Citation

Williams MJ. Accid. Anal. Prev. 1981; 13(2): 133-145.

Copyright

(Copyright © 1981, Elsevier Publishing)

DOI

unavailable

PMID

unavailable

Abstract

The traffic conflicts technique was developed as a tool for estimating accident potential at intersections and for indicating methods of reducing hazardous conditions. A review of evaluation studies fails to confirm that the method can perform these tasks. This results partly from methodological problems in the studies. Differences in the definitions of both accidents and conflicts have produced results which often are incomparable simply because different pairs of variables have been used in the analyses. Other theoretical inconsistencies appear to limit the likelihood of predicting accidents from conflicts. It is suggested that a hierarchy of traffic events ranging in severity from slight conflicts to fatal accidents exists. Certain fundamental characteristics of these events (including time of occurrence, type of manoeuvre, location and probable cause) differ so markedly that the prediction of one from the other may not be possible. Evidence is presented which indicates that neither accident nor conflict data, recorded using present methods, is of much value in predicting future accidents or conflicts, respectively. It is suggested that a new method of recording conflicts which overcomes the conceptual problems of previous definitions may be useful in evaluating countermeasures at least as an experimental tool. This would require a change in the criterion used for assessing the benefits of countermeasures from one of the reduction in accidents to one of eliminating predetermined improper driving behaviour or operational problems.