The effect on the preference-reversal phenomenon of using choice indifferences

Abstract

Preference reversal appears to be non-transitive choice behavior, inasmuch as subjects choosing between two gambles with similar expected values typically select the one with a larger chance of winning (a P-bet), yet they place a higher certainty equivalent, or judged indifference point (JIP), on the one with the larger amount to win (the $-bet). In the present experiments, we explore a pair of methods for determining money indifference points that are procedurally closer than is common in this research to the choice phase of the preference reversal experiment. The two new approaches for determining money indifference points for gambles (called choice indifference points, or CIP) are as follows: a version of the classical psychophysical up-down method and a variant called PEST (i.e., Parameter Estimation by Sequential Testing). With both methods, CIPs reduced the number of reversals, as compared to JIPs, to the point with PEST that the remaining ones may be due primarily to chance fluctuations in choices between gambles. Subjects who reversed frequently usually judged the gambles to have similar values; those who did not reverse at all, judged them as different. The typical subjects exhibits JIP>CIP for $-bets, but not for P-bets. This bias appears to underlie the observed intransitivity when JIP is used.