Would you pay money to avoid an eight-minute conversation about politics with someone from the other party? Nearly a third of Americans in a recent experiment would. In a recent working paper, Luca Braghieri (Bocconi University), Peter Schwardmann (Carnegie Mellon University) and Egon Tripodi (Hertie School) recruited roughly 1,000 U.S. Democrats and Republicans, had them take an incentivised quiz on politically charged facts (on topics such as immigration, policing, healthcare) and paired them at random with a co-partisan or a counter-partisan for an unstructured video conversation, after which they could revise their answers for real money. Before talking, participants were willing to pay $2.45 on average for a conversation with a co-partisan but only $1.81 for one with a counter-partisan, citing two concerns: there would be less to learn, and it might be unpleasant. Given the choice, people would self-select into echo chambers.
Were they right to expect less learning? Objectively, no, because knowledge is spread across party lines, participants had more to learn from counter-partisans. Yet they converted that potential into actual learning at a significantly lower rate. Strikingly, recordings and transcripts reveal that co- and cross-partisan conversations barely differ in content or tone. The barrier lies not in what is said but in how it is received: participants believe counter-partisans are less informed, and this distrust survives the conversation itself, blocking them from crediting the information they hear.
On the social side, the news is better. Despite their apprehensions, participants found cross-partisan conversations as enjoyable as co-partisan ones, and this positive surprise significantly reduced affective polarisation: the gap in warmth people feel toward their own party versus the other. The effect persisted in a follow-up survey some 100 days later. Depolarisation was concentrated in emotionally engaged conversations, where partners disclosed personal experiences, expressed feelings, and validated each other.
The upshot is a mixed message for the idea that talking across the aisle will mend America’s divisions. Contact durably warms feelings toward the other side, but it does little for information sharing as long as people remain convinced the other side has nothing to teach. Policies encouraging contact can build social cohesion; unlocking the informational benefit may require tackling misperceptions about the other side’s competence directly.
Link (pdf): Talking Across the Aisle
