No introductory course on risk analysis is complete without a review of the pitfalls and biases that could ensnare the unsuspecting risk estimator. These tend to come in two categories: those that occur as a result of individual cognitive limitations and those that manifest through group dynamics. A properly trained facilitator should be reasonably capable of identifying these threats, and a properly developed expert elicitation framework reasonably well equipped to manage them. In spite of this, heuristics and biases continue to be highlighted as the Achilles’ heel of the Potential Failure Mode Analysis (PFMA) methodology, or cited as a contributing factor when a dam safety incident occurs. This paper describes the expert elicitation framework used by Reclamation and argues that by design this framework is relatively insensitive to the adverse effects of such biases. This is a result not only of the “strong facilitator” model used by the agency but also the fact that dam safety decisions are not being made solely on the basis of the numbers generated in a risk analysis. The paper discusses some commonly cited biases and explains how each one can be dealt with through good risk analysis practice. It concludes with a discussion of how and whether to account for uncertainty in subjective probability estimates.
