Citation

Cownie AR, Calderwood JH. OR Oper. Res. Q. (1953) 1966; 17(3): 253-262.

Copyright

(Copyright © 1966, Pergamon Press)

DOI

unavailable

PMID

unavailable

Abstract

It is argued that accidents are the product of a basically simple closed-loop process. The forward causal chain is essentially observations-decisions-accidents, the relationship between decisions and accidents being a stochastic one. The loop is closed by the feedback of information from decisions to observations, which occurs when the consequences of past decisions help to determine a decision-maker's interpretation of his observations. The effect on a hazardous activity of a change in its properties, such as the introduction of a safety measure, can be widely diffused both in time and space because a change in a person's interpretation of his observations can affect his decisions throughout his subsequent participation in the activity. In practice it is very easy to observe only a local part of the effect of making a change to the system, such as when a dangerous cross-road is improved. It is suggested that material causes of accidents may determine the distribution of accidents without greatly affecting their total number. An operational game is proposed for investigating the extent to which subjects playing the game regulate the level of risk of incurring a penalty which they allow themselves to take.