Report numberRA-2005-74
TitleThoroughfares in Flanders
SubtitleA risk analyses based on road and environment characteristics
AuthorsKurt Van Hout
Elke Hermans
Erik Nuyts
Tom Brijs
Published byPolicy Research Centre for Traffic Safety 2002-2006
Number of pages108
Document languageDutch
Work packageOther: Infrastructure and space

This report is the second of a 3 report cycle concerning traffic safety in the built-up environment. The first report was a literature review in which the influence of several infrastructure characteristics on traffic safety in the built-up environment was studied. In this study a risk analysis is made for throughroads in relation to certain elements of the cross section. A next report will study traffic safety applied on a throughroad as a whole.


The data used in this study come from some inventories made in 2000 for some other projects. The database contains a large amount of road segments spread all over the Flanders Region. Accidents from 1996 till 2001 are used.


The study aims at determining the impact of specific characteristics of traffic, road and surroundings, which are readily available, on traffic safety. This is done in a cross sectional study. The result will be a number of models that give the number of accidents (total number as well as number of accidents with bicyclists or mopeds) related to traffic intensities and road and surrounding characteristics.


Within the statistical software package SAS models are built, by means of the procedure GENMOD, that give the number of accidents and the number of accidents with bicyclists (or mopeds) as a function of vehicle intensity, bicycle (+moped) intensity (only for accidents concerning bicyclists or mopeds), vehicle and bicycle (+moped) intensity and lastly vehicle and bicycle (+moped) intensity as well as road and surrounding characteristics. All models are based on the same basic dataset, no division is made between road or accident type (unless for accidents with bicyclists or mopeds).


Additionally, the existence of different groups of road segments within the data is checked by means of latent class analysis. This seems to be the case only to a limited extent. A fairly small group of segments is characterized by a significantly larger number of accidents. These road segments therefore deserve some extra attention. Anyway, the differences with the models derived earlier are limited.


The models explain the variance in number of accidents for existing road segments. The connections found aren’t necessarily causal. The models are therefore not suitable to predict changes in the number of accidents following a street renovation. The models can be used for estimating the number of accidents in relation to intensities, road and surrounding characteristics. As such they are useful in the creation of a comparison group in Before-After studies and for the detection of dangerous zones.


The results show that the connection between vehicle intensity and the number of accidents isn’t as easy as usually supposed. For large vehicle intensities the number of accidents is flattened down. The number of bicycle (=moped) accidents grows as expected with a growing number of bicyclists and mopeds.


Apart from the traffic intensities, road characteristics like the number of lanes, the presence of foot and bike paths or parking lanes, amongst others, but also surrounding characteristics as building density, morfological location type and the presence of certain functions along the road influence traffic safety. Interpretation of the coefficients of the different explaining variables however is complicated by the probable presence of correlations between the variables.


The impact of road and surrounding characteristics differs according the road users involved and the location of the accidents. Some variables influence stronger the crossroad accidents, while other variables have a stronger impact on road segment accidents. A striking difference exists in the expected number of bicycle accidents on roads with bicycle paths for both directions on one side of the road and roads with one way bicycle paths on both sides. Cyclists are more at risk on two way cycle paths compared to one way cycle paths.

DownloadPDF icon RA-2005-74.pdf


The Policy Research Centre for Traffic Safety carries out policy relevant scientific research under the authority of the Flemish Government. The Centre is the result of a

cooperation between Hasselt University, KU Leuven and VITO, the Flemish Institute for Technological Research.


Leuven vito