PROPER CRIME DATA COLLECTION is difficult, but we think that the eventual benefits of a better understanding of crime, improved techniques (developed using those data) for law enforcement to address crime, and increased accountability of and trust in law enforcement entities outweigh the challenges. Moreover, crime results in substantial costs to the nation: Improved measures of crime are imperative for establishing policy priorities across the full range of crime types and for evaluating policy effectiveness in a wide range of areas.
That said, there is a perennial risk that appeals for major change could be considered unrealistic, or simply naive, without grappling with a major underlying question: How can one have confidence in implementing major, new data collections and procedures—on anything approaching a timely basis—when past change has taken so long or proven unsuccessful? The historical record (see Appendix B) abounds with missed opportunities for major change and improvement—the 1958 suggestion that the Federal Bureau of Investigation (FBI) Uniform Crime Reporting (UCR) Program’s firm adherence to distinguishing between “Part I” and “Part II” offenses would unduly marginalize the latter and the 1985 call to phase in detailed incident-based crime reporting among them. An obvious further example is that federal law enforcement agencies are mandated by law to report UCR-defined crimes to the UCR Program. This mandate became law in 1988, yet it was not until 2015—following a year-long concerted effort to get a very small set of those federal agencies (including units of the FBI itself) to submit data—that federal
submissions warranted publication as a special section in that year’s Crime in the United States tabulations. This chapter covers our panel’s final two formal conclusions. It is about the bigger picture: identifying what our research and our discussions with the wide array of crime data users and stakeholders suggest to be the overriding impediment to change, and suggesting ways to overcome that impediment.
In our assessment, the greatest single barrier to change in the nation’s crime statistics system is that there is (and has been) no clear, direct answer to the question of who owns—who is ultimately responsible for—the nation’s crime statistics, for the full range of crime offenses. Both the FBI and the Bureau of Justice Statistics (BJS) clearly have a strong stake in crime statistics and in understanding and countering crime, generally. But the culture that has evolved over the decades is essentially that the FBI administers the UCR program; it collates the input data; it filters the data through its edit checks and anomaly detection routines. The FBI also publishes tabular summaries of the data—historically, as depicted in Section B.4, with strong disclaimers characterizing the national collection as little more than collation, though the FBI has shown more engagement in recent decades. Meanwhile, BJS supports the UCR program through its grantmaking authority and its sponsored methodological studies; it analyzes the data (along with that from BJS’s own National Crime Victimization Survey [NCVS] and other resources) and, through its network of state Statistical Analysis Centers, it interprets the data for a wider array of stakeholders. The culture that has evolved over time for national crime statistics collection is one in which no party asserts or has asserted complete program management for data collection on all crime.
In several important ways, the development of this form for U.S. crime statistics collection is completely predictable and understandable:
- A reluctance to impose requirements is directly related to what is certainly a cornerstone of the nation’s crime statistics apparatus: the voluntary nature of data contributions to the national/federal level, which itself owes directly to the reservation of most criminal justice and police functions to the states under the U.S. federated system and constitutional government.
- Relatedly, a strong tendency toward decentralization of authority pervades the U.S. justice system. Not only is authority for investigating and combating crime dispersed across thousands of state, local, and tribal agencies, but even enforcement of laws against federal offenses is rarely vested exclusively in a single federal law enforcement agency.
For example, a single offense type like counterfeiting is the investigative/enforcement province of (at least) the U.S. Secret Service, the FBI, and the Federal Trade Commission. Controlled substance–related crimes may be covered (based on their circumstances) by many more agencies ranging from the Drug Enforcement Administration to U.S. Customs and Border Protection to the criminal investigations unit of the Food and Drug Administration.
- The United States is also fairly extreme relative to other countries in the decentralization of its federal statistical system. A great deal of federal statistical activity is vested in 13 principal statistical agencies of the federal government (of which BJS is one), and not a central national statistical office as in some other countries. Moreover, dozens more bureaus and agencies in the federal government devote appreciable resources to conducting statistical activities.
This unique structure is not inherently bad, but the collective consequences are considerable. Deference to the voluntary nature of data contributions means that the federal level cannot directly impose systems or requirements on local law enforcement agencies. Historically, this has suppressed the notion that use of systems or implementation of requirements might be motivated or incentivized, positively or negatively. Put another way, there has been a historical reluctance to apply either “carrots” (e.g., direct financial support) or “sticks” (e.g., linking statistical reporting completeness to eligibility for grant funds) to sway state and local participation in crime statistics. Likewise, dispersed authority for law enforcement has strong virtues. Yet a necessary consequence is that authority for reporting or recording those offenses is just as, if not more, dispersed as a result, making it difficult to prioritize measurement of specific offense types or to ensure uniform measurement of concepts.
