Poisson-Based Regression Analysis of Aggregate Crime Rates

• NCJ Number: 183161
• Journal: Journal of Quantitative Criminology Volume: 16 Issue: 1 Dated: March 2000 Pages: 21-43
• Author(s): D. Wayne Osgood
• Date Published: March 2000
• Length: 23 pages
• Grant Number(s): 94-JN-CX-0005
• Sponsoring Agency: Office of Juvenile Justice and Delinquency Prevention (OJJDP)
• Publication Format: Article
• Publication Type: Report (Study/Research)
• Language: English
• Country: United States of America

Annotation

This article introduces a statistical approach to analyzing aggregate crime rates to solve problems arising from small populations and low base rates; the regression models based on the Poisson distribution resolve these common problems.

Abstract

When the population size of an aggregate unit is small relative to the offense rate, crime rates must be computed from a small number of offenses. Such data are ill-suited to least-squares analysis. Poisson-based regression models of counts of offenses are preferable because they rest on assumptions about error distributions that are consistent with the nature of event counts. A simple elaboration transforms the Poisson model of offense counts to a model of offense rates per capita. Using this method to analyze juvenile arrest rates for robbery in 264 nonmetropolitan counties in 4 States revealed the method’s advantages. The negative binomial variant of Poisson regression effectively resolved difficulties that arise in ordinary least-squares analyses. Figures, tables, and 18 references (Author abstract modified)