Ratio-Scale Elicitation of Degrees of Belief
29 Pages Posted: 31 Oct 2008
Date Written: September 1988
Most research on rule-based inference under uncertainty hasfocused on the normative validity and efficiency of variousbelief-update algorithms. In this paper we shift the attentionto the inputs of these algorithms, namely, to the degrees ofbeliefs elicited from domain experts. Classical methods foreliciting continuous probability functions are of little use in arule-based model, where propositions of interest are taken to becausally related and, typically, discrete, random variables. Wetake the position that the numerical encoding of degrees ofbelief in such propositions is somewhat analogous to themeasurement of physical stimuli like brightness, weight, anddistance. With that in mind, we base our elicitation techniqueson statements regarding the relative likelihoods of various cluesand hypotheses. We propose a formal procedure designed to (a)elicit such inputs in a credible manner, and, (b) transform theminto the conditional probabilities and likelihood-ratios requiredby Bayesian inference systems.
Suggested Citation: Suggested Citation