When assessed with the standard gamble, the validity of a utility score is partially dependent on the respondent's numeracy skills. People who lack basic numeracy skills, such as knowledge of probability, may be unable to respond in a manner that reflects their true strength of preference for a health state. In this study, we pilot-tested a new approach to the standard gamble that uses qualitative rather than quantitative descriptions of probability.

Students enrolled in an undergraduate business course rated the utility of oropharyngeal cancer using two different representations of the paper standard gamble. In one representation, the probability of perfect health or death was described quantitatively (e.g., 80% chance of perfect health). In the other representation, the probabilities were described qualitatively (e.g., perfect health is likely.) For the qualitative representation, gambles were constructed using the terms: always, extremely likely, likely, probable, possible, unlikely, rarely, and never. Afterwards, subjects assigned numerical values to the qualitative terms, directly and with visual analog scales (VAS). These numerical values were then used in the determination of the utility. Subjects also completed an extended version of the Lipkus numeracy survey and demographic questions.

Utilities derived using the quantitative SG and qualitative SG were significantly correlated (r = .41, p<.0001); however mean utilities estimated with the quantitative SG (M=.81, SD=.18) were significantly higher than those estimated with the qualitative SGs (direct estimation, M=.68, SD=.23; VAS estimation, M=.68, SD=.22). The qualitative utilities derived from direct estimation or the VAS were not statistically different. Differences between the quantitative and qualitative utilities were unrelated to the subject's numeracy score. On average, participants answered nearly 80% of the numeracy questions correctly.

Overall, participants were highly numerate and demonstrated, via direct estimation and the VAS method, a consistent understanding of the qualitative probability terms. When given the opportunity to choose between gambles in which their own subjective interpretations served as the index of risk, subjects took higher risks (i.e., had lower utilities) than they did when the risks were presented numerically. The inclusion of additional qualitative terms may be necessary to allow participants to provide more precise estimates of their utility for the health state.