✍ Scribed by Jean-Paul Caverni
No coin nor oath required. For personal study only.
Preference reversal has been a well-established phenomenon since the initial work by Slovic and Lichtenstein (1971). Although people choose to participate in gambles with a high probability of winning small amounts, they assign larger selling prices to gambles with a low probability of winning large amounts. This phenomenon is usually held to be a bias (e.g. Hogarth, 1987, pp. 110 and 221).
Ganzach (this issue) points out that preference reversal still occurs in equal-probability gambles (in which all payoffs have the same probability), whether subjects possess or d o not possess the gambles. Although they choose low-variance gambles (LVG, in which all payoffs are moderate), subjects put a higher price on high-variance gambles (HVG, with very high as well as very low payoffs). According to Ganzach, 'dependence of preference on response mode in equal-probability gambles can be explained by an anchoring and adjustment process in which the outcome with the larger payoff receives the heaviest weight in judgment (i.e. in pricing) but not in choice. These results cannot be explained by expression theory, compatibility, or prominence.'
While one might agree that Ganzach's results cannot be explained by expression or prominence (insofar as these theories deal with probabilities), it is unclear why they cannot be explained by compatibility. A strong argument in favor of anchoring concerns compatibility theory. In the pricing task, the payoffs are compatible with the cardinal response scale, but they are not compatible with the 0-1 scale used in the choice task. Under this argument, the only way to test whether anchoring is the right explanation for preference reversal would be to compare choice (to participate) and pricing (to sell) using the same kind of 0-1 scale in each case. For example, in a selling condition, subjects could be asked to indicate those games they would be willing to give up. If preference reversal can be explained by anchoring but not by compatibility, then the phenomenon will occur in the selling condition even if stating a price is not required.
Broadly speaking, our understanding of the processes underlying preference reversal might be increased by taking into account all the features in which the two tasks differ. For example, the response required for choice is a comparative one, while the response required for the pricing task is an absolute one. Why not ask subjects to make absolute choices and do comparative pricing?
But an even more important question that might be raised concerns the very existence of the so-called 'preference reversal phenomenon'. One can indeed wonder whether it is legitimate to contend that preference reversal is a manifestation of a bias in subjects. On what grounds can we base the assumption that preference should be the same in choice and pricing situations?
Several plausible interpretations can be proposed. It is reasonable to assume that a subject will not buy a ticket if its price is above the highest payoff he or she can obtain from it. In this case, it is not unreasonable to propose a higher price for an HVG gamble, which may have a larger payoff.
Why should the fact that subjects prefer low-variance gambling lead them to assume that most individuals gamble like they do? It is only in the case where the gambler is aware that a subject gambles like most individuals d o that one can consider his or her behavior as inconsistent: why put a higher price on a gamble that is less popular? The problem is that, with the task as it is presented, nothing allows us to assume that the subject is indeed aware of this fact.
There is another, final, way of interpreting the phenomenon which also appears perfectly coherent. Subjects choose to give up the gambles they d o not wish to participate in and sell them for a higher price, thereby maximizing their gain. Is this so surprising?
📜 SIMILAR VOLUMES