Often, it has been the case that when new topics of keen policy interest arise and demand quantification, the process of determining what the nation needs to know about the topic and how best to measure it has been done ad hoc. A common approach has been to create new offices or agencies to collect the relevant information or to coerce the new topic into existing data collection structures, more for convenience than accuracy. The confluence of all these decentralizing forces and consequences for crime measurement is probably most strikingly illustrated by the offense of arson. The section on arson in Appendix C gives considerably more detail but, in brief, the United States has no fewer than three designated repositories of national statistics on arson, administratively nested within two separate executive departments—all with explicit legal mandates. Specifically, federal law explicitly authorizes collection of data on arson and criminal misuse of explosives in the FBI, the U.S. Fire Administration of the Federal Emergency Management Agency, and the U.S. Bomb Data Center of the Bureau of Alcohol, Tobacco, Firearms,
and Explosives. The three data resources all use slightly differing concepts and rely on voluntary data contributions from different sources of largely unknown accuracy and coverage—and, given the inherent difficulty that may arise in determining whether a fire is deliberately or maliciously set, are all likely inadequate measures of total arson activity in different ways.
In short, because no single entity owns or assumes direct responsibility for the nation’s crime statistics as a whole, there is neither a strong coordinating body nor a defined governance structure dedicated to the generation of data on all types of crime. In our assessment, real improvement in national crime statistics hinges critically on the creation and maintenance of both these functions.
We use both terms, coordination and governance, because the two functions have distinctly different foci and responsibilities. As we will discuss in more detail, delineating the roles separately does not mean that they have to be performed by two different agencies/entities, nor that they necessarily have to be consolidated into one.
Expectations for a strong, national-level coordinating agency for crime statistics require that the agency be actively and directly involved in routine, daily data collection and processing operations. It would bear primary responsibility for regular engagement with source data providers (motivating input to the greatest extent possible while still respecting the ultimately voluntary nature of data contribution to national collections) and managing the resulting data products. Roughly, the key roles of a coordinating body/coordinating agency for crime statistics would include but are not limited to:
- Cultivating an “honest broker” role in interactions with partner agencies, data providers, and providers of computer hardware/software solutions, including negotiation of data-sharing arrangements and memoranda of understanding for data sharing and easing the challenges of implementing necessary changes in data collection content or procedure;
- Maintaining credibility as a source/provider of data in its own right, while providing for analysis (and limited interpretation) of the coordinating agency’s own data resources along with those it builds from partner contribution;
- Ensuring and, to the extent possible, enforcing data quality from data provider agencies—deciding what are adequate data as input to the files and what data products are credible for release;
- Managing inflow of data provided by partner agencies and external sources, including providing protections for confidentiality of person, household, business, or organizational records in the data and ensuring
that data quality channels are being followed (e.g., that possible errors detected at the national or state levels are in fact reviewed, revised, and resubmitted); and
- Managing outflow of data and statistical products, including providing feedback and products to source data providers and brokering access to data and summaries to all stakeholders (other agencies, the media, qualified researchers, and so forth).
Again, where the coordination role has tended to be somewhat passive in the past, the great need in a modern system is for top-level, national commitment to the data collection to be continually demonstrated, integrating and assimilating data from a wide variety of sources, and working to ensure their quality. Accordingly, we conclude that this strong national coordination in national crime statistics is essential:
Conclusion 3.1: A stronger federal coordination role is needed in the production of the nation’s crime statistics: providing resources for information systems development, working with software providers to implement standards, and shifting some burden of data standardization from respondents to the state and federal levels. The goal of this stronger role is to make crime data collection a product of routine operations.
If coordination is ideally a close-up, hands-on (to the extent possible) role in national crime statistics, then the similarly, critically important role of a governance structure is to take a broader, whole-system view. A core duty of a governance structure is to seek—and act upon the needs of—the broader data provider and stakeholder communities, setting policy for the crime statistics system writ large and providing overall program leadership. Expanding somewhat on these points, the ideal roles of a crime statistics governance structure include but are not limited to:
- Decision-making authority for content (based on stakeholder input), including, in particular, managing a mechanism to maintain and regularly update the underlying classification of crime for statistical purposes and its integral list of attributes;
- Providing voice for (and attention to feedback from) a full range of stakeholders, being responsive to data users’ concerns, and involving partner agencies in each others’ related data collection efforts;
- Taking a leadership role in promoting data quality, promulgating best practices, and fostering a wider role for analysis of crime data and fuller understanding of crime data among providers and public alike; and
- Facilitating systemwide methodological research and development, including the development and refinement of imputation procedures for
missing data and establishment of protocols for preserving data providers’ privacy and the confidentiality of their responses.
In formulating a formal conclusion, on this topic, we emphasize the absence of such a whole-system view at present and urge that such a governance structure—applying to “traditional” and “new” types of crime alike—be established:
Conclusion 3.2: Having an effective governance structure for the complete U.S. crime statistics enterprise is critical. There is currently no entity responsible for reporting on the full range of crimes in the proposed classification (most notably for top-level categories 6–11).
To move forward, it is critical that there be an organizational home for chief coordination and governance roles. In keeping with our panel’s charge, we concluded it best not to focus on the exact who associated with these roles—whether BJS, FBI, both, or neither—but rather to focus most heavily on the what and how of crime statistics coordination and governance. To that end, what is of paramount importance is that the coordination and governance of modern crime statistics collection and dissemination be done in accordance with the principles and practices—and the general sensibilities—expected of a statistical agency.
