THE CURRENT, NATIONAL CRIME DATA INFRASTRUCTURE rests on two principal pillars, the Uniform Crime Reporting (UCR) Program administered by the Federal Bureau of Investigation (FBI) and the National Crime Victimization Survey (NCVS) sponsored by the Bureau of Justice Statistics (BJS). Beginning operations in 1930 following the guidance and structures outlined by a working group of the International Association of Chiefs of Police (1929), the UCR Program has morphed into a set of data collections that operate on a principle of voluntary contribution of information from local law enforcement agencies, nearly all channeled through a network of state-level coordinating agencies, for aggregation by the FBI. Within the set of UCR collections, two are central: the Summary Reporting System (termed SRS or UCR Summary in this report) that directly carries on the data series created in 1930, compiling counts or tallies of offenses reported to or otherwise known by law enforcement agencies, and the newer National Incident-Based Reporting System (NIBRS) that began operations in the late 1980s and that captures incident-specific detail on a larger set of offenses than the UCR Summary.1 The second pillar of current national crime statistics is the NCVS, a household survey that—by eliciting both count and detailed
1 As in the panel’s Report 1, we maintain an important distinction between two basic units of measurement related to crime, between offenses and incidents. Offenses are the specific actions or behaviors—generally deemed socially unacceptable or subject to some sanction or punishment—that are partitioned into unique categories in one part of the crime classification recommended in Report 1. Incidents are events that can comprise one or more offenses that happen during that particular instance of “crime.” A premise of our arguments is that ideal data collection should identify all of the offenses in an incident rather than collapse detail within an incident to a single offense (that is deemed to be most important or severe) as is done by the SRS.
information on personal victimization experiences—is positioned to generate estimates of crime offenses whether they are reported to law enforcement or not. The NCVS began operations in the early 1970s as the National Crime Survey and took its current name in 1992 after a thorough redesign.
The UCR Summary still adheres heavily to definitions and concepts laid down in 1929, with relatively little change—and, to important degrees (and for important purposes, including basic comparison across the sources), the conventions of NIBRS and NCVS were patterned after the UCR Summary at their formation. Thus, the current data collection mechanisms are relatively ill-suited—by construction—to cover new or emerging types of crime and so their limitations cannot be addressed simply through more complete collection of the same information. Some 30 years after NIBRS data collection began, and for a variety of reasons, NIBRS-format data are reported to the FBI from only roughly one-third of jurisdictions; likewise, the NCVS is enduring the same challenges as other federal surveys, trying to achieve reasonable response rates within tight cost constraints. The point is that even if these barriers to respondent participation were suddenly lifted—if the nation’s current major crime data sources were to continue exactly as-is, just with full cooperation—the major gaps in content knowledge would remain poorly covered, simply because the topics concern offenses and require contextual information that the NIBRS and NCVS are not currently equipped to address in their current configurations.
The only way to start to answer these and other important questions about crime in the United States in a comprehensive way is to fundamentally modernize and redesign the underlying crime data infrastructure. It is amidst this mix of factors—the hardening of concepts and concepts of crime measurement over decades of practice, and the recognition of major, substantive gaps in the nation’s quantitative understanding of crime—that the Panel on Modernizing the Nation’s Crime Statistics has carried out its work.
This panel’s origin is described more completely in our first report (National Academies of Sciences, Engineering, and Medicine, 2016:Ch. 1; hereafter referred to as Report 1), but to recap in brief: This study is a successor to and extension of a previous, comprehensive Panel to Review the Programs of the Bureau of Justice Statistics conducted between 2007 and 2009. The BJS Review panel produced a first report (National Research Council, 2008) on options for conducting the NCVS—the bureau’s most costly but one of its most important data collections—before focusing on the overall BJS portfolio in its final report (National Research Council, 2009). That review naturally touched on the UCR Program as a logical counterpart to the NCVS and given
the BJS’s legislatively assigned task of providing comparable data series of crime. However, that study was focused squarely on the BJS; more direct assessment of the UCR and its component programs, and the manner of “crime statistics” beyond the purview of the BJS (or the FBI), was explicitly outside its study scope.
