There are two national crime series which have data on crime rates and trends. The National Crime Victimization Survey (NCVS) is based upon a sample of households and includes both crimes that are reported to police and those that are not reported. The Uniform Crime Reports (UCR) is based upon local police reports which are compiled by the FBI. See (link) for a more detailed explanation of each data set. The two data series complement each other and both are “right” in terms of measuring what they are designed to measure. Please see the report, The Nation’s Two Crime Measures (NCJ 246832, BJS web), for more information.
Number of victimizations per 1,000 persons or households in a given population that occurred during the year.
Many surveys have a specific margin of error because all of the questions are asked of every person. For example if you ask an opinion question all of the respondents in the sample, everyone can give an opinion. The margin of error is different for different crimes and different findings in NCVS because questions are asked only of people who are victims of those crimes. There is a large margin of error around statistically rare crimes, such as rape/sexual assault. The margin of error is smaller around crimes which occur more frequently, such as property theft. Within a crime category such as violent crime, the margin of error around specific characteristics such as hospitalization may be larger than that for overall violent crime, since the estimates are based only on violent crime victims who experienced that characteristic.
NIBRS stands for National Incident-Based Reporting System. Unlike the FBI's Uniform Crime Reporting (UCR) Program, which only collects data on the most serious offense that occurs during a criminal incident, NIBRS collects data on each reported offense occurring during criminal incident. You can learn more about NIBRS in the NIBRS Edition of the FBI's CJIS newsletter.
Though the NCVS was originally designed to provide national level estimates of criminal victimization, BJS has recognized an increasing need for victimization data at the state and local level. BJS has developed multiple approaches for obtaining subnational NCVS estimates, see NCVS Redesign: Subnational for additional information about these approaches.
UCR stands for the Uniform Crime Reporting Program, a project of the FBI. You can learn more about the UCR program on the FBI website.
The NCVS and UCR both offer important information on criminal victimization; however, the two programs were created to serve different purposes. The primary objective of the UCR is to provide a reliable set of criminal justice statistics for law enforcement administration, operation, and management. The NCVS was established to provide previously unavailable information about crime (including crime not reported to police), victims, and offenders. More information on the similarities and differences between the NCVS and UCR can be found in Nation's Two Crime Measures
We develop national estimates from sample cases of interviews with victims. We take the data we get from these interviews and weight it to represent the nation as a whole. All of the published data from the survey represent weighted estimates. When the national estimate is based on 10 or fewer actual sample cases, we make note of this and encourage caution in interpreting results.
The Census Bureau has regulations to protect confidentiality of data, which prevents the public release of any information from small areas that might make it possible to identify individuals who participated in the survey. Geographically identified data from the NCVS can be made available to researchers on a restricted basis through the Census Bureau's Federal Statistical Research Data Centers (FSRDC), following the approval of a proposal detailing how the data will be used. For more information see: https://www.census.gov/fsrdc.
Though the NCVS was originally designed to provide national level estimates of criminal victimization, BJS has recognized an increasing need for victimization data at the state and local level and has developed multiple approaches for obtaining subnational NCVS estimates. (See NCVS Redesign: Subnational for additional information about these approaches.) However, caution should be used when working with the restricted use files since sample at the local level may be limited and is not necessarily representative of the area as a whole.
When national estimates are derived from a sample, as with the NCVS, caution must be used when comparing one estimate to another estimate or when comparing estimates over time. Although one estimate may be larger than another, estimates based on a sample have some degree of sampling error. The sampling error of an estimate depends on several factors, including the amount of variation in the responses and the size of the sample. When the sampling error around an estimate is taken into account, the estimates that appear different may not be statistically different.
One measure of the sampling error associated with an estimate is the standard error. The standard error can vary from one estimate to the next. Generally, an estimate with a small standard error provides a more reliable approximation of the true value than an estimate with a large standard error. Estimates with relatively large standard errors are associated with less precision and reliability and should be interpreted with caution.
Data users can use the estimates and the standard errors of the estimates provided in NCVS reports to generate a confidence interval around the estimate as a measure of the margin of error. A confidence interval around the estimate can be generated by multiplying the standard errors by ±1.96 (the t-score of a normal, two-tailed distribution that excludes 2.5% at either end of the distribution). Therefore, the 95% confidence interval around an estimate is the estimate ± (the standard error X 1.96). In others words, if different samples using the same procedures were taken from the U.S. population, 95% of the time the estimate would fall within that confidence interval. See the NCVS Methodology for an example.