“…While the census block group is a commonly used unit of analysis in neighborhood level studies of violence, and is better theoretically grounded than the use of census tracts, zip codes, or counties, it contains artificially imposed boundaries that are used to capture neighborhood boundaries. The use of street segments that are comprised of street blocks, or street corners and block faces (e.g., Braga et al, 2011;Groff and Lockwood, 2014;Smith et al, 2000) might better represent natural neighborhood boundaries and reduce the likelihood of pattern obfuscation (Andersen and Malleson, 2011), making it a more appropriate spatial unit of analysis for estimating the association of alcohol outlets and neighborhood robbery occurrences. Fourth, our study does not control for the influence of other retail establishments that might serve as crime attractors (e.g., gas stations, laundromats, pawn shops, check cashing services, vice-related incidents) on neighborhood robbery levels, or directly test for availability of vulnerable targets (e.g., number of intoxicated individuals).…”