NCJ Number
193667
Journal
Journal of Quantitative Criminology Volume: 17 Issue: 4 Dated: December 2001 Pages: 291-327
Editor(s)
David McDowall
Date Published
December 2001
Length
37 pages
Annotation
This article provided a methodological evaluation of the Cantor-Land approach to the times series analysis of the crime and unemployment relationship through the discussion of both the theoretical ideas and the equations used to represent the ideas of C-L.
Abstract
In 1985, David Cantor and Kenneth Land (C-L) analyzed the relationship between annual unemployment rates and crime rates in the United States which became an example for future time series analyses. Their conceptual model reflected the transient relationships between aggregate unemployment and crime rate fluctuations. The C-L study argued that prior research had not accounted for the ways unemployment might influence crime. This article questioned the procedures used by the C-L study and set-out to update the C-L data set with nationally aggregated data, add additional variables, and analyze homicide and robbery rates in the United States from 1946-1997. By analyzing nationally aggregated data, numerous problems were identified when attempting to study the impact of unemployment on crime rates. The regression equations used by C-L did not adequately represent the theoretical ideas. In addition, the variables in the theory were not sufficiently represented by those available in official unemployment statistics. This created a lack of confidence in the conclusions drawn by the studies. It was suggested that most sociological analyses of crime rate time series had serious methodological deficiencies. The methods used failed to show long-run tendencies or relied on misspecified models and improper estimation procedures. References