Race, Space, and the Spread of Violence Across the City

A central explanation for elevated violence in urban African American neighborhoods is that the relationship can be accounted for by race-related differences in socioeconomic conditions. Yet recent studies have shown that predominantly African American neighborhoods experience higher rates of violence relative to other ethno-racial communities, even when socioeconomic conditions can be held constant. Explanations for these differences center on the deleterious effects of extensive residential segregation and the existing racial order privileging whites and their communities. This study examines whether the concentration and spread of urban violence can be characterized as a racially invariant process over half a century. In a detailed case study of the prototypical "rustbelt" city of Buffalo, New York, between the 1950s and the 1990s, negative binomial regression analyses fail to support the racial invariance thesis for predicting intra-neighborhood homicide rates. Multinomial logit regression models show that racial composition is also associated with vulnerability to inclusion in a cluster of highly violent neighborhoods as homicide diffused across the city, especially in recent decades. Thus, the spatial organization of urban African American neighborhoods generates a unique racialized vulnerability to the extralocal diffusion of violence; there is, in effect, an ecological racial variance in the spread of violence across the city. Moreover, this process has become increasingly entrenched over time, undermining any assertion that the urban experience reflects a post-racial America.