Compared with the United States and other developed countries, the fatality crash rate of 100 million vehicles per kilometer on the freeways in China is still at a severe level, indicating it is of considerable significance to improve the safety of freeways. By developing a Bayesian generalized ordered logit model, the study analyzed the crash data of the Shenhai Freeway. The study adopted the crash severity determined based on crashes casualties and economic loss as the dependent variable. We fit a generalized ordered logit model and compare it with a traditional ordered logit model via Bayesian inference. The generalized model indicated the superiority by its better model fit as well as the statistical significance of the spatial term. Time of day, season, vehicle type, weather condition, crash cause, emergency medical services (EMS) time, and road geometry were found to have significant effects on crash severity propensity. The average marginal effects of the contributing factors on each crash severity level are also calculated. On the basis of estimation results, several countermeasures regarding traffic management rules, driver education programs, vehicle and roadway engineering, incident detection and reporting system, and emergency services are proposed to mitigate freeway crash severity.

CICTP