摘要: |
A major challenge of risk and reliability analysis for human errors or hardware failures is the need to enlist expert opinion in areas for which adequate operational data are not available. Experts enlistedin this capacity provide probabilistic estimates of reliability, typically comprised of a measure of central tendency and uncertainty bounds. While formal guidelines for expert elicitation are readily available, they largely fail to provide a theoretical basis for expertise and judgment. This paper reviews expertise and judgment in the context of risk analysis; overviews judgment biases, the role of training, and multivariate judgments; and provides guidance on the appropriate use of atomistic and holistic judgment processes. |