This report outlines the methods that can be used in the evaluation of road safety treatments, and discusses the data requirements, and the advantages and disadvantages of each approach. It is designed to assist selection of the most appropriate evaluation method and make users aware of the limitations associated with different methods. The report considers four different aspects of report validity: construct validity, internal validity, external validity, and statistical conclusion validity. It also considers the threats to report validity, the most pervasive of which is the regression to the mean effect. Different methods for conducting evaluations based on crash reductions are described, and their advantages and disadvantages discussed. These include cross-section studies, simple beforeafter studies, before-after studies with control sites, Empirical Bayes, and randomised control trials. The practical implications of the different methods in terms of assessing the validity of studies and of conducting evaluations are discussed. The report concludes with six case studies which illustrate the methods described. (Author/publisher)
Abstract