Of Uniform Crime Reports (UCR) in Nation
Total Page:16
File Type:pdf, Size:1020Kb
State of West Virginia Department of Military Affairs & Public Safety Criminal Justice Statistical Analysis Center Division of Criminal Justice Services July 2006 Report Highlights • The methodology introduced in this report offers a valid approach for Establishing the “Statistical Accuracy” assessing the statistical accuracy of UCR crime statistics in WV and the of Uniform Crime Reports (UCR) in nation. West Virginia • Of the 31,084 reported crimes in the population, 1,297 were estimated to contain a classification error. James Nolan, Ph.D. - West Virginia University Stephen M. Haas, Ph.D. - Criminal Justice Statistical Analysis Center • An estimated 4.17% of all records Theresa K. Lester, M.A. - Criminal Justice Statistical Analysis Center reported for the population of 12 law enforcement agencies in this study Jeri Kirby, M.A. - West Virginia University were misclassified. Carly Jira, B.A. - West Virginia University • The UCR Part I crimes of aggravated assault, burglary, larceny, The Uniform Crime Reporting studies have sought to ascertain the and robbery contained a statistically significant amount of classification (UCR) Program is a national initiative magnitude of error resulting from the error. involving more than 17,000 city, county, erroneous classification of crimes. and state law enforcement agencies That is, few studies have assessed the • A significant level of error was found for nonindex crimes such as which voluntarily report crime data to amount of error found in crime totals simple assault/intimidation, unfounded the FBI (FBI, 2004).1 The main objective due to the misclassification of crime offenses, and general incidents. of the UCR Program is to generate a types on the part of law enforcement • A great deal of overlap in valid set of crime statistics for use in officers and agencies. Classification classification error was found between law enforcement administration, error (or the misclassification of crime some individual crimes such as operation, and management. types) and the impact of this error on aggravated-simple assaults and Over the years, however, the official UCR statistics provides the basis larceny-burglary offenses. information gathered and reported for this report. • Both the Violent Crime and Index through the UCR Program has become It is anticipated that an totals for the state were significantly a social indicator for the nation. The understanding of classification error and undercounted in reported UCR statistics. public looks to the statistics generated its consequences for crime reporting will from the UCR Program for information have notable implications for both state • The Violent Crime Total for WV on fluctuations in the level of crime. and federal UCR Program was undercounted by 22.49% compared to the Index Total at 2.35%. Meanwhile, criminologists, sociologists, administrators. The identification of legislators, municipal planners, the classification error and its sources may • The Property Crime Total for the media, and other students of criminal provide a basis for the modification of state did not contain a significant justice use the statistics for varied law enforcement training curricula. In amount of classification error at 0.18%. research and planning purposes (FBI, addition, such information may lead to 2004). more accurate crime reporting as well • The differentiation between Such widespread use of UCR as provide a means for adjusting future aggravated and simple assault crimes accounted for a disproportionate information has underscored the crime data based on known error. amount of classification error in importance of ensuring its accuracy. This report seeks to assess the reported UCR statistics in WV. While considerable attention has been statistical accuracy of crime reporting • The inclusion of simple assault into focused on errors associated with victim in West Virginia. In this pursuit, Violent Crime and Index Totals reduced reporting and missing data (Hart and however, a central purpose of this report classification error for these categories Rennison, 2003; Maltz, 1999), few is to introduce a methodology for to nonsignificant levels. assessing the “statistical accuracy” of strict accounting of all crimes, it remains occur when state UCR programs and crime estimates produced by the UCR a tremendously useful resource for the FBI make decisions about how to Program. Currently, no uniform method gaining knowledge about crime. deal with problems of missing data at exists for identifying and assessing the As a result, the value of UCR is not the time of publication and reporting impact of classification error in WV or contingent on FBI and state UCR (Maltz, 1999). Multiple methods exist the nation. Thus, this report applies an program officials eliminating all errors for dealing with problems associated original methodology designed to in reported statistics. Instead, it is more with missing data. For instance, some examine classification error which may important that the managers of the jurisdictions may choose to not report in turn