Observing Violations of Transitivity by Experimental Methods.

The preference reversal phenomenon is usually interpreted as evidence of nontransitivity of preference, but has also been explained as the result of the difference between individuals' responses to choice and valuation problems; the devices used by experimenters to elicit valuations; and the "random lottery selection" incentive system. This paper reports an experiment designed so that none of these factors could generate systematic nontransitivities; yet systematic violations of transitivity were still found. The pattern of violation was analogous with that found in previous preference reversal experiments and is consistent with regret theory. Copyright 1991 by The Econometric Society.