Reading focus
Syllabus: How Do Social Influence and Power Work?
Compare agreement with accuracy and confidence.
Why you’re reading it. Test what a consensus can establish.
A question to consider. How would you change the forecast exercise to distinguish useful new information from repeated versions of the same estimate?
Why this reading
A group can become more confident and less diverse without becoming more accurate. This experiment separates those outcomes instead of treating consensus as evidence of correctness.
Before you read
An aggregate combines individual estimates. Diversity here means numerical dispersion, not demographic representation. Arithmetic and geometric means combine values differently; extreme estimates affect them differently.
Read for
- Compare no feedback, average feedback, and access to all estimates across repeated rounds.
- Separate individual accuracy, aggregate error, estimate diversity, and confidence.
- Check why the authors transform estimates and how that affects their claim.
Consider afterwards
- How would collecting private estimates before discussion help you judge whether agreement improves a group's accuracy?
Editorial context
The experiment used 144 Swiss university students, factual estimates, incentives, and restricted communication. Its conclusions depend on tasks and aggregation choices. It does not show that deliberation, shared expertise, or all social learning makes groups less accurate.