ABSTRACT
If errors are no longer to be considered as random occurrences, then it follows that we should be able to identify them and predict their likelihood. The impetus to achieve this has been fuelled in the wake of several recent and significant incidents, most notably in the nuclear industry, and there now exist several human error identification (HEI) techniques. What is lacking up to now is a rigorous quantitative and qualitative evaluation of these techniques. The present paper addresses the issue of prediction of error form and readers interested in the separate issues of error likelihood and error occurrence are referred to Kirwan (1994) for a recent review of the area.