influence how such assessments program understand these errors and missing data while others may produce are conducted in other states and the make every attempt to measure them. estimates for missing data and adjust nation. By doing so, program managers will crime statistics based on those In an effort to illustrate the merit become more cognizant of the limitations estimates. Missing information and the of this methodology, this report utilizes in UCR and can begin to engage in handling of such data is a common a random sample of crimes reported to efforts to improve the accuracy of crime source for error found in the reporting the UCR Program in WV. An analysis reporting. of statistical data and is not unique to was conducted to identify the under- and Toward this end, considerable UCR statistics. overreporting of incidents across various attention in recent years has focused on However, there have been no true crime types. Both the sources of errors that result from victims deciding systematic studies of classification classification error and the most not to report crimes and by the police errors in the UCR statistics. As a result, common reasons for misclassifications electing not to record them (Bachman, little is known about the degree to which are discussed. To begin, this report 1993; Black, 1974; delFrate and classification error influences the provides a brief overview of Goryainov, 1994; Gove, Hughes, and statistical accuracy of crime statistics. classification error and its impact on the Geerken, 1985; Greenberg and Ruback, Moreover, research has not progressed accuracy of official crime statistics. 1987; Greenberg and Ruback, 1992; to the point of establishing a single Schwind and Zenger, 1992; Shah and methodology for assessing classification Assessing the Accuracy of Pease, 1992; Skogan, 1976; Warner and error in UCR data. For these reasons, Crime Statistics Pierce, 1993; Wexler and Marx, 1986; the current study may not only assist Since the inception of the UCR White and Mosher, 1986). For example, UCR Program administrators in WV, but Program more than 70 years ago, in a study based on the 2000 National may have implications for how statistical police records have been the primary Crime Victimization Survey (NCVS), accuracy is examined on a national source of data in these national crime Hart and Rennison (2003) found that basis. To provide a foundation for this statistics. Although the quality of police nearly 4 in 10 property crimes and 6 in study, the following discussion provides records has often come into question, 10 violent crimes were never reported a brief description of classification error UCR has been found to be a valid to police. Hence, it is clear that not all and why it is often present in crime indicator of the “index” crimes (Gove, crimes are reported to law enforcement statistics. Hughes, & Geerken, 1985). and not all crimes are accounted for in It is well known, however, in the official reporting. This type of error Classification Error: fields of criminology and criminal justice occurs at the input stage of the process Definition and Sources that UCR is a “statistical program,” and is due to nonreporting of crimes by A classification error occurs when meaning that it is not an actual victims and law enforcement. the police officers record the facts of accounting of all crimes. Nevertheless, On the other hand, error can also an incident correctly, but misclassify the UCR has been invaluable to police and occur at the latter stages of the process. crime type. For example, an criminologists alike in their efforts to Recently, there has been an investment “aggravated assault” that involves a understand the nature and extent of on the part of the U.S. Department of weapon is sometimes recorded by the crime locally and nationally. Therefore, Justice to learn more about errors at the police as a “simple assault” when the in spite of the fact that UCR is not a output stage (Maltz 1999). These errors victim is not seriously injured. Such a 2 “Statistical Accuracy” of UCR Crime Statistics crime classification may be correct for it actually had (McCoy, Matza, & agency revealed that 3.00% of all larceny criminal prosecution, but not for Fazlollah, 1998). records contained a record error. Yet, “statistical” purposes. Given that this Lastly, errors can occur at the only a fraction of this 3.00% actually incident involved a weapon, it should be point of automation or when resulted in a reclassification of UCR recorded as an aggravated assault. automated systems are upgraded or crime. This is because many of the record Classification errors can occur for revised. Automated systems are errors were discovered to be the result of multiple reasons. Some of the most often programmed to allow for the slight discrepancies in the dollar amounts common sources of error include the automatic translation of reported recorded for theft loss (i.e., incorrectly inaccurate interpretation of UCR crimes to UCR definitions. In these recorded as $400.00 as opposed to the definitions, purposive actions on the part instances, reported crimes are actual amount of $450.00).