Our panel’s administrative parent, the Committee on National Statistics of the National Academies of Sciences, Engineering, and Medicine, published the first edition of its cornerstone document, Principles and Practices for a Federal Statistical Agency, in 1992 and later adopted the standard of issuing a revised edition every 4 years. Box 3.1 lists the core principles and practices as expressed in the sixth edition of the report, released earlier in 2017. It is no accident that the ideal coordination and governance roles listed evoke the language and spirit of these principles and practices, in that they (and many other arguments made throughout this report) were made with the tenets of Principles and Practices very much in mind. All four principles, and the practices related to them, are particularly vital in the politically charged area of crime statistics, being issues that the existing crime statistics structures have grappled with in the past as well as foreshadowing important developments to come:
- Relevance to policy issues means much more than positing that crime is an important policy topic; it is for adherence to this principle that we emphasize sensitivity to data user and stakeholder feedback, and active engagement with data providers, as strongly as we do.
- Credibility among data users may be the area in which national crime statistics have, historically, fallen shortest, with the overly broad branding
- To its credit, adherence to the principle of cultivating trust among data and information providers may be said to be a strong point of the current national crime statistics apparatus. Decades of high and genuinely impressive levels of local-agency voluntary contribution of UCR Summary data is undoubtedly a partial function of crime reporting, but it could not endure without grounding in trust relationships between local agencies, the state UCR coordination network, and the national UCR program. Likewise, BJS’s impressive record of methodological innovation in the NCVS is anchored in gaining trust and rapport with survey respondents—to most effectively yet respectfully motivate them to share details about often painful and possibly traumatic experiences. Going forward, modern crime statistics will depend critically on nurturing the same kind of trust relationships with holders of more detailed,
of Crime in the United States and insufficient documentation of uncertainty in counts (and completeness of underlying source data) in the UCR police-report data.
incident-specific information and data from diverse administrative-type data maintainers.
- The principle of independence from political and other undue external influence is a very important one. Consistent with this principle, the true value of national crime statistics is their production in a climate of impartiality and objectivity, working to ensure that definitions and concepts are implemented consistently and uniformly with data quality as the ultimate end objective rather than advancement of any political or administrative agenda.
The practices listed in Box 3.1 are essentially the means by which the principles are met and reinforced; particularly important practices to stress in the development of new crime statistics structures include building/maintaining openness about sources and limitations of the provided data, cooperation with (and responsiveness to) data users, and establishment of an active research program (including methodological development and, substantively, work contributing to the periodic update of the underlying classification and definitions of crime offenses).
Neither of the major agencies that currently collect national crime data, BJS and the FBI, has all of the necessary requirements to fulfill the coordination and governance roles that we envision, though both have some relevant attributes. This observation stems from several things that plain reading of the current statute and regulations justifying the collection of national crime statistics permit stating as basic fact; see Box 3.2 for excerpts of the relevant text. Among these basic facts is that BJS has a very lengthy and detailed list of specific duties etched into its authorizing statute, including a key clause directing the bureau to develop and provide consistent, comparable indicators of crime and victimization. In fact, fair reading of these detailed duties shows very strong correspondence with the ideal coordinating and governance functions we articulated above; legally, BJS has the proper scope and range already built into its legislation, including (respecting the legislative intent when BJS was established in 1979) an obligation to focus on such “new” offense types as corruption and other white-collar offenses. But a significant caveat must be noted: As our predecessor National Research Council (2009) panel reviewing the full suite of BJS programs observed, BJS has never been resourced and funded commensurate with the extensive nature of the duties assigned to it. Then, as now, an overwhelming share of the essentially flat-line funding BJS is allocated is applied directly to conducting the National Crime Victimization Survey (NCVS).
The law and regulation authorizing the UCR Program is decidedly vaguer in letter. As noted in Report 1 and in Box 3.2, it was only 6 decades after the FBI began collecting UCR data that a definition (short of explicit authorization) of the UCR was written into law. That said, the very concise yet very broad
record collection and maintenance functions vested in the FBI have endured and been generally respected for those many decades. Moreover, the FBI’s statutory authority to acquire crime records has been supplemented over time, including being designated by regulation as the Justice Department’s collector of hate crime statistics in 1990. The argument that the FBI has cultivated relationships with its “peer” law enforcement agencies at the state and local levels, particularly through its cautious, deliberate Advisory Policy Board (APB) structure, remains a strong one that cannot and should not be upended casually.
In both cases, addressing the crime categories that have long been neglected in national crime statistics will require serious attention to topic domains that may not mesh neatly with the traditional highest priorities of either agency—and there are grounds for concern about the capability of either BJS or the FBI to effectively tap the data resources of external agencies. BJS can legally enter into data collection agreements with all manner of governmental and private agencies. But, practically, there are limits to the amount of suasion that a small bureau, nested within a subunit of the Department of Justice, can be expected to have in negotiations with agencies in other cabinet departments or with pertinent independent agencies such as the Federal Trade Commission, the Environmental Protection Agency, the U.S. Postal Service (and its Postal Inspection Service), and the Securities and Exchange Commission. Work on the “new” crime types will also involve creating interfaces with federal and state Offices of Inspector General (law enforcement agencies but not in the traditional police mold) and a variety of state-level public welfare agencies. BJS might have the legal authority built into its enabling law to broker arrangements with the whole gamut of agencies, but it would likely need additional clout to do so efficiently. The FBI might reasonably be expected to have more clout with some agencies—yet, as observed in opening this chapter, it took nearly 30 years to achieve minimal compliance with explicit,
binding, federal law mandate on the FBI’s peer federal law enforcement agencies (including the FBI itself) to report UCR-type offenses to the FBI’s UCR Program. The FBI’s purview has been limited to police-report (or, more properly, law enforcement-report) data, so it would enjoy no special suasion with entities collecting crime-relevant information but who do not view their role as crime investigation. Indeed, some agencies, businesses, or operations might balk at sharing their records and information with a clearly constituted law enforcement agency, even if strictly for statistical purposes.
