An Expert Opinion Elicitation Method Based on Bayesian Intervals Estimation and Computational Searching Algorithms: an Application to Oil Refinery Risk Analysis

摘要

This paper proposes an expert opinion elicitation method based on the theory of Bayesian intervals estimation and computational searching algorithms. Each step of the proposed elicitation procedure involves two types of questions: firstly, the expert is asked about which of two subintervals holds the unknown quantity of interest (a procedure similar to a computational searching algorithm in an ordered list). Then, the expert assigns a credible level on which the decision was based. As a result, a set of Bayesian interval estimates of the unknown quantity is defined and inferences about the probability density function underlying the expert beliefs about the unknown are performed. The elicitation method is illustrated by means of an example in the context of quantitative risk assessment of a new oil refinery under development in the Northeast of Brazil.