A spatial analysis of risk factors associated with road collisions in Ciudad Juarez, Mexico and using a geographically weighted regression approach

Abstract

This study examined changes in the spatial patterns and determinants of road collisions in Ciudad Juarez, Mexico in 2019 and 2020, encompassing the initial period of COVID-19 mobility restrictions. Kernel density estimation and local indicators of spatial association were used to compare collision distributions and identify significant clusters between years. A geographically weighted negative binomial regression model then generated local coefficients to analyze how demographic, socioeconomic, land use, and road network factors influenced collision probability spatially. Results show collisions decreased 13.18% in 2020 but clustered differently, validating restrictions' impact. Population aged 15–64 and industrial land uses significantly increased risk spatially, whereas commercial uses decreased. Lower socioeconomic conditions also correlated with higher risk. Younger populations presented varying collision likelihoods intra-urbanly. This research thus emphasizes how local contexts shape risk factors’ effects and informs data-driven safety policies accounting for place-specific issues. By adopting an explicitly spatial modeling approach, localized risk patterns were uncovered not detectable through traditional methods.