Citation

Elvik R. Transp. Res. Rec. 2009; 2103: 18-24.

Copyright

(Copyright © 2009, Transportation Research Board, National Research Council, National Academy of Sciences USA, Publisher SAGE Publishing)

DOI

10.3141/2103-03

PMID

unavailable

Abstract

This paper discusses how accident modification functions describing the effects of road safety measures can be developed on the basis of evaluation studies. An accident modification function is a continuous function that describes the effect of a road safety measure as a function of one or more characteristics of the measure or the context of its use. Accident modification functions should be based on data that are easily available before a measure is implemented and that can be used to predict its effect. The following steps in developing accident modification functions are outlined: (a) development of hypotheses regarding systematic variation in effects of a road safety measure (i.e., identifying the independent variables of accident modification functions); (b) exploratory analysis of patterns in an evaluation study designed to propose a first version of an accident modification function; (c) testing of the accident modification function on a new evaluation study employing the same study design as the one used to develop the first version of the accident modification function; (d) synthesis of results obtained for the two evaluation studies by means of a revised accident modification function; (e) testing of the revised accident modification function by means of a third evaluation study; (f) synthesis of results obtained from three evaluation studies by means of a revised and possibly refined accident modification function; and (g) repetition of Steps 5(e) and (f) each time a new evaluation study is reported. The procedure is illustrated using evaluations of the road safety effects of bypass roads and the conversion of intersections to roundabouts as examples. In the first of these cases, an accident modification function was successfully developed. In the second case, development was not successful.