Estimating the Effects of Safe Streets Baltimore on Gun Violence 2007–2022

By Daniel W. Webster, Carla G. Tilchin, Mitchell L. Doucette

  Background Safe Streets Baltimore (Safe Streets) is a community violence intervention (CVI) program designed to reduce gun violence in neighborhoods with high levels of gun violence. Frontline workers are recruited for their ability to connect with individuals at highest risk for involvement in gun violence and mediate disputes, promote nonviolent norms for settling disputes, and connect program participants to services. Baltimore has fully implemented the program in 11 neighborhoods between 2007 and 2021. Six of these sites have been fully operational for less than three years and have not been previously evaluated. Prior evaluations of Safe Streets have shown mixed results across the sites and over time. Study Methods To estimate program effects, we analyzed variation in neighborhood-level monthly counts of homicides and nonfatal shootings for the period January 1, 2003 through July 31, 2022. The primary analyses were augmented synthetic control models for each site. This method generated a “synthetic” comparison for each Safe Streets site using a weighted combination of data from neighborhoods that did not implement the program but had similar levels and trends of violence before program implementation. We calculated program effects comparing treated sites to their synthetic controls, estimating what would have happened if Safe Streets had not been implemented. Because confidence in forecasts from statistical models tends to decrease over long periods of time, we generated estimates for the first four years of program implementation for the longer running sites in addition to estimates of the entire time a Safe Streets site has been in operation. We calculated average effects across all sites and within strata of site tenure (longer running and new sites) weighted by the precision of each site’s estimated effects. Key Findings: During the first four years of program implementation across the five longer-running sites, Safe Streets was associated with a statistically significant average reduction in homicides of 32%. Over the entire study period among these longer-running sites, homicides were 22% lower than forecasted if the program had not been implemented. Three of the five sites had significant reductions ranging from 28% in McElderry Park to 48% in Lower Park Heights. In SandtownWinchester, Safe Streets implementation was associated with a significant increase in homicides. Estimates of Safe Streets effects across the six new sites varied with an average reduction of 8% that was not statistically significant. Over the entire study period across all sites, Safe Streets was associated with a statistically significant 23% reduction in nonfatal shootings. Eight of the 11 sites had program-related reductions in nonfatal shootings. Four sites had significant reductions ranging from 29% in Lower Park Heights to 84% in Franklin Square. Sandtown-Winchester’s site was associated with a 53% reduction in nonfatal shootings over a period of more than seven years

Baltimore: Johns Hopkins University, Bloomberg School of Public Health, Center for Gun Violence Solution, 2023. 34p.