When intuitive Bayesians need to be good readers: The problem-wording effect on Bayesian reasoning

Area of Science:

• Cognitive Psychology
• Decision Science
• Behavioral Economics

Background:

• Humans often fail to apply Bayesian reasoning to textbook problems, despite appearing intuitive Bayesians when updating probabilities from experience.
• Previous research focused on statistical formats (e.g., natural frequencies vs. single-event probabilities) but largely overlooked the impact of problem wording.

Purpose of the Study:

• To investigate the independent effects of statistical format and wording on Bayesian reasoning in medical screening problems.
• To determine if improved wording can enhance Bayesian reasoning, particularly with natural frequency formats.

Main Methods:

• Conducted seven experiments with 4909 adult participants evaluating Bayesian reasoning in medical screening scenarios.
• Manipulated the statistical format (natural frequencies, single-event probabilities) and the wording (standard vs. improved) of the problems.
• Analyzed the proportion of Bayesian answers generated under different conditions.

Main Results:

• Improved wording significantly increased Bayesian answers, especially when using natural frequencies.
• The effect of improved wording on natural frequencies was more substantial than the effect of using natural frequencies over single-event probabilities.
• Improved wording facilitated Bayesian reasoning by clearly cueing mathematical operations, unambiguously linking numbers to their reference class, and simplifying language.

Conclusions:

• Wording is a critical, often underestimated, factor influencing Bayesian reasoning in textbook problems.
• Improved wording can overcome limitations in statistical format, leading to more accurate Bayesian inference.
• Effective problem-solving, even for intuitive Bayesians, requires strong reading comprehension skills for statistical problems.