Document Type

Working Paper

Publication Date



Since 1968, violence and other crimes in New York City have followed a pattern of recurring epidemics. There have been three consecutive and contiguous cycles characterized by sharp increases in homicides and assaults to an elevated rate followed by equally steep declines to levels near the previous starting point. The most recent epidemic, from 1985-96, had the sharpest rise and steepest decline of the three epidemics. Popular explanations of the current epidemic fail to account for both the rise and fall of the decline, or for the repetitive pattern of these epidemics. In this article, we use public health data to identify factors associated with the cyclical rise and fall of homicides and non-lethal injury violence in this most recent of the three epidemics. Homicides in this period were concentrated among minority males, ages 15-24, while victimization rates for females of all ages remained stable. Gun homicides and non-lethal gun assaults accounted for all the increase and decline in interpersonal violence from 1985-95; intentional injuries caused by other means also were stable or declining over this period. Next, we use hierarchical linear regression models, with a rich set of time-varying covariates and controls for both temporal and spatial autocorrelation, to identify whether the rise and fall of homicide are explained by processes of diffusion across adjacent neighborhoods. We estimate the probabilities of homicides and assaults in a neighborhood controlling for rates of homicide or assault in the surrounding neighborhoods in the preceding year, and find that gun violence diffuses across neighborhoods over time. Epidemic patterns of violence disproportionately affected African Americans as victims of gun violence, both homicides and non-lethal gun assaults. Diffusion was strongest in neighborhoods where social control was compromised by extreme poverty and concentrated racial segregation. Concentrations of immigrant households in neighborhoods were a protective factor in suppressing the violence epidemic. We then use social contagion theories to link contagion of violence across neighborhoods to individual data on social interactions that may animate the transmission and diffusion of violence. Analyses of the social contexts and interpersonal dynamics of violent events reveals how street interactions in interpersonal disputes link individuals within and across social networks in a competition for status that is skewed by the presence of firearms. Decisions to carry, show and use weapons are based on perceptions of threat and danger, which are shaped by the presence of firearms and the potential stigma associated with non-action. These events occur across multiple contexts - bars, streetcorners, drug markets - to reinforce perceptions of risk and expectations of the threat or reality of lethal violence.


Criminal Law | Law | Law and Society