The Relationship Between Judgment and Choice (Decision, Preference Reversal Phenomenon).

Abstract

Holistic and dimensional processing are two broad classes of multiattribute decision strategies which are distinguished by the order in which aspects of the alternatives are processed. A dual-process theory has been suggested, in which judgment relies on holistic processing while choice employs dimensional processing. For each processing strategy, a mathematical model is proposed and axiomatized. A thought experiment demonstrates that preference for simple gambles will violate a necessary property of the dimensional processing model if gambles are psychologically characterized by probabilities and outcomes. Therefore, if gambles are processed dimensionally, the psychological dimensions are not probabilities and outcomes. A single-process theory is introduced and axiomatized, in which both judgment and choice rely on dimensional processing. Discrepant judgment and choice orderings are attributed to reference effects. A special case of the theory predicts that for alternatives with 3 or more dimensions, judgment and choice orderings will be discrepant if and only if choice is intransitive. Incompatible judgment and choice is studied in 3 experiments. Experiment 1 investigates the "preference reversal phenomenon," an observed discrepancy between preferential choice and minimum selling price judgments of simple gambles. The phenomenon is found to be more than just an example of discrepant judgment and choice. The "assessment modes" of minimum selling price and willingness to play the gambles are also discrepant. Moreover, the predominant direction of reversals between judgment and choice depends on the assessment mode. Experiment 2 investigates the assessment mode reversals. For gambles with fixed minimum and maximum outcomes, assessment mode reversals are no longer significant. Moreover, the pedominant direction of reversals between judgment and choice no longer depends on the assessment mode. Therefore, within this domain, the relationship between judgment and choice can be studied with some generalizability across assessment modes. In Experiment 3, the single-process theory of judgment and choice is found to fail. Reasons for the failure are discussed. A single-process theory might be salvaged, but multi-process theories seem more promising.