Changes in the legislation on the use of daytime running lights by motor vehicles and their effect on road safety in Hungary.

As of 1 March 1993, a partial (relating only to semi-motorways and main roads), and as of 1 June 1994, a total (relating to all roads) legal obligation to use DRL came into force outside built-up areas in Hungary. This paper examines the effectiveness of this regulation as an accident countermeasure. In view of the fact that it became possible to use one part of the road network as a comparison group during the period of partial obligation, a comparison group investigation has been carried out in addition to the analysis based on the odds ratio method familiar from earlier studies. The different methods provided different results, the probable causes of which are discussed in the present study. The result of the comparison group analysis are consistent not only with the hypothesis that DRL reduces certain types of collisions, but also with the results of a meta-analysis of 17 studies that have evaluated the effects on accidents of using DRL on cars. By using this method, it was possible to detect in the experimental group a 13.0% reduction in the number of frontal and 'crossing' vehicle collisions in daylight, in good visibility conditions, during the after period. The number of frontal and crossing collisions decreased without a simultaneous increase in rear-end collisions: the number of the latter has not significantly changed. Thus, the hypothesis that it is more reliable to estimate the collision-reducing effect of DRL on the basis of accidents only occurring in good visibility conditions, rather than on the basis of the total number of daylight accidents, has been justified by this study. The causes of this, as well as the difficulties and confounding factors arising when the odds ratio method is used, are dealt with in this work. By analyzing over a longer time series, it was possible to verify that the positive road safety effect is not a mere novelty effect, and is not simply due to the 'regression to the mean'.