Ranking and selecting dangerous accident locations: correcting for the number of passengers and Bayesian ranking plots

Abstract

In Flanders (Belgium), approximately 1014 accident locations are currently considered as ‘dangerous’. These ‘dangerous’ accident sites are selected by means of historic accident records for the period 1997-1999. More specifically; a combination of weighting values, respectively 1 for each light injury, 3 for each serious injury and 5 for each deadly injury (1_3_5), is used to rank and select the most dangerous accident locations. In the first part of this paper a sensitivity analysis is performed to investigate the influence of the number of passengers on the ranking of the accident locations. Secondly, the use of Bayesian ranking plots in order to visualize the probability that a location will be ranked as dangerous, based on estimates from a hierarchical Bayes model is evaluated. Results show that giving weight to the severity of the accident instead of to all the injured occupants of the vehicle does have important consequences for the selection and ranking of dangerous accident locations. Government should therefore carefully decide whether to rank accident locations by means of the severity of the accident or the severity of the injured occupants. Secondly, probability plots can provide policy makers with an graphical instrument to select dangerous road locations on a statistically sound basis.