An Interval Approach to Measuring Decision Maker Preferences

A method for constructing interval measurements of decision makers' absolute risk-aversion functions is presented. Under this new procedure, the form of the absolute risk-aversion function is not restricted, and the precision of the interval measurement can be determined by the analyst. Interval measurements are used with the criterion of stochastic dominance with respect to a function to order uncertain choices. An empirical test shows that interval measurements exclude preferred choices from consideration less often than do single-valued utility functions and are more discriminating than first- and second-degree stochastic dominance.