n.Newfeed

Politics · 7 min read

The Current Evidence on Citizens’ Assemblies and Public Decisions

What assemblies can support in participant deliberation, and why representation and policy influence depend on design.

Assemblies promise a different kind of public judgment

A citizens' assembly brings a relatively small group of residents together to learn, deliberate, and make recommendations on a public question. Members are usually selected through a lottery followed by stratification intended to approximate selected characteristics of the wider population. Facilitators structure discussion, specialists and affected groups provide evidence, and the assembly reports to a government, legislature, referendum process, or public audience.

The current debate is not whether participants can sit through complex evidence. Many documented assemblies show that they can. The harder questions concern representation, agenda control, public legitimacy, and influence. A group of 50 or 100 people can reason carefully and still fail to represent every politically relevant experience. A government can commission an assembly and still ignore its recommendations.

As of 2026, public authorities continue to use deliberative processes, especially for environmental and long-term policy questions. The OECD's Government at a Glance 2025 reports that 60 of 148 recent cases in its dataset focused on environmental issues. That count documents institutional interest; it is not an estimate of effectiveness or a census of every assembly worldwide.

Selection aims at diversity, not a perfect miniature

Organizers often use a two-stage process. They send many invitations at random, then select from respondents to match targets such as age, gender, geography, education, or socioeconomic position. Compensation, childcare, transport, accessible venues, and language support can make participation possible for people who cannot attend conventional public meetings.

This method can produce a group more demographically varied than a self-selected hearing. It does not create a statistically perfect sample. Response rates may be low, initial address lists may exclude people, and the final group is small. Stratification can balance only characteristics organizers record and choose. Political identity, disability, race, religion, migration status, or lived experience may be missing or legally difficult to collect in some jurisdictions.

A 2025 critical analysis by Arvid Lindell and colleagues identifies four challenges to strong claims of representativeness: small samples, group effects during deliberation, bias in sampling frames, and low acceptance rates. Their argument is not that assemblies are useless. It is that organizers should specify what kind of representation they seek rather than describing a few dozen people as “the public.”

Advocates respond that electoral institutions are not perfect demographic mirrors either, and that the assembly's purpose is considered judgment under better informational conditions, not a survey estimate. That response changes the claim. It presents the body as a diverse deliberative forum whose members can offer reasons, not as an instrument that automatically measures majority opinion.

Learning and facilitation create capacity and power

Assembly members typically receive briefing materials, hear witnesses, question experts, deliberate in small groups, and revise recommendations. Well-designed processes give participants time, balanced materials, transparent evidence selection, and opportunities to request additional information. The OECD's 2020 good-practice principles emphasize a clear public purpose, accountability, transparency, representativeness, inclusion, information, group deliberation, time, integrity, privacy, and evaluation.

These design choices can improve the quality of discussion. Participants often report learning, and observational research documents people revising views after hearing evidence and one another. Deliberation can surface tradeoffs that a yes-or-no poll compresses. It can also generate recommendations with explanations, minority positions, and implementation conditions.

The same learning phase gives organizers substantial power. Someone frames the remit, selects initial evidence, hires facilitators, schedules speakers, and decides which claims require rebuttal. Neutral facilitation is an aim, not a measurable absence of influence. Experts may define feasible options before citizens evaluate them.

Recent systematic work on expert input asks whether this influence becomes undue. Early findings presented to the European Consortium for Political Research synthesize 92 empirical studies and suggest that longer deliberation, competing expert perspectives, neutral facilitation, and participant autonomy help members question expert claims. Because this review is recent and conference-stage evidence may change before final publication, it should be treated as a signal rather than a settled field estimate.

The practical implication is straightforward: an assembly needs an auditable information process. Publishing briefing documents, witness-selection rules, funding, facilitator roles, transcripts, and dissenting views makes influence visible. Transparency cannot remove framing, but it allows scrutiny.

Recommendation quality differs from political effect

Assemblies can be evaluated at several levels. One can ask whether recruitment was fair, whether participants learned, whether deliberation was respectful and reasoned, whether recommendations were coherent, whether the wider public considered them legitimate, and whether policy changed. Success on one level does not guarantee success on another.

