The impact of diversity on the quality of collective prediction

Abstract

Nowadays, there appears to be sufficient evidence that a collective of uninformed individuals can outperform single individuals (even individual experts) in solving some difficult judgment and prediction problems. In some situations, the predictions given by collective members on the outcome of a future event can be modified (such as by updating or removing predictions) to reach a higher consistency level (less diversity) in a collective. The main concern of this paper is to investigate the impact of diversity on the quality of collective prediction by taking into account such modifications within a collective. Diversity is understood as the variety of individual predictions on the same problem. The simulation experiments (with different collective cardinalities and levels of diversity reduction) have revealed that such modifications are not useful in improving the quality of collective prediction. Instead, they have emphasized the important role of diversity in leading to a better collective prediction. That is the more diverse the collective, the better the collective prediction.