Investigation of Contributing Factors of Wrong-Way Driving Crashes on Partially/ Uncontrolled-Access Divided Highways

Abstract

Wrong-way driving (WWD) crashes are more likely to pose fatal and serious injuries than other types of highway accidents. This study focuses on differentiating the confounding factors that contribute to WWD crashes from other type of crashes on Alabama divided highways by performing statistical analysis. Crash data from 2009 to 2013, including 112 verified WWD crashes, are compared with 49,599 other type of crashes on the same class of roads during the same period. A simple descriptive data analysis was conducted to identify different explanatory variables contributing to WWD crashes. The results illustrate characteristics of WWD crashes, including temporal distribution, driver characteristics, vehicle characteristics, and environmental conditions. Also, the study provides a comparison of characteristics of WWD crashes between Alabama divided highways and freeways. The Firth’s penalized-likelihood logistic regression model was used to identify the statistically significant contributing factors. Odds ratios (OR) of different variables were calculated to measure how each factor affected WWD crashes when compared with other types of crashes. The results show that wrong-way (WW) drivers are more likely to be older and driving under influence (DUI) than other types of crashes on Alabama divided highways. Furthermore, WWD crashes were found to be more prevalent in urban areas and dark road conditions. Dark roadways with no lighting conditions were found to have the largest OR. Additionally, 18 crash entry points are confirmed according to the narratives on crash reports. Among these 18 WWD crashes with known entry points, eight of them occurred at median openings, and six WW drivers entered from parking lots of gas stations or other business area. To complement the analysis, the contributing factors of the fatal WWD crashes were also investigated. Finally, to mitigate WWD activities, countermeasures of three different groups, education, enforcement, and engineering were discussed based on the data analysis results.