Externality of risk and crash severity at roundabouts

Abstract

The severity of 1491 crashes on 148 roundabouts in Flanders-Belgium was examined in order to investigate which factors might explain the severity of crashes or injuries and to relate these factors to the existing knowledge about contributing factors for injury severity in traffic. Logistic regression and hierarchical binomial logistic regression techniques were used. A clear externality of risk appeared to be present in the sense that vulnerable road user groups (pedestrians, bicyclists, moped riders and motorcyclists) are more severely affected than others. Fatalities or serious injuries in multiple-vehicle crashes for drivers of four-wheel vehicles are much rarer. Injury severity increases with higher age. Crashes at night and crashes outside built-up areas are more severe. Single-vehicle crashes seem to have more severe outcomes than multiple-vehicle crashes. However, systematic differences in the reporting rate of crashes are likely to exist and may have affected the stated results. Correlations with important, but unobserved variables like the impact speeds in the crashes might exist as well and could provide an alternative explanation for some results.