Effects of Signalization on Intersection Crashes

This research performed a study of newly signalized intersections in Florida to evaluate the impacts of signalization on crashes. Crash data from the Florida Department of Transportation (FDOT) crash database were used. Statistical crash prediction models were developed to estimate average number of crashes in terms of all crashes and types of crashes at intersections before and after signalization. Seven predictor variables were finally considered for the models, which included Average Daily Traffic (ADT), surrounding land use, location type, number of lanes, posted speed, median, and shoulder type. Poisson regression and negative binomial regression were used during modeling. The goodness-of-fit of developed models were evaluated based on Pearson's R-square and likelihood ratio index. Furthermore, average number of crashes at an intersection before and after signalization could be estimated with the developed models given the intersection characteristics to determine the impacts of signalization through the changes of crash frequencies.