Het programma voor de herinrichting van de gevaarlijke punten op gewestwegen in Vlaanderen: een effectevaluatie

Abstract

In an attempt to work to a better traffic safety in Flanders and reach the goals of the decrease in numbers of severely and deadly injured, the Flemish Government decided in 2002 to select the most dangerous spots in Flanders and adapt their infrastructure. Spots were selected on the basis of the number and severity of injured as a consequence of traffic crashes in Flanders during three consecutive years. Up to now, the black spot program is still running, however already a lot of spots have been adapted. The Flemish Minister of Mobility and Public Works commissioned the Policy Research Centre Mobility & Public Works, track Traffic Safety to evaluate the effectiveness of those adapted spots on traffic safety. To examine the traffic safety effectiveness, the number of crashes before the adaptation of the infrastructure were compared with the number of crashes after the adaptation of the black spot. This comparison was made for all injury crashes on the one hand, and for severe crashes, which encompasses only crashes with severely and deadly injured, on the other hand. More specific the Empirical Bayes method was used, which is internationally accepted as the best standard in the evaluation of traffic safety measures. This method compares the number of crashes after the implementation of a measure, with the number of crashes before, taking different confounding variables into account. Those confounding variables are general crash trend, general changes in traffic volumes, and the regression to the mean phenomenon. In this research the crash trend was taken into account trough a comparison group. This group encompassed the dangerous spots which were adapted after 2008 or are not adapted yet up to now. This comparison group was handled as these locations are comparable to the research locations for different characteristics, since these are selected from the same set of dangerous spots. However they differ from the research locations because no traffic safety measure was executed during the research period. Secondly, the total crash trend for Flanders was used as an additional control. Next to the crash trend also the regression to the mean effect, which is an important confounding variable, was controlled. To evaluate the effectiveness of the adaptation of a traffic safety measure at a certain location, at least one year of crash data before and after the adaptation of the black spot is necessary. To make a selection of the crashes around the black spot, those crash data have to be geographically located. These located crash data are available for Flanders until 2008. Subsequently only black spots that were adapted and open for traffic until 2007 could be evaluated. This leaded to a final research group of 134 black spots, which were all intersections. The first comparison group, that encompassed all dangerous spots that were adapted after 2008, consisted of 211 black spots. A meta-analysis of the 134 black spots, found a decrease in crash rates of 24% to 27%, dependent on the comparison group that was used. This decrease is attributable to the adaptation of the black spots, as possible confounding factors were controlled. Also for the severe crashes a significant decrease was found. Dependent on the comparison group, a decrease of 40 to 52% in the number of severely and deadly injured was found. Next to this, a comparison was executed to examine whether there was a difference in effectiveness, dependent on the characteristics of the intersection and the type of adaptation. However, only a limited number of useful conclusions could be made. An almost significant difference was found dependent on the type of signalisation in the period before the adaptations were executed. The spots that were priority controlled showed a higher effectiveness compared to spots with traffic lights. Also a significant higher effectiveness was found for locations with a lower traffic volume compared to locations with a higher volume. Next to the crash rates, it was also examined whether there was a difference in the number of injured road users. Here a distinction was made between the different kinds of road users. For every road user, that is car driver and passengers, moped rider, cyclist, motorcyclist, pedestrian and truck driver, a decrease was found from the period before to after the measure.