Severity analysis of vehicle crashes on highway-rail grade crossings: ordered response logit modeling

The purpose of this paper is to develop an ordered response logit (ORL) model to explore the impact of various variables on three different severity levels of vehicle-related crashes at highway-rail at-grade crossings in the U.S.A. An ORL model is developed and marginal effects are also calculated. The ORL modeling results indicate that vehicle-rail crashes occurring with higher vehicle and train speeds, at higher temperatures (greater than 80oF), male drivers, and drivers with an age above 25 years, all have detrimental effects of increasing the chance of fatal and injury crashes. On the other hand, it is also shown that truck-trailer vehicles, vehicle-rail crashes occurring under rain and snow weather conditions, on HRGCs located in commercial areas, residential areas, industrial areas and institutional areas, with asphalt, and asphalt and flange crossing surface types, and with AADT being greater than 20,000, all decrease the likelihood of injury related and fatal crashes.