Figure 3.

(A) Relationship of within-subject risk premium values across the gains and losses domains. The dashed red line visualizes the correlation predicted by the theoretical reflection effect, with a slope of −1 for the risk premium metric (left) and +1 for the power function metric (right). (B) Cross-domain predictive comparison, percentage of choice behavior correctly predicted by each risk preference, across both domains. Randomized within-domain power function values were obtained through bootstrap analysis, randomly resampling risk preference value and participant's choice sets independently (N = 10,000 iterations, with replacement). The standard error measurement (SEM) value is the median SEM across iterations.