TO DERIVE STATISTICS ABOUT CRIME—to estimate its levels and trends, assess its costs to and impacts on society, and inform law enforcement approaches to prevent it—a conceptual framework for defining and thinking about crime is virtually a prerequisite. Developing and maintaining such a framework is no easy task, because the mechanics of crime are ever evolving and shifting. As one example, the public disclosure in early 2016 that two hospitals in California and Texas had both made ransom payments in difficult-to-trace Bitcoin currency in order to regain control of their own internal computer networks (Cowley and Stack, 2016) raises major conceptual challenges. In such attacks involving “ransomware,” in which hackers cut off network access or basic functionality until payment is made, it is certainly intuitive that a “crime” has occurred, but few of the related questions have easy answers:
- What is the criminal action(s)? The authoring of the threatening software? The transmission of it? The causing of harm (if any) to victims’ existing networks or data resources? The demand for, or the acceptance of, the ransom?
- Who is the victim(s)? The hospital or its owners? The hospital’s (or hospital group’s) insurance company? Current patients, whose effective access to therapies may have been encumbered by the system “failure”? Future patients (or their insurers), who may face higher insurance premiums as a result of the attack?
- Who is the offender(s), and where did the offense(s) take place? Is it proper to think of the local hospital as the “scene of the crime,” or the location from which the hacker issued the attack (if such is ever determined)? Or is it more proper to think of “cyberspace” as a location outside of conventional geographic space, or even of the crime as truly “locationless?”
Assuming that answers to the above conceptual questions are decided upon, the next step—developing statistical measures for such ransomware attacks—is challenging in its own right. A “simple” count of incidents is anything but: Does embedding malicious code in an email or on a webpage constitute one incident, thousands of incidents (based on the number of email recipients targeted), or potentially millions of incidents (based on webpage browse attempts)? If an incident count fails as a metric, does an estimate of incurred loss or harm (if feasible) fare any better?
This is but one example, but it is illustrative of larger concerns. “Crime in the United States” in 2016 is, arguably, at least as much about corporate fraud as about armed robbery, harassment via the Internet as about breaking and entering, and endangering health through environmental pollutants as about assaults and muggings. But, for decades, the nation’s perspective on crime has been dominated by so-called “street crime”—violent crime and some types of property crime—to the general exclusion of non-street crime, of which the fielding of ransomware is certainly an example. The lack of systematic information about non-street crimes makes it very difficult to develop sound judgments about whether adequate resources are being devoted to these types of problems. A conceptual framework that encompasses the full range of crime is essential for drawing attention to important issues that may be ignored because they do not have the necessary statistical indicators for comparative purposes.
It is useful to begin with a brief history of how the nation’s crime statistics arrived at their current state, but to make the story short upfront: U.S. statistics on crime in 2016 are, by and large, still measured following concepts outlined nearly a century earlier in 1929, making use of a list of defined crimes that evolved from what was most feasible and tractable to measure. As we describe later in this chapter, this study is an attempt to step back and rethink the approach to the entire enterprise of crime data collection—beginning, in this report, with development of a proposed classification of criminal offenses to serve as a broad, conceptual framework for what “crime in the United States” means. This new classification and framework would then be a useful blueprint for constructing new measures of crime.
We return to the high-level conceptual questions—what exactly is “crime” and what is the value of classifying it—shortly. But first, as orientation, it is important to consider how the scope and content of today’s U.S. crime statistics came to be. To do that, one must start at a seemingly unlikely source: the U.S. decennial censuses of population of the 19th century. The American census has never been strictly a simple head-count, and the censuses of the 1800s evidenced an increasing and insatiable curiosity (of the young nation and its Congress) for information on all manner of topics, from education to industry to housing conditions. The 1850 census was particularly pivotal in both the content and the increasing professionalization of census conduct: It was the first to provide enumerators with formal rules and instructions on who to count and where to count them, and the first to make a systematic effort to enumerate the population in jails and prisons. Because the new instructions directed enumerators to record “the crime for which each inmate is confined, and of which each person was convicted” (Gauthier, 2002), the 1850 census became a critical step in giving the nation its first indicators of the nature of crime in the United States. It was, to be sure, a very tentative step—so much so that, as events transpired, the compendium volume from the 1850 census makes no apparent use of the detailed offense information, instead analyzing overall levels of convictions by race, native/foreign-born status, and other variables. Nonetheless, crime emerged as a topic to be studied—and measures of it as something to be improved upon—in subsequent decades.
The novelty of census statistics related to crime could not long mask a crippling conceptual shortfall inherent to them: Being based on the reports of prisoners already convicted, the data were necessarily unable to describe current crime conditions. This lack of timeliness was exacerbated by the basic fact that, in the pre-computer age, tabulating and analyzing detailed census items simply took a long time—to the extent that the 1880 census report covering crime content was not published until 1888 (Wines, 1888). Amidst these frustrations, calls for systematic collection of crime data closer to the source began to surface. A resolution passed at the 1871 convention of the National Police Association—forerunner to the International Association of Chiefs of Police (IACP), founded in 1893—is commonly considered the originating spark for systematic collection of statistics of crimes reported to or otherwise known by the police in the United States (Maltz, 1977). One year earlier, more contemporaneous collection of relevant data—likely from police—appeared to be the legislative intent behind a provision in the act that created the permanent U.S. Department of Justice. That enabling law directed that the Attorney General’s annual report to Congress on the business of the Department should include “any other matters appertaining thereto that he may deem proper”—
particularly “the statistics of crime under the laws of the United States, and, as far as practicable, under the laws of the several States” (16 Stat. 164).1 However related, or not, to these external pushes, the 1880 census brought with it an attempt to obtain crime/offense data nearer the source. It included among its “social statistics” schedules a report intended to be completed by every police department; the form asked dozens of specific inquiries, among them a request for a complete accounting of arrests and dispositions for 24 categories of crime in the preceding year (Wright, 1900:218–221). In the end, this ambitious 1880 police statistics schedule yielded a tabulation for all cities of population 5,000 or greater, but one that provided specific offense-detail for only two crime types asked about in a separate question (namely, homicide and “fires supposed to have been incendiary in origin”/arson);2 a somewhat pared-down version of the schedule was repeated in the 1890 census.
