ADDITIVE CONJOINT-MEASUREMENT FOR MULTIATTRIBUTE UTILITY

This paper shows that the role of risky alternatives can be greatly reduced in the elicitation procedures of multiattribute utility. This reduction can be achieved by invoking methods from additive conjoint measurement; it is desirable because risky choices involve more cognitive problems, thus more biases and unreliability, than riskless ones. Existing results of multiattribute utility are generalized to obtain a complete axiomatization of the new elicitation procedure. The approach has been developed in a medical decision analysis project to advise on the choice between surgery and radiotherapy for laryngeal cancer. (C) 1994 Academic Press, Inc.