Selection with Variation in Diagnostic Skill: Evidence from Radiologists.

Area of Science:

• Decision analysis
• Medical diagnostics
• Behavioral economics

Background:

• Systematic differences in decision-making exist across professions like physicians and judges.
• Existing models attribute these differences solely to variations in preferences.

Purpose of the Study:

• To develop and apply a framework that accounts for both preferences and diagnostic skill variation.
• To analyze how skill differences impact diagnostic decisions in a real-world setting.

Main Methods:

• Developed a novel framework incorporating preferences and diagnostic skill.
• Applied the framework to study pneumonia diagnoses by radiologists under quasi-random assignment.
• Estimated a model to quantify the impact of skill and preferences on diagnostic thresholds.

Main Results:

• Radiologists exhibit variation in diagnostic skill, with skill accounting for 39% of decision variation.
• Less skilled radiologists optimally adopt lower diagnostic thresholds due to perceived costs of misdiagnosis.
• Uniform decision guidelines are less effective than skill-enhancing policies.

Conclusions:

• Diagnostic skill is a critical, often overlooked, factor in decision-making variations.
• Failing to account for skill differences can lead to misleading research findings.
• Policies should focus on improving diagnostic skill for better outcomes.