Redefining Neighborhoods: Built Environment’s Multiscale Impact on Intimate Partner Violence

• In Event: Rethinking Place and Spatial Dynamics

Abstract

This study addresses the critical challenge of defining neighborhoods for built environment variables to better understand their association with intimate partner violence (IPV) rates. Conventional approaches aggregate built environment data using administrative units (e.g., census block groups), which may misalign with the conceptual “proper neighborhood”—characterized by individual-centered, overlapping, and multiscale spatial dynamics. We propose the generalized additive smoothing (GASS) framework to define proper neighborhoods and compare its utility against traditional administrative definitions. Focusing on Chicago, we analyze IPV rates at the census block group level while controlling for socioeconomic confounders. Built environment variables are aggregated using both administrative units (aligned with outcome geography) and GASS-derived neighborhoods, enabling direct comparison of regression model performance. Results reveal that GASS-defined neighborhoods uncover distinct spatial scales at which specific built environment factors influence IPV rates. Some variables lose significance as predictors at administrative units but become significant at GASS-defined neighborhoods. Notably, some variables (e.g., liquor stores) emerge as proxies for socioeconomic conditions when modeled at their optimal scales. This underscores the importance of spatial definition choices: variables at improper neighborhood definitions may conflate built environment effects with socioeconomic pathways.

Authors

• Mengyu Liao, University of Maryland, College Park

• Taylor Oshan, University of Maryland, College Park