There are cases that can be made for which agency, or coalition of agencies, should play the strong coordination and governance roles we envision, and we discussed the assignment of the functions extensively in our deliberations. We recognize the legal and regulatory mandates, and general justification, for national crime data collection, and we appreciate the breadth of terrain that must be traversed to begin to come close to a complete picture of crime in the United States. However, we concluded that there is no solid basis for suggesting that any particular “organizational chart” arrangement or designation of duties is superior to any other; we are aware of no relevant research literature or other scientific footing that we could point to in making such a ruling.
Our core argument is that U.S. crime statistics should be coordinated and governed according to the expected sensibilities of a federal statistical agency—which is not the same as saying that a federal statistical agency (and only a federal statistical agency, such as BJS) should play those roles. It is difficult to imagine a U.S. crime statistics system without substantial involvement of both BJS and FBI. Yet both would require departures from their status quo operations. BJS is extraordinarily effective within its historically tight resources but would require more resources and likely more legal authority to be successful. And, in support of not only its UCR/NIBRS work but also its maintenance of criminal history records and other data, the FBI should also strongly consider positioning part of its Criminal Justice Information Services (CJIS) Division to directly function as a statistical unit. This change may (perhaps) involve seeking designation and certification as a recognized statistical unit of the federal government pursuant to the Confidential Information Protection and Statistical Efficiency Act (CIPSEA).
Our bottom line is we lack a firm scientific basis for suggesting any particular assignment of strong coordination and governance roles in a new crime statistics infrastructure—and yet we emphatically believe that establishing such roles is critically important, and that there finally be a clear, comprehensible answer to the question, “Who ‘owns’/takes responsibility for the nation’s crime statistics?” Under the highly decentralized U.S. federal statistical system, such centralized oversight as exists in the system is vested in the U.S. Office of Management and Budget, in the Executive Office of the President. Accordingly, our report’s central recommendation is addressed to
the single entity that is the most sensible (if not the only realistic) adjudicator of the issue:
Recommendation 3.1: The U.S. Office of Management and Budget (OMB) should explore the range of coordination and governance processes for the complete U.S. crime statistics enterprise—including the “new” crime categories—and then establish such a structure. The structure must ensure that all of the component functions of generating crime statistics are conducted in concordance with the sensibilities, principles, and practices of a statistical agency. It should provide for user and stakeholder involvement in the process of refining and updating the underlying classification of crime. The new governance process also needs to take responsibility for the dissemination of data products, including the production of a new form of Crime in the United States that includes the “new” crime categories.
It stands to reason that the designation of a coordination and governance structure by OMB should be predicated upon a wide variety of stakeholder input and perspectives from the potential data-providing and data-using agencies, and that statutory language may be necessary to achieve the necessary changes. We will comment further on the final sentence of this recommendation in closing this chapter in Section 3.3.
We began this chapter with reference to the federated nature of law enforcement in the United States—the reservation of much police and law enforcement functions to the states and thence to local governments. Accordingly, given the vital role of the states in American law enforcement, another important observation about the coordination of national crime statistics is in order: It is difficult to imagine major change in U.S. crime statistics being successful without national crime data collection being recast as a strong federal-state cooperative or partnership program.
In the UCR’s police-report data context, the pivotal role of the states was clear from the outset—if not fully acted upon for a number of years. In the handbook that defined the initial parameters of the UCR Program, the International Association of Chiefs of Police (1929:14) put matters bluntly:
If crime statistics are ever to be complete for the entire nation, or even for a single state, the state governments must lend a hand. They alone possess the necessary power to enact statutes requiring the submission of uniform crime reports. . . . If the states all follow one uniform plan, the
[U.S.] Department of Justice can then lift bodily the appropriate parts of the state compilations.
As established within the following year, the UCR Program did not entirely follow this guidance, making broader appeals for data submissions to local agencies instead. Still, the very first UCR bulletin invited attention “to the practice prevailing in certain States whereby the State Bureau of identification or other similar agency compiles the crime returns and transmits the reports to this Bureau.” The bulletin claimed that “greater accuracy and comprehensiveness will be assured” through this state coordination, and noted that “the results have been encouraging” from the first two states to adopt such a model, Ohio and Utah (Bureau of Investigation, 1930a:1). Thereafter, attention was devoted in the bulletins to new participation by local agencies, and less so about interactions with the states. However, this changed abruptly at the end of the 1960s, at which time more formalized relationships with the states were actively sought. As the 1970 issue of Crime in the United States summarized (Federal Bureau of Investigation, 1970:57):
The last several years the FBI has been actively assisting individual states in the development of statewide programs of police statistics compatible with the national system. These statistical programs have been given impetus by developing statewide computerized law enforcement information systems, of which they are an essential part. Through such mandatory state programs more complete and a better quality reporting is envisioned. Likewise. through coordination with the state agency the data is available for the use of the state, and the collection machinery to the national agency is substantially streamlined.
