AT THE MOST BASIC LEVEL, “classification” is a means for grouping like things alike. The difference between classification schemas is the manner by which things are judged “like enough” to be put in the same category, and the full extent to which a set of categories is delineated. There are abundantly many classifications that can be developed surrounding some particular phenomena and just as many specific, worked-out examples, ranging from very crude definition lists for a few specialty categories of interest to intricate, hierarchically structured “trees” covering the whole range of experience of a particular group or phenomenon.
We initially touched on this conceptual issue in Section 1.2, but it is natural to delve into it again here: A key underlying question to this chapter and report is why should “crime” be developed into a classification? More specifically, why is it important to classify crime as a prelude to discussing improvements in crime statistics collection? The answer is that a proper classification provides a strong basis for structuring and organizing information, and so provides a useful blueprint for operationalizing actual information collection. In the parlance of official statistics, a classification is a form of standard meant as reference across related data collections (and ideally across various countries or jurisdictions operating those collections). For the United Nations Statistics Commission, Hoffman and Chamie (1999:2) defined a statistical classification as one having “a set of discrete categories, which may be assigned to a specific variable registered in a statistical survey or in an administrative file, and used in the production and presentation of statistics”—emphasizing standardization in variable listings
and presentation. But—cast more expansively—a classification scheme can serve as a map of the entire space of a complex phenomenon. Writing specifically of full implementation of the National Incident-Based Reporting System (NIBRS), Lynch and Addington (2007a:317–318) lamented that “using all of the information in NIBRS simply to re-create the [Uniform Crime Reporting (UCR) “index crime” categories] is a lost opportunity to learn more about crime.” They argued for a thorough reimagining of NIBRS’s purpose and, along with it, experimentation with alternative crime classifications “that can [each] tell us something important about crime and that can be reasonably estimated within each system”—regardless of whether those estimates flow naturally from current data systems. To better mobilize crime statistics with the end goal of conveying the risks of criminal victimization to the public, for instance, they suggested that a classification could be built around any of three dimensions, each of which would put a slightly different lens on the problem (Lynch and Addington, 2007a:318–319):
- Classifying crime between those “involving intimates and acquaintances and those involving strangers,” because “stereotypic crime [between strangers] is what we are afraid of and what we demand that our police address, whereas crimes among persons who know each other are more complex and require different responses”;
- Distinguishing between “crimes occurring in public as opposed to private places,” in which expectations and assumptions of law enforcement control vary; and
- Splitting “crime by activity at the time of victimization,” where even a four-way split of activity among “home, school, work, and leisure” would be profoundly useful to public understanding.
In this chapter, we review some exemplar classifications of crime that have been proposed at differing times, drawing mainly from international experience but also including some U.S. variations as well. To this mix of examples, we can also further mention some embedded classifications we have already discussed—and it is useful to do so, inasmuch as they extend the set of extreme approaches to classification. To wit, the original set of UCR offenses delineated by the International Association of Chiefs of Police in 1929 (Box 1.2) is a quintessential definition- or listing-based classification schema, trying to use similarity in explicit offense definitions across jurisdictions and balance those with identifying major crimes of particular public interest. This was underscored by the format of the original Uniform Crime Reporting manual (International Association of Chiefs of Police, 1929)—a thick volume, with most of its heft coming from a small-print compilation of extracts from state criminal codes. As the UCR definitions have remained relatively unaltered over the decades—with expansion when NIBRS was implemented—so too has the UCR Program’s maintenance of a primarily (penal/criminal) code-based
classification scheme. By contrast, the National Crime Victimization Survey (NCVS) arrives at a similar list of covered crimes but does so via entirely different means; where the UCR Program requires local police departments or state coordinators to parse available information and decide which definition best suits a particular incident/offense, the NCVS never puts the burden on the respondent or the interviewer to “label” the crime at hand. Rather, the NCVS’s event- or roughly attribute-based classification asks a number of general questions and combines that information—formulaically, yet flexibly—to construct the proper classification label.
