This report contains a detailed comparative study into differences and similarities between alternative risk and prognostic models for a number of developed countries. These countries are: The Netherlands, Great Britain, the former Federal Republic of Germany, Japan, and the United States. An attempt is made to explain the number of fatalities by both time development and by traffic volume as a measure of exposure. No predictions for the number of fatalities are made. No effort is given to explain the theoretical basis for the models. One of the goals of the research was to establish a framework in which it is relatively easy to estimate models and to compare them with alternatives. Once models are estimated, alternatives. Once models are estimated, a number of statistical measures are derived that can be compared. A detailed description is given of the selected statistical measures. Based on the five countries considered, six models are evaluated and compared. Four of these models are in fact exponential variants of the exponential development: the "Exp0 model"; the "Exp model"; the "ExpV0 model"; and the "ExpV model". The two other models are in fact logistic variants of the logistic development: the "Logis model"; and the "LogisV model". None of the models seems globally acceptable. Probably, the set of models should be extended. For the Netherlands, the logistic model seems to fit best of the six models considered. It fits substantially better than the exponential model used in the Policy Information System for Road Safety (BIS-V). It is suggested to replace the exponential model with the logistic model.
Abstract