Should We Expect Expected Value?

Expected value (EV) is commonly treated as the central determinant of risky choice, either directly or as a reference point for higher‐order representations of risk. However, people's valuations of gambles often deviate from EV, and it is unclear whether this reflects biases in representing and integrating numerical information or preferences for other properties of the outcome distribution. We compared explicit EV estimation with certainty equivalent–based attractiveness evaluation across four instructional conditions in a sample of 137 participants. Participants evaluated two‐outcome gambles varying in EV, standard deviation, and skewness. Analyses showed that participants estimated EV accurately when instructed to do so, and these estimates were largely unaffected by variance or skewness. In contrast, subjective valuation was strongly influenced by skewness but not by standard deviation. Positively skewed gambles were valued above their EV, whereas negatively skewed gambles were valued below it. The findings indicate a dissociation between EV computation and subjective valuation: people can estimate EV directly, yet broader evaluations of risky options rely on additional psychologically salient features, particularly skewness. Thus, biases in processing the displayed numbers explain only a minor part of the deviations of valuations from EV; most of the deviation arises when gambles are valued.