Deliberative polling

A Deliberative Poll asks what the public would think about an issue if it had the time, information and opportunity to reason about it together. A conventional opinion poll records top-of-the-head reactions; a Deliberative Poll first surveys a random, representative sample, then brings a subset of that sample together (usually over a weekend), gives them balanced briefing materials, lets them question competing experts and politicians, and has them discuss the issue in small moderated groups. At the end the participants answer the same questionnaire again. The difference between the before and after answers is presented as an estimate of "considered" public opinion.

The method was devised by the political scientist James Fishkin in the late 1980s and is a registered trademark associated with what is now the Deliberative Democracy Lab at Stanford University. It is one member of the wider family of mini-publics, alongside citizens' juries, consensus conferences and citizens' assemblies, but it differs from most of them in three ways: the sample is larger (typically a few hundred people rather than a few dozen), participants answer questionnaires privately rather than negotiating a joint report, and the design is quasi-experimental, often with a control group that is surveyed but does not deliberate.

Fishkin frames the method as a way to reconcile three values that usually trade off against each other: political equality (everyone's view counts equally, approximated by random sampling), deliberation (views are formed through weighing arguments) and non-tyranny or mass participation. Mass referendums satisfy equality but not deliberation; elite committees deliberate but are unrepresentative. A random sample that deliberates is meant to provide a "microcosm" of the public in a deliberative condition.

The causal claim behind the method is that opinion change during the event comes from information and reasoned exchange rather than from social pressure. Researchers test this by checking whether participants' knowledge rises, whether opinion change correlates with knowledge gains, and whether views become more internally consistent. Fishkin and Luskin report knowledge gains and sizeable opinion shifts across many polls, and argue that the changes are not explained by group polarisation or dominance by high-status participants.

Design elements

• United Kingdom, 1994. The first Deliberative Poll was held in Manchester in April 1994 with about 300 randomly selected Britons, on crime and punishment. It was filmed and broadcast by Channel 4. Participants became less punitive on several measures, for example in support for sending more offenders to prison.
• Texas electricity planning, 1996–1998. Under state integrated resource planning rules, several Texas utilities commissioned Deliberative Polls of their customers. Support for paying more for renewable energy and for efficiency measures rose after deliberation. Advocates credit the polls with helping to push Texas utilities towards wind power; the extent of that causal role is debated, since resource quality, transmission and state policy also mattered.
• United States, 1996 and 2019. The National Issues Convention in Austin (1996) was an early national application; Merkle's review in Public Opinion Quarterly questioned how far its results could be generalised. In September 2019 "America in One Room" brought several hundred registered voters, sampled by NORC, to a venue near Dallas. The organisers report that on many of the most polarising proposals Democrats and Republicans moved closer together, and that affective polarisation fell.
• Mongolia, 2017. A 2017 law required a Deliberative Poll before parliament could consider constitutional amendments. In April 2017, 669 randomly selected citizens deliberated in Ulaanbaatar on proposed changes. Support for creating a second parliamentary chamber fell sharply during deliberation, and the proposal was not pursued. The constitution was amended in 2019. Most published accounts are by participants in the process.
• Other uses. Polls have been held on local budgets, energy, European issues ("Tomorrow's Europe", 2007), in Chinese townships, on Japanese pension reform and on online platforms with automated moderation.

• Scale and influence. A few hundred participants can be representative, but the rest of the public does not deliberate. Unless an authority commits in advance to consider the results, a poll can become a sophisticated consultation that is cited when convenient and ignored otherwise.
• Agenda and materials. Whoever frames the questions and writes the briefing shapes the outcome. Advisory boards mitigate but do not remove this power, and participants rarely set the agenda themselves.
• What causes opinion change. Sturgis, Roberts and Allum found that knowledge gains and attitude change are only partly linked, which complicates the claim that information drives the shifts. Social dynamics in small groups, the novelty of the setting and the desire to please organisers are hard to rule out completely.
• Persistence. Some follow-up studies find that opinion changes fade after participants return to their usual information environment.
• Self-selection. Not everyone invited attends. Organisers compare attendees with the full sample on attitudes and demographics, but unmeasured differences can remain.
• Commercial and trademark concerns. The method is closely tied to one research centre, which helps quality control but concentrates methodological authority.

Deliberative Polls have been run in dozens of countries and are among the most systematically evaluated democratic innovations, thanks to their survey design. The consistent findings are that participants learn, that opinions often change, and that changes are usually not explained by simple group polarisation. Direct policy impact is rarer: Mongolia is the only known case where a national law made the method compulsory, and the Texas energy polls are the most cited instance of downstream policy change. The broader lesson is that random selection plus structured information can produce measurably different public judgements, but that the institutional link to decision-making decides whether those judgements matter.

• Fishkin and Luskin (2005), "Experimenting with a Democratic Ideal", Acta Politica
• Fishkin, Siu, Diamond and Bradburn (2021), "Is Deliberation an Antidote to Extreme Partisan Polarization?", APSR
• Fishkin and Zandanshatar, "Deliberative Polling for Constitutional Change in Mongolia", ConstitutionNet
• Deliberative Democracy Lab, Stanford University: first Deliberative Poll on crime, UK 1994
• James S. Fishkin, When the People Speak: Deliberative Democracy and Public Consultation (Oxford University Press, 2009)