Application of Multinomial Regression Model to Identify Parameters ImpactingTraffic Barrier Crash Severity

Background:Run Off The Road (ROTR) crashes are some of the most severe crashes that could occur on roadways. The main countermeasure that can be takento address this type of crash is traffic barrier installation. Although ROTR crashes can be mitigated significantly by traffic barriers, still trafficbarrier crashes result in considerable amount of severe crashes. Besides, the types of traffic barriers, driver actions and performance play animportant role in the severity of these crashes. Methods:This study was conducted by including only traffic barrier crashes in Wyoming. Based on the literature review, there are unique contributoryfactors in different crash types. Therefore, in addition to focusing on traffic barrier crashes, crashes were divided into two different highwayclasses: interstate and non-interstate highways. Results:The result of proportional odds assumption was an indication that multinomial logistic regression model is appropriate for both non-interstate andinterstates crashes involved in traffic barriers. The results indicated that road surface conditions, age, driver restraint and negotiating a curve weresome of the factors that impact the severity of traffic barrier crashes on non-interstate highways. On the other hand, the results of interstate barriercrashes indicated that besides the types of barriers, driver condition, citation record, speed limit compliance were some of the factors that impactedthe interstate traffic barrier crash severity. Conclusion:The results of this study would provide the policymakers with the directions to take appropriate countermeasures to alleviate the severity of trafficbarrier crashes