Estimation of side-impact crash risk at signalized intersections

Side impact accidents are considered to be the most dangerous of all types of intersection accidents due to their high severity. In-depth investigation of accident occurrence could be a valuable means of mitigating these accidents. Based on the relationship between the distribution of disturbances near intersections and driers reactions, this study developed a logic for three major types of signalized intersections side impact crashes were considered for this study - right-angle, and left and right-turning crashes. The study developed models to understand the relationship between accident frequency and some explanatory variables that represent, driver, vehicle, traffic flow and intersection design characteristics. Negative binomial regression with maximum likelihood estimation of parameters is applied to address the overdispersion usually found in accident count data. The models explain the mechanism of side-impact accident occurrence and could be used to assist safety management agencies to devise countermeasures aimed at divers, and the physical roadway environment.