In order to analyse the effects of different measures on the number of road traffic accidents, realistic statistical models are required. Two road safety problems are studied. The fit between some statistical models and empirical data is tested. The first problem is to find the effects of speed restrictions. Two different experiments with somewhat different structure are analysed. The purpose is to show that the models that are applied have a reasonable fit to the empirical data, to find estimates of the size of the effects and to give confidence levels for these estimates. The first experiment is concerned with the effects of the speed limits 90 km/h and 100 km/h relative free speed during the summers of 1961-1966. It is estimated that the speed limits have lowered the expected number of accidents by approximately 15-30%. The other experiment was made to study differentiated speed limits. Using data from 1968 and 1969 it is estimated that the effect of lowering the speed limits on some roads from 110 km/h to 90 km/h decreases the number of accidents by approximately 30%. In the report, confidence levels for the estimates are given. The second problem is how the traffic flow influences the distribution of accidents in single and multiple vehicle accidents. It is shown that models that are based on some simple theoretical consideration cannot be satisfactorily fitted to empirical data. This is probably due to the fact that our knowledge of the traffic and of the roads is not sufficient to construct realistic models of the variations between the accident types.
Abstract