Group Discussions Improve Competence Calibration: Making Self-Perceived Competence Valuable When Harnessing the Wisdom of Crowds
Goedde-Menke, Michael; Diecidue, Enrico; Jacobs, Andreas; Langer, Thomas
Abstract
This paper experimentally demonstrates that group discussions can serve as an instrument to improve competence calibration, which in turn allows getting more wisdom out of the crowd through competence weighting. While the alignment of individuals’ estimation accuracy and self-perceived competence is typically poor and competence-weighted aggregates do not even match the accuracy of simple averaging, we find that preceding group discussions on unrelated judgment problems enhance competence calibration. Consequently, the subsequent performance of competence-weighted aggregation schemes raises to and beyond prediction market levels, suggesting an easy-to-implement approach for effectively exploiting crowd wisdom.
Keywords
estimation accuracy; wisdom of the crowd; calibration; competence weighting; prediction markets