In the early 1980s, the Police Executive Research Forum (PERF) developed a simplified Crime Classification System (CCS) for testing and deployment in various sites, notably Peoria, IL (Police Executive Research Forum, nd). PERF’s proposed classification was innovative in both its most fundamental principle—deciding up front that the classification should be “non-legal [and] victim-focused,” rather than offense-focused—and its stark brevity. In fact, the classification sought to resolve all criminal behavior into a set of only five categories, three of which were meant to be “straightforward” and “based on the type of harm experienced by the victim” (Police Executive Research Forum, nd:2–3):
- Injury (only)—“reported crimes in which at least one victim received physical injury or was threatened with physical injury (e.g., assault, rape, etc.);”
- Loss (only)—“reported crimes in which at least one victim had property stolen, damaged or destroyed (e.g., burglary, larceny, fraud, etc.);” and
- Injury and Loss—“reported crimes in which at least one victim was physically injured or threatened with injury and property was stolen, damaged or destroyed (e.g., robbery).”
Some time later, a separate task force that developed the Blueprint for the Future of the Uniform Crime Reporting Program (Poggio et al., 1985:91) restated the CCS (citing an unpublished summary). In the restatement, the victim-harm categories were expanded to include two combinations:
- Threat plus loss—intended to cover UCR-defined robbery, not involving injury; and
- Injury plus loss—intended to cover violent crimes like homicide or rape that also included theft of property as well as the crime of robbery with injury.
The original CCS’s final two categories covered cases where either the victim or the harm can not be specified:
- Regulatory—“crimes in which the harm is not to an individual or business but against society or governmental order,” including “crimes such as perjury, treason, and runaways.” “As a practical matter, most of the regulatory crimes reported by the police are typically vice offenses such as prostitution, gambling, drunken driving, and narcotics offenses.”
- Incomplete—“reported crimes that have an identifiable victim but neither injury nor loss,” including “crimes that are planned and perhaps begin but not to the point that the victim is harmed” and commonly referring to “attempts and conspiracy-type offenses” (so much so that the Blueprint task force revised this category to simply “Attempts” of any crime type).
Anticipating concern over the level of detail possible with such a small number of “crime” categories, the CCS developers sought to minimize “any inconvenience incurred as a result of giving up the familiarity of legal categories” by prescribing a battery of offense and victim characteristics to be collected along with the simplified crime type. The CCS was intended to be something that could be compiled directly from law enforcement records systems, and the detailed characteristics were described as data that are “widely available in contemporary police incident reports but [not] usually presented to the public” (Police Executive Research Forum, nd:4). The original CCS suggested collection of victim characteristics as precise as “level and type of injury” and type of medical treatment obtained, and such offense characteristics as the victim/offender relationship and the “extent of force” used. Also included among the CCS offense characteristics was a variant of the “seriousness score” calculation proposed by Sellin and Wolfgang (1964); the seriousness calculation is not described in any detail in the Peoria CCS testing, but was still characterized as relatively easy for records clerks to implement using information in departments’ regular incident reports.
The report summarizing the performance of the CCS in testing by the Peoria Police Department suggested some unanticipated difficulties, notably that 27 percent of crime incident reports in the coding sample could not be assigned a geographic location as precise as a census tract. Tellingly, the report also included a cross-tabulation of the CCS categories versus the UCR category (including both Part I and II offenses) that would apply to the incident. Inclusion of such a table in the report is not surprising, as an examination of the performance of a new classification scheme relative to an older one. However, by the time of the restatement of the CCS by the Blueprint authors, “UCR category” was subtly but explicitly added to the list of “offense characteristics” to be coded when using the CCS—effectively, making the CCS more the generation of a top-level, grouping code rather than a substitute for the UCR’s classification scheme (Poggio et al., 1985:92). In essence, the Blueprint authors
concluded, the auxiliary “characteristics” data required by the CCS (including the UCR category) were too integral to deriving useful information from the CCS to make it feasible; for law enforcement agencies with records systems that lacked all of the CCS “characteristics,” CCS-typed classifications "are insufficiently precise and cannot substitute for the current UCR classification,” and so the framework was abandoned.
