers in law enforcement) and timely (short time between collection and dissemination). Our intent in proposing this system is not to be duplicative of effort in the existing UCR data collection. Rather, the idea is to maximize the use
OCR for page 205
Ensuring the Quality, Credibility, and Relevance of U.S. Justice Statistics of those data systems that local police departments prepare and maintain on their own, as a course of doing business and, in several cases, to promote public transparency by providing up-to-date crime information for public view on websites. Rather than put the burden for processing, coding, and interpreting results on the local departments (through completion of summary report questionnaires), the intent of this system would be to create means by which available electronic data could be transmitted to BJS with minimal effort on the part of local departments (save, perhaps, for stripping personal identifiers). The burden of sample design, data processing, and compilation would be on BJS. However, the availability of the data in electronic form and the development of routines for handling specific varieties of local data types would, ideally, yield a system in which some summary measure or index could be generated and disseminated quickly. The value of BJS involvement in compiling crime data from local departments (and, in some cases, publicly posted online as well as used in internal meetings and assessments) would be rigor in design (documenting that index measures are representative of some larger whole, if not the entire nation then something like urban areas of a particular size), consistency in definition and coding, and attention to data quality. Some further comments along these lines may be useful: Though the objective of such a system would be to be minimally invasive—making use of data that departments already have, and have in electronic format—success in implementation will depend on building ties with and commitments from individual departments. Making use of consortia such as PERF (given its initial work in the area), the Major City Chiefs Association, and IACP would be instrumental. A necessary first step in constructing such a system is to assess its basic feasibility: studying individual departments’ technical capability and the availability of suitable data resources. Short of a complete inventory of technical systems status (such as BJS performed for state corrections departments and individual facilities; Bureau of Justice Statistics, 1998b) or its SEARCH-conducted Survey of Criminal History Information Systems (Section 4–B), a module of questions on the LEMAS survey would be a useful start. The sample panel of departments would likely emphasize the largest cities, given that they are most likely to have amenable record systems. The preliminary work we just described would be useful in determining whether the content must be limited to such large cities (say, above 250,000 population) or whether a more nationally representative sample including smaller agencies is feasible. In constructing the sample, BJS should take into account the distribution in crime with respect to population size and changes in that distribution. Crime is not as highly
OCR for page 206
Ensuring the Quality, Credibility, and Relevance of U.S. Justice Statistics concentrated in the most populous of cities as in past decades, with historically high-rate large cities having been some of the beneficiaries of major declines in the 1990s and 2000s. For jurisdictions that organize their collections by their relevant state criminal codes, and whose definitions of standard crimes are not consistent with the FBI’s UCR definitions, BJS will need to learn and apply the techniques used by the localities in “converting” their data in order to report UCR summaries. A crime type such as aggravated assault, for instance, can vary substantially in scope across the states, and so a common metric would need to be defined. As a sample-based equivalent to the UCR, and for keeping collection tractable, an initial focus on UCR Part I crimes is sensible. Over time, an interesting question is whether to expand beyond that scope. Data on arrests with no “crimes known to police” equivalent in the UCR, such as drug, weapons, and DWI arrests would likely have strong policy interest. Using a panel of reporting departments should make it possible to develop quarterly estimates. However, at least two technical challenges would need to be resolved in order to smooth the introduction of a new data collection effort. The first is the exact timing and coordination of collection from “live” incident files. Department incident files may be subject to revision after initial entry, “closing” for reporting purposes at some defined data period (such as a month or quarter). Protocols for timing access to and collection of data would need to be developed, as well as mechanisms for updates as necessary (e.g., when assault turns into homicide when a victim survives a shooting for a long period). The second technical challenge is the resolution of the data that the local departments could, and could most feasibly, give to BJS: whether incident-level files as in existing NIBRS or aggregate summary reports like the agencies now send to the FBI. In some ways, aggregated summary figures would be the least demanding on both suppliers and BJS, but may require additional care (and timing) in coding to reflect federal definitions. Raw incident-level files would obviously be analytically useful, and BJS could request them geocoded, but this would be substantially harder to negotiate and would impose a greater burden on the data collector and on BJS as aggregator. BJS and its data collectors would have to anticipate a two-track plan, working with data of either type, until one became more universally preferable. The implementation of such a sample-based analogue to the UCR would enable great strides in making BJS’s law enforcement data more timely and relevant to data users, and would well complement a reengineered core-supplement design for LEMAS. In time, success in building two relatively
OCR for page 207
Ensuring the Quality, Credibility, and Relevance of U.S. Justice Statistics nimble systems could prove useful for law enforcement agencies and policy makers alike. For instance, a LEMAS supplement could provide a “quick response” capability for documenting local agencies’ experience with some new or emerging crime problem; the records-based selection could begin to show continued growth of the problem (or hint at signs of resolution); and the UCR and, particularly, the NCVS would be poised to provide richness of information on contexts and possible causes that are not possible with the interim indicators. All of this—well-designed data systems working at multiple resolutions—would contribute greatly to BJS’s core mission to assist state and local governments and agencies.
OCR for page 208
Ensuring the Quality, Credibility, and Relevance of U.S. Justice Statistics This page intentionally left blank.