The objectives of this research project included the specific evaluation of a variety of design improvements for safety as well as the development of estimation techniques by which accident frequency and severity could be predicted and used for allocation of safety improvement funds. The former objective was achieved by studying 326 sites in Alabama, both test and control, to determine the effect of 1)4 different types of improvements. Both parametric and non-parametric statistical tests were applied, resulting in estimates of accident frequency and severity reduction. The ratio of PDO to total accidents, defined in this project as the severity ratio, was used for the latter estimate. Generally, the confidence intervals for these estimates were quite wide. A procedure was developed to translate such estimates into budget allocation policy by the use of dynamic programming. This volume is the first of a three-volume final report. It contains a summary of the research methodology and the essence of the findings. Volumes II (Research Report - FHWA/RD-81/180) and Volume III (Users Manual - FHWA/RD-81/181) are unpublished. (Author/publisher)
Abstract