This study was launched at the National Academies of Sciences, Engineering, and Medicine when the U.S. Office of Management and Budget (OMB) encouraged the BJS and the FBI to jointly sponsor this work—with the direct mandate to take a full view of the entire enterprise of data collection on “crime” in the United States and assess its relevance to modern information needs. The full statement of task of the current Panel on Modernizing the Nation’s Crime Statistics is presented in Appendix A; developed in collaboration with all parties (BJS, FBI, OMB, and the Academies), it is a detailed yet broad mandate, to cover three main topic planks in two operational phases.
Our Report 1 (National Academies of Sciences, Engineering, and Medicine, 2016) was designed to cover the first, substantive plank in the panel’s broader charge. The core of Report 1 is the presentation of a proposed, new classification for criminal offenses; said classification is a detailed revision of an International Classification of Crime for Statistical Purposes, maintained by the United Nations Office on Drugs and Crime and to which the panel remains indebted for providing a solid starting point. Our study was deliberately constructed in this fashion—starting with a classification system and only then moving into the full range of methodological and implementation issues, in this report—because it has been many decades since core concepts of U.S. crime statistics have been reviewed in full and from first principles. Traditionally, classification in U.S. crime statistics has been treated solely as a post hoc labeling and editing procedure; this study’s basic premise is that there is considerable advantage in achieving clarity in concept first and building new indicators and data systems using the classification as a blueprint, rather than forcing existing statistics into a new post-processing mold. Another main objective of this final report, referenced in the full statement of charge and combining methodological and implementation elements, is to consider the appropriate governance structure—the most productive division of labor between the BJS and FBI, other data providers, (volunteer) law-enforcement-agency contributors, and other entities.
The centerpiece of the panel’s data-gathering work, particularly for the first phase, was the convening of two workshop-style sessions in June and July 2014—supplemented by presentations and discussions at the panel’s subsequent meetings in 2015 and 2016. While the workshop-style sessions were primarily intended to elicit user and stakeholder views on the need for crime statistics, they naturally and inevitably raised central implementation concerns as well, and so provided a trove of information underlying this report.
Collectively, the panel’s two reports represent a complete reconceptualization and redefinition of crime, in the context of U.S. crime statistics. At the outset of this report on methodological and implementation issues, it is useful to restate some basic precepts from Report 1 to explain our approach—in particular, ideas that we expressed in detail in Section 1.2 of Report 1’s examination of “What Is Crime?”
At the outset in 1929–1930, the UCR Program established definitions based on the basic notion of “crime” as actions that are unlawful (counter to the language of criminal or penal codes) and so subject to punishment or sanction. This literal notion of crime motivated the initial restriction of UCR content to a small number of offenses, precisely because these major violent and property crimes were thought to be defined relatively consistently across the criminal codes of the various U.S. states (and because they were believed to be relatively well reported to local law enforcement agencies). However, consistent with our charge to adopt a broad perspective, we favor a behavioral concept of crime, defining it as a class of socially unacceptable behavior that directly harms or threatens harm to others. In Report 1, we argue in greater detail that the behavioral definition is less constraining and more dynamic in nature than the letter of legal text, befitting a phenomenon in crime that is constantly evolving based on shifts in technology and social norms. We also note in Report 1 that it is impossible to completely sever ties between the text of federal and state criminal code and our behavior-based definitions, and that the code remains an essential anchor. This is the reason why we exclude some behaviors that might be rude or unpleasant but that do not directly cause or threaten harm. It is also the reason why we insert the word “unlawful” in many of our specific offense distributions, because activities such as gambling or use/possession of some controlled substances are not criminal in all jurisdictions.
This broader perspective on crime differs from decades of crime statistics practice, and so it may be jarring to some. In particular, it necessarily blurs the line between things that are literally dubbed “crime” or “criminal” because they are adjudicated in criminal courts and those that are tortious or civil in nature (and so are adjudicated in other courts or proceedings). But this is the fundamental point of this panel’s work; we will argue in this report that the nation’s picture of crime is limited unless it considers all types of crime. Another critical point that should be made in this regard is that the task of classifying crime in Report 1 was to divide the entire space of behaviors that could be considered crime into unique offense categories; it was not to rank order them by importance, severity, or any other such scale. Homicide, robbery, harassment, environmental dumping, and tax evasion are all individual categories in the crime classification—and so we view each of them as a crime offense—but it would be a serious error to interpret our work as suggesting that
they are equally important or that any of them is wholly unimportant relative to the others. Again, the interest is in a complete picture of crime in the United States, and that picture is incomplete without all perspectives.
