The nation’s two official crime measures: NCVS and UCR

This week, the Bureau of Justice Statistics (BJS) released official numbers on the nation’s two crime measures: the Uniform Crime Report (UCR) and the National Crime Victimization Survey (NCVS). The UCR measures crime that is reported to the police, and the NCVS data is based on individuals’ reporting of past victimizations. The main strength of the NCVS is that it can measure unreported crime, and together the two datasets can be helpful for determining reporting rates. The report released by BJS shows crime rates measured by the UCR and the NCVS for the years 2011 through 2020. When comparing the estimates, reporting rates tend to be higher for violent crimes than property crimes, with the exception of motor vehicle theft. 

Data

The UCR’s Summary Reporting System (SRS), developed in 1929, tabulates crimes reported to the police occurring by month and year. In 1982, the UCR’s National Incident-Based Reporting System (NIBRS) was introduced in an attempt to modernize the UCR, emphasizing incident-level data collection to provide more detail and context about each incident (e.g. details on victims or offenders of crime, characteristics of the incident). However, participation rates for NIBRS are still not adequate enough to obtain accurate national-level estimates. Thus, the UCR’s SRS is the primary measure of crimes reported to the United States at the time of this report. Per the BJS, 2020 is the last year that the UCR will present SRS estimates, so we can expect 2021 estimates in the NIBRS format. That being said, the UCR estimates in the current BJS report are based on SRS.

The UCR SRS provides a measure of the number of crimes recorded by law enforcement agencies in the United States. It collects incident information on ten offenses: murder, rape, robbery, aggravated assault, human trafficking, burglary, larceny, motor vehicle theft, and arson. Monthly crime data are compiled from law enforcement agencies who submit their data to the UCR program. Data are compiled to tabulate national crime counts as well as individual counts for states, counties, and agencies. The UCR also collects Supplemental Homicide Reports to understand the extent and nature of homicides in the nation. A few drawbacks to the UCR SRS are: 1) it relies on the hierarchy rule, which means that only the most serious crime per incident is reported (e.g., if there was a rape and a murder, only the murder would be reported); 2) it does not measure unreported crime; 3) law enforcement participation is voluntary; and 4) there is no detailed information on offense characteristics. 

The NCVS is the primary source of unreported crime and includes data on the characteristics of nonfatal victimizations. The NCVS collects detailed information on the frequency and nature of several offenses, including: rape, robbery, aggravated assault, simple assault, larceny, burglary, trespassing, motor vehicle theft, and other types of household theft. The NCVS also periodically conducts supplemental surveys that have questions on additional crimes, such as school crime, identity theft, stalking, and financial fraud. A few drawbacks to the NCVS are: 1) it does not capture murder offenses; 2) it only interviews people age 12 and older; and 3) people do not always accurately recall victimizations or know the exact time that they occurred.

Results

Rates for violent crime and property crime are both higher when tabulated by NCVS rather than UCR, suggesting a fair amount of underreporting. Violent crimes were somewhat more likely to be reported than property crimes, with the exception of motor vehicle theft. These trends were consistent across all ten years of data. There are no clear changes in reporting rates over time, as they fluctuated up and down from 2011 through 2020 without any discernable pattern. 

In 2020, the reporting rate for violent crime was 24.4%, with UCR and NCVS estimates of 4 and 16.4 per 1,000, respectively. The reporting rate was higher for certain violent crimes though, such as robbery (47.3%) and aggravated assault (96.6%). When it comes to property crime, the reporting rate was 20.7%, with UCR and NCVS estimates of 19.6 and 94.5 per 1,000, respectively. For certain types of property crime, this estimate was higher. For example, reporting rates were 32.6% for burglary and 58.1% for motor vehicle theft.

Taken together, aggravated assault had the highest reporting rate (96.6%), followed by motor vehicle theft (58.1%), robbery (47.3%), and burglary (32.6%). 

Limitations

Unfortunately, because the NCVS data presented here are aggregated at the national level, it is impossible to calculate reporting rates for individual states or localities. This is problematic because it prohibits our ability to understand what factors contribute to the “dark figure of crime.” Because different places have different policies and local dynamics, it would be helpful to know how reporting rates differ from place to place.

The NCVS began collecting subnational estimates to help with this problem, but the data are not accessible to the public, nor are they reported here. While it is possible to apply for access to restricted data, the process is so arduous and expensive that it is impractical. Just a few months ago I tried to do so, and I learned that the process includes submission of a three-page project proposal, followed by submission of a more detailed proposal, and an estimated cost of $20,000 per year to access the lab where data are located. 

Further, when comparing UCR and NCVS data, it is important to note that they differ in a few important ways. First, they serve different purposes, as discussed above. Second, the programs measure an overlapping but nonidentical sets of crimes. Third, the definitions of some crimes (e.g. burglary) differ between the UCR and the NCVS. Fourth, the programs calculate crime rates based on different populations: the NCVS rates are based on households (for property offenses) and the population age 12 and older (for violent offenses), while the UCR rates (for property and violent offenses) are based on the total population. That being said, differences in statistical estimates could result. 

Conclusion

This report provides us with a basic idea of reported and unreported crime in the United States, but the level of detail provided is unsatisfying. It would be helpful if micro-level estimates (e.g., state or county) were more accessible for NCVS data, so that it could be broken down to a unit akin to that provided by the UCR. Similarly, the UCR data also suffers from a severe lack of detail in terms of details on the crime provided. While this is expected to improve with NIBRS, the participation rate in NIBRS is still less than ideal and varies widely by region. At present, the current state of crime data in this country is still haphazard and disappointing, but it has improved slightly over time and I hope it continues to do so.