The objectives were to create a dose-response model of real-life vehicle collisions and to study the effect of measurement errors in real-life accident data. Errors in impact severity measurements can influence the understanding of injury risks and projections of safety benefits. A theoretical dose-response model of real life accidents is used to understand the complex links between impact severity and injury outcome. A simulation of the effects of errors was performed. Synthetic errors in the impact severity of different magnitude were introduced in a data set. In accident data with errors the number of injuries and fatalities will be overestimated at high and low severities. The slope of the risk curve will decrease, influencing the injury risk for a given severity. The consequences of using data with low accuracy may be severe, suboptimization is possible. Impact severity measurements with low accuracy will influence the understanding of the distribution of impact severity and injury risk. New methods, such as crash recorder techniques, used to increase the accuracy of impact severity measurements, are and will be more important in accident analysis. (Author/publisher).
Abstract