I will be giving a talk on April 6 to the Seminar on Risk and Decision Sciences sponsored by the Harvard Center for Risk Analysis. Hopefully, Joseph Pliskin and Milton Weinstein will be at the talk (Weinstein invited me so I expect him to be there). I will be talking on a topic that has been influenced by the research of these two men over my entire career, so I'm looking forward to presenting my ideas to this group. The following is the abstract that I have sent to them.
The Quality Adjusted Life Years Utility (QALY) model is the most influential and widely used utility model in health utility analysis. Unfortunately, two obstacles have impeded efforts to test whether preferences for health risks actually satisfy the QALY model. First, the Pliskin, Shepard, and Weinstein (1980) axiomatization of the QALY model was developed within the framework of expected utility (EU) theory. A large body of empirical research has demonstrated that the behavioral assumptions of EU theory are violated by actual preferences. Evidently it is difficult to interpret tests of QALY assumptions that are carried out within the EU framework if EU theory is descriptively invalid. Second, empirical tests of preference axioms pose difficult problems in statistical analysis and experimental design. For example, the transitivity axiom would be easily testable if preferential choices were free from random variation, but special designs and analyses are required to test it under realistic conditions that include random variation in choices. This talk will review research that addresses both of these concerns. Mathematical research by Bleichrodt, Johannesson, Miyamoto and Wakker has developed axiomatic foundations for QALYs that are compatible with non-EU theories like cumulative prospect theory (CPT) and rank dependent utility (RDU) theory. I will not dwell on the mathematical technicalities of this research, but the discussion of non-EU axioms for QALYs will let us see what are critical tests of QALY assumptions in a non-EU framework. I will review several psychological experiments that have attempted to test QALY assumptions in a non-EU framework. Non-EU tests of QALY assumptions require small but important alterations of the experimental designs that test QALY assumptions in the EU framework. The talk will conclude with a few general remarks about the current empirical status of the QALY utility model. .