Accident prediction models have been used to identify factors that may contribute to crashes at a particular location. Models have been developed separately for highway locations and for intersections. Few accident prediction models have been developed specifically for truck crashes on roadways affected by traffic signals. On arterial roadways with a large number of intersections, it is difficult to isolate the relationship between accidents and specific geometric features, including intersections, because of the proximity of many of these elements to each other. For this reason, an accident prediction model that treats intersections as a geometric feature on the roadway may allow for both the impact of the intersection and of the geometric features of the adjacent roadway segments to be accounted for in one model. By using a database developed for truck accidents in New Jersey for 1998 and 1999, an overall analysis of truck accidents at signalized intersections along US-1 in New Jersey was performed. A Poisson regression and negative binomial model approach was used, and an accident model for estimating accidents on arterial roadways influenced by signalized intersections was determined.
Abstract