Risk Terrain and Multilevel Modeling of Street Robbery Distribution in Baltimore City

Abstract

The structure and functions of neighborhoods determine the impact of measures used to estimate the distribution of street robbery across space and time. The risk of street robbery could vary between different sections of the area. Prior research has typically relied on one-dimensional analysis, which sparsely accounts for the hierarchical configuration of a neighborhood’s influence on spatial crime distribution. Much less is known about how the predictor variables moderate each other at different neighborhood levels. Data was collected from 13,789 US Census blocks ( N = 13,788) aggregated to administrative neighborhoods ( N = 278) to examine how neighborhood structure affects street robbery distribution at two spatial levels. A Risk Terrain Model was adopted to develop a physical environment risk measure—Aggregate Neighborhood Risk of Crime (ANROC), which, alongside sociodemographic risk factors, predicted the outcomes of the Negative Binomial General Linear Regression models. Study findings suggest that individual-level (block) and group (neighborhood) level predictors influenced street robbery. Both group-level physical environment and total population moderated the impact of the individual-level physical environment on street robbery incidents.