(DP 1984-01) Multidimensional Choice and Preference Reversals

Abstract

Results of various experiments reported in recent literature have shown repeated violations of expected utility theory. In the model of this paper, expected utility is only first of three criteria that determine choices under uncertainly--the second criterion is the probability of doing better than or at least maintaining the status quo, and the third is the maximum possible gain. Choice being thus multidimensional, the observed violations get accounted for: the Allais and Ellsberg paradoxes, probabilistic insurance, risk seeking with negative prospects, translation effects, preference reversals, and Tversky intransitivities