In 1975, the SEARCH Group1 set out to construct a prototype records system—capable of exploiting increased computerization of law enforcement departments—to bring fully attribute-based classification to police report statistics. Put most succinctly, attribute-based reporting of crime was intended as “a means of systematically capturing crime event data in the basic detail required to classify the particular offense according to any classification system” (SEARCH Group, Inc., 1976:v). The fine-grained detail—like the answers to individual questions on the NCVS—could be pieced together to match a variety of other classifications, be they a single state’s criminal code or the reporting requirements of a UCR-type program.
The crucial parameter in defining the SEARCH Group’s classification scheme was determining the number of attributes/variables necessary to completely span the crime-type coverage of five selected existing schemes:
- The existing UCR offense lists and definitions;
- The Uniform Offense Classification (UOC), now maintained by the Federal Bureau of Investigation’s National Crime Information Center, that essentially defines the offenses coded on criminal history records;
- The Model Penal Code (MPC) developed by the American Law Institute (originally published in 1962, current full release being American Law Institute, 1985) that provided suggested text for standardized crime definitions that would be used, in part if not in full, in several states’ individual efforts to restate their criminal codes;
- The Texas Penal Code; and
- The California Penal Code.
Ultimately, with some revision, the group found that 28 attributes, derived from parsing the texts of these five sources, were adequate in reconstructing the detail from any of the source codes. Due to the length of the resulting
1The SEARCH Group, founded in 1969, is now more properly known as SEARCH, The National Consortium for Justice Information and Statistics. It is self-described as “a nonprofit organization governed by a Membership Group of governor appointees from the 50 States, the District of Columbia, and the territories” to support “the information sharing, interoperability, communications, information technology, high-tech crime investigative and criminal records systems needs of State, local and tribal justice and public safety agencies and practitioners nationwide” (http://www.search.org/about-search/company-background/).
code list, we refer the reader to Section C.1 in the report appendixes. Of particular note in that listing is the use of 128 verbs—the 26 actions perpetrators can take against property (attribute B5), the 72 actions perpetrators can take against person victims (D5), and the 30 crime-type actions perpetrators can take that are not directed toward either a person “victim” or property (E7)—that are pivotal to determining crime type in a pure attribute-based classification. Akin to NCVS processing, the SEARCH Group developers then set about determining roughly 200 rules—called equations in their report—that built a specific code’s offense list by combining the presence/absence or value of various attributes.
Ten city or state police departments were recruited to provide a sample of source documents—500 each, including the detailed incident report as well as how the local agency would code the incident for UCR, UOC, or state penal code purposes (two agencies each participated from California and Texas, because their codes were used in defining the classification). The SEARCH Group classification was run on this training set, and the generated values were compared with the agencies’ “actual” codes to validate the methodology.
Though acknowledging classification errors “rang[ing] from 22 to 31%,” the SEARCH Group, Inc. (1976) expressed general satisfaction with the results. The error rates could largely be explained, developers argued, by variability in the skill of the human coders of the incident reports—as well as two longstanding concerns with crime classifications. First, assault is inherently difficult to classify, particularly the distinction between aggravated assault and simple assault. Second, the multiple sources for the attribute-based scheme overlap substantively with each other for some crime types and don’t clearly achieve mutual exclusivity of concept; for instance, “the unauthorized signing of someone else’s name to a credit card slip prepared with a stolen card” could fall into any of two UCR codes or three UOC codes (SEARCH Group, Inc., 1976:14).
The SEARCH Group produced human coder performance rates and computer processing times that it felt were impressive, and argued that the next logical step was to move from prototype to implementation in live operations in one or more individual departments. Alas, the record suggests that there were no local-agency takers for the live operational trial.
