Key factors contributing to crash severity at highway-rail grade crossings

Abstract

The purpose of this paper is to develop and compare the preferred multinomial logit (MNL) and ordered logit (ORL) model in identifying factors that are important in making an injury severity difference and exploring the impact of such explanatory variables on three different severity levels of vehicle-related crashes at highway-rail grade crossings (HRGCs) in the United States. Vehicle-rail crash data on USDOT highway-rail crossing inventory and public crossing sites from 2005 to 2012 are used in this study. Preferred MNL and ORL models are developed and marginal effects are also calculated and compared. A majority of the variables have shown similar effects on the probability of the three different severity levels in both models. In addition, based on the Akaike information criterion, it is found that the MNL model is better than the ORL model in predicting the vehicle crash severity levels on HRGCs in this study. Therefore, the researchers recommend the use of MNL model in predicting severity levels of vehicle-rail crashes on HRGCs.