PREDICTING YOUR OWN CHOICE POLICY: RELATIONS BETWEEN RISK ATTITUDE, MARGINAL VALUE, AND STRATEGIC FACTORS IN RISKY CHOICE

This study examined the descriptive validity of the utility scale. The scale is viewed as consisting of psychologically separable operations, or productions, which are strategically deployed. Several hypotheses derived from this viewpoint were tested.;The idea that productions compete for cognitive resources led to the first hypothesis: best simulations would include a pair of scaling operations at the expense of the asset integration operation, and would include the asset integration operation only at the expense of one of the scaling operations. The pattern of operations combining to form best simulations was consistent with this hypothesis. A second hypothesis was that the simulations would be consistent with choices only when the choice set included gambles offering the same chances of meeting an experimentally induced aspiration level. ANOVA strongly supported this hypothesis. Finally, canonical correlation analysis revealed that insight into their own risky choice behavior was less among subjects who described themselves as risk seekers in gains.;Results were discussed in terms of two dimensions of inconsistency between subjects' risk attitudes and their actual choice policy. The first dimension arises when subjects strategically limit the set of quantitative scaling productions used to make choices. The second dimension arises when subjects change the problem representation; these changes include setting an aspiration level, and adding or deleting attributes. The study demonstrates the use of simulation to construct models of peoples' choices, and to examine peoples' insight into their own choice.;Outcomes in gambles were assignments of arithmetic problems expressed as changes to a reference assignment that was changed at points in the experiment. Subjects provided value and utility scales that served as the basis for a variety of simulations, and then made pairwise choices among gamble pairs. Operations of value measurement, utility relative to measured value, and asset integration were either present or absent in simulations of the choices.