On the use of judgment in probabilistic risk analysis

Abstract

The extensive use of judgment in risk studies creates several issues related to the perception of probabilities and the value of formal theories for uncertainty analysis. The direct assessment of probability distributions may lead to biased curves that do not represent the state of knowledge of the assessors. The use of formal methods and the calibration of the analysts help alleviate some of these biases. Case studies involving the assessment of probability distributions for human error rates and for failure rates of components exemplify the importance of the careful use of expert opinions in risk studies.