America in One Room (2019 national Deliberative Poll)

Background

By the late 2010s American politics was marked by rising "affective polarisation": partisans not only disagreed but increasingly disliked and distrusted each other. A large literature suggested that discussion across party lines could backfire, entrenching positions. James Fishkin, the Stanford political scientist who invented Deliberative Polling in the late 1980s, and his colleagues set out to test the opposite expectation at national scale ahead of the 2020 presidential election.

America in One Room was organised by Stanford's Center for Deliberative Democracy (now the Deliberative Democracy Lab) with the Helena Project, a non-profit, and By the People Productions; NORC at the University of Chicago did the survey work.

How it is governed

• Sampling. NORC drew a stratified random sample of registered voters to form a microcosm of the electorate. 523 "citizen-delegates" came to a hotel in Grapevine, Texas, near Dallas, from 19 to 22 September 2019. A control group of 844 people who did not attend answered the same questions at the same times.
• Materials. A 55-page briefing document, intended to be balanced, set out arguments for and against 26 or more proposals on five issues: immigration, the economy and taxes, health care, the environment, and foreign policy.
• Process. Delegates alternated between small moderated groups and plenary sessions with competing experts, and questioned five presidential candidates — Bill Weld, Joe Walsh, Mark Sanford, Julián Castro and Michael Bennet. In the Deliberative Polling format, moderators do not steer the substance and questions for the plenaries are formulated in the small groups.
• Measurement. Delegates answered a detailed questionnaire before and after; a follow-up a year later reached 463 of the 523.

Economy and public goods

The event was funded privately; the sources reviewed for this entry do not report its cost or what delegates were paid or reimbursed. The product was public information: published before-and-after results, media coverage (including a New York Times feature with portraits of 519 delegates and a documentary series) and a peer-reviewed article in the American Political Science Review.

Conflict and disputes

• Selection into attendance. Although the sample was random, those who chose to travel may differ from those who declined; the control group helps but cannot fully remove this.
• Artificial setting. Critics question whether a weekend of balanced information, free of partisan media, tells us much about everyday politics; Fishkin's team argues it shows what an informed public would think.
• Researcher involvement. The method's designers also produced the main analysis; independent replication of such national events is rare because of cost.
• Political reading of results. The one-year follow-up found delegates 15 points more likely than the control group to support Joe Biden, rising among moderates and independents — a result that opponents could read as partisan effect rather than depolarisation.

Outcomes

The APSR article reports that deliberators showed large, depolarising changes in policy attitudes and large decreases in affective polarisation, while the control group did not. On 26 highly polarising proposals, Democrats and Republicans moved closer on 22, significantly on 19. Examples include Democratic support for a $15 federal minimum wage falling from 82.5% to 59.4%, and Republican support for rejoining the Trans-Pacific Partnership rising from 23% to 62%. Each party's favourability towards the other rose by about 13–14 points, and agreement that people with opposing views have good reasons for them rose from 34% to 54%. 98.2% of delegates found the experience valuable.

The APSR article closes by discussing how elements of this "antidote" could be scaled, for instance through online deliberation, to reach far more people than a single in-person weekend.

Lessons for institution design

1. Structure, not just contact, matters. Balanced materials, moderated small groups and access to competing experts seem to be what turns cross-party contact into depolarisation.
2. Control groups make claims credible. Comparing participants with non-participants answers the obvious objection that opinion was shifting anyway.
3. One-off events do not change institutions. Without a connection to a decision-making body, the effects remain on participants and on public debate.
4. Scale is the bottleneck. National in-person events of this kind are expensive; online deliberation tools are the main proposal for reaching more people.

Further reading

• Fishkin, Siu, Diamond and Bradburn (2021), "Is Deliberation an Antidote to Extreme Partisan Polarization?", American Political Science Review
• Stanford Deliberative Democracy Lab, America in One Room
• Wikipedia, America in One Room