(英) |
When it is difficult to make judgments individually, people sometimes aggregate other people’s opinions and make group judgments (e.g., majority rule). In such situations, it is generally expected that aggregating individuals with high confidence will be a good strategy. However, to what extent can individuals’ confidence in one current task set predict group judgmental performance in another task set that they will face in the future? Our computer simulations based on behavioral data revealed that, compared to the groups constructed from randomly aggregated individuals regardless of confidence levels, the groups constructed from individuals with high confidence (i) generally showed better accuracy of group judgments; (ii) however, when few current tasks were available (i.e., highly uncertain situations), their accuracy tended to largely decrease in future tasks. |