Determining quantitative road safety targets by applying statistical prediction techniques and a multi-stage adjustment procedure.

Abstract

Due to substantial progress made in road safety in the last ten years, the European Union (EU) renewed the ambitious agreement of halving the number of persons killed on the roads within the next decade. In this paper we develop a method that aims at finding an optimal target for each nation, in terms of being as achievable as possible, and with the cumulative EU target being reached. Targets as an important component in road safety policy are given as reduction rate or as absolute number of road traffic deaths. Determination of these quantitative road safety targets (QRST) is done by a top-down approach, formalized in a multi-stage adjustment procedure. Different QRST are derived under consideration of recent research. The paper presents a method to break the national target further down to regional targets in case of the German Federal States. Generalized linear models are fitted to data in the period 1991-2010. Our model selection procedure chooses various models for the EU and solely log-linear models for the German Federal States. If the proposed targets for the EU Member States are attained, the sum of fatalities should not exceed the total value of 15,465 per year by 2020. Both, the mean level and the range of mortality rates within the EU could be lowered from 28-113 in 2010 to 17-41 per million inhabitants in 2020. This study provides an alternative to the determination of safety targets by political commitments only, taking the history of road fatalities trends and population into consideration.

DOI: 10.1016/j.aap.2012.06.006

Cite this paper

@article{Wittenberg2013DeterminingQR, title={Determining quantitative road safety targets by applying statistical prediction techniques and a multi-stage adjustment procedure.}, author={Priv. -Doz. Dr. Ralf H. Wittenberg and K Sever and Sven Knoth and Nergis Şahin and Julia Bondarenko}, journal={Accident; analysis and prevention}, year={2013}, volume={50}, pages={566-77} }