“…In contrast with other strategies based on classification variability (Weiss & Shanteau, 2003), which require repeated observations, the agreement-in-confidence heuristic enables people to learn an advisor’s accuracy even in one-shot interactions as a basis for weighting their advice in future interactions. This strategy is consistent with various social psychological phenomena, including the “false consensus” effect, naïve realism and social judgment theory’s “latitude of acceptance,” which all indicate a tendency for people to discount disagreeing opinions, underweight advice as a function of distance from one’s own opinion, and consider one’s own opinions as more objective or frequent than others’ (Ecken & Pibernik, 2016; Liberman, Minson, Bryan, & Ross, 2012; Minson, Liberman, & Ross, 2011; Ross, Greene, & House, 1977; Schultze, Gerlach, & Rittich, 2018; Sherif, Sherif, & Nebergall, 1965; Soll & Larrick, 2009; Yaniv, 2004). Importantly, however, our approach differs in suggesting that these phenomena are part of a normatively justified strategy that enables people to discern advisors’ features without the benefit of feedback: When a judge and advisor are independent, their rate of agreement varies as a simple monotonic function of their respective accuracies, so that agreement rate can be used to infer an advisor’s accuracy.…”