Crash prediction models for national and county roads in Norway were developed in order to calculate predicted numbers of injury crashes, slightly injured, seriously injured, fatalities, and the total number of killed or seriously injured, as a function of traffic volume, segment length, road category, speed limit, number of lanes, number of intersections and other road characteristics. Models were calculated as generalized negative binomial models (negative binomial models with a variable overdispersion parameter). Results can be used in evaluations of road safety measures with the empirical Bayes method which is regarded as state of the art with respect to control for regression to the mean. The models can also be applied in conjunction with several tools of the Norwegian Public Roads Administration for road safety analyses and estimation of expected effects of road safety measures. (Author/publisher)
Abstract