For decades, a relative stalemate continued: The 1870–1871 calls for routine and systematic police-report statistics went effectively unanswered, while the census continued to mine prisoner-reported offenses most extensively (albeit at the slow, decennial rate). To a large degree, this stunted progress was due to a common conceptual difficulty, hinted at by the language of the Justice Department’s establishing act: State criminal laws vary so greatly that a common definition of “crime” and crime types is not particularly practicable. The special agent in charge of generating statistics on prisoners and other institutionalized persons in the 1880 and 1890 censuses, Wines (1888:XLVI–XLVII) bemoaned the “deplorably meager and inadequate” state of U.S. crime statistics, which he attributed principally to “the relegation of this particular function [dealing with crime] to the governments of the several States.” His review of the resulting state criminal statutes “shows the most striking and illogical variations, not only in respect of the definition of crime, but in the methods of dealing with them.”3 In the absence of common definitions across states, census-takers erred on the side of faithfully capturing and rendering the information reported to them—with no eye toward standardization or correcting for redundancy. Hence, by the 1880 census,
1Over time, this Justice Department reporting function morphed into the more general authorization for the Attorney General to “acquire, collect, classify, and preserve identification, criminal identification, crime, and other records” found in today’s 28 U.S.C. § 534(a)—which remains the full authorizing text of the Uniform Crime Reporting program.
2The tabulation included a collection of totals from other questions on the special police-department schedule, including overall number of arrests, counts of liquor saloons and houses of prostitution, and rough annual cost of police department operations.
3“None of the founders of the republic seem to have appreciated the benefit to posterity of complete information respecting crime and criminal procedure, in tabular form, year-by-year, upon a uniform system, such as to admit of easy and instructive comparison,” Wines (1888:XLVII) wrote, adding his voice to the calls for systematic crime data collection. “I venture to express the wish that the government, through its Department of Justice, or some special bureau of the Department of the Interior, could be induced, even at this late date, to begin the collection of criminal statistics from all the states and territories, and their publication in an annual blue-book.”
the special Report on the Defective, Dependent, and Delinquent Classes (Wines, 1888) covering prisoner-reported offenses tallied 231 separate offense categories, disaggregated by race and sex at the national level and reported as state-level counts. These offenses were categorized into six large groups—offenses against the government, against society, against the person, against property, on the high seas, and miscellaneous—that were described only as very general headings with no further rationale for their construction. A decade later, the 1890 census Report on Crime, Pauperism, and Benevolence in the United States (Wines, 1896) attempted a rough gradation of prisoners’ reported offenses and retained the same broad tiers but made a more systematic (if similarly undocumented) attempt to designate 3–9 main groups within those tiers—and dozens of highly specific offense categories within each group.4 Under the heading of “assault, all sorts” alone, the 1890 census tallied charges for numerous adjectives attached to assault (e.g., atrocious, felonious, indecent, and riotous assaults), by all manners of weaponry (e.g., assault by ax, by vitriol, or by pistol), by apparent intent (e.g., to kill, to commit rape and murder, or to rob, rape, and murder), by whether there was actual battery or wounding—and numerous combinations thereof. The end result may be described as page after page of neatly rendered but very sparse tables—all of which could spur an endless succession of questions but none of which could really provide useful information about the level and extent of criminal behavior nationwide.
Interest in understanding crime surged in the 1920s, which proved to be a pivotal decade for the collection of nationwide crime statistics. Now established as a permanent agency, the Census Bureau had gradually stepped up its work in the general field—beginning to examine commitments to prison/jail (and the charges associated with them) on a full calendar year basis (rather than point-in-time snapshot), and issuing as-yet-unprecedented detail on flows into and out of corrections in a series of special reports in 1923 (preceded by “preparatory” surveys in 1917 and 1922; Cahalan, 1986:2–3). Emboldened, the Census Bureau commissioned the drafting of a manual for preparing crime statistics—intended for use by the police, corrections departments, and courts alike. Releasing the manual, the Bureau noted that—“in response to a general demand for more complete and satisfactory statistics relating to crime and to criminals”—it would be “undertaking to collect such statistics to a limited extent annually hereafter.” The focus of this series would remain collection of data from prisoners, but the new work was meant to “supplement rather than supersede” the decennial
4Noting that the six categories were the same as used in 1880, Wines (1896:139) opined that “this classification, like all other attempted classifications of crime, is only partially satisfactory. The classifications in the criminal codes of the several states do not correspond with each other, and in a number of codes all attempt at classification has been abandoned. . . . The offenses charged are stated in detail precisely as reported by the enumerators [and are] not improbably copied from the commitment papers by the courts” onto official prison records, used to count most of the prisoner population.
- Manslaughter (all degrees)
- Murder (all degrees)
- Abuse or carnal abuse of infant, child, etc.
- Assault with intent to commit rape
- Burglary with intent to commit rape
- Defiling child, minor, etc.
- Indecent assault
- Indecent liberties with child
- Rape, ravishing, or statutory rape (all degrees)
- Assault with intent to rob
- Automobile banditry
- Highway robbery
- Hold up
- Robbery (all degrees)
- Robbing mail, post office, bank, etc.
