Modeling Severities of Gravel Road Crashes Using Random Parameters

Abstract

Gravel roads crashes are a roadway safety hazard and such crashes have higher probability of fatalities and serious injuries when compared to paved roads crashes. Previous studies investigated the contributing factors that affect gravel roads crash severities. There were widely range of contributory factors such as road conditions, weather conditions, vehicle type, driver behavior and characteristics, road geometric features, terrain, and environmental conditions that were investigated. However, no previous studies accounted for unobserved heterogeneity in the crash data while modeling gravel roads crash severities. This study employed random-parameter binary logit model also known as mixed logit model to model severity of gravel road crashes in Wyoming. Gravel roads crash data from Wyoming Department of Transportation from 2010 to 2019 were studied and analyzed.  Both binary logistic regression model and mixed binary logistic regression model were developed. Mixed logistic binary regression model was found to be better fit the data in terms of goodness of fit statistics.  The results showed that crashes involving motorcycles had the higher risk of fatalities and serious injuries. In addition, horizontal curves and male drivers were found to have significant effect on the gravel roads crash severities. Passenger cars were found as a random parameter from the results of mixed binary logistic regression model. In addition, existence of horizontal curve was found to be the most contributing factors increase the probability of having minor or possible injuries in gravel roads accidents. Furthermore, having a snow on the road surface was found to have significant effect on increasing the probability of resulting in property damage only accidents. The results found in this study will be helpful to identify effective countermeasures which will reduce gravel road crash severity. Thus, applied variable speed limits at the horizontal curves location and at snow falling time will enhance the gravel roads safety.