A Test of Rank-Dependent Utility in the Context of Ambiguity

Experimental investigations of nonexpected utility have primarily concentrated on decision under risk ("probability triangles"). The literature suggests, however, that ambiguity is one of the main causes for deviations from expected utility. This paper investigates the descriptive performance of rank-dependent utility in the context of choice under ambiguity. We use the axiomatic difference between rank-dependent utility and expected utility to critically test the two models against each other. Surprisingly, rank-dependent utility does not provide any descriptive improvement over expected utility. Our data suggest other, "framing," factors that do provide descriptive improvements over expected utility.