Politics · 6 min read
What Citizens' Assemblies Change—and What Their Influence Depends On
Assemblies often change participant knowledge and views, while public and policy effects depend on institutional design.

What Citizens' Assemblies Change—and What Their Influence Depends On
A citizens' assembly brings a relatively small group of residents together to learn about a public question, hear competing evidence, deliberate, and make recommendations. Members are usually selected by lottery and then stratified to resemble the wider population on chosen characteristics. The form promises something elections and opinion polls do not: time for ordinary citizens to consider a complex issue under structured conditions.
Research supports a modest version of that promise. Participants often gain knowledge, revise opinions, and produce reasoned recommendations. Evidence that an assembly changes the wider public or government policy is less consistent. Its influence depends on recruitment, information design, facilitation, media attention, and a clear institutional route from recommendation to decision.
A mini-public is designed, not simply sampled
Pure random selection can produce a group that is unbalanced by age, region, education, or other relevant characteristics, especially when many invitees decline. Organizers commonly send a large random set of invitations, recruit volunteers from respondents, and use stratified selection for the final group. This improves descriptive diversity but does not create a statistically precise miniature electorate.
Participation also requires time. Stipends, childcare, travel, disability access, and translation affect who can accept. If the invitation and support process excludes people with fewer resources, the final lottery cannot repair the imbalance.
Once assembled, members need a question, schedule, witnesses, written materials, facilitators, and voting rules. Each choice can affect the outcome. An oversight group may balance advocates and technical experts; members may request additional witnesses; fact-checkers may review claims. Transparency about those decisions is central because the public cannot evaluate deliberation it cannot inspect.
Participants usually learn and sometimes change position
Studies of deliberative mini-publics commonly compare surveys before and after participation. They often find increases in factual knowledge and shifts in policy attitudes. A 2025 meta-analysis in the European Journal of Political Research reported strong evidence that participants in democratic innovations change opinions about the policy under discussion, while effects on capabilities such as internal political efficacy vary by design and measure.
Opinion change is not proof that deliberation produced a better answer. Participants may update in response to evidence, but they can also follow persuasive speakers, perceived group norms, or the framing of the agenda. Researchers therefore examine speaking equality, reasons given, information balance, and whether minority positions remain expressible.
Facilitation can reduce domination by confident speakers and keep discussion relevant. It can also steer a group through subtle choices about when to intervene or summarize. Expert input can improve factual understanding while giving credentialed speakers disproportionate authority. The right comparison is not an influence-free assembly—no information process has one—but a process in which influences are diverse, disclosed, and contestable.
Ireland shows both reach and institutional mediation
Ireland's deliberative bodies are among the most studied cases. The 2012–2014 Convention on the Constitution combined 66 citizens with 33 politicians and a chair. A later Citizens' Assembly of 99 citizens and a chair considered the constitutional provision on abortion, among other issues. Its 2017 report recommended substantial reform.
Those recommendations entered an established constitutional sequence. A parliamentary committee considered the issue, the Oireachtas legislated for a referendum, and voters approved repeal of the Eighth Amendment in May 2018. The assembly did not itself change the constitution. Political institutions translated its work into a proposal, and the electorate made the legal decision.
The distinction prevents two opposite overstatements. It is too strong to say the assembly caused the referendum result by itself; campaigning, party decisions, social movements, court cases, and changing public attitudes also mattered. It is also too weak to call the assembly ceremonial. It created a structured public record and recommendations that officials used in moving a difficult question into the formal process.
Later Irish experience shows that the chain is not automatic. Government can modify, delay, or decline recommendations. Voters can reject constitutional proposals even when an assembly addressed related issues. An assembly's influence depends on what decision-makers promise in advance and how faithfully subsequent proposals reflect the deliberation.
Wider spillovers remain difficult to demonstrate
Assemblies involve tens or hundreds of people; democracies involve millions. The democratic case therefore depends partly on spillover: can nonparticipants learn from or trust the considered judgment of the group? A 2023 systematic review by Ramon van der Does and Vincent Jacquet found the evidence for many spillover effects tentative because the empirical literature was small.
Media coverage is one mechanism, but detailed deliberation competes poorly with conflict-driven news. Official reports can be long and little read. A recommendation may become a partisan cue, causing citizens to respond to the endorsing party rather than to the assembly's reasons. Public knowledge that members were randomly selected can increase perceived fairness in some settings, but acceptance varies with composition, issue, and prior views.
Policy impact is also hard to code. A government may adopt part of a recommendation years later for several reasons. Counting any overlap as success inflates influence; requiring verbatim adoption understates it. The OECD's 2020 survey documented hundreds of representative deliberative processes and proposed standards for good practice, but a case inventory is not a controlled estimate of effects.
Legitimacy depends on the assignment
A citizens' assembly can advise elected institutions, set an agenda, draft options, or in rare designs receive delegated authority. These roles require different standards. An advisory body can broaden evidence without claiming to replace representation. A body with binding power needs clearer authorization, accountability, and review.
Random selection contributes a kind of equality: residents who would not run for office can participate. Yet a small assembly cannot embody every identity or interest. Stratification categories involve judgments about which differences matter. Deliberation can generate informed views, but it cannot convert a temporary sample into “the people” as a whole.
Good design therefore begins with an explicit remit and response commitment. Organizers should publish selection methods, evidence, funding, conflicts, transcripts or recordings where appropriate, and reasons for accepting or rejecting recommendations. Participants need enough time and support. Opposing evidence should be answerable. Independent evaluation should measure process and downstream effects without assuming that policy adoption is the only success.
What the evidence permits
Citizens' assemblies can change their participants. The research literature supports learning, considered opinion change, and the production of recommendations under well-resourced conditions. Bounded cases such as Ireland show that assemblies can become consequential parts of a constitutional process.
The evidence does not support a general claim that assemblies reliably move mass opinion, improve trust, or cause governments to enact their proposals. Selection, agenda control, facilitation, expert influence, and institutional follow-through remain sources of both variation and contest.
The central question is therefore not whether an assembly is democratic in the abstract. It is what job the body has been assigned, how its members and evidence were chosen, what route connects its work to public authority, and whether that route was honored. Those conditions determine whether deliberation becomes informed advice, political cover, or something in between.
Sources
5 sources, every one with a link you can follow.
- OECD, Innovative Citizen Participation and New Democratic Institutions: Catching the Deliberative Wave (2020).
- The Irish Citizens' Assembly, Final Report on the Eighth Amendment (2017), a primary institutional record.
- van der Does and Jacquet, “Small-Scale Deliberation and Mass Democracy”, Political Studies (systematic review).
- “A survey of methods for evaluating mini-publics”, a systematic review of evaluation practice.
- “A meta-analysis of the effects of democratic innovations on participants”, European Journal of Political Research (2025).
