Safety of rural highway users in the United States has become a critical issue, as such highways account for majority of fatal crashes. This paper presents the details of an investigation aimed at identifying critical factors contributing towards increased crash severity in rural areas. Crash data from KARS (Kansas Accident Reporting System) database was analyzed and crash severity was modeled using ordered choice models. These ordered choice models are capable of distinguishing the differences between two ranked levels qualitatively, in this case, between two crash severity levels. The estimated probit model provided satisfactory results in predicting crash severity as illustrated by model fitting statistics. The results indicated that many driver related factors, such as alcohol involvement, lack of seat belt usage, excessive speed, and driver ejection or being trapped due to the crash are contributory towards increased severity of crashes in rural highways. It also showed that, severities of single vehicle crashes are higher as compared to two vehicle and animal-vehicle crashes. However, when two vehicles collide, head-on, angle, rear-end and sideswipe collisions have higher propensity of resulting in higher severities. Roadway geometry related parameters, such as curved and graded roads are also contributory towards increased crash severity in rural areas. In contrast, under wet road surface conditions, probability of having a more severe crash is low. Driver cautiousness under such conditions resulting in reduced speeds might have led to this situation. Based on the identified contributing factors, countermeasure ideas are suggested for improving highway safety in rural areas (A). For the covering abstract of the conference see E217780.
Abstract