Opinion Science #116: Change Beliefs, Change Opinions? with Yamil Velez
10 snips
May 4, 2026 
Yamil Velez, assistant professor of political science at Columbia who studies when and why opinions resist change, discusses belief versus attitude differences. He explains why fact corrections often fail. He shares how AI-generated, personalized arguments can shift beliefs but only some belief changes move attitudes. The conversation covers tailored persuasion, focal reasons, and implications for polarization and AI research.
AI Snips
Chapters
Transcript
Episode notes
Attitudes Are Weighted Sums Of Beliefs
- Opinions are computed from many beliefs weighted by perceived importance rather than single facts.
- Fishbein's expectancy-value model explains why correcting a peripheral belief (e.g., hammers vs guns) won't shift an opinion rooted in other considerations.
Knowledge Alone Doesn't Predict Behavior
- Many people acknowledge factual negatives (e.g., smoking is unhealthy) but those beliefs may not be the ones shaping their attitude.
- If a belief isn't pivotal for someone, correcting it rarely changes their behavior or stance.
Rally Claim Debunked But Opinion Unmoved
- A viral clip showed a man claiming 'more people are killed with hammers than guns' and refusing to budge after being corrected.
- The debunked statistic didn't alter his Second Amendment opinion because it wasn't central to his stance.