The report continued by naming specific instances of new state programs for UCR collection—including the strident statement that “in 1969 the FBI seized the collection of all data directly from municipalities and counties in New Jersey and Michigan.” Programs were also said to have been established in 1969–1970 in Kentucky, Rhode Island, Pennsylvania, Minnesota, Nebraska, Florida, and Wisconsin, with some partial implementation in California as well. In the decades that have followed since then, what has evolved in the national UCR program is the development of a network of state UCR programs—through which virtually all local-agency-submitted reports are channeled before forwarding on to the FBI and national collection. (Exceptions to reporting through the state are now rare, typically coinciding with technical difficulties at the state level or—as is currently the case for the state of Mississippi—lack of a recognized state program.) Collectively, this network of state agencies has a support arm/service organization of its own, in the Association of State Uniform Crime Reporting Programs (ASUCRP).
In Appendix E, we compile extracts of crime reporting laws and regulations defined in state legal and administrative codes (as distinct, whenever possible, from related law on the maintenance and use of criminal history records, for instance); the appendix also summarizes the status of incident-based crime reporting in each state. The summaries of state law are admittedly somewhat cursory in nature, being limited to immediately obvious sections of code and support documentation, and not an exhaustive canvass of every state’s legal documents. Even at that, the material in the appendix is both dense and difficult—but impressions from it are absolutely essential to understanding some features of current crime data collection and the prospects for future approaches.
Above all, reference to the language of state laws makes abundantly clear that partnership with the states is not simply desirable—if only to reduce the number of data input streams from almost 20,000 individual law enforcement agencies to roughly 50—but virtually essential. Throughout this report, we have tried to take care to characterize the voluntary nature of local agency crime reporting as effectively being voluntary to the national program. But the same is not true of reporting from local agencies to the state. By our reading and count in Appendix E, the local law enforcement agencies operating within 43 states are legally obligated to submit crime report data to a state-level entity—typically, the state police/highway patrol, justice department, or other entity that is designated as the state UCR program/coordinator. An additional three states have legal code that suggests “cooperation” with the national UCR program but does not explicitly require transfer of data to the national level; four states’ laws appear to leave the issue unresolved. By comparison, it is hardly unprecedented for local agencies to be explicitly required by state law to contribute data to the national UCR program—eighteen are so required, and another seven are encouraged to contribute to the national collection without it being explicitly mandated (the other half of state codes appear to be silent on the issue).1
Another key observation from the state law appendix is the very encouraging signals that come from some of the most recent revisions of state crime reporting laws to update—signs of intent to not only upgrade to more detailed incident-based crime reporting, but also to do so on a quicker pace than the national program. As we will discuss further in Section D.1, one of the critical barriers to rapid adoption of the National Incident-Based Reporting System (NIBRS) by states and localities was the lack of early NIBRS-format adoption by large, major jurisdictions whose commitment to the new reporting standard would serve as a motivational example to other jurisdictions. Accordingly, the fact that both California and Texas have enacted laws committing to
1 Indeed, it is a curious quirk in the letter of Ohio state law that appears to put higher priority on reporting crime to the national UCR program than to the state coordinator.
replace UCR Summary submissions with NIBRS format ahead of the FBI’s 2021 sunset date is remarkable—Texas even more so given its longstanding previous reputation as having no plans for NIBRS implementation whatsoever. Indiana has also required all local agencies to report in NIBRS format by mid-2018. Illinois’s new crime reporting statute hints at conversion statewide, Chicago having long been a holdout and having made substantial investment in overhauling its internal information management systems.
Going further along these same lines, our review of the state language does not purport to be exhaustive; it may miss references to information and data requests that are peppered throughout the definitions of offenses in state law, for instance. Still, we list in the appendix offense types that are identified as particularly high priorities for crime measurement when those are apparent in state reporting law. Domestic violence, drug-related offenses, computer-involved crime, human trafficking, and other specific offenses are all issues that one or more states have legally insisted be addressed in the state’s crime reporting program. These and other emerging examples—perhaps most notably New York’s statutory call for environmental offenses and insurance fraud to be treated commensurately with other crime types in the state UCR program’s annual reports—suggest that state-level interest in the substance of modernized crime statistics (and the “new” crime types) in particular may be encouraging grounds for more active partnership work.
