Race, crime, and public housing in Atlanta: Testing a conditional effect hypothesis

This article tests a conditional effect hypothesis which predicts that the strength and magnitude of the association between racial composition and mi me rates will dissipate with increasing distance of neighborhoods from publ ic housing projects. We examine this hypothesis with 1990-92 gee-coded mime incident data matched with 1990 block-group-level census data for Atlanta. The hypothesis is supported in models predicting murder, rape, assault, an d public order crime, but not robbery and property crime. Confirmation of o ur conditional effect hypothesis for several major types of crime suggests the potential for bias in interpretations of estimated race effects in mult ivariate neighborhood-level models that do not control for public housing.