- Assault (all degrees)
- Assault and battery
- Assault with intent to commit murder or manslaughter
- Assault with intent to kill
- Attempt to commit murder or manslaughter
- Attempt to kill
- Pointing gun
- Shooting with intent to kill
- Wife beating
- Breaking and entering
- Burglary (all degrees and without degree to time of day or night, or nature of structure entered)
- Having burglars’ tools
- Safe blowing
- Unlawful entry
- Bogus checks
- False checks
- Forgery and forging (all degrees)
- Possessing forged instrument
- Larceny [. . .]
- Embezzlement, all degrees, including—Appropriating money, etc.; Breach of trust; Fraudulent conversion; Larceny by agent, bailee, trustee, or other fiduciary; Misappropriation
- Fraud, including—Cheating; Defrauding; Drawing check without funds; False pretenses; Illegal sale; Obtaining property by false pretenses; Selling property held under conditional sale; Swindling; Using mail to defraud; Worthless check
- Having stolen property, including—Automobile theft; Bringing stolen property into State; Buying stolen property; Receiving stolen property
- Larceny, all degrees, including—Common theft; Grand larceny; Petit larceny; Pocket picking; Shoplifting; Stealing (including stealing any animal or other property); Theft
- Carrying weapons
- Carrying weapons
- Carrying concealed weapons
- Carrying dangerous weapons
- Unlawful possession of weapons
- Sex offenses, except rape
- Bigamy and polygamy
- Crime against nature
- Keeping house of ill fame
- Securing or transporting women for immoral purposes
Nonsupport or neglect of family
- Abandoning children, family, wife, etc.
- Contributing to delinquency of child
- Failing or refusing to provide for children, family, wife, etc.
- Nonpayment of alimony
- Wife desertion
- Violating drug laws
- Possession of narcotics
- Selling cocaine, morphine, or other drugs
- Violation of Harrison Act
- Violating liquor laws
- Carrying liquor
- Possession of liquor, unlawful
- Prohibition law, violating
- Selling liquor
- Driving while intoxicated
- Drunken driver
- Drunken driving
- Disorderly conduct and drunkenness
- Drunk and disorderly
- Habitual drunkenness
- Disorderly conduct
- Breach of peace
- Disorderly conduct
- Disorderly conduct and vagrancy
- Disorderly person
- Soliciting alms
- Suspicious character
- Violating traffic or motor-vehicle laws (Include here all traffic or motor-vehicle law violations except those involving larceny or attempted larceny of motor vehicles, and possession of stolen automobiles (which should be placed under heading 7); and except driving while intoxicated (which should be placed under heading 13)
- Violating municipal ordinances (Include here those violations of ordinances which cannot be assigned to any of the headings above)
- Contempt of court
- Criminal anarchism or syndicalism
- Drug addiction
- Malicious mischief
- Quarantine (venereal)
- Using another’s property
- Violating parole
- Violating revenue laws
- Violating United States penal laws
- All other
NOTE: The 19 main items, dubbed an “abridged list of offenses,” were intended as the main body of the classification; the specific items listed under each are “important specific offenses” that would be grouped under the headings.
SOURCE: Excerpted from U.S. Bureau of the Census (1927).
census of prisoners; the annual survey would “cover only a few outstanding facts [and] leave the more detailed investigations to be made once in 10 years” (U.S. Bureau of the Census, 1927:ii).
Importantly, the new manual sought to solve a perennial problem by suggesting a standard taxonomy of crime, as shown in Box 1.1. The suggested taxonomy suggested 19 top-level offenses, each with some subset of specific offenses beneath it; the manual characterized the 19-item list only as “about as short as a significant list can be made, though not all the headings are needed by every agency which may use the list,” giving no further insight into its derivation. Accordingly, the manual raised “no objection to subdividing the offense groups” as desired but strongly discouraged violating the basic categorization “as that would make it impossible to compare accurately or to combine the statistics from different States.” We will return to this terminological point later on, but it should be noted that the 1926 manual described the proposal as a “classification”—which it is, in the general sense of “classifying” things being a process of assigning labels to them in some structured way, yet it falls short of the full set of expectations for a rigorous “classification.” What the manual did stress was the importance of uniformity in concept, noting that “officials concerned with law enforcement, who are well acquainted with the wide variation in the terms used in different parts of the country in describing similar offenses, will appreciate the importance of a standardized classification like this and of strict adherence to the classification” (U.S. Bureau of the Census, 1927:5). The taxonomy offered radical simplification of some detail relative to the crime “code lists” of the 1890 and other censuses—consolidating legally defined degrees of crimes like manslaughter and robbery into single categories—while stopping short of a unified, nonredundant set of crime types. The Bureau’s proposed taxonomy also attempted to list the 19 major crime types in rough descending order of seriousness—a first attempt at a hierarchy rule, under which “an offender who is imprisoned for two or more offenses should be classified under the heading which stands nearest to the top of the abridged list,” which is to say the most serious of the committed offenses (U.S. Bureau of the Census, 1927:6).
Developments accelerated on police-report data, as well. Shortly after the Census Bureau issued its manual, the IACP in convention adopted a resolution to create a Committee on Uniform Crime Records—to begin the process of describing what a national system of data on crimes known to the police might look like, nearly 60 years after the first calls for the same. That committee of police chiefs issued its report in 1929, arguing stridently for the necessary link between police-report data and crime statistics (International Association of Chiefs of Police, 1929:vii):
Compilations of the number of persons tried, convicted and imprisoned do not, and cannot, provide an index of crime and criminality. Only a police
Part I Classes
- Felonious homicide
- – Murder and nonnegligent manslaughter
- – Manslaughter by negligence
- Aggravated assault
- Burglary—breaking or entering
- – $50 and over in value
- – Under $50 in value
- Auto theft
Part II Classes
- Other assaults
- Forgery and counterfeiting
- Embezzlement and fraud
- Weapons; carrying, possessing etc.
