Inferences derived from Starmer's (1992) experimental evidence concerning Expected Utility (EUT), Fanning Out (FO), and Fanning In (Fl) theories are both incomplete and incorrect A subjectivist Bayesian approach based on calculating posterior probability distributions for experimental outcomes is used to quantify the degree of support for each theory and to make coherent inferences about the relative performance of FO and H theories in explaining violations of EUT.