Work on a standardized classification of crime in Australia began in the late 1970s, culminating in a draft classification (apparently opened for stakeholder discussion in 1980) that was circulated as the Australian National Classification of Offences (ANCO) in 1985. In use over the next decade, gaps and deficiencies were noted in the ANCO, and a more thorough revision and reassessment led to publication of the formal first edition of the Australian Standard Offence
Classification (ASOC) in 1997. Again, the new system was subjected to years of work and experience before another major round of revisions was applied and a second edition was released—this time, with the “substantial review” said to be supplemented by “extensive consultation, predominantly with government agencies responsible for formation and administration, and non-government analysts of crime and justice issues.” An unchanged (save for typographical fixes) third edition was published in 2011, albeit now rebranded as the Australian and New Zealand Standard Offence Classification (ANZSOC) after adoption for use by New Zealand entities and in keeping with efforts to harmonize classification schemes between the Australian Bureau of Statistics (ABS) and Statistics New Zealand (Australian Bureau of Statistics, 2011:vii).
The ANZSOC definition document lists six “criteria” used to identify the individual categories in its classification scheme (Australian Bureau of Statistics, 2011:6):
- Violence: To convey whether violence is involved and—if so—its nature and extent;
- Acquisition: To convey whether the basic intent of the offense is acquisitive, or to obtain property;
- Nature of Victim: To characterize “the nature and vulnerability” of victims, whether the victims are “persons, property [or] the community”;
- Ancillary Offences: “Whether the offence only exists as an extension of, or in relation to, another offence”;
- Seriousness: The severity of offense, based on such factors as aggravating circumstances/conditions and vulnerability of the victim; and
- Intent: Whether the offense was intentional or negligent/reckless.
The ANZSOC authors take pains to note that “the divisions of the Classification are not ranked by seriousness,” though seriousness is one of the criteria. ABS maintains a separate listing—the National Offence Index (NOI), based in part on levels of reported crimes in the different categories and the length/severity of sentences given to offenders—that “enables selected ANZSOC groups to be ranked in order of seriousness” (Australian Bureau of Statistics, 2011:6).
Further, the ANZSOC designers note that “certain design considerations were deemed to be of particular importance in determining the structure of the Classification” (Australian Bureau of Statistics, 2011:7):
- Conformance with jurisdictional criminal codes: Definitions constructed to “encompass, as far as possible, the various legal definitions and criminal codes” used throughout both nations;
- Correspondence with “usual” data collection: Crimes should be “representative of information available and typically reported on when collecting data”;
- Relevance: Definitions should allow collection of relevant data to “address important areas of social concern”; and
- Statistical balance: Crimes should be distributed “relatively evenly” across the full classification, or at least not overly concentrated in just a few categories.
The basic ANZSOC structure is listed in Section C.2, in the appendixes, due to its length. The ANZSOC developers note up-front that capturing the full range of attributes of interest is “beyond the current capabilities” of the current classification. In particular, the developers concede that the scheme itself does not allow for grouping or analysis by three important concepts: (1) family and domestic violence offenses (attributes of victim/offender relationship and location of incident, among others); (2) “e-crime” or cybercrime (attributes of the mode of the offense, meaning computer involvement); and (3) terrorism-involved offenses (as distinct from non-terrorism-related assaults, homicides, and other crimes) (Australian Bureau of Statistics, 2011:8–9).
