II. Measuring Robbery in the United States
The National Crime Victimization Survey (NCVS) and the FBI’s Uniform Crime Reports (UCR) seek to measure the annual number of robberies in the United States. These methods of data collection have different strengths and weaknesses for describing robbery patterns. Nevertheless, each contributes valuable information to the understanding of robbery.
The NCVS is a survey composed of a nationally representative sample of respondents who are asked directly about types of victimizations they may have suffered. This method, by virtue of directly soliciting information from citizens, therefore uncovers a substantial number of crimes that might otherwise go unreported to authorities. In particular, robberies by acquaintances and attempted robberies are much more prevalent in NCVS data when compared with official data sources. Clearly, some victims would seek to protect acquaintances from legal entanglements. Others may seek to avoid potential retribution from the perpetrator for informing police. Similarly, unsuccessful robberies, where no property is lost, may not be worth the victim’s time in terms of reporting. In addition, the robberies recorded in the NCVS involve far fewer guns than those recorded by the police in the UCR. The official police data compiled in the UCR clearly suffer from nonreporting, which appears to be somewhat of a problem as only about 60% of victims report robbery victimizations to police (Hart & Rennison, 2003). Thus, the picture of robbery in the UCR tends to involve more gun robberies, many more completed robberies, and more robberies by strangers than the NCVS data depict.
To illustrate the data source differences, in 2006 the NCVS estimated that there were 645,950 personal, noncommercial robbery victimizations, and the UCR estimated about 261,000 such events for the same year. Thus, the nonreporting by victims and nonrecording or “unfounding” of robberies by police, where the investigation of the incident leads them to the belief that the event brought to their attention should not be classified as a robbery, both contribute to the differences in observed robbery levels in the United States across the two series.
Another important distinction between the NCVS and the UCR data is that commercial robberies are not captured as robbery victimizations in the NCVS and are only captured in the UCR. Thus, the UCR offers a more expansive picture of victimization in the sense that approximately one quarter of robberies recorded in 2006 involved commercial interests as the direct victim. Since commercial robbery is likely to have a higher level of reporting, this does not translate into a large knowledge gap regarding the extent of this particular type of robbery.
Despite the differences in the two data collection efforts, the contemporary trends from 1990 to 2006 indicate that robbery has declined. The NCVS indicates that during that period, robbery fell from levels of approximately 600 per 100,000 to approximately 250 per 100,000 residents over the age of 12 in the United States. The UCR tracked a decline from 250 recorded robberies per 100,000 U.S. residents down to levels of approximately 150 per 100,000. Thus, both measures agree that the crime of robbery was much rarer in the middle of the 2000s than it had been during the early 1990s.
Overall, individuals should be cautioned that neither the UCR nor the NCVS offers a complete picture of robbery prevalence in the United States. The concept of convergence of the two—that is, UCR and NCVS measuring the same underlying phenomenon—is unlikely to occur precisely because many crimes go unreported and because of coverage and reporting issues that emanate from the UCR data collection effort. Taken together, however, the two data series’ complementary treatment of the crime of robbery offer a good sense of the nature, trends, and extent of robbery.