Ireland's constitutional processes are frequently cited because assembly or convention recommendations helped place issues before voters, including marriage equality and abortion. Those outcomes depended on a larger chain involving elected officials, advocacy, legal design, campaigns, and referendums. The assemblies were consequential institutions within that chain, not solitary causes of constitutional change.

Elsewhere, recommendations enter a report and receive a formal response without implementation. Governments may accept easy items and defer costly ones. Political conditions can change after an assembly. A recommendation may also be influential without immediate adoption by changing language, evidence, or coalition positions—effects that are difficult to measure and easy to exaggerate.

An assembly's mandate matters. Advisory bodies allow elected representatives to retain accountability but risk symbolic consultation. Binding authority increases consequence while raising questions about delegation to a small unelected group. Linking recommendations to a referendum expands the decision to the electorate but subjects the result to campaign resources and simplified ballot wording.

Public legitimacy cannot be borrowed automatically

Members may become informed through weeks of work, while the wider public sees only a final list. This creates what political theorist Cristina Lafont and others identify as a democratic concern: people outside the room are asked to defer to a process they did not experience. Descriptive diversity may make the assembly relatable, but it does not make participants authorized representatives in the electoral sense.

Supporters propose “coupling” assemblies to wider institutions. Public submissions can inform the agenda; assembly sessions can be broadcast; members can explain their reasoning; legislatures can debate each recommendation; and voters can decide specified questions. Permanent or recurring assemblies may build institutional knowledge and predictable routes to response.

These designs trade depth against scale. An assembly creates conditions for sustained attention that millions of citizens cannot share simultaneously. Wider participation increases reach but may reduce time per person and change who speaks. Digital tools can broaden input while introducing access, moderation, verification, and platform-governance problems.

Cost is another constraint. Representative recruitment, compensation, facilitation, evidence preparation, translation, accessibility, and evaluation require money and months of work. That may be justified for a consequential, clearly scoped question. Routine use for decisions with little room for influence can become expensive theater.

The field is moving toward institutional fit

The earlier case for assemblies often emphasized novelty: ordinary citizens selected by lot could reason across disagreement. The current debate is increasingly about institutionalization. The OECD has documented recurring models at local and regional levels, and practitioners are designing standing panels, legally required government responses, and links to legislatures.

This shift improves the research question. Instead of asking whether citizens' assemblies “work” in general, evaluators can compare recruitment methods, information governance, facilitation, mandates, response duties, and policy domains. Pre-registered evaluation, participant and nonparticipant surveys, independent observation, and long-term tracking can separate immediate satisfaction from durable political effects.

Evidence is strongest for a bounded proposition: under supportive conditions, diverse lay participants can learn about complex issues, exchange reasons, and produce considered recommendations. Evidence is weaker and more variable for claims that assemblies represent the whole population, reduce polarization outside the room, increase general trust, or cause policy adoption.

The signals to watch are therefore institutional. Do invitations reach and retain people usually absent from consultation? Can participants choose or challenge evidence? Are minority reasons recorded? Must authorities respond item by item? Does the wider public understand the process? Are outcomes tracked after the launch publicity ends?

Citizens' assemblies are neither replacements for elections nor elaborate opinion polls. They are designed deliberative bodies whose value depends on how they connect to public authority. The open issue in 2026 is less whether citizens can deliberate than whether institutions will give that deliberation a fair input, a transparent mandate, and an accountable route to consequence.

Sources and review note

2 sources, every one with a link you can follow.

  1. Current practice is documented in the OECD's Government at a Glance 2025 chapter and its 2020 comparative report and principles.
  2. Representativeness critiques are synthesized in Lindell and colleagues' 2025 analysis.

Claims about policy effects remain case-dependent; process quality and government uptake must be assessed separately.

Want the reader to remember this spot? Sign in — it is free.

Next Ranked-Choice Voting: What Municipal Elections Show About Turnout and Ballot Exhaustion

The app

Keep going, or take it with you.

Free accounts read twenty pieces a day. Pro is US$1.99 a month or US$11.99 a year for unlimited reading.