Academic journal article The Psychological Record

Preference Reversal: A New Look at an Old Problem

Academic journal article The Psychological Record

Preference Reversal: A New Look at an Old Problem

Article excerpt

One of the most frustrating findings in the classical view of preferences is the preference reversal (PR) phenomenon (Stalmeier, Wakker, & Bezembinder, 1997). For more than three decades psychologists and economists have been intrigued by the PR anomaly. The classic example of preference reversal involves the choice-pricing discrepancy in the evaluation of two lotteries with equal expected value. One lottery typically has a high probability of winning a modest cash amount (a P bet). The other, riskier lottery has a small chance of winning a large monetary amount (a $ bet). Participants in a typical preference reversal task tend to state a higher cash equivalent for the $ bet, but they tend to prefer the P bet, when asked to choose between these two lotteries. For example,

    Lottery A provides a 9/12 chance of winning $110 and a 3/12 chance
    of losing $10. (P bet)
    Lottery B provides a 3/12 chance of winning $920 and a 9/12 chance
    of losing $200. ($ bet)

When individuals are asked to choose between these two lotteries, they tend to choose Lottery A over Lottery B. When asked to select a cash equivalent price for each lottery separately they tend to place a higher selling price on Lottery B. This pattern is usually interpreted as indicating an inconsistency, because both measures, choice and pricing, are assumed to be measures of people's well-articulated preferences. The PR phenomenon was first reported by Lichtenstein and Slovic (1971) and by Lindman (1971). Since then, three waves of studies of preference reversal have been described (for a complete description, see Tversky, Slovic, & Kahneman, 1990).

There have been three alternative interpretations of PR. These alternatives arise from the violation of one of three principles important to decision theory, namely (a) violations of transitivity (Fishburn, 1985; Loomes & Sugden, 1983); (b) violations of the independence axiom (Holt, 1986; Karni & Safra, 1987); and (c) violations of procedure invariance (Goldstein & Einhorn, 1987; Tversky et al., 1990). Psychological models have been developed to account for the phenomenon. The majority of the models have focused primarily on the idea that the different psychological processes take place in one context versus the other. For example, expression theory (Goldstein & Einhorn, 1987) postulates that PR is caused by changes in the mapping from a gamble's components to the response. This transformation is assumed to differ predictably for each gamble and for each response mode. Contingent weighting theory (Tversky, Sattath, & Slovic, 1988) attributes PR to variations in the stimulus weighting. The exponential weight of the probabilities is low under the pricing condition and high under the choice or attractiveness judgment conditions, while monetary outcomes are assumed to be weighted more heavily in pricing than in choice. The work by Slovic, Griffin and Tversky (1990) further argued that the shift in weights results from the compatibility between the attributes of the options and the response scale. Gonzalez-Vallejo and Wallsten (1992) used the contingent weighting model to explain PR between both choices and prices, and also between choices expressed with numerical probabilities and those expressed with verbal probabilities. Change-of-process theory (Mellers, Ordonez, & Birnbaum, 1992) attributes PR to variations in the decision strategies used to combine information. The bids are a multiplicative function of probability and amount, whereas ratings are an additive function of probability and amount. According to Tversky et al. (1990), however, the violations of transitivity and independence axiom can account for only a small fraction of observed preference reversals. Therefore, violations of procedure invariance (usually considered as arising from effects of scale compatibility) are nowadays the most widely accepted explanation: Due to the identical scale, the payoffs of a lottery are weighted more heavily in pricing than in choice. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.