- Sex offenses (except rape)
- Offenses against the family and children
- Drug laws
- Driving while intoxicated
- Liquor laws
- Disorderly conduct and vagrancy
- Traffic and motor vehicle laws
- All other offenses
SOURCE: International Association of Chiefs of Police (1929:24–25).
record of known offenses will do this. Likewise, the police alone know the number and nature of persons apprehended for committing offenses. Thus at the start we are compelled to recognize that crime statistics must originate with the police and that without police support there can be no crime statistics. To launch upon a project directed towards that end without consulting police needs and capacities, is utterly futile. This has been repeatedly demonstrated.
The IACP committee noted intent to base its own proposed uniform classification “directly upon that employed by the Bureau of the Census in compiling statistics of prisoners” while simultaneously criticizing “certain manifest weaknesses” in the Census classification. Chief among these weaknesses was “its failure to recognize differences in statutory definitions” across the states (International Association of Chiefs of Police, 1929:5). Curiously, in laying out its own proposed classification—the taxonomy shown in Box 1.2—the committee would go even further in smoothing over state-level variations in definition than the Bureau. The IACP committee’s “classification” was even further removed from a rigorous classification for statistical purposes in that it pointedly did not attempt to cover the entire range of what might constitute “crime.” Rather, the committee’s stated rationale was strictly pragmatic: Its “Part I” offenses were those that “seem to make possible the compilation of uniform and comparable returns of offenses known to police,” being both salient in the public eye and having variations in state-by-state
definitions that, “though rather wide in certain cases, were not irreconcilable” (International Association of Chiefs of Police, 1929:180,4). In short, the offenses branded “Part I” crimes in 1929 were so designated because of their perceived severity and because it was believed that some common-denominator definition—spanning sometimes considerably wide definitions and degrees—could be derived for them. The listing was also meant to “[set] apart those grave offenses which experience has shown to be most generally and completely reported” to police (International Association of Chiefs of Police, 1929:6). Conversely, “Part II” offenses were judged to fall short of that standard: “Either they are concealed when committed so that the police frequently do not know they occur, or else their statutory definitions vary to such a degree that it would be impossible to obtain reasonably comparable results” (International Association of Chiefs of Police, 1929:180). Importantly, different reporting criteria applied to the two groupings of crime: The Part I offenses would be the ones for which “offenses known to the police” and completed arrests would be recorded; only arrests would be logged for Part II offenses.
The principal reason for the highly pragmatic, relatively limited scope of the identified Part I and II crimes in the 1929 IACP plan was the intended architecture for data collection. Specifically, the committee argued for a system by which local departments would voluntarily contribute police-report data directly to a coordinating entity within the federal government, for compilation and analysis. As to which federal agency should fill this role, the IACP in convention in 1929 “decided to invite the National Division of Identification and Information in the Department of Justice” to do so (International Association of Chiefs of Police, 1929:12)—a division housed within a Bureau of Investigation, a Bureau whose director (J. Edgar Hoover) had held the post for roughly five years and that was still six years removed from acquiring its current designation as the Federal Bureau of Investigation (FBI). “The only other unit which might conceivably operate the [crime statistics] system,” wrote the IACP committee, “is the Bureau of the Census”—but that idea was deemed impractical (International Association of Chiefs of Police, 1929:12–13):
While the statistical techniques which [the Census Bureau] controls are of a superior order, it lacks those intimate police contacts which the Department of Justice can provide and upon which the success of a national system for reporting offenses and offenders must ultimately depend. Since the police of this country may either grant or withhold crime returns, it is clear that without police cooperation there can be no crime statistics. The real problem with which we are here confronted is one of maintaining [sic] police cooperation. Lack of it has doomed all earlier efforts to failure.
Two years later, the National Commission on Law Observance and Enforcement—better known as the Wickersham Commission, after chairman and former U.S. Attorney General George W. Wickersham—issued a “Report on Criminal Statistics” that arrived at the opposite conclusion. The commission argued that “the plan of direct sending of voluntarily gathered material from the local police to a central Federal bureau can not be the ultimate plan.” It favored gradual adoption of mandatory reporting (under state laws)—and favored statistical expertise in processing the returns (National Commission on Law Observance and Enforcement, 1931:16–17):
It is not only important to provide criminal statistics, it is quite as important to see to it that misleading information is not sent out under official auspices and with the imprimatur of the Government. Statistics require experts to analyze, interpret, and compile them, as well as to provide and revise the plans for gathering them. A Bureau of Statistics in the Department of Justice would do for Federal criminal statistics what a Central Bureau in each State should do for State statistics. But there would remain the task of working all into a unified system for general purposes [and reasons] for committing this task ultimately to the Bureau of the Census seem decisive.5
However, the commission’s arguments did not affect wheels already set in motion: Collection of the first set of Uniform Crime Reporting (UCR) program data was initiated by the IACP in 1929, with authority transferred to the eventual FBI in 1930, where it has remained since. In the meantime, the Census Bureau’s regular collection of data on crime continued into the early 1930s before being discontinued, and the special Census of Institutions connected to the 1940 decennial census made it the last decennial count to ask criminal or offense status.6
Importantly, the 1929 UCR guidelines and designations—and their relatively unchanging nature over time—influenced complementary and successor crime data collection efforts:
- In 1957, the FBI engaged a three-member Consultant Committee chaired by sociologist Peter Lejins “for the purpose of making suggestions” concerning the UCR Program, with the charge to the committee
5Specifically, the reasons were outlined in a companion paper by Warner (1931), presented in the same volume as the commission’s criminal statistics report; Warner was also the lead developer of the Census Bureau’s crime statistics instructions in 1926.
