City-level urban form and traffic safety: A structural equation modeling analysis of direct and indirect effects

While most existing studies have examined the impact of the community- and street-level built environment on traffic safety, few have provided empirical evidence on the associations between urban form characteristics and traffic safety at the city level. To this end, this article first created a detailed list of 23 variables to measure city. level urban form of 100 major Urban Areas (UAs) in the United States and then applied factor analysis to construct five latent variables which describe urban form. Factor analysis is also used to define mediator variables reflecting citywide transportation network features and dependent variable of traffic safety. Structural Equation Modeling (SEM) is then used to investigate how city-level urban form, directly and indirectly (through mediators of transportation network features), affects traffic safety. Based on the statistical results, urban traffic is safer in UAs with more uniform job-housing balance among their different tracts, more polycentric design, and less low-density sprawl. In addition to spatial variation in employment and urban density that have significant direct effect on traffic safety, improving transportation network connectivity and increasing the supply of public transit facilities and upper-level transport infrastructures can decrease traffic fatalities indirectly, through encouraging the use of non-driving transport modes. It is estimated that a 10% increase in urban density as well as a 10% increase in even spatial distribution of employment can reduce the rate of fatal crashes by > 15%, on average. These findings demonstrate the importance of incorporating city-level land-use policies into the planning practices, in terms of traffic safety.