Empirical Bayes Procedure for Ranking Sites for Safety Investigation by Potential for Safety Improvement

Abstract

The identification of sites requiring investigation for possible safety treatment is one of the most important aspects of infrastructure safety management and has been the subject of considerable research aimed at improving the efficiency of the process. The more recent techniques use the empirical Bayes (EB) method for estimating the safety of specific sites. A refinement of the EB method that is conceptually sound and inherently simple to apply is the focus of this research. With this refinement, the EB estimate of the safety of a site is compared with its expected safety to determine the potential for safety improvement, which forms the basis for ranking sites for safety investigation. A comparative evaluation is provided of the proposed method against other EB methods and the more conventional ones, with data for signalized intersections and two-lane rural highway sections. The refined EB method is shown to be comparatively efficient.