The U.S. Department of Justice administers two statistical programs to measure the magnitude, nature, and impact of crime in the United States: the Uniform Crime Reporting (UCR) Program and the National Crime Victimization Survey (NCVS) (formerly known as the National Crime Survey). Because of differences in methodology and crime coverage, the results from the two programs are not strictly comparable nor consistent. By complementing each other's findings, the two programs enhance our understanding of the nation's crime problem.
The Federal Bureau of Investigation's Uniform Crime Reporting (UCR) Program, which began in 1929, collects information on the following crimes reported to law enforcement authorities: homicide, forcible rape, robbery, aggravated assault, burglary, larceny-theft, motor vehicle theft, and arson.
The UCR data are compiled from monthly law enforcement reports made directly to the FBI or to centralized state agencies that then report to the FBI. Each report submitted to the UCR Program is examined thoroughly for reasonableness, accuracy, and deviations that may indicate errors. Large variations in crime levels may indicate modified records procedures, incomplete reporting, or changes in a jurisdiction's boundaries. To identify any unusual fluctuations in an agency's crime counts, monthly reports are compared with previous submissions of the agency and with those for similar agencies.
Law enforcement agencies active in the UCR Program represent approximately 240 million United States inhabitants -- 98% of the total U.S. population.
The UCR Program provides crime counts for the nation as a whole, as well as for regions, states, counties, cities, and towns. This permits studies among neighboring jurisdictions and among those with similar populations and other common characteristics.
UCR findings for each calendar year are published initially in a preliminary release in the spring followed by a detailed annual report, Crime in the United States, issued in the summer following the calendar year. In addition to information on crime counts and trends, this report includes detailed data on crimes cleared, persons arrested (age, sex, race) for a wide range of crimes, law enforcement personnel (including the number of sworn officers killed or assaulted), and the characteristics of homicides (including age, sex, and race of victims, victim-offender relationships, weapon used, and circumstances surrounding the homicides). Other special reports are also available from the UCR Program.
Following a 5-year redesign effort, the UCR Program is currently converting to a more comprehensive and detailed reporting system, called the national Incident-Based Reporting System (NIBRS). NIBRS will provide detailed information about each criminal incident in 22 broad categories of offenses.
The Bureau of Justice Statistics' national Crime Victimization Survey (NCVS) (formerly known as the national Crime Survey), which began in 1973, collects detailed information on the frequency and nature of the crimes of rape, personal robbery, aggravated and simple assault, household burglary, personal and household theft, and motor vehicle theft. It does not measure homicide or commercial crimes (such as burglaries of stores).
Interviews are conducted by U.S. Census Bureau personnel with all household members at least 12 years old in a nationally representative sample of approximately 49,000 households (about 101,000 persons). Households stay in the sample for 3 years and are interviewed at 6-month intervals. New households rotate into the sample on an ongoing basis.
The NCVS collects information on crimes suffered by individuals and households, whether or not those crimes were reported to law enforcement. It estimates the proportion of each crime type that was reported to law enforcement, and it details the reasons given by victims for reporting or not reporting.
The survey provides information about victims (age, sex, race, ethnicity, marital status, income, and educational level), their offenders (sex, race, approximate age, and victim-offender relationship), and the crimes (time and place of occurrence, use of weapons, nature of injury, and economic consequences). Questions recently added to the survey cover the experiences of victims with the criminal justice system, details on self-protective measures used by victims, and possible substance abuse by offenders. Periodically, supplements are added to the survey to obtain detailed information on special topics such as school crime.
Findings from the NCVS for each calendar year are published in a press release the following April (preliminary data), in a BJS Bulletin in the fall presenting summary final data, and in a detailed report the following June covering all NCVS variables. Each year BJS staff develop Special and Technical Reports on specific crime topics.
As the previous description illustrates, there are significant differences between the two programs. The NCVS, for example, includes crimes both reported and not reported to law enforcement, but it excludes homicide, arson, commercial crimes, and crimes against children under age 12 (all of which are included in the UCR Program). The UCR captures crimes reported to law enforcement, but It excludes simple assaults from the Crime Index. Moreover, even when the same crimes are included in the UCR and NCVS, the definitions vary.
Another difference is the way that rate measures are presented for crimes such as burglary, household theft, and motor vehicle theft in the two programs. The UCR rates for these crimes are largely per-capita (number of crimes per 100,000 persons), whereas the NCVS rates for these crimes are per-household (number of crimes per 1,000 households). Because the number of households may not grow at the same rate each year as the total population, trend data for rates of household crimes measured by the two programs may not be comparable.
In any large-scale data collection program, there are many possible sources of error. For example, in the UCR Program a police officer may classify a crime incorrectly, and in the NCVS a Census Bureau Interviewer may incorrectly record the responses given by a crime victim. Crime data may also be affected by how the victim perceives and recalls the event. Moreover, as data are compiled and processed, clerical errors may be introduced at any stage. Both the UCR and NCVS programs employ extensive accuracy checks at various stages in the data collection process to minimize errors.
As noted above, the NCVS is based on an extensive, scientifically selected sample of American households. Thus, every crime measure presented in NCVS reports is an estimate for the nation based on results obtained from the sample. Estimates based on a sample have sampling variation, or a margin of error (which defines a confidence interval) associated with each estimate. This means that if another sample is drawn, there is a certain probability that the resulting estimate would be somewhat different from the original one. If the survey were repeated many times with different samples, the resulting estimates would converge around the actual measure for the entire population. Rigorous statistical methods are used for calculating the magnitude of the sampling variation associated with the NCVS estimates. Trend data in NCVS reports are described as genuine year-to-year changes only if there is at least a 90% certainty that the measured changes are not the result of sampling variation. The UCR data are based on the actual count of offenses reported by law enforcement jurisdictions. In some circumstances UCR data are estimated for nonparticipating jurisdictions or those reporting partial data.
Some differences in data from the two programs may result from the fact that NCVS estimates are subject to sampling variation. Apparent discrepancies between statistics from the two programs can usually be resolved by comparing NCVS sampling variations (confidence intervals) with UCR statistics. Year-to-year changes in individual crime categories reported by the UCR usually fall within the confidence intervals of the NCVS estimates, indicating no statistically significant differences between the output of the two programs. Even should the UCR changes fall outside the intervals, incompatibility of statistics should not be assumed. To illustrate, when differences between UCR and NCVS occur, there is a 1 0% chance they are due to sampling variation because of the 90% confidence level established by NCVS. It should also be noted that definitional and procedural differences between the UCR and NCVS programs can account for apparent discrepancies in data output.
As has been discussed throughout, the results of UCR and NCVS are not strictly comparable for a variety of reasons. Data users, however, possessing the basic understanding of each program's objectives, methodology, and coverage, can use the output from each in a complementary manner to better assess crime occurrence, losses, law enforcement involvement, arrestee descriptive information, and victimization data. By properly utilizing both programs in tandem, the crime issues in this country can be viewed in a much broader, more complete scope.
Bureau of Justice Statistics: Four measures of Violent Crime
Bureau of Justice Statistics: The nation's Two Crime Measures (pdf file)