This report documents the results of the first year of work on a two-year project to develop a practical procedure for identifying potentially high accident locations throughout the Texas highway system. The procedures considered in this report are basically two: (1) control chart methodology and (2) the statistically smoothing of traffic accident rates by means of Empirical Bayes methodology. The Empirical Bayes methodology, is discussed in detail and demonstrated by means of three example problems. The use of Empirical Bayes methodology to detect location that (1) may be over represented in traffic accidents or (2) may be over represented in accident severity (or, more accurately, societal costs) is explained. Plans for the development and implementation of Empirical Bayes methodology into the State's procedure for identifying and rank-ordering highway locations that are potential candidates for safety treatments or other remedial countermeasures are discussed.
Samenvatting