Abstract

We develop a formal computational model to analyze how differences in influence structure and interests shape organizational decision-making involving information aggregation. Our model reproduces several important patterns from experiments and empirical observations, namely that small groups may fail to surface privately held information, that the quality of decision making is sensitive to order of speech effects, that groups may fall victim to the “illusion of consensus” (such that public and private votes diverge sharply), as well as sometimes succumb to “groupthink”. Critically, we also identify the conditions when social influence and diverging interests can improve decision quality such that organizations outperform crowds, teams and even the best individual in the group. Our analysis suggests that these different outcomes can all arise under different settings of the same basic mechanism: a tradeoff between certification and censoring of information, which is shaped by the distribution of influence patterns and preferences in the group.