In order to conduct its work, the panel had to make assumptions, recognize limitations, and make decisions on approach up-front. In order to calibrate expectations, we note several of these in this section as preface for the arguments that follow.
To begin, the role of this report is not and cannot be to serve as a precise, intricately detailed implementation plan. Rather, its role is to be a general blueprint—to speak to the shape, the contours, and the basic principles of a modern crime data infrastructure, and suggest general paths for implementation. This is the case for a variety of reasons, not least of which is that panels of the National Academies of Sciences, Engineering, and Medicine are generally precluded from describing or advocating specific federal budgetary priorities or recommending specific bureaucratic adjustments unless explicitly tasked to do so. Another important reason is a theme that we return to throughout this report: that there is formidable inertia in the enterprise of national crime statistics, and that the imperative is to make a start and refine based on the results rather than become mired in initial details. We cannot presume that our collective perspective, or even the wide range of user and stakeholder perspectives we sought in our meetings, is sufficiently thorough as to prescribe what offense measures should be improved first or what alternative data sources are more or less reliable than others. Arriving at an ideal and improved set of measures will require extensive iteration and refinement, based on prototype and preliminary measurement attempts and reflecting an even broader sweep of stakeholder input. We understand that some will look at our guidance in this report and wish for more specificity, but we conclude it to be best and most appropriate to maintain a fairly high-level view in suggesting revision of the crime statistics enterprise.
Another reason for not being too prescriptive or detailed on any specific point of implementation or methodology is, itself, a general constraint of this study. This constraint owes directly to the sheer extent of the panel’s charge and the vastness and continual evolution of crime itself: We are obliged to emphasize breadth over depth on any specific area of crime or aspect of methodology. The panel’s workshop-style sessions, the presentations at meetings, our deliberations, and our readings and research covered an extensive amount of terrain—and yet barely scratched the surface of what could be discussed. Some topics that we discuss at a high level—for instance, illegal drugs (and data on them), the relationship between firearms and crime, and the
measure of elder abuse and mistreatment—have been the focus of panel studies in the past, and any number of others are ripe for in-depth review on their own.
Relatedly, the panel had to deliberately keep its focus on crime statistics—the material for crafting measures of levels and trends in the occurrence of specific offenses—as contrasted with justice statistics writ larger. The primary front on which we faced this challenge was addressed in Report 1, which defined what is meant by crime while minding the traditional distinction between what is typically processed in civil justice (or purely regulatory) proceedings versus criminal justice proceedings. There, we established definitions based on behaviors rather than statutory code language—arguably blurring the functional civil/criminal line in some instances but all in service of identifying behavior that should properly be labeled “crime.” But there are further ramifications of a purely crime-statistics focus in this implementation- and methodological-centric report:
- The UCR Program has evolved over the years to become a set of data collections, as described in Box 1.1. Some of these are specialized crime/offense collections and so are germane to crime statistics; these include the UCR data collections on hate crimes and arson, as well as the Supplementary Homicide Reports (SHR). But the UCR’s collection of data on arrests and clearances of arrests/cases—intended more as a measure of law enforcement activity and effectiveness, and of characteristics of the arrestee population, than of crime occurrence—and its collection of data on staffing and resources available to local law enforcement agencies are issues that we concluded out of our scope. This is not to exclude the possibility that arrests might be the most suitable or salient measure for certain specific offense types, because arrest might be the earliest measurable outcome in justice-system processing for those offenses. This might include some white-collar offenses such as embezzlement; we discuss such matters further in Chapter 2.
