Analyzing types of freeway crashes using nested logit model

Road crashes and their consequences are one of the most important problems that affect people's lives. In order to reduce the fatalities and related cost of crashes, traffic safety researchers are continuously investigating approaches to reduce the occurrence and consequences of crashes. Crash type modeling is one of the most common tools for road safety goals in transportation facilities, and the purpose of crash type modeling is to establish a relationship between the frequency of crashes based on its type and other effective variables. One of the advantages of crash type models is that with the help of these models, it is possible to identify the places where there is a possibility of a certain type of crashes and to examine the effect of different variables on different types of crashes. In this research, using the data of freeway accidents in Iran, the type of crash was identified with a new approach called nested logit model. To this aim, crashes were initially divided into two categories of single-vehicle and multi-vehicle crashes, and then single-vehicle crashes were divided into three categories of collision with a fixed object, off-road crashes, and overturning crashes, and multi-vehicle accidents were divided into two categories of collision with a vehicle and multi-vehicle collision crashes. Then the effect of different variables of environment, road, driver, and causes of crashes with different types of crashes were investigated and the effect of significant variables on each type of crash was explored with marginal effect.