In Section D.2, we will discuss in greater detail the notion of data quality as the primary “value added” to local law enforcement stakeholders (and data users)—as the thing that makes it beneficial to local agencies’ routine operations and thus motivates cooperation with the national crime statistics program. Along this line, the review of state legislative language is also useful to consider; data quality assurances may resonate well with states like Georgia, Illinois, and South Carolina, which have particularly strong record-audit or quality-control-inspection provisions built into their state codes.2
Two other comments are motivated indirectly by the summaries of state law in Appendix E. First, an observation based on what is not covered by the appendix is that similar brokering arrangements with state-level coordinators might be appropriate to manage data inputs from relevant state, local, or private authorities, if those records aren’t immediately or already directly shared with some federal counterpart. For instance, state environmental quality or consumer protection agencies may be important sources of records on environmental and consumer fraud offenses, drawing their information from local/municipal or other sources. In some cases, these agencies may be legally required to report to a state-level entity in much the same way as local police are
2 It is useful to see some other examples spelled out in state code, including Connecticut’s provision that makes the FBI (and the national UCR program) the arbiter of whether crime data submitted to the state are incomplete or incorrect.
commonly obligated to report UCR data to the state. Similarly, there are some offenses such as fraud against the government (e.g., fraudulent medical benefit claims) in which state attorneys general or inspectors general may be the only potential, viable source—likewise for highly state-differing policies on gambling or alcohol enforcement.
The difficulty of parsing the language of state law occasions the second comment, which is that the construction and maintenance of crosswalks between offense codes in different data systems are essential functions. Our proposed classification uses behavior-based definitions as a workaround to the intricate detail in state criminal code language; we define offenses based on the behavior and not on trying to find some lowest-common-denominator wording across dozens of state criminal codes. We believe in this approach, with an ideal vision for crime statistics being one in which data providers (including police officers generating reports or survey respondents answering questions) never have to go through the cognitive process of recalling codes and deciding which classification labels are most appropriate to the incident at hand. That said: To be useful in daily operations, every state and local law enforcement information management system has to be able to handle and work with offense codes, based on state (and local) law, for those offenses for which officers have arrest power and departments have booking power. Difficult as the task of finding and referencing passages of state law related to crime reporting is, it pales in comparison to the task of building crosswalk files between offense codes based on state law and different coding systems (e.g., NIBRS codes or our classification’s codes, plus associate attributes)—and keeping those crosswalk files up to date with enactment of new law. Building these crosswalk files exclusively at the federal level and filtering them down would be difficult and likely fraught with error. Building and maintaining them is also not something that should be imposed directly on information management system vendors/solution providers, though they will obviously be essential to incorporating them into the systems. Crosswalk construction and maintenance may be a very daunting task, but one that is only really tractable through partnership with the states.
The coordination of data transfers between multiple federal agencies is a difficult undertaking, much less the management of data inputs from a much wider variety of state, local, and private entities. This is arguably an even harder task when data contribution and sharing is premised on said contributions being voluntary in nature. We have argued that treating the general problem as a federal-state cooperative arrangement is virtually essential to making a new system work. But it follows that there may be lessons that might be learned
from other statistical agencies in other policy domains, in which federal-state compacts are fairly frequent. We believe it appropriate to mention a few illustrative programs here that might be useful to consider. In doing so, we reiterate that we know of no example federal-state cooperative that is a direct, obvious, interchangeable match to the problems of crime data collection—nor do we know of any federal-state arrangement that has uniquely solved all the myriad problems associated with data transfers—but merely that they may provide useful perspectives for framing implementation and methodological work.
A major example of a federal-state cooperative data collection program that might be a useful model to consider is the Bureau of Labor Statistics (BLS) Survey of Occupational Injuries and Illnesses (SOII).3 Businesses and establishments are required to keep records of work-related injuries and illnesses by Occupational Safety and Health Administration (OSHA) regulations, and the SOII program has served as a means of regularly compiling information from these records. Each year, a sample of employers is selected, and is requested to submit a mix of summary-count data and incident-specific detail based on their in-house OSHA records—the summary-count data applying to all workplace injuries and illnesses and the incident-level detail being requested about those cases that lead to one or more days of missed work. The responses are channeled through BLS’s network of state coordinators. One possible weakness of the SOII model for crime data collection is the frequency of collection—the data are requested once a year, in January, rather than on a monthly (more continuous basis) throughout the year. But what makes the SOII model most useful as an illustrative example is that BLS (together with its state partners) assumes the burden of classification and standardization of responses from the responding employers. In the survey’s data collection forms, critical variables including the injured worker’s actual occupation (or job title) and the nature of the injury/illness are free-text responses; the reporting business or establishment is not asked to internalize a set of classification rules and self-apply them. Instead, BLS and its partners assume the responsibility for working with reported job title and other contextual information and assigning the proper code from the U.S. Standard Occupational Classification and standardizing injury type, among others. It is a time-intensive process—businesses are alerted to their selection in the SOII sample and asked for data in January (being asked to provide data for the previous calendar year), and estimates are typically produced in October or November. Hence, the survey’s ongoing work to carry out these tasks makes it a useful case to study; the
3 Additional information on the survey is available at http://www.bls.gov/iif and https://www.bls.gov/respondents/iif/faqs.htm. The most recent data collection instruments for the survey, and additional descriptive information, is included as part of BLS’s filing for clearance by the U.S. Office of Management and Budget to conduct the survey, available at https://www.reginfo.gov/public/do/PRAViewICR?ref_nbr=201706-1220-005.
agency is constantly trying to improve computer-based autocoding of the free-text responses to make the program more efficient (see, e.g., Measure, 2014). To the extent that BLS is able to directly obtain administrative record-type data in lieu of filled survey forms—a more direct analogue to receipt of local law enforcement records rather than summary “reports”—its work to use such data in the SOII will be useful to monitor as well.
