Evidence for the rapid construction of preference during utility assessments.

Subjects often construct preferences during the elicitation process. This could have implications for the processes automated utility assessments use to find indifference points. In particular, if subjects frequently shift preferences during elicitation processes, then reversible and irreversible procedures might reach different results. We analyzed series of choices made by two groups of subjects during computerized standard gambles for monocular and binocular blindness, comparing the utility assessed in a reversible search process with the utility that would have resulted from an irreversible search process. A minority of subjects reversed their choices during the assessments. The mean differences between these utilities and the predicted results of an irreversible search were quite small. Consequently, automated standard gamble surveys can terminate quickly, using irreversible searches, with little likelihood of skewing population results. Clinical decision support systems can use simple mechanisms to accommodate infrequent preference reversals, such as restarting a search.