Toward a more nuanced understanding of probability estimation biases

Area of Science:

• Cognitive Psychology
• Decision Science
• Behavioral Economics

Background:

• Humans frequently make probability judgments with uncertain evidence.
• Previous research shows significant misestimations in these judgments, with various proposed causes like base rate neglect or overweighting individuating information.

Purpose of the Study:

• To quantitatively assess the contribution of different factors to probability estimation errors.
• To determine if single factors can fully explain observed judgment biases.

Main Methods:

• Non-professional subjects estimated probabilities in four real-world scenarios: cancer detection, drunkenness, and sniper detection.
• Statistical analysis quantified the influence of various explanatory variables on probability judgments.

Main Results:

• Explanatory variables accounted for 30-45% of response variance across scenarios.
• No single factor explained more than 53% of the explainable variance.
• No single factor disproportionately contributed more than its 'fair share' to the explained variance.

Conclusions:

• Attributing probabilistic judgment errors to a single cause, including base rate neglect, is statistically untenable.
• Biases in probability estimation reflect a weighted combination of multiple factors.
• The specific mix of contributing factors varies depending on the problem scenario.