Judgment errors in naturalistic numerical estimation.

Area of Science:

• Cognitive Psychology
• Judgment and Decision Making
• Numerical Cognition

Background:

• Everyday numerical estimations influence behavior and well-being.
• Estimates are prone to scaling errors (reporting mistakes) and belief errors (knowledge application mistakes).
• Prior models often conflated these error types or focused narrowly on scaling errors.

Purpose of the Study:

• To quantitatively model numerical estimates and associated errors.
• To investigate the role of semantic memory in forming beliefs about numerical quantities.
• To differentiate and model scaling errors versus belief errors in naturalistic judgment tasks.

Main Methods:

• Developed quantitative models of numerical estimation incorporating semantic memory insights.
• Tested models against participant data in everyday judgment tasks (Studies 1-2).
• Validated models using semantic judgment, free association, and verbal protocol tasks (Studies 3-8).

Main Results:

• Belief error models demonstrated high out-of-sample accuracy in predicting estimates and errors.
• Belief error models significantly outperformed scaling error models.
• Best-fitting belief error models replicated patterns previously attributed to scaling errors, suggesting they are a form of belief error.

Conclusions:

• Belief errors, informed by semantic memory, provide a more accurate account of numerical estimation than scaling errors.
• The cognitive underpinnings of judgment errors in numerical tasks are better explained by belief error models.
• Findings suggest that apparent scaling errors may represent nuanced belief errors.