Unravelling the multiple effects of multilevel neighborhood characteristics on traffic crash risk from a spatiotemporal heterogeneity perspective

Although extensive research has explored the relationship between traffic crashes and neighborhood factors, the multiple effects—global, local, and interactive—of these factors on traffic crash risk (TCR) remain unclear. This study addresses this gap by developing a comprehensive analytical framework that integrates detailed crash data with sociodemographic, macroscale built environment (BE), and microlevel street environment (SE) factors. Using optimal parameter geographic detectors (OPGD) and multiscale geographically weighted regression (MGWR), we systematically explore the spatiotemporal heterogeneity and interaction effects of neighborhood factors on TCR in Hong Kong. The results reveal that BE and SE factors exert a stronger and more direct influence on TCR than traditional sociodemographic variables, with significant spatiotemporal variations. Specifically, fire hydrant and intersection density, as well as visual complexity, display strong localized effects, while factors such as street greenery and life facility density exhibit stable global patterns. Furthermore, interaction analysis uncovers nonlinear synergies, with certain factor combinations—such as healthcare facility and intersection density—substantially amplifying nighttime TCR. These findings enhance the theoretical understanding of TCR and provide policymakers with data-driven insights for developing spatiotemporal adaptive traffic safety strategies, ultimately fostering safer and more resilient urban environments.