Elaborating or Aggregating? The Joint Effects of Group Decision-Making Structure and Systematic Errors on the Value of Group Interactions

Abstract

We explore when group interactions will have a positive effect on the accuracy of quantitative judgments. The results from three laboratory experiments revealed that the value of group interactions, compared with a statistical aggregation of individual judgments, differed strongly between groups in which decisions were made based on consensus, compared with groups with a randomly assigned group leader, and that this effect was moderated by the level of systematic error among group members. In particular, when systematic errors were low, group interactions generally provided little value, and groups’ decision-making structure (consensus versus leader based) did not have a significant effect on the value of group interactions. However, when the level of systematic error was high, the value of interactions in groups with a randomly designated leader was strongly positive and significantly higher than in groups with a consensus decision-making structure where interactions still provided only little value. Moreover, we found that this effect was mediated by information elaboration—which was higher in leader groups but only had a significant effect on the value of group interactions when there was a high level of systematic error. Finally, we also identified two important boundary conditions for these results. In particular, leader groups did not provide more value than consensus groups in the presence of strong systematic errors when leaders were assigned based on prior individual performance rather than randomly or when group members had only little expertise in the task. This paper was accepted by Yuval Rottenstreich, decision analysis.