6That said, Title 13 of the U.S. Code—current census law—was codified in 1954, based on the legislation that had authorized the 1930 and 1940 censuses, and many sections of the law have remained untouched. Hence, the oddity that the Census Bureau arguably has a stronger basis in current law for the collection of crime data than the FBI: 13 U.S.C. § 101 authorizes the Secretary of Commerce (and, correspondingly, the Census Bureau) to “collect decennially statistics relating [to] crime, including judicial statistics pertaining thereto” and, further, to “compile and publish annually statistics related to crime.”
adding that “these suggestions may cover any phase of the collection or publication of the data that comes to mind in the light of the experience gained thus far” (Federal Bureau of Investigation, 1958:9). However, that committee largely constrained its review of the existing Part I and Part II UCR classification to some options suggested by the FBI, recommending that manslaughter by negligence and petty larceny (value below $50) be stricken from the existing Part I crimes (from the 1929 classification shown in Box 1.2) and that arson not be added to either listing.7 These changes—motivated by the objective of constructing a simplified “crime index” indicator—were adopted in UCR tabulations beginning in 1958.
- The ongoing National Crime Victimization Survey (NCVS) organized by the Bureau of Justice Statistics (BJS) was first fielded in full in 1972 (then as part of the National Crime Surveys), fulfilling a recommendation of the President’s Commission on Law Enforcement and Administration of Justice (1967:x) “to develop a new yardstick to measure the extent of crime in our society as a supplement to the FBI’s Uniform Crime Reports.” The survey was intended precisely to shed light on the so-called “dark figure of crime” comprised of incidents not reported to the police (Biderman and Reiss, 1967), and would over time add supplements and questions to measure phenomena uniquely suited to the personal interview approach. Though some argued for tailoring the survey content to crime types where police records are not suitable (see, e.g., Biderman, 1966), the survey designers opted to pattern the main core of the survey’s underlying concept of crime after what was then being collected by the UCR Program. In part, this decision was made to satisfy one of the new survey’s primary objectives: to act as a complement to (and potential calibration against) UCR-based, police-report crime totals.
- When researchers and the FBI developed a “blueprint” for a next-generation UCR—what would become the still-not-yet-fully-realized National Incident-Based Reporting System (NIBRS)—the list of crimes covered by the new, more detailed system was substantially augmented (particularly the array of Part II offenses), and some conceptual changes suggested (e.g., expanding the rape category in UCR’s Part I listing to include sexual assault more broadly). However, in this effort, wholesale overhaul and modernization of the crime classification was briefly considered but not adopted (Poggio et al., 1985).
7Manslaughter by negligence was singled out for removal from the Part I list mainly because “over 99 percent of all [such] cases . . . are made up of traffic fatalities which are attributable to culpable negligence”—arguably, a considerably different phenomenon than a component of a violent crime-intensive “crime index” (Federal Bureau of Investigation, 1958:25). The inclusion of petty larceny was challenged mainly on grounds of “consistency of reporting,” whether shopkeepers or other victims would actually report very minor thefts to the police (Federal Bureau of Investigation, 1958:27).
To measure “crime” one must first define it—and that is no easier a conceptual task in 2016 than it was in 1929 or earlier. There are at least two general approaches to defining crime, and both sound simple while masking bewildering complexity. This general divide is illustrated in the language of the Model Penal Code (MPC)—published by the American Law Institute in 1962 in an effort to suggest standardization in U.S. state criminal codes and which at least informed the revisions of several states’ codes. Explicitly, the MPC holds that crime is “an offense defined by this Code or by any other statute of this State, for which a sentence of [death or of] imprisonment is authorized” (American Law Institute, 1985:§ 1.04).8 The first clause of this definition is the simple, obvious, literally legal definition—“crime” is activity that is unlawful, either the commission of something that is explicitly banned or the failure to do something that is explicitly mandated by letter of the law. But complexity sets in with the second clause, which modifies the first and narrows its focus: the full thought becomes “crime” is that activity that is both unlawful and subject to certain punishments or sanctions. Black’s Law Dictionary puts the point more succinctly, defining crime as “an act that the law makes punishable” (Garner, 2014).
What this basic legal definition leaves open are the questions of exactly which law and which degree of punishment are used to define crime, and the answers to those questions vary greatly in the U.S. system. Roughly, the challenge is delineating “criminal” law, procedure, and adjudication from “civil” or “regulatory” concepts, and that line is far from sharp. The MPC language—echoed in some states—uses the punishment of incarceration as the criterion: behavior that is deemed punishable by incarceration is crime, but behavior punishable only by other means (e.g., fine or forfeiture) is not. However, for example, Wisconsin broadens the scope of crime to include those with only financial punishment, specifying that crime is “conduct which is prohibited by state law and punishable by fine or imprisonment or both” but adding that “conduct punishable only by a forfeiture is not a crime” (2011–12 Wis. Stats. § 939.12). California’s definition, which dates back to 1873, goes further, melding the concept of crime with that of a “public offense” and so including offenses of a political variety (California Penal Code § 15):9
8This definition is etched, in full or in part, in some states’ criminal codes. For example, Wyoming Criminal Code 6-1-102(a) stops with the first clause (“No conduct constitutes a crime unless it is described as a crime in this act or in another statute of this state.”), while Washington Criminal Code § 9A.04.040 invokes the full text less the death penalty option (“An offense defined by this title or by any other statute of this state, for which a sentence of imprisonment is authorized, constitutes a crime.”)
