EFFECTS OF PROBLEM REPRESENTATION AND FEEDBACK ON RATIONAL BEHAVIOR IN ALLAIS AND MORLAT‐TYPE PROBLEMS

This paper reports the results of an experiment which investigated the effects of various ways of structuring decision problems under uncertainty (viz, matrix, tree, versus word representations) and various forms of feedback (discussion versus non-discussion of prepared answers conforming to and deviating from the irrelevance of identical outcomes postulate) on consistent choice behavior (i.e., in accordance with this postulate) in the context of Allais and Morlat-type (A & M) problems. The special structure of A & M problems is especially useful for studying such behavior in that it presumably leads to a set of choices that violates this fundamental norm of decision theory. The experiment involved 134 industrial management undergraduates who gambled to improve their grades in a statistics course. The results showed that: (1) structuring such problems as “decision trees” or “decision matrices” had either a detrimental or no beneficial effect respectively on the number of consistent responses; while (2) discussion led to some improvement in consistency over no discussion. The implications of the findings for training in rational decision making are discussed.