In the 1997 first edition, the ABS suggested that the ASOC would be “progressively introduced”—adopted and phased in over time, because “each State and Territory will need to undertake a detailed mapping program that will link every individual criminal offence within their statutes to a single ASOC category. When completed this linkage will be compiled into a single detailed concordance by the ABS” (McLennan, 1997:1). Though the ANZSOC remains an official ABS standard, several Australian states use their own customized—and, generally, extended—classifications for crime recording and reporting. The government of Queensland defines a formal “extension” to ASOC/ANZSOC, adding a fourth hierarchical level to the classification for several categories. For instance, the “Queensland Extension” (QASOC) subdivides the ANZSOC group “serious assault resulting in injury” into five subgroups—assault occasioning grievous bodily harm; torture; wounding; assault occasioning actual bodily harm; and serious assault resulting in injury (remainder)—and the “unlawful entry with intent/burglary, break and enter” group into eight subgroups depending on type of premises (dwelling, shop, other building), whether violence or threats were used, and whether both breaking and entering were involved (Office of Economic and Statistical Research, Queensland Government, 2008). South Australia’s JANCO crime classification goes even further in allowing more detailed hierarchical layers, providing for 10-digit offense codes—and is further distinguished from the ANZSOC because, as its name suggests, it is premised on the original 1985 ANCO draft rather than the more recent ASOC and ANZSOC taxonomies. The JANCO documentation notes only that JANCO codes “continue to be used in South Australia for State based crime reporting” despite ASOC’s issuance in 1997, and that the “two systems are not directly comparable, with ASOC having sixteen divisions for criminal offences rather than ANCO’s
eight divisions” (Office of Crime Statistics and Research, Government of South Australia, 2014:Cover, viii).
In particular, the state of Victoria’s newly independent (as of January 2015) Crime Statistics Agency (CSA) maintains its own crime classification, “largely based on the structure and principles” of the ANZSOC but that “has been modified to suit the legislative environment” of the state. Its modifications are described as following “similar logic” to the UNODC’s proposed international crime classification (Crime Statistics Agency, Government of Victoria, 2015:1), which we describe below in Section 4.1.5. We single out Victoria’s classification as another exemplar for classification—and provide a full listing of the classification in Section C.3—because it makes some interesting changes to both the ANZSOC and UNODC structures.
Prior to 2000, Irish crime statistics published by An Garda Síochána (the Irish national police) were categorized into two broad categories—“Indictable” and “Non-Indictable”—based on the level of court (circuit or high court versus district court) that would have jurisdiction. In line with efforts to modernize police business processes, crimes were recategorized in 2000 but still into two basic categories: “Headline” and “Non-Headline” offenses, intended to correspond more closely to the distinction between serious and less serious crimes that had come to be informally attached to the previous classification. About five years later, authority for production of official crime statistics was transferred from the Garda to Ireland’s Central Statistical Office (CSO; its national statistics office), on the recommendation of a government-wide Task Force on Crime Statistics; the transfer was completed in 2006 (Carey, 2008:1). Shortly thereafter, stakeholder consultations began to assess user/research needs and develop a fuller taxonomy of crime. The resulting Irish Crime Classification System was promulgated in April 2008 (Central Statistics Office, 2008:6).
The CSO’s stated rationale in constructing their specific classification scheme include:
- Accessibility and Clarity to Users: Recalling that their previous system relied heavily on “legalistic” jargon that either clashed with or was entirely unknown to users outside the criminal justice system, the CSO notes that the classification uses common terminology “to the greatest extent possible,” the primary exception being reference to the specific titles of some legislative acts (Central Statistics Office, 2008:6).
- Data Availability and Coverage: The Irish developers acknowledge using concepts and approaches from the Australian and New Zealand Standard Offence Classification System as a base, adapting approaches from that
classification “to make them more relevant in an Irish context.” The Irish classification is meant to be comprehensive—spanning more than just violent, street crime and crime reported to the Garda—yet “a major design consideration was data availability” (Central Statistics Office, 2008:6). That is, the scheme was developed knowing that much of the classification would remain blank or incomplete for at least the early phases of implementation; crimes reported to the Garda would continue to be the core of Irish crime data for several years. With its broad scope, the Irish classification envisions “incorporating offences investigated and processed by other agencies” such as tax and environmental authorities—and “reference needs to be made to other sources, such as Crime and Victimisation surveys”—for a “fuller picture” of crime levels in the country (Central Statistics Office, 2008:7). Knowledge of the type of information that could be acquired directly from Garda records played some role in the construction of categories—and the fact that Ireland has a single, national police force (in the Garda) made uniformity of definitions across local jurisdictions a moot point.