A second, longstanding federal-state cooperative program is one that is arguably closest in content to some crime data: the nation’s core collection of vital statistics. The National Vital Statistics System is the oldest and most successful example of intergovernmental data sharing in public health and is a potentially instructive example to consider because it faces some of the same structural problems as faced in crime statistics. Information from vital records are provided through contracts between the National Center for Health Statistics (NCHS) and vital registration systems operating in the 57 vital registration jurisdictions (the 50 states, New York City, the District of Columbia, and the territories) that are legally responsible for the registration of births, deaths, fetal deaths, marriages, and divorces. Unit records with information on births, deaths, and fetal deaths are, for the most part, sent to the registration jurisdiction using electronic systems from hospitals and funeral directors. They are then sent electronically from the jurisdictions to NCHS. The development of model state laws, standard certificates, and standardized methods of data collection and processing support the creation of national data files. While automating the registration of vital events increased the timeliness and quality of the data, the adoption of different hardware and software systems by the jurisdictions has made it much more complicated and expensive to change the content of the data collection. Work is underway to address the challenges of maintaining and improving electronic systems for collecting and transferring information.
A final example to consider for its organizational aspects is the dense network of regional and state field offices maintained by the National Agricultural Statistics Service (NASS). A single “state” office covers several of the New England states; otherwise, every state has a NASS state field presence. The day-to-day function of NASS field staff is to work sufficiently in concert with state agricultural departments—collecting data and items on a vast array of crop- and practice-specific collections—that they are practically embedded in the state agencies. Moreover, the National Association of State Departments of Agriculture (NASDA) that functions as an organizational arm for the state agriculture agencies has also served as a corps of field representatives and interviewers for NASS censuses and surveys.
A final note that must be made in this context is that, in the current crime statistics system, the states have had two fairly well-defined roles. The first, the state UCR programs, has already been discussed and should be expanded upon in new collection strategies. But it is also important that the second be continued as well—the network of Statistical Analysis Centers (SACs) that receive some support funding through BJS grant monies and that have long served a critical role as interpreters and analysts of criminal justice system data, including both BJS and FBI/UCR data. The SACs publish reports and research results for the public, the media, and the broader research community, but their primary (and crucial) constituency is state legislative and executive decision makers. The National Crime Statistics Exchange (NCS-X) initiative will provide useful insights that can be used to demonstrate the utility of incident-based data and encourage still further NIBRS participation. The SACs are likely better poised to perform these kinds of analyses than many individual police departments, and so their role in examining the first full NIBRS data will be vital.
Managing and enhancing the outflow of data and data products is an important part of the ideal, beginning-to-end coordination and governance roles we envision for national crime statistics. Issues related to publishing annual reports and other tabulations are crucial to the credibility of the data collection enterprise as a whole; this is an area, after all, in which the annual publication of a volume entitled Crime in the United States has constrained the scope of what is commonly viewed as crime and given an undue impression of quality and completeness. Hence, the final sentence of Recommendation 3.1, on data products and their dissemination, involves issues that deserve serious attention.
Weighing all that we have heard from data users and stakeholders, we conclude that new crime data products—and data dissemination more generally—must necessarily progress along two lines. In our assessment, there remains a necessary role for an annual, comprehensive report that reflects all crime types—yet the issuance of data and analyses should not be inextricably linked to the preparation cycle for the full-fledged annual report. Hence, the aim for crime data products should extend beyond a printed volume, and instead aim for more dynamic presentation and availability of a crime data repository that is updated on a flow basis.
There is a certain inevitability to an annual report as a partial dissemination vehicle for crime statistics. Per our examination of state crime reporting laws in Appendix E, 34 of 50 states are explicitly required to issue a crime statistics report with some frequency, typically annually, to state officials (executive or
legislative) or to the general public; though law and regulation appear silent about production of a report in the remaining states, several of them do issue annual reports nonetheless. At the national level, both Crime in the United States (UCR) and Criminal Victimization (NCVS) have been annual fixtures for decades. Issuance of the annual report is one way (though certainly not the only way) to construct time series and permit annual benchmarking/comparisons.
