NCJ Number
136261
Journal
Criminology Volume: 30 Issue: 1 Dated: (February 1992) Pages: 125-132
Date Published
1992
Length
8 pages
Annotation
Menard disagrees with Blumstein, Cohen, and Rosenfeld who say that Uniform Crime Report (UCR) and National Crime Survey (NCS) data behave similarly over time, and this paper argues that such controversy is inevitable due to data limitations and problems in defining comparability.
Abstract
Annual UCR (1946-1989) and NCS (1973-1989) robbery and burglary rates have confidence intervals that assume large samples. Further, correlations between the two measures assume serially independent observations. Both of these assumptions are false, and little weight can be placed on the estimates. Most variations in UCR data occurred in the 1960's before NCS data collection began; in addition, NCS rates have a higher level and a larger variance than UCR rates. Despite differences in levels and variances, however, fluctuations in the two data sets follow remarkably similar time paths. The correspondence of time paths suggests that UCR and NCS data are more strongly related than their correlations imply. Given the data limitations and uncertainties in defining comparability, it is not surprising that the Blumstein research and Menard find many reasons to disagree. Differences between UCR and NCS data raise the possibility of substantial bias in one or both of the measures, and studies of the relation between the two data sources will likely continue. The current paper argues that identifying substantive differences is the most promising approach to understanding the relation between UCR and NCS rates and that adjustments to remove the differences will always be debatable. 14 references and 2 figures