9Oklahoma Criminal Code § 21-3 stipulates a nearly identical definition of “crime or public offense,” save for minor syntax differences. Michigan Penal Code § 750.5 takes a similar approach
A crime or public offense is an act committed or omitted in violation of a law forbidding or commanding it, and to which is annexed, upon conviction, either of the following punishments: 1. Death; 2. Imprisonment; 3. Fine; 4. Removal from office; or, 5. Disqualification to hold and enjoy any office of honor, trust, or profit in this State.
Several states demur on stating a core definition of “crime” and instead delve directly into distinguishing offenses based on severity of offense (including aggravating or contributing factors to the act, such as weapon involvement) or the extent of imprisonment or punishment. Hence, the common differentiation between felonies, misdemeanors, petty offenses, or some other general infractions occurs, as well as the typically numbered degrees attached to offenses. But the criteria for these gradations of offense types vary by jurisdiction, and so the concept of what behavior is thought of as crime (or perhaps most crime-like) varies as well. For instance, Vermont statute holds that “[a]ny offense whose maximum term of imprisonment is more than two years, for life or which may be punished by death is a felony. Any other offense is a misdemeanor” (13 Vermont Stat. 1).10 Meanwhile, Virginia code uses “felony” to denote a crime subject to a prison term of any length, and explicitly excludes the broad class of traffic offenses from designation as crime (Virginia Criminal Code § 18.2-8):
Offenses are either felonies or misdemeanors. Such offenses as are punishable with death or confinement in a state correctional facility are felonies; all other offenses are misdemeanors. Traffic infractions are violations of public order [defined elsewhere in law] not deemed to be criminal in nature.
Still other state codes take different approaches: Connecticut Penal Code § 53a-24 distinguishes between “crimes” and “offenses,” dividing the former into felonies and misdemeanors and setting aside “violations” as “every offense that is not ‘crime’ ”;11 Indiana’s criminal code adds a clause including “a delinquent act” as “crime” for purposes of the Victim Rights article of the code (IC 35-40); Colorado Revised Statutes 18-1-104 take “offense” and “crime”
but does not include the concept of public offense; under that state’s law, “‘crime’ means an actor omission forbidden by law which is not designated as a civil infraction,” punishable by one or more of the sanctions mentioned in the California language save that “death” is not an option (“other penal discipline” is) and “fine” is clarified to “fine not designated a civil fine.”
10Between 1948 and 1984—when the passage was repealed—U.S. federal criminal code maintained a similar distinction: “any offense punishable by death or imprisonment for a term exceeding one year is a felony;” “any other offense is a misdemeanor;” and “any misdemeanor, the penalty does not exceed imprisonment for a period of six months or a fine of not more than $500, or both, is a petty offense” (62 Stat. 684). Law enacted on October 30, 1984, revised the $500 criterion to “$5,000 for an individual and $10,000 for a person other than an individual” (98 Stat. 3138)—but the entire definitional section (18 USC § 1) had been repealed 18 days earlier via a continuing appropriations bill (98 Stat. 2027).
11“Conviction of a violation shall not give rise to any disability or legal disadvantage based on conviction of a criminal offense” (Connecticut Penal Code § 53a-24).
as synonymous and subdivides offenses among one of 18 felony (drug or nondrug), misdemeanor (drug or nondrug), petty offense (drug or nondrug), or unclassified categories.
Recitation of the legal text may be dry, but it is undeniably vital in defining crime—and it makes clear the challenges of working within a measurement framework defined strictly by the language of federal and state criminal codes. Hence, the second approach illustrated by the language of the MPC, which is to emphasize the general type of behavior that might be said to constitute crime. The first-stated guiding principle given by the authors is that the Code is intended to “forbid and prevent” “conduct that unjustifiably and inexcusably inflicts or threatens harm to individual or public interests” (American Law Institute, 1985:§ 1.02). This language serves as an implicit, behavioral-rather-than-legal definition of crime: roughly speaking, “crime” is a class of socially unacceptable behavior that directly harms or threatens harm to others. Like the legal definition, this thread has also frequently been woven into state criminal codes—though substitutions in wording hint at the complexity inherent in this behavioral application. For instance, Texas Penal Code § 1.02 repeats this language, albeit substituting “causes” for “inflicts” and revising “individual or public interests” to “those individual or public interests for which state protection is appropriate.” Washington Criminal Code § 9A.04.020 states the first principle of construction of the code as forbidding “conduct that inflicts or threatens substantial harm to individual or public interests.” Florida Criminal Code § 775.012 eschews “unjustifiably or inexcusably,” stating that the code is meant to prohibit “conduct that improperly causes or threatens substantial harm to individual or public interest.” More than mere semantics, these substitutions in language raise difficult questions in operationalizing a common definition. How “substantial” must the real or threatened harm be before the action constitutes a “crime”? How palpable or immediate must the threat of harm be to qualify as “crime”? And—akin to the blurred line between action that is criminal and that which is civil/regulatory in the purely legal definition—how broad are “individual or public interests,” and which qualify as those “for which state protection is appropriate”?
The point may seem basic but is undeniably important: The definition of “crime” is and must be dynamic in nature, because crime is tied to shifts and development in technology, society, and legislation. Just as some crimes such as motor vehicle theft and carjacking were made possible only by the invention of the automobile, so too did the creation and emergence of the Internet spur all manner of new behaviors—some of which are indisputably “crime.” Other incidents, such as those known as “hate crimes,” existed long before there were legislative efforts to recognize and designate such incidents as criminal acts, but come into sharper focus with shifts in social norms and expectations. Some behaviors, such as marijuana possession and use, have involved many legislative
actions to both criminalize and decriminalize such acts over time, and these laws currently exhibit important variations across jurisdictions.
