Repository hosted by TU Delft Library

Home · Contact · About · Disclaimer ·

The analysis of Dutch national data on heavy truck accidents: The necessary extension of traditional frequency counts with logistic regression analysis

Publication files not online:

Author: Vries, Y.W.R. de · Mooi, H.G.
Institution: TNO Wegtransportmiddelen
Source:ESV 17th International Technical Conference on the Enhanced Safety of Vehicles, held June 4-7, 2001 in Amsterdam. Session 11 'Safety of Heavy Trucks and Buses (Lanekeeping, Stability, Tire/Road)
Identifier: 362644
Article number: Paper No. 153
Keywords: Traffic · accidents · analysis · heavy vehicles · Netherlands · trucks


For the 1993-1997 Dutch national accident data, logistic regression analysis was used to find the most important factors, that influenced the outcome of an accident with a truck involved. Frequency counts were used to identify factors that occurred most frequently. The combination of these two methods led to the most important factors influencing the number and severity of truck accidents. An important extension with respect to only frequency counts is that significance levels were taken into account to check whether differences are really distinguishable. It was concluded that the combination of frequency counts and logistic regression is a necessary extension to prevent the presentation of artifacts when basing conclusions solely on frequency counts and to find factors with high risk. Furthermore, it was found that national statistics are not detailed enough to find the real underlying causes and can therefore hardly be used to find points for improvement on vehicle and road design. Therefore, in-depth investigation of truck accidents is necessary to identify real causes.