Procedure for Ranking Unsignalized Rural Intersections for Safety Improvement

Abstract

This paper presents a procedure for ranking rural unsignalized intersections that uses quantitative safety evaluations performed as part of the safety inspection process. The procedure might be effective for the selection of cost-effective treatments at intersections and might be quite helpful for administrations that do not have high-quality crash data and for those that manage low-volume roads for which crash data cannot give enough information to help prevent future crashes. The procedure evaluates a safety index (SI) that can be used to rank intersections for further investigation. The SI can be assessed whether crash data are available or not. If crash data are available and their quality is good, the SI can be effectively used in conjunction with the empirical Bayes (EB) estimate of frequency as ranking criteria. If crash data are not available or poor, the SI can be used as a proxy for crash data and becomes the only ranking criterion. Validation of the SI procedure was performed by comparing the results with EB safety estimates. The SI was assessed in 22 three-leg intersections in Italy. In the same intersections, a safety performance function was calibrated and the EB refinement technique was used to obtain a better estimate of existing safety performance. Correlation between SI values and EB safety estimates was highly significant, with 84% of the variation in the estimated number of crashes explained by the SI value. The results from a Spearman's rank correlation show that rankings from the SI and the EB estimate agree at the 99.9% significance level.