The Bayesian framework for statistical inference offers the possibility of taking expert opinions into account, and is therefore attractive for practical problems concerning inspection and replacement of technical systems. However, the use of a single prior distribution fails to indicate the amount of information on which subjective probabilities are based, and leads to problems when combining the opinions of several experts.
The introduction of imprecise prior probabilities solves these problems, and leads to simpler and clearer elicitation of prior information. Problems concerning elicitation of lifetime distributions, combination of opinions, introduction of statistical models and calculation of bounds on expected loss within Bayesian decision theory have been analyzed and solutions proposed.
In the current paper application of the concept to an age replacement problem is described. The entire process, from elicitation and model assumptions to reaching a final decision, is discussed.