- Shifting from “Total Crime” to a Fuller Picture: One historic problem in the designation of the two massive categories in earlier classifications—either Indictable/Non-Indictable or Headline/Non-Headline—is that temptation grew to treat these two main numbers as the “total crime” level in the country. Purposely, the new Irish classification was designed so as not to provide for a grand total, but rather to emphasize the 16 main groups (and totals for subgroups) for a more nuanced picture and a deliberate “shift away from the notional concept of a total crime level” (Central Statistics Office, 2008:7).
- Capacity for Revision: Developers acknowledged that different agencies—not just the Garda (supplying information to the CSO)—might overlap in their coverage of particular incidents, and wanted a system flexible enough to accommodate evolution over time based on that input. The desire for general continuity of concept over time led to the construction of a “condensed” classification (listed elsewhere in these materials) for some reporting purposes. The condensed classification collapses some detailed offense types (and so would be somewhat robust to addition/subtraction of individual offense types under those headings).
The Irish crime classification is certainly not directly transferable to the U.S. experience, due mainly to the highly centralized nature of the Irish (national) Garda relative to the highly decentralized American law enforcement system, and accordingly its focus on such events as traffic violations. Still, it is a useful model to consider, in several respects—not least of which because the Irish experience of trying to break from the longstanding Headline versus Non-Headline Crime divide parallels the U.S. problem of expanding focus wider
than the small set of UCR Part I crimes. Structurally, a key feature of the Irish system is the way in which it reconciles the desire for relatively fine-grained classification with public demand for a relatively smaller number of defined offenses/statistics. The CSO introduced, and commonly uses for tabulation, a “condensed” classification scheme—selectively collapsing some detail levels and grouping them slightly differently to produce more intelligible summaries of crime for public consumption. The structure of the condensed classification is depicted in Appendix Section C.4. Second, upon introducing the classification in April 2008, the CSO took care to “back cast” recorded crime for the previous five years—that is, recalculating recorded crime totals for 2003–2008 using the new scheme—to ease the transition (Carey, 2008:1).
For good or ill, the Irish experience is also a very telling one for a reason to which we will return in closing this report and explore in more detail in our second report: It offers the cautionary tale that even an excellent classification scheme is for naught if system implementation issues are not addressed in tandem. As part of a wider ranging review of agency practices, the Garda Síochána Inspectorate (2014:9–11) noted “systemic failures in recording practices and non-compliance with crime counting rules” in the Police Using Leading Systems Effectively (PULSE) system, the Garda’s incident recording system. PULSE is effectively analogous to the UCR Program’s databases—it is a recording system, not directly derived from or populated by the Garda’s records management system. An audit sample of PULSE records examined by the Inspectorate suggested misclassification rates of 30 percent or greater, including an unusually large number of entries coded in PULSE as “Attention and Complaints” or “Property Lost”—both of those being technically non-crime categories. It is PULSE crime records that are transferred to the CSO for classification and tabulation—and so flawed data inputs necessarily led to flawed national crime data. Following the release of the Inspectorate’s report, the CSO suspended the publication of Irish police-report crime statistics for six months and initiated its own review. In so doing, Ireland followed the recent example of the United Kingdom Statistics Authority (2014a,b), which pointedly decertified police-report statistics—barring them from bearing the label of “National Statistics” or official statistics—due to similar data quality concerns. Upon completing its review, the Central Statistics Office (2015) agreed to resume publication of police-report crime statistics, while working with the Garda on data improvements and simultaneously beginning a Crime and Victimisation Survey (akin to the U.S. National Crime Victimization Survey and British Crime Survey) as a separate check on the level of crime. The basic and condensed Irish Crime Classification Systems are listed in Section C.4, in the appendixes.
