Evaluation of a Comprehensive and Coordinated Program to Reduce Community-Level Violence

Introduction: Residents of urban communities experience more violent crime than suburban or rural residents. However, urban crime varies tremendously by neighborhood within large urban cities, largely concentrated in neighborhoods with high levels of poverty and low resource availability. There is no simple or single answer to reducing violence. The problem is multiply determined and requires coordinated effort at multiple levels and across systems, focused within neighborhoods of highest risk. This paper presents data on the implementation and results of a 5-year project designed to evaluate the impact of a comprehensive violence prevention strategy among 10-24 year old youth in Chicago, IL that included: (a) CeaseFire, a multi-component intervention using a public health approach to reducing gun violence; working to change the behaviors of the highest risk youth and young adults who are most likely to be involved in gun violence.; (b) family-focused interventions targeting families of high risk children in early elementary and middle school grades; (c) a community safety committee including partners from social service agencies, business leaders and schools; and (d) public education and technical assistance related to program implementation and evaluation.

Methods: Using crime incident data from the police department for the years 2005-2015, pre-post and growth curve models were used to evaluate the impact of this comprehensive strategy when compared to other similar neighborhoods based on propensity score matching on characteristics of the neighborhood including indicators of economic disadvantage, violent crime, and gang involvement.

Results: Pre-post analyses found a 17% decrease in homicides over the course of 5 years, at the same time homicides increased by 10% in other similar communities and 9% across the city. A discontinuous growth model (Singer & Willett, 2003) was used to more rigorously assess the impact of implementation of the set of interventions on linear rate of change in crime incidents. Because the project was not expected to demonstrate an immediate impact on crime, a slope-only discontinuity was modeled. The dependent variables for these analyses were counts of incidents of violent crimes (i.e., all violent crimes and shootings). To account for this over dispersion in the outcome measures, we selected a negative binomial distribution for the regression models. Analyses found significant differences between the target neighborhood and comparison neighborhoods from pre-implementation to post-implementation, suggesting significant decreases in community-level violent crime.

Conclusions: Implications for violence prevention strategies to impact community-level, as opposed to individual-level, violence prevention will be discussed.