The Market for Overconfidence

Researchers distinguish overprecision as both the most robust and the least understood form of overconfidence. Such overly precise judgments claim more certainty than is objectively warranted and often result in expressions of confidence that far exceed the accuracy of those judgments. In this dissertation, I investigate the social underpinnings of overprecision by examining two key relationships: the relationship between producers (i.e., individuals forming judgments) and receivers (i.e., individuals utilizing those judgments for some purpose) and the relationship between rival producers. Prior research has examined certain elements of these exchanges but has not adequately addressed the exchange system as a whole. To fill this gap, I create an experimental framework in which judgment advisors (producers) interact with guessers (receivers) in a market environment, where producers compete with one another for guessers' attention. I argue that confidence is intensified when competition for influence exists between producers. Moreover overprecision, customarily viewed as a detriment to effective decision making, may actually aid producers since its benefits (such as higher rates of selection) often exceed its costs (such as tarnished reputations).;Two initial studies contrast between the competitive market and a more static producer-receiver exchange. In Study 1, guessers completed an estimation task and chose advice from one of four advisors. Advisors display a high degree of overprecision in their estimates: their expressed confidence levels far exceed their hit rates. The expressed confidence of advisors increased over time with no corresponding gains in accuracy. Guessers did punish advisors for being confident and wrong by selecting them less often in subsequent rounds. However, this cost is outweighed by the benefit of continuing to express high confidence, namely that it attracts more customers. In Study 2, guessers completed the same estimation task with the aid of advisors, but in this case guessers and advisors were paired for the duration of the session. Advisors again display overprecision in their estimates. Despite similar costs and benefits associated with the advisors' actions, there are no significant changes to their confidence levels and distributions across rounds as had occurred in the first study.;In Studies 3 and 4, I examine how changes in certain features of the market affect these patterns of confidence. In Study 3, I varied the initial information provided to advisors. "High" information advisors received a stronger signal as to the correct answer. These high information advisors possess some inherent advantages over less informed advisors (e.g., greater accuracy). However, they remain overconfident to a similar magnitude, potentially due to the competitive pressure of the other advisors. In Study 4, I demonstrate the importance of behavior transparency for the documented patterns of advisor results. When advisors have information about the decisions made by their competitors, they are more overconfident and escalate their confidence at a higher rate than when this information is undisclosed.;I tested whether certain interventions to the market can help reduce overconfidence and improve performance in Study 5. A Control condition mirrored the basic framework of Study 1. In a Payoff condition, advisor earnings were based upon both their selection by guessers and their accuracy. For a Feedback condition, guessers received enhanced information about the performance of advisors at the time in which they made their selection decisions. The proposed interventions did little to change advisor behavior, as the degree of overprecision and the levels of expressed confidence were similar across all three conditions. However, the interventions did help receivers achieve higher levels of performance than in the Control condition.;Combined, these studies help further our understanding of this pervasive yet enigmatic form of overconfidence.