Discrete-choice modelling of patient preferences for modes of drug administration
Lancsar and Savage argue that current methods of deriving welfare estimates, using discrete choice experiments, are inconsistent with random utility and welfare theory. In this paper I show that this not the case. The general formula proposed by Small and Rosen for estimating welfare, which Lancsar and Savage claim should be used, reduces to the method used by health economists for state of the world models. The important question then becomes when are state of the world models, as opposed to multiple alternative models, appropriate?