A Study of Expert Judgment on Human Error Probability
An experiment was conducted to derive empirically human error probabilities from a task performed under 12 different conditions. The task was to control a simulated flexible manufacturing scenario (FMS) under three Performance Shaping Factors (PSF): Incentive, workload and event frequency of breakdowns. Six experts with background in human factors assess the relative contribution of each PSF in affecting the likelihood of failure with the multi attribute decomposition technique. The conversion of the assessment values to probabilities was achieved by the use of an empirically derived calibration equation. Results indicate a poor match of empirical HEPs and their estimates and increase the doubts that subjective estimation is a solution to the missing data problem in reliability measurement.