ABSTRACT
The aggregation of judgments across individuals in a group has been shown to lead to a group estimate that is better than most of the individual estimates. Demonstrations of this effect have focused on tasks where individuals produce subjective probability or magnitude estimates (e.g., Ariely et al., 2000; Budescu & Yu, 2007; Steyvers, Wallsten, Merkle, & Turner, 2014; Turner, Steyvers, Merkle, Budescu, & Wallsten, 2014; Wallsten, Budescu, Erev, & Diederich, 1997). In a now classic study, Galton (1907) asked over eight hundred individuals to estimate the weight of an ox. The median weight estimate, which corresponds to a simple form of aggregation, came within a few pounds of the true answer. This group estimate was much closer to the truth than the vast majority of individual estimates, a phenomenon that has become known as the Wisdom of Crowds effect (WoC; reviewed in Surowiecki, 2004). The most basic explanation of this effect is that the averaging across individuals reduces the noise associated with each individual decision—some individuals overestimate and others underestimate the underlying quantity—and aggregating cancels out some of these errors in judgment. The benefits of aggregating across individuals have also been demonstrated in more complex tasks involving rank-ordering judgments (Lee, Steyvers, & Miller, 2014), and optimization problems (Yi, Steyvers, & Lee, 2012). Recently, it has been shown that the benefits of averaging also extend to judgments within an individual (Vul & Pashler, 2008; Hourihan & Benjamin, 2010).