But the indispensable role of a comprehensive annual crime report, for purposes of nurturing a new crime statistics system, is simply that what is most necessary for the “new” and heretofore neglected-in-national-statistics offense types in our proposed classification is getting them into the same national conversation about crime as the “traditional,” index, violent and property crimes. Accordingly, the best and most logical way for that to happen is for the new and traditional offenses to be presented and discussed, with commensurate attention, in the same forum. For that purpose, an annual report that strives for some synthesis rather than simple presentation of counts is ideal. No data user or stakeholder is well served by a static, incomprehensible mass of tables published solely for publication’s sake. In contrast, a report that captures some indicators of how, when, and why both “traditional” and “new” crimes occur is beneficial, even at the expense of some more preparation time.4
Dissemination of modern crime data should also be paired with modern, dynamic data access tools, capable of handling a broad range of user data queries. The important part in introducing the two tracks for development of dissemination approaches is decoupling them—not tying release of data through online means solely to the production cycle of an annual report. Hence, an online crime data repository and access tool should be updated as often as practicable. It should be capable of generating standard tabulations, but do so in a way that conveys the degree of uncertainty and types of error associated with the estimates. It should also facilitate researchers’ access to data files and associated metadata, permitting intensive data users to conduct their own analyses offline. Ideally, it could also permit user-constructed data visualizations and graphics. These are all lofty objectives—it is easy to hope for a high-quality, national-level, at- or near-real-time dashboard, but that is truly a distant wish. As the range and content of national crime statistics grow, from an NCS-X–supplemented national NIBRS file to the fuller, multisource collection enabled by and based upon a new classification of crime, there will inevitably be a learning curve. Challenges will be encountered and iterative improvements
4 An admitted wild card in consideration of annual reporting is that a comprehensive annual report should, necessarily, combine insights from police-report, survey, and administrative-type/external sources in tandem—yet the UCR-based Crime in the United States and the NCVS-based Criminal Victimization have traditionally had staggered release dates. Over the years, the temporal separation between the reports has created the potential for confusion or mixed messages. However, whether it is possible to more directly synchronize the releases (or whether it is worth doing so) is an open question that we did not attempt to resolve.
made, but the point is to place prototypes (annual report and data access tool alike) into production and improve from there.
Finally, as the traditional UCR Summary is phased out in 2021, it is important that the annual dissemination of information that has been routinely presented in Crime in the United States continue to allow for the possibility of assessing changes over time in rates of traditionally measured crimes in ways that preserve the continuity of the statistical series. Variation across agencies and states in full NIBRS adoption should be anticipated and planned for, and information about changes in agency participation and data completeness should be fully documented so that sources of potential errors in crimes such as homicide are transparent and understood by the public.
Appendix D includes longer-form discussions of some additional important topics related to methodology and implementation, mentioned in brief in the main text but fuller consideration of which might otherwise disrupt the narrative flow. These include:
- Examining the commonly cited barriers that impeded adoption of NIBRS for 3 decades, accompanied by discussion of what has changed now to make success more likely;
- Arguing that improved data quality should be the major goal of voluntary contribution of data to national crime statistics;
- Observing that—rather than warn against overly simplistic comparisons of crime rates across jurisdictions and time—the dissemination platform for modern crime statistics should also provide easy access to the kind of contextual and community data that enables effective comparison; and
- Describing the challenges of getting a better reading on crime involving businesses and organizations as actors.
The path ahead for U.S. national crime statistics is neither easy, inexpensive, nor short in terms of time to completion. If BJS and the FBI are successful in getting a designed sample of NIBRS-responding agencies from which sound inference can be drawn to a larger whole—and if the FBI is successful in its goal in sunsetting the UCR Summary Reporting System by 2021 in favor of full NIBRS—the fact will remain that it took more than 35 years to reach that goal. To modernize crime statistics and to equip the nation to better understand problems of crime and victimization will require a large investment in collecting data on issues that are currently all but ignored in current crime
measurement efforts. An adequately resourced NCVS and a full-participation (and fully representative) NIBRS are important steps along the way, but real benefit will arise from finally starting to shed light on too-long neglected or obscured offenses, consistent with a rigorous classification.
In this report, we have suggested adding a third core crime data component: administrative-records-type data and information on regulatory violations, amassed through a crime statistics clearinghouse function. Constructing this clearinghouse is a difficult but essential task. However, in closing this report, we would be remiss if we did not at least mention an important point along these lines. Some particularly difficult-to-measure offenses (e.g., trafficking in persons or migration offenses) are very likely to elude detection in third-party, administrative-type data resources, just as many traditional offenses elude detection in the police-report, UCR/NIBRS data. In the same way that survey data (i.e., the NCVS) play an indispensable role in providing an estimate of crime not reported to police, there is a necessary role for alternative methodology for validating counts or characteristics of relevant offenses that are not reported in regulatory data. A modern crime data infrastructure should enable long-term research and development on innovative technologies and methodology for measuring crime just as it should enable such research on the utility of specific metrics, as in our Conclusion 2.2 on the harms or cost of crime. We have focused in this report on the three, major, essential pieces of a new crime data infrastructure—already a difficult goal to attain—but the longer-term research interests are vital as well.
Attribute-driven and incident-based reporting using the classification proposed by our panel means upending concepts that have been at the heart of crime data collection for decades—not least of which are a shift away from defining crime by the strict language of state penal code violations and the treatment of some offenses that carry administrative/regulatory sanctions in the same manner as those punishable by corrections. It also challenges crime data providers, collectors, aggregators, processors, and users alike to embrace some markedly different approaches—the most fundamental, and simple, being the parsing of multiple offense codes for simple incidents, relative to the cognitive oversimplification of the traditional UCR Summary that collapses each incident to a single offense. Full implementation will mean major changes in data collection forms, software, and processes, all of which take time and commitment. Yet such radical transformation becomes justified if the principles on which this classification and subsequent data structures are borne out, providing a much more robust understanding of the nature of crime and firmer ground for developing policies to combat it.
This page intentionally left blank.