Citation

Cooper PJ. Accid. Anal. Prev. 1994; 26(2): 263-275.

Affiliation

Insurance Corporation of British Columbia, North Vancouver, Canada.

Copyright

(Copyright © 1994, Elsevier Publishing)

DOI

unavailable

PMID

8198695

Abstract

At least some of the controversy surrounding the level of effectiveness of seat belts has arisen due to the difficulty in calculating a direct impact on police-reported accident data. These data are notoriously biased in the direction of belt-wearing overestimation. Furthermore, not enough information on nonwearer overrepresentation in crash involvement is currently available over the entire range of accident severity levels. The work reported in this paper was undertaken in order to develop a methodology for calculating nonwearer overrepresentation at various levels of crash severity and for adjusting police-reported belt-wearing data. The underlying premise was that police officers make reasonably consistent and valid judgements related to probable fault or causor status in casualty-producing accidents that they attend. Separating reported wearer and nonwearer crash involvements into culpable and nonculpable situations led to a procedure for adjusting police-reported belt-wearing totals to account for likely overestimation and also enabled the assessment of probable levels of nonwearer overrepresentation in crashes at all severity levels. The results, in terms of both nonwearer overrepresentation and seat belt effectiveness, were comparable with those reported in the literature for fatal crash involvement. Seat belt effectiveness was found to average 49% for fatal crashes, 35% for serious-injury crashes, 18% for moderate-injury crashes, and 11% for minor-injury crashes. Belt nonwearer overrepresentation was found to increase with increasing levels of belt wearing in the driver population from a factor of between 2 and 5 (depending on crash severity) at b = 0.5 to between 5 and 10 as b approached 0.9.