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

This article investigates overconfidence in subjective probability distributions (SPDs), focusing on how people combine aleatory uncertainty (inherent randomness in outcomes) and epistemic uncertainty (imperfect knowledge about the distribution). Across four experiments involving judgments about everyday domains (e.g., commute times, housing prices), the research finds that people’s SPDs are consistently overprecise—that is, they concentrate probability mass too narrowly and underestimate their own epistemic uncertainty. Although participants recognize their limited knowledge, they fail to appropriately widen their probability distributions to reflect this uncertainty, a phenomenon termed "confidence leakage," where confidence about knowing the distribution’s shape improperly influences confidence about specific outcomes. The study introduces novel measures of distribution concentration and calibration, revealing that overprecision varies by domain and persists despite manipulations designed to highlight epistemic uncertainty, underscoring challenges in debiasing overconfidence in probabilistic judgments.