Epidemiology of Crime in the United States

National Crime Victimization Survey

Victimization Incidents Versus Offenses Known to Police

The FBI Uniform Crime Reports measure crime from the somewhat limited perspective of the police. Instead of asking what police know about criminal incidents, what if we could ask victims of crime everything they know about their "victimization incidents" and the offenders who committed them? We could also ask victims whether or not they reported these offenses to the police, and we could then learn something about differences between incidents reported to police and incidents that victims neglected to report. In other words, interviews with victims of crime could tell us a lot about offenses that never come to the attention of police.

The National Crime Victimization Survey (NCVS) is designed for this very purpose--to measure the total volume of crime in the United States, both reported and unreported to the police. The Bureau of Justice Statistics (BJS) and the U.S. Census Bureau conduct approximately 150,000 interviews annually with members of households that have been selected by probability sampling methods, which make them representative of all households in the United States. These respondents are asked whether they or their households have been the victims of a number of violent or property crimes. If respondents or their households have been victimized, they are asked additional questions about the nature of the victimization incidents (e.g., time and place? was a weapon used?), about characteristics of offenders (gender, age, race? stranger or non-stranger?), and about their response (report to police or not?). The interview also includes questions about characteristics of the victims and their household. Data from these victimization interviews have been reported annually since 1973. The NCVS was substantially redesigned in 1993 to improve and broaden its measurement of victimization incidents.

As shown in the following figure from the BJS website (http://www.ojp.usdoj.gov/bjs/), the total volume of violent victimization incidents is considerably greater than the numbers of crimes known to police and, especially, arrests for violent crime. Throughout the 1970s and 1980s, there were approximately three to four million violent victimization incidents, whereas violent crimes recorded by the police (UCR violent crime index minus murder) ranged from less than one million in the early 1970s to a peak of less than two million in the early 1990s. This graph also shows that victims said they reported violent victimizations to the police in only about half of the incidents. Of course, police may neglect to officially record some violent incidents reported by victims. As indicated earlier and suggested in the graph below, an increasing tendency for police to respond to victims by officially recording violent offenses may be a major factor in the "wave" of violent crime shown in the UCR violent crime index during the 1970s and 1980s.

For more information on the UCR and the NCVS, and the differences between these two measures of U.S. crime rates, see the following two brief reports that are available on the BJS website:

  • Department of Justice, "The Nation's Two Crime Measures" (click here)
  • Michael R. Rand and Callie M. Rennison, "Two Crime Stories? Accounting for Differences in Our National Crime Indicators" (click here)

Trends in Violent Victimization Incidents

Whereas the figure above shows trends in the estimated absolute numbers of victimization incidents, data from the NCVS are usually adjusted for population growth and reported as rates per 1,000 people in the U.S. population age 12 or older. The following graph shows the rate over time of violent victimizations as measured in the NCVS since 1973. The NCVS violent crime rate excludes murder, but includes rape against both sexes (the UCR only measures rape of women), robbery, aggravated assault, and simple assault (which is excluded from the UCR). Throughout the 1970s and 1980s, the rate of violent victimizations remained relatively stable at a rate of approximately 50 violent victimizations per 1,000 population age 12 and over. Recall that the UCR rate of violent offenses known to police increased during this same period to a peak of approximately 750 per 100,000 population in the early 1990s (which would amount to a rate of 7.5 per 1,000). Again, the discrepancy between the upward trend in the UCR violent crime rate and the stable trend for NCVS violent victimization rates suggests that victims were increasingly reporting violent incidents to the police and that the police were increasingly recording incidents as "official" crimes. However, since the early 1990s, the UCR and NCVS measures of trends in violent crime are in full agreement. There is no question that rates of violent crime in the United States have dropped steadily from the early 1990s into the initial years of the 21st century. More specifically, the NCVS indicates that the risk of violent victimization has been cut by over half, from more than 50 per 1,000 to slightly more than 20 per 1,000, since 1994. This dramatic decrease in violent crime appears to be due to a variety of factors, including structural changes in U.S. society, such as an aging population and an improved economy, decreases in drug use and drug-related crime, and more extensive law enforcement, gun control, and incarceration of offenders.

Trends in Household Property Crime Victimizations

Unlike the NCVS data on rates of violent victimizations (and UCR rates of offenses known to police), the rates of property crime victimization incidents in the NCVS are based on households rather than on individuals in the population. Similar to the UCR Property Crime Index, the property crimes included in NCVS household victimization rates include theft, burglary, and motor vehicle theft; however, the NCVS does not count property crimes against businesses. For additional details on how these two sources of evidence differ, see the following report from the U.S. Department of Justice:"The Nation's Two Crime Measures" (click here)

As shown below, the long-term trend in the NCVS victimization rate for property crime is quite clear: it has fallen dramatically from over 500 victimizations per 1,000 households in the early 1970s and to less than 200 per 1,000 households in the early 2000s. Put another way, the annual risk of a household experiencing a property crime incident has been cut in three decades from one out of two to less than one out of five. At least in the 1980s, the UCR rate of property crimes known to police remained at a fairly high level while the NCVS rate of household property victimizations was steadily declining. This discrepancy suggests that citizens and/or police were "officially reporting" a greater share of this type of victimization incident. However, the UCR and NCVS data are in full agreement that the rate of property crime has dropped significantly during the 1990s and early 2000s.

NCVS Property Crime Rates
Source: U.S. Department of Justice, Bureau of Justice Statistics Next Page