In 2007–2008, several parties revived a notion that had surfaced as early as 1951 and separately began to raise the idea of extending the set of international standards to include a new classification of crime. The United Nations Statistical Division assembled an expert group in September 2008 to discuss the idea, and the European Commission likewise indicated interest in developing a Europe-specific crime classification scheme. A conference in Vienna in October 2008 led to the proposal that the United Nations Office on Drugs and Crime (UNODC), with the United Nations Economic Commission for Europe, establish a task force under the Conference of European Statisticians to develop such a system. From the outset, the task as outlined was difficult (UNECE Secretariat, 2009:1):
Any [crime] classification for statistical purposes could not easily be imposed for use at local level and would not necessarily solve difficulties of cross-national comparability. Rather, in addition to a national legal classification system, a parallel, behavioural/event based classification—based on but not restricted by legal definitions—would be useful for grouping data for statistical purposes at the international level.
Initial task force work included an informal survey of other nations’ existing data systems and further discussions (in which the United States participated, via BJS). The first phase of work led to production of a “principles and framework” document in September 2011 (UNODC/UNECE Task Force on Crime Classification, 2011), in which the task force committed to developing a classification intended for and satisfying the demand of statistical purposes and to using the event/incident as the standard unit of analysis (we discuss the principles of a classification for statistical purposes further in Section 5.1). Significantly, the task force also worked out a first-cut classification that included a companion, coequal set of “tags” or attributes meant to be completed for each crime event/incident. In doing so, the task force recognized that the tags would give eventual data users the possibility to recover great specificity in deriving incident counts—to the point of suggesting that the tags could be concatenated to form labels akin to the defining “equations” of the SEARCH attribute-based classification (Section 4.1.2). So, for instance, the task force suggested that “1.1.At.Fi.FV.OC” (using the numbering scheme they used at the time) would represent a member of an organized criminal group (OC tag, for organized crime involvement) shooting (Fi tag, for firearm) at a female (FV tag, for female victim) with intent to kill or seriously injure (1.1, the category for “intentional homicide”) but missing (At tag, for attempted). (UNODC/UNECE Task Force on Crime Classification, 2011:16).
After reporting back to the Conference of European Statisticians in June 2012, the UNODC and Mexico’s national statistical office (the National Institute of Statistics and Geography, or INEGI) set out to put the work in
greater context, drafting a “road map” for improvement of international crime statistics with the new classification as a starting point. Meanwhile, the task force was renewed (expanding to a broader “expert group”) and continued working and meeting periodically, with meetings in 2013 and 2014 set with the goal of submitting a first official version of an International Classification of Crime for Statistical Purposes for approval by the appropriate United Nations’ commissions in early 2015. The group’s final preparatory/development meeting in Vienna in May 2014 included a first round of “testing”—feedback from participating countries on the degree of concordance between the draft ICCS and the primary criminal statistics repository in a country. From the U.S. perspective, the UCR was treated as the nation’s primary repository—mainly using the Summary Reporting System as the reference, but also referencing NIBRS as necessary. That meeting also involved considerable testing and discussion on how best (or whether) to further extend the classification to better recover information on homicide (United Nations Office on Drugs and Crime, 2014).
After a period of further editing and refinement, including a reworking of the general numbering scheme, Version 1.0 of the ICCS was published in March 2015 (United Nations Office on Drugs and Crime, 2015). Presented for approval by the two United Nations bodies, the ICCS was ratified by the Statistical Commission during its March 3–6, 2015, meeting and the Commission on Crime Prevention and Criminal Justice during its May 18–22, 2015, meeting.
As we discuss in more detail in the next chapter, we use the Version 1.0 ICCS as the base for our own suggested classification of crime—and so do not go into detail here about the structure or content of the ICCS, since that will naturally occur in outlining our own suggestion. For reference, the short-form headings (excluding detailed definitions and specific legal inclusions or exclusions from categories) is rendered in Section C.5 in the Appendixes.
This page intentionally left blank.