U.S. flag

An official website of the United States government, Department of Justice.

Natural History of Neighborhood Violence

NCJ Number
Journal of Contemporary Criminal Justice Volume: 20 Issue: 2 Dated: May 2004 Pages: 127-147
Jeffrey Fagan; Garth Davies
Date Published
May 2004
21 pages
This study estimated the effects of neighborhood disadvantage on cyclical and nonlinear patterns of violence in New York City from 1985 to 2000.
The neighborhoods were composed of several census block groups that were spatial units constructed by Jackson and Manbeck (1998) based on interviews with neighborhood residents and physical examination of naturally occurring neighborhood boundaries. These neighborhoods were small social areas where the effects of local social and economic contexts influenced both social control and crime opportunities. There were 285 neighborhoods in the final study sample. The study's ecology variable was a latent construct that represented the convergence of the components of concentrated disadvantage (poverty, labor market, segregation, supervision, anonymity, immigration, and housing structure). To represent neighborhood conditions more concisely, principal components analyses were used to construct a factor score for each dimension. Data on homicides and nonfatal hospitalized injuries were obtained from the Vital Statistics records and hospital admissions registers of the New York City Department of Health and Mental Hygiene. Individual growth curve models were developed for the neighborhoods, first estimating an unconditional growth curve model and then conducting a second estimation with covariates representing susceptibility. Models were estimated with time-varying covariates where both slopes and intercepts varied, and residual observations within neighborhoods were correlated through the within-tract error-covariance matrix. A measure was included to account for the endogeneity of crime and social disadvantage, which was the predicted value from a Poisson regression for the count in the initial year in each series (1985 for homicide, 1990 for violence) predicted from the ecology measures. The study found several elements in a natural history of violence, including its onset and increase, its persistence over time, the endogeneity of violence, and structural disadvantage within some neighborhoods, along with the diffusion of violence from one area to the next. During the homicide epidemic in New York City beginning in 1985, there were differences in the natural history of homicide but fewer differences when nonlethal violence was included. There were also differences in race-specific victimization patterns. The dynamics of the rise, spread, and decline of homicide differed for African-Americans, the racial/ethnic group that suffered disproportionately during this epidemic. Patterns of racial residential segregation and its collateral concentration of disadvantage in neighborhoods with high African-American populations combined to isolate homicides and minimize the spread to surrounding areas. At the same time, these same patterns sustained homicide at a higher rate in these areas, even as homicide rates were declining in other areas. Concentrated disadvantage was a consistent factor that made neighborhoods susceptible to elevated rates of homicide and violence, but segregation might have dramatically limited the spread of violence to surrounding neighborhoods. Gun violence is likely to spread across neighborhoods. The role of guns in homicide may explain why gun homicides are more likely to show a contagious pattern than violence generally. Guns have proven to be an agent in the transmission of violence and a "cancer" on social norms. 4 tables, 9 notes, and 72 references

Date Published: May 1, 2004