Proceedings of the Annual Meeting of the Cognitive Science Society

The crowd wisdom effect is a well-established phenomenon that is widely employed for predicting and estimating variables across various domains. Previous research has focused on enhancing the wisdom of crowds by improving individual estimates while maintaining some of the initial opinion diversity. However, it is theoretically possible to increase collective accuracy by largely increasing the diversity in a crowd. In this study, we propose a method that leverages the anchoring effect to extremize individual judgments and thus increase the diversity of opinions in a crowd. This is achieved by dividing the crowd into two groups, anchoring each group to either a low or high value, and aggregating all estimates. We use a mathematical model of anchoring to determine when this strategy is expected to outperform the crowd wisdom effect. Results from three experiments provide converging evidence that the proposed approach outperforms traditional methods in estimating and forecasting unknown quantities.