A Prescriptive Approach to Elicitation of Decision Data

Abstract

Most current decision analytical tools and elicitation methods are built on the assumption that decision-makers are able to make their probability and utility assessments in a proper manner. This is, however, often not the case. The specification and execution of elicitation processes are in the majority of cases left to the discretion of the users, not least in user-driven cases such as public information and e-democracy projects. A number of studies have shown, among other things, that people’s natural choice behaviour deviates from normative assumptions, and that the results display an inertia gap due to differently framed prospects. One reason for the occurrence of the inertia gap is people’s inability to express their preferences as single numbers. Instead of considering this as being a human error, this paper uses the gap in order to develop a class of methods more aligned to the observed behaviour. The core idea of the class is to acknowledge the existence of the gap and, as a consequence, not elicit single point numbers.