Binary Choice Probabilities Between Gambles: Interlocking Expected Utility Models.

摘要

This report discusses two models for describing an individual's binary choice probabilities between gambles whose outcomes are measured along a unidimensional scale on which a larger outcome is preferred to a smaller outcome. The models presume an underlying von Neumann-Morgenstern utility function. They view each pair of gambles in an interlocking fashion rather than as two separate entities. The first model is an incremental expected utility advantage model designed for the context in which the outcome probabilities are not associated with specific chance events. The second model is a generalization of Edwards' RELM rule that is intended for contexts in which outcome probabilities are tied to designated chance events. Data from previous studies lends some support to the models as descriptors of binary choice probabilities in risky decision making.