Considering “crime” as a unit of analysis raises further complexity. In the simplest case—thinking of “crime” as an action by one party against some other actor (whether another person, another business/institution, or society at large)—one has to recognize an essential duality: Legal criminal codes may appropriately treat “crime” and “offense” as synonymous, yet “crime” and “victimization” are also inherently linked. Accordingly, thinking of crime from the perspective of its victims can affect what one labels “crime” and how one tries to measure it. Of course, “crime” is not a strictly one-to-one action; broader “incidents” of crime may involve one (or multiple) offender(s) taking one (or multiple) crime-type action(s) against one (or multiple) victims. Hence, measuring “crime” is not completely equivalent to measuring (or counting) crime “incidents.” Moreover, some specific “crime” types are serial in nature and may best (or only) be thought of as processes over time, stalking or harassment being examples of such pattern-of-conduct crimes. A very fundamental measurement concept relative to crime—one with major implications for the scope of this study—is that “crime” also takes place in the context of a broader justice system, in the various stages of which different labels and structures may apply. So the same “crime(s)” become “arrests,” booking “charges,” “arrestable offenses,” or investigative “cases” to law enforcement officers; judicial “cases,” counts of “charges,” and grounds for sentencing criteria in the courts; and charges of conviction in the correctional system. Each of these additional different labels are countable and capable of analysis, but each has different scope and potentially different underlying definitions—and arguably serve as better seeds for measuring other phenomena (such as law enforcement effectiveness or judicial system throughput) than for measuring the level of criminal activity. Yet it is also true that some of these alternative labels and corresponding measures might be the best, if not the only, way to get a reading on some types of criminal behavior—for instance, white-collar offenses such as embezzlement or some types of fraud may only appear to be potential “crime” when charges are rendered or arrests made, much later in the law enforcement and investigatory processes than crimes such as homicide or burglary.
Given this extensive degree of diversity of concept and potential misalignment in definition, one thing that must be said clearly of the historic (and existing) UCR program standards is this: The degree of standardization in concept and reporting style that the UCR program was able to impose beginning in 1929, first under the IACP and continuing under the FBI, across widely disparate law enforcement agencies contributing information on a strictly voluntary basis, was and remains a phenomenal accomplishment. That said, the problem is that the “uniformity” that the UCR has achieved has been greater in concept than in practice—and, in concept at least, has arguably
worked too well. Useful standardization rigidified over time—highlighting, in the UCR’s short list of Part I offenses, a set of important crimes to be sure, but not necessarily the most important or most salient (to the American public) crimes—and so the basic features of UCR measurement remain largely the same nearly 90 years later. The FBI annually releases tabulations of UCR data in the broad-titled Crime in the United States series, and has done so for decades—so it is not surprising that discussions of “crime in the United States” tend to put great weight on the UCR numbers and largely follow the contours of UCR’s Part I offenses. As described in detail in Chapter 2, changes to Part I offenses (and UCR content, generally) typically require either enacted legislation or years of vetting through an elaborate advisory process. Consequently, it has been difficult for the UCR program (and corresponding crime statistics) to nimbly adapt to the wider range of actors (e.g., offenses by and against businesses and institutions) and offense types that characterize contemporary crime.
Some summary points from the preceding discussion are useful in outlining the directions we take in this report:
- For purposes of developing a modern crime classification, we think it most appropriate to take the criminal offense as the most fine-grained unit of analysis—and, specifically, to emphasize behavioral definitions of individual offenses rather than rely exclusively on the language of statute and criminal law.
- Though we may classify and think of crime in terms of specific offenses, the practical unit of analysis on which we concentrate is the incident. Incidents of crime can be very complex—comprising one or more criminal offenses and “linking” one or more offenders with one or more victims, typically but not necessarily occurring within a tight window of time and physical space.
- While “simple” counts of offenses or incidents are the most common end statistics, the framework we propose does not rule out other measures besides counts as being more useful or applicable for some crimes. That is, there are other measures related to offense behaviors (such as estimates of damage or financial loss inflicted, or even the perception of victimizations or occurrences) that may have greater salience for some crime types.
In addition to the above points, we argue that it is both beneficial and essential to recast the enterprise of crime data collection by constructing a rigorous modern classification of criminal offenses. We will argue further, in Chapters 4 and 5, for a classification structure that is attribute-based to the greatest extent possible—most prominently through collection of measurable attributes related to the offense as a coequal part to the structured listing of offense categories. The combination of a classification table with the collection
of attribute information provides the essential flexibility to work with a variety of data resources and collection methodologies (not just reports generated by law enforcement officers or questionnaires completed through personal interviews) that we describe in Chapter 2 and to attempt to meet the myriad demands and desires of users of crime statistics that we describe in Chapter 3.
