2011), “The Accuracy of Small-Group Estimation and the Wisdom of Crowds,” working paper

We measure the ability of people to estimate the price of familiar household items in a variety of contexts. We manipulate whether estimation is done alone or with others, whether it is done independently or with the knowledge of the estimates of others, and whether it is done in a cooperative or competitive environment. From these basic estimation data, we construct a series of aggregated group estimates, exploring the conditions under which a small group of three people provide the most accurate information. We compare the performance of various small-group estimates to standard Wisdom of Crowds analysis, and find that priming people, or placing them in a cooperative group setting, is less effective than averaging the independent estimates of individuals. We also find, however, that it is possible to extract relatively more information from the decisions people make in a competitive group setting, using cognitive models of their decision-making