This report provides a detailed technical description of the proposed methodology for the evaluation of developments in Dutch road safety in the SWOV projects ‘Verkeersveiligheidsbalansen’ (Road safety assessments) and ‘Verkeersveiligheidsverkenningen’ (Road safety outlooks). The methodology is based on structural time series models by state space methods, in order to accommodate the dependencies between observations consisting of repeated measurements over time. A dedicated bivariate local linear model, called the basic evaluation model, is proposed, in which the development of road safety is assumed to be the product of the developments of two latent, unobserved variables: exposure and risk. The methodology is illustrated with a simultaneous analysis of the annual number of Dutch road traffic fatalities and motor vehicle kilometres driven in the period 1948-1998. The results are found to compare favorably with those obtained with standard analysis techniques, and the report concludes with a number of possible extensions of the basic evaluation model.
The basic evaluation model
Paper presented at the ICTSA meeting, 27-28 May 2004, INRETS, Arceuil, France