Modeling spatial relationships between multimodal transportation infrastructure and traffic safety outcomes in urban environments

The interest in multimodal transportation improvements in urban areas is increasing in cities across the U.S. Improved access to multimodal transportation attracts new users, but can possibly increase their exposure to risk from crashes, particularly in areas where the “safety in numbers” phenomenon does not exist. The relationship between access to multimodal transportation and safety in urban environment is complex, as non-motorized user vulnerability becomes a predominant risk factor when they are involved in a crash, even at lower vehicle speeds. This paper aims to evaluate the relationship between multimodal transportation infrastructure, expressed through the presence of multimodal facilities and user exposure, and traffic safety outcomes. Using Chicago as a case study, a comprehensive dataset is developed that significantly contributes to the existing literature by including socio-economic, land use, road network, travel demand, and crash data. Area-wide analysis on the census tract level provides a broader perspective about safety issues that multimodal users encounter in cities. Negative-binomial regression models with fixed and random effects are estimated to account for data overdispersion and spatial effects. Total vehicle-only crashes, total crashes with non-motorized users, and fatal vehicle-only crashes are modeled. The results show strong association between the variables related to multimodal transportation availability and usage, and both motorized and non-motorized crashes. Although simplified in terms of some spatial correlation assumptions, demonstrated methods prove to be a beneficial and computationally efficient tool for estimating and easily interpreting modeled relationships. Further research efforts to address the limitations of the presented approach are proposed.