Carnegie Mellon University
April 01, 2022

Polarized Speech: A Function of Self-Persuasion

By Stacy Kish

A new study finds competitive debaters, randomly assigned a position, persuade themselves to the superiority of their side, even if it falls contrary to their own personal beliefs. This suggests self-persuasion is a significant and resilient contributor to polarization and disagreement on policy. The results are available in the April 1 issue of the journal American Economic Review.

“Politicians are in the business of persuasion,” said Peter Schwardmann, assistant professor in social and decision sciences at Dietrich College of Humanities and Social Sciences and contributing author on the study. “This work gives us a window into politics and how a politician’s beliefs may evolve.”

Polarization in society is on stark display in the U.S. Congress. No matter how much people communicate, we seem to be growing further and further apart.

While self-persuasion has been studied in the laboratory setting, Schwardmann and a team of researchers know that this phenomenon is not confined to the lab. They used data gathered during international debate competitions to explore how self-persuasion influences an individual’s factual belief and confidence when defending a position.

“We find that competitive debates lead to polarization, because people persuade themselves that their side is right even before the debate starts,” said Joël van der Weele, associate professor at the Center for Research in Experimental Economics and Political Decision Making at the University of Amsterdam. Van der Weele is a contributing author on the study. “The debate itself does not lead to convergence of opinions, so the initial polarization persists, even when we ask them a day after the debate.”

Data collection took place at four competitions (2019, 2020 and 2021) that involved more than 400 participants from 58 countries. At the beginning of every debate, each team was presented with a topic and randomly assigned to either the supporting or opposing position. Teams received 15 minutes to prepare their defenses — without time for research — before engaging in an hour-long debate that followed the procedures of British Parliamentary debating rules.

Schwardmann and his colleagues gathered three types of surveys to evaluate participants’ thoughts on a topic throughout each competition. They took a baseline survey before the event, a second survey before each debate and a final survey after each debate. The surveys evaluated the participants’ factual beliefs in the motion being argued, confidence in the strength of their position and how personal attitudes aligned with the argued motion.

The researchers found self-persuasion occurs despite incentives for accuracy and persists even after exposure to opposing views. In addition, participants were inclined to believe a statement was true if it strengthened their argument for an assigned position.

“We like to think that we are rational people who base [our] opinions on fact, but  we often end up with the opinions that are ‘convenient’ or strategically useful in a given context,” said van der Weele. “The apparent ease with which we do this, even in a setting where these opinions have been induced in an explicitly random manner, should lead us to question our own opinions much more, or simply take them less seriously.”

Self-persuasion can drive political beliefs and limit the ability to resolve conflict. Schwardmann is interested in exploring this topic further, with a focus on whether greater confidence in a position actually helps with persuading others.

“The exchange of ideas during a competitive debate does not lead people to reach consensus,” said Schwardmann. “A useful strategy to avoid self-persuasion may require a more collaborative approach to arrive at the truth.”


Schwardmann and van der Weele were joined on the project, “Self-Persuasion: Evidence from Field Experiments at International Debating Competitions,” by Egon Tripodi at the University of Essex. The project received funding from the CRC TRR 190 Rationality and Competition, the Research Priority Area Behavioral Economics at the University of Amsterdam, the Dutch Science Foundation, the European University Institute and the Russell Sage Foundation.