Characteristics of U.S. cities with extreme (high or low) crime rates: Results of discriminant analyses of 1960, 1970, and 1980 data

Why do some cities have higher or lower crime rates than others? In this study we attempt to answer this fundamental question by identifying the theoretically motivated structural covariates which differentiate U.S. cities with extreme high and low crime rates. We apply discriminant function analysis across comprehensive samples of all U.S. cities with populations of greater than 25 000 in the periods 1960, 1970, and 1980 and then posit four question: are empirical findings from regression-based studies of city crime rates replicated in discriminant studies with extreme rates? Are the covariates that predict high or low crime rates unique to specific time periods? Which covariates are better able to discriminate high or low rates for specific crimes? And are specific covariates distinguishable in crime rate changes across time periods? Among our general findings, it appears that those cities with extreme high (low) crime rates tend to be the largest (smallest), most (least) economically deprived, and most (least) socially disintegrated. Associations with these latter two characteristics appear to have grown stronger over the past three census periods. The theoretical importance of these and other findings uncovered here are discussed and interpreted through Wilson's (1987) notion of concentration effects and social isolation which may have transformed inner-city areas in recent years.