An uncertainty evaluation technique for determining probability estimates

The development of a technique to use expert judgments for constructing probability estimates is presented. The framework of using the eigenvector method to obtain point estimates of probabilities is discussed. These point estimates, however, contain some degree of uncertainty. The proposed technique estimates the variability of point estimates of probabilities by analyzing the inconsistencies in the expert judgments. This is achieved by sequentially addressing the following issues; representation of inconsistencies in expert judgment by log-normal error factors; development of quotient index as the proper index of inconsistency in the expert judgments; and estimation of variance of probabilities from the inconsistency in expert judgment. The need for and importance of such uncertainty estimates are explained, and the application of the technique is illustrated with a numerical example.<
>