- Similarly, a major issue that has vaulted to the forefront in public debate during the lifetime of this study is data on the use of force by law enforcement, particularly resulting in injury or death of arrestees or suspects. Both the BJS and the FBI offered ideas for the generation of systematic use-of-force data as a standalone collection; at this writing, the FBI’s proposed collection from local law enforcement partners appears to be going forward with the support of the Justice Department. But, having discussed the concept of justifiable homicide (and its role relative to our proposed crime classification) in Report 1, we do not believe it necessary to comment further on the disposition of either a standalone database on use of force by or involving law enforcement, or on the UCR Program’s existing Law Enforcement Officers Killed or Assaulted (LEOKA) collection.
- Our predecessor National Research Council (2009:Finding 3.1) panel reviewing BJS programs critiqued current U.S. justice statistics for taking an overly cross-sectional approach—collecting detailed data “on an institution-by-institution basis (police, courts, corrections)” but in a way that “makes it difficult or impossible to answer questions about the flow of individuals from arrest through eventual exit from the system.” However, that panel also recognized that this approach was necessarily the case, noting a critical need for greater capacity for linkages within and between crime and justice datasets. Such linkages include connections through all the steps in processing in the criminal justice system—from police report through arrest through charges through adjudication through corrections processing—as well as linkages to external social and economic data, as would be ideal for studies of recidivism. The current panel’s recommended classification of crime is certainly intended to foster agreement on definitions and units of measurement, which is one part of the general problem—but beyond that, we do not believe it our role to go further. We agree that it would be ideal to develop a system of identifiers that would make it easier to trace the progress of crime offenses and incidents (and suspects/offenders and cases in general) through each stage in the criminal justice system, but the construction and implementation of such is beyond our scope.
- Related to this latter point, the FBI’s Criminal Justice Information Systems (CJIS) division is responsible for maintenance of national criminal history record databases as well as the UCR; criminal history records improvement is an area in which BJS and the FBI have historically cooperated, leveraging BJS’s authority to issue grants that the FBI lacks. There is obviously overlap between definition and measurement of “crime” and the promulgation of “criminal history” records. But delving into the content of criminal history record databases and fully dealing with the issues related to transmitting, disclosing, and linking person-level records in multiple datasets are all major tasks worthy of detailed research and debate in their own right, and not subjects that this panel is constituted to address.
It should be noted that both the FBI and BJS continued to develop their data collections during 2017, yielding changes that may not be documented in this report because they occurred after the panel’s active data-gathering phase.
Finally, in terms of style and execution, we follow the practice of Report 1 in deliberately being very sparing in our designation of formal findings or recommendations. We chose that approach in Report 1 to spotlight a single recommendation—the panel’s proposed classification of crime for statistical purposes. This report offers considerable commentary related to past and future crime reporting processes, but the panel opted to focus on a small
number of broad, main messages that we designate as formal conclusions and a recommendation, rather than dilute this guidance with a longer checklist of highly specific recommendations.
Chapter 2 begins where we left off in Report 1, comparing the coverage of the classification of crime we recommended in our Report 1 to the current national crime statistics, and to NIBRS in particular. Based on that coverage match, we outline the basic structure of a new system of multiple data collection efforts for crime statistics, using our proposed classification as a base. Chapter 3 rounds out our formal findings and suggests directions for coordination and governance within a modernized crime statistics infrastructure.
As previously noted, the full text of the panel’s detailed charge is reprinted in Appendix A. As important reference for a focus on implementation and methodology issues, this volume also includes some extensive appendixes. Appendix B highlights important themes and lessons in the evolution of current U.S. national crime statistics, a review that provides important context about retooling the crime data infrastructure. Appendix C is a longer exposition of a rough, substantive match between our proposed classification of crime and its associated set of attributes/contextual variables with the nation’s existing crime data resources, particularly NIBRS and NCVS. Appendix D offers longer-form commentary on some important, remaining issues that crosscut issues of crime statistics coverage and governance, important issues but ones for which full discussion in the body text might be unduly disruptive to the flow of arguments. To inform meaningful discussion of the role of the states in crime data collection, Appendix E presents state-level profiles summarizing legal requirements for crime reporting defined in state law or regulation and the status of adopting incident-based crime reporting. A final, shorter Appendix F sounds a cautionary note from international experience, using modernization experiences in the United Kingdom to underscore that some of the changes we suggest are not simple panaceas.