NCJ Number
134811
Journal
Journal of Research in Crime and Delinquency Volume: 28 Issue: 4 Dated: (November 1991) Pages: 426-429
Date Published
1991
Length
4 pages
Annotation
The argument is presented that the splitting of independent variables for crime rate cannot be substituted for structural modeling in the analysis of unemployment crime relationship by Cantor and Land.
Abstract
Examples of misspecification in the equations used by Cantor and Land are provided. In particular, for robbery and burglary, Cantor and Land use models estimated by the Cochrane-Orcutt method to deal with autocorrelated errors. This is equivalent to imposing untested nonlinear restrictions on a higher order dynamic model and should be treated with care. Within this framework, the preferred results for larceny and theft are the only ones with exception to the objections outlined. Counterarguments focus on the fact that levels of integrations and misspecifications apply to other works. The final argument in defense of the 1985 analysis of unemployment crime relationship by Cantor and Land is the use of splitting the independent variables which might influence crime rates into two groups as a substitute for structural modeling. However, because differencing is no substitute for structural modeling their equations are misspecified by the exclusion of other variables which influence the fluctuation of crime rates around their trends. The question of whether the rate of unemployment has a positive or negative impact on crime rates in the United States is still not fully elucidated. 9 references