Skip to content
Citizen Infrastructure

The three levels of assembly impact

What does a citizens’ assembly actually do to the world? A 2025 evidence brief from People Powered — synthesising 66 empirical studies across five continents, led by Maria Milosh at the University of Chicago — proposes a clean answer: any given assembly’s impact lands on three distinct levels, and the evidence base for each is very different.

The three levels

Section titled “The three levels”
  • Personal impact. What happens to the members themselves. Deliberation reliably shifts what people know, how they weigh differing views, their sense of political efficacy, and often their subsequent participation. This is the strongest of the three evidence bases, though much of it is self-reported and short-term; the fewer multi-wave designs that exist find effects that persist six months out and further, showing up in voting behaviour and continued civic engagement. Assemblies that build in structured learning — expert briefings, Q&A with room to sit with what was heard, small-group discussion — amplify these effects most.
  • Spillover impact. What the wider community — people who never set foot in the room — takes away from the assembly. When an assembly is well-communicated, its mere existence lifts trust in government, reduces perceived polarisation, and improves social cohesion among non-participants. This is the case for treating public communication as part of the assembly’s work, not an afterthought.
  • Institutional impact. What actually changes in the institutions the assembly was addressing — laws passed, agencies restructured, decision routines opened up. This is the level with the thinnest evidence, both because openness of institutions is notoriously hard to measure across contexts, and because most assemblies are one-off and their recommendations advisory — making causal claims hard even when things do change. This is the field’s clearest open research frontier.

Two known gaps

Section titled “Two known gaps”

The brief and its authors are candid about where the evidence doesn’t yet reach:

  • Durability. Most personal-impact studies are short-term. Whether the shifts last — years out, across changes in participants’ circumstances — is largely unknown; longitudinal work is expensive and rare.
  • Geographic skew. Assemblies are expensive, so the evidence base is dominated by wealthy Europe and North America. Hyper-local work in Latin America and Africa, including projects in People Powered’s own network, is under-documented — a live gap for anyone who wants to know whether the personal / spillover / institutional pattern holds in resource-constrained contexts.

Why the taxonomy matters

Section titled “Why the taxonomy matters”

Judging an assembly only on institutional uptake — “did the government pass its recommendations?” — writes off the two levels where the evidence is actually strongest, and hands the field’s opponents an easy dismissal. The reverse mistake is also live: counting a personally-transformative but politically-inconsequential assembly as a success risks participation washing. The taxonomy is useful precisely because it forces the honest question — impact on what? — before the argument about whether an assembly “worked.”

It also sharpens the design brief. If a commissioner cares most about spillover, invest in public communication and open sessions to observers. If personal impact matters, design the learning phase harder. If institutional embedding is the goal, that is a different problem entirely — and one no single assembly on its own is likely to solve.

Sources

Section titled “Sources”