Dr. Brian Soller
Associate Professor of Sociology, Anthropology, & Public Health (SAPH) UMBC

Spatial clustering in crime is often treated as a statistical nuisance—modeled as spatial autocorrelation rather than explained. This talk reframes spatial dependence as a social process largely generated by routine human mobility. Integrating methods and insights from sociology, geography, and social network analysis, I use high-resolution GPS location data from a large panel of Baltimore-area residents to construct street-level mobility networks that capture patterns of street use by locals and non-locals, as well as network ties between street blocks formed through shared movement pathways. I show how these mobility-based connections help explain both crime concentration within blocks and spillover effects across connected blocks. No prior knowledge of R or advanced programming is required; rather than focusing on technical mechanics, the talk emphasizes how integrating theory and methods across traditionally siloed fields allows computational social science to identify the social processes that generate spatial patterns.