The key distinction between the rigorous classification we will propose and the “classifications” that have come before in U.S. crime statistics is that it is intended to partition the entirety of behaviors that could be considered criminal offenses into mutually exclusive categories. The census-related collections of crime data in the 19th century were premised on grouping prisoner-reported offense descriptions into rough, unstructured categories. The UCR framework developed in 1929 set out to impose more structure based on perceived seriousness of offenses and attempts to derive some “common denominator” across state criminal codes; more categories came into play when NIBRS and NCVS began development, but the coverage of today’s U.S. crime statistics remains a distinct subset of the broader world of “crime.” This comprehensive approach to a rigorous classification is at once its greatest strength and liability: Mapping out the full terrain of “crime” (and building a new set of measures out of the classification) means that the mapping is necessarily broad but sparsely filled with data, including offenses that simply defy measurement of any type, much less cost-effective measurement. These “gaps” in the classification usefully illustrate the types of information that would be important to have, but they could also be interpreted with undue pessimism, as being sufficiently numerous as to undermine the whole enterprise. We argue that the benefit of a fuller framework for thinking about “crime in the United States” outweighs the “cost” of appearing to be very incomplete in filling out that framework—particularly to the extent that data collection based on the classification continues to improve, and that the classification scheme itself is accompanied by a mechanism for future (and regular) revisions, modifications, and adjustments.
Between 2007 and 2009, the Committee on National Statistics (CNSTAT) of the National Academies of Sciences, Engineering, and Medicine (in collaboration with the Committee on Law and Justice [CLAJ]) undertook a comprehensive Review of the Programs of the Bureau of Justice Statistics. While specifically tasked to spend its first year suggesting specific strategies for conducting BJS’s ongoing National Crime Victimization Survey, the main goal of the study was to consider the full suite of BJS data collections, to identify gaps and suggest methods of improving the relevance and utility of BJS products. The BJS Review panel produced an interim report on NCVS options (National Research Council, 2008) and a final report addressing the
broader data collection portfolio (National Research Council, 2009a). Because one major use of the NCVS is as a counterpart to UCR measures, and because UCR crime statistics are certainly relevant to the broader question of collecting statistics on the entire justice system, the BJS Review panel was obliged to at least sketch out the role of the UCR. However, more direct assessment of UCR conduct and concepts was outside its study scope.
Four years later, the U.S. Office of Management and Budget (OMB) encouraged BJS and the FBI to jointly sponsor a study that would take an arms-length look at the entire enterprise of data collection on “crime.” The full statement of task of our 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. This first report of the panel covers the first phase: the substantive plank, which directs the panel to suggest a new classification of “crime”—reflecting development of classifications in other fields such as public health as well as historical and international efforts at classifying criminal activity—as a basis for discussion of a new crime measurement system.
The second phase of the panel’s work (to be covered in its final report) will address the methodological and implementation planks of the charge. The methodological and implementation challenges associated with generating crime statistics—not the least of which is a governance structure or consideration of a cost-effective division of labor between the BJS and FBI, other data providers, (volunteer) law-enforcement-agency contributors, and others—are sufficiently daunting that one may question the utility of a first report focusing expressly on classification and definition issues. This concern, in turn, derives at least in part from a traditional view of crime classification as a post hoc, labeling and editing procedure. But the letter of our charge positions the broad activities appropriately—using a classification as the blueprint for constructing/rebuilding a modern set of crime indicators, rather than coercing existing statistics into a new post-processing structure. The concern also underplays the value of a new classification of crime as a product in its own right—just as the International Classification of Diseases is useful for discussing the full range of human health problems, or the North American Industry Classification System is itself a useful guide to understanding sectors of the economy, so too could a new crime classification have value in organizing thought about a uniquely complex and important set of behaviors.
In carrying out this first phase of the panel’s work, the panel focused attention on enlisting a broad variety of perspectives—various constituencies of crime data users and providers—to participate in two workshop-style sessions in June and July 2014. In its other meetings, the panel also invited other experts and practitioners to discuss their work and perspectives, including emerging studies of “white-collar” or regulatory offenses, mechanisms for
sharing experiences of cybercrime attacks and formulating responses, and considering corporate security (e.g., internal security teams employed by large retailers) and the resulting “crime” (e.g., pilferage from store stocks) that may not be reported to the police. Participants in the panel’s workshops and other data collection activities, and the discussion questions that drove the workshop-style meetings, are listed in Appendix B. Significantly, the panel became aware early in its work of efforts sponsored by the United Nations Office on Drugs and Crime (UNODC) on formulating an International Classification of Crime for Statistical Purposes (ICCS). As will be described more completely in Chapter 5, the UNODC work would prove a solid base on which to base our defined task of suggesting a modern taxonomy of crime.
Having explored basic premises of the definition of “crime” and its classification in this chapter, our argument in this first report builds toward a recommended initial classification scheme via several integral, detailed parts. Chapter 2 reviews the current, major nationally compiled sources of crime statistics, building from the historical development outlined in this introduction and establishing the vocabulary for what follows. Here, summaries of current collections of data not commonly thought of as sources of “crime statistics” are provided and our discussion is focused on the coverage and scope of these collections rather than their specific benefits and weaknesses. Our charge is very clear in obliging us to weigh the unique demands and requests of the user constituencies for U.S. crime statistics, and Chapter 3 synthesizes the results of our panel’s workshop-style sessions on the same. Likewise, our charge directs us to consider alternate and existing classifications of crime, including those being developed by other countries and international organizations. Chapter 4 begins the process of articulating a recommended modern classification, reviewing several historical and contemporary examples of classification efforts. The argument then takes full form in Chapter 5—delineating some core principles for deriving the recommended initial classification, offering some guidance on what this classification is (and what it is not) relative to current data systems, and closing with some initial thoughts on next steps of work.
This material is supplemented by five appendixes, two of which we have already introduced: Appendix A presents the statement of task to our panel in full detail and Appendix B lists the participants and motivating questions behind the panel’s workshop-style meetings in 2014, our primary “data-gathering” activities. Two subsequent appendixes are presented as back-matter in this report only because their volume would overwhelm the main-text chapters; Appendix C lists out the detailed structure of some of the exemplar
This page intentionally left blank.