Overconfidence in Probability Distributions: People Know They Don’t Know but They Don’t Know What to Do About It

Abstract

Quantifying uncertainty in the form of a probability distribution is a critical step in many managerial decision problems. However, a large body of previous work has documented pervasive overconfidence in subjective probability distributions (SPDs). We develop new methods to analyze judgments about variables which entail both epistemic and aleatory uncertainty and, in three experiments, study the quality of people’s SPDs in such settings. We find that although SPDs roughly match the aleatory concentration of the real-world distributions, people’s judgments are consistently overconfident because they fail to spread out probability mass to account for their own epistemic uncertainty about the location and other properties of the distribution. Although people are aware of this lack of knowledge, they do not know how to appropriately incorporate it into their SPDs. Our results offer new insights into the causes of overconfidence in real-world judgment domains and shed light on potential ways to address this fundamental bias.