Community violence and academic achievement: High-crime neighborhoods, hotspot streets, and the geographic scale of "community".

Numerous studies have demonstrated a negative relationship between community violence and youth academic achievement, but they have varied in their geographic definition of "community," especially as it relates to proximity to students' residences. We extend this by considering the independent relationships between academic achievement and violent events (from 911 dispatches; e.g., gun shots) at the neighborhood (i.e., census tract) and street-block levels. We use data from standardized Math and English Language Arts (ELA) tests from Boston, MA for 2011-2013. Exposure to community violence was partially independent between streets and tracts, with some students living on low-crime streets in high-crime neighborhoods or high-crime streets in low-crime neighborhoods. Initial regression models found that differences in a neighborhood's violent crime predicted up to a 3% difference in test scores on both Math and ELA tests. Students living on high-crime streets scored an additional 1% lower than neighbors on safer streets. Subsequent models with student-level fixed effects, however, eliminated these relationships, except for the effect of neighborhood-level violence on Math scores. These findings suggest that future work should consider community violence at both geographic scales, but that in this case the impacts were only consistent at the neighborhood level and associations at the street level were seemingly due to spatial segregation of households.