Safety Impacts of Converting Stop-Controlled Intersections in Ottawa to Roundabouts

Abstract

Roundabouts are becoming increasingly popular in North American road networks. They are used as a method for reducing traffic conflicts and enhancing road safety. This paper provides a statistical assessment of the impacts of roundabouts as modern safety treatment in the City of Ottawa. The assessment uses two statistical methods, which are Negative Binomial (NB) regression and Empirical Bayes (EB) before-and-after study to account for Regression to the Mean and time trend effects. The intersections within the City of Ottawa that were re-constructed as roundabouts within the period of 2012-2016 were taken as the study sample in this study. The results of the NB analysis showed significant Roundabout variable in total collisions, property damage only (PDO) collision severity and all collision impact type except for single motor vehicle (SMV) collisions. The results showed an increase in total, PDO, rear-end, and sideswipe collisions and decrease in angle collisions. The EB before-and-after study findings confirmed the increase in total, SMV, Sideswipe, rear-end and PDO collisions and the reduction of angle collisions, although the increase in SMV collisions was not statistically significant. The EB analysis also showed a 42% reduction in the number of injury+fatal collisions, which is very close to the estimated 48% reduction in angle collisions. In conclusion, this study shows an overall positive safety impact of roundabouts because of an approximately 42% reduction of the severe collisions that result in injury or fatality despite an increase of around 15% in the total number of collisions.