Citation

Nicholson AJ. Aust. Road Res. 1991; 21(1): 30-39.

Affiliation

University of Canterbury, NZ

Copyright

(Copyright © 1991, Australian Road Research Board ARRB)

DOI

unavailable

PMID

unavailable

Abstract

This paper discusses the stochastic nature of accident occurrence and the parallels that exist between accident occurrence and a commonplace random process - throwing a die. It is argued that there is a need for those involved in interpreting accident count data to employ 'common sense', based upon knowledge gained from observations of throwing a die. It is shown that the die-throwing analogy is appropriate, whether accident occurrence is governed by a stationary or by a non-stationary process. In addition, it is shown that a recommended simple statistical test can lead to erroneous conclusions regarding the statistical significance of accident count data. A more sophisticated and sound statistical test is described, and the discrepancy between the results of the two tests is explained. It is argued that the credibility of those involved in accident analysis and prevention depends upon their demonstrating a good understanding of the stochastic nature of accident occurrence.

Language: en