THE DISTRIBUTION OF HOUSEHOLD PROPERTY CRIMES

Previous studies have established that repeat victimizations occur mor e frequently than would be expected if households within a particular area were victimized randomly. This implies that characteristics of th e household affect the victimization rate. Even controlling for these characteristics, we find that a Poisson model does not capture the dis tribution of victimizations because repeat victimizations are more con centrated than it would indicate. This leads us to adopt the negative binomial generalization of the Poisson model. Our analysis uses sociod emographic attributes of the household and community-level characteris tics to predict victimizations, with the victimization data being the observed number of property crime victimizations from the 1992 British Crime Survey. The negative binomial generalization is found to be hig hly statistically significant and the crime concentration it implies b ecomes much more marked as the predicted number of victimizations incr eases.