Sociologists and criminologists are ambivalent about crime statistics like those generated by the Uniform Crime Reporting System (UCR). On one hand the data are characterized as incomplete and inaccurate, with one frequently cited source saying they are “not worth the paper they are written on.” On the other hand, UCR data are widely used with little regard to validity because they are the only data that provide national coverage and information about local areas. I argue that neither approach is productive. What is needed is research on the sources of measurement error so that the data can be used reasonably. I describe some of the problems with UCR homicide data, then demonstrate that with adjustments that they are extremely reliable and provide the basis for precise estimates of patterns and trends.
Colin Loftin is Professor of Criminal Justice at SUNY-Albany and Co-Director of the Violence Research Group. He received his PhD in Sociology from UNC-Chapel Hill, and has held faculty positions at Brown, U. of Michigan, and U. of Maryland. His research interests include understanding violence as a social process extending beyond individual action, improving the quality of data on the incidence and nature of crime, the design and evaluation of violence prevention policies, and population risk factors for violence.