Evaluating Groups With Roles Versus Groups Without Them

A funder asks you to justify why a structured group programme costs more per participant than a drop-in social club. The honest answer is not “because more staff time is involved.” It is that the evidence increasingly distinguishes between groups that give people something to do and be responsible for, and groups that simply put people in a room together. If your evaluation plan doesn’t measure that distinction, you cannot answer the question a funder is actually asking.

This matters more than it sounds. Two randomised trials in older adults, run by the same research programme, tested different mechanisms against a common comparator and got different results. The comparator both times was befriending — regular, warm, unstructured social contact, usually delivered by a volunteer. In the first trial, structured prosocial engagement and volunteering roles outperformed befriending on loneliness reduction among lonely older adults in Hong Kong. In the second, telephone-delivered behavioural activation and mindfulness, given in eight defined half-hour sessions to more than 1,100 older adults living in poverty and digital exclusion, also beat a befriending control at 12 months. Separately, a 2025 trial found that befriending itself is not nothing — it reduced loneliness scores meaningfully in residential aged care at 8 and 16 weeks. Read together, these three results say something specific: contact alone works a little; contact plus a defined role or a defined activity works more.

The roundtable problem

Call it the roundtable problem. A group with no assigned roles — everyone arrives, everyone talks, nobody is responsible for anything beyond showing up — tends to produce a wide spread of experience. Some members find each other and form the relationships the programme hoped for. Others sit at the edge of the conversation every week, attend three sessions, and stop coming. Because there is no role structure, there is no natural way to notice this happening, and no natural intervention when it does. Attendance numbers look fine for months before the dropout shows up in the data, if it shows up at all, because unstructured groups rarely track individual trajectories.

A qualitative synthesis of social prescribing participants found this same pattern from the participant side: people describe benefit that goes beyond contact itself, toward restored participation and purpose, and structured, purposeful group activity is reported as more effective than contact alone. That is a participant-reported perception, not a trial outcome, and it should be treated with the caution that implies. But it lines up with what the trial evidence shows, which is unusual enough in this field to be worth taking seriously.

What “structure” actually means, operationally

Before you can evaluate for it, you need to define it in terms a programme can actually build. Structure in a group setting typically means one or more of the following:

1. A defined role each member holds — not “group member” but something with a name and a function: peer mentor, note-taker, activity lead, greeter for new arrivals, rotating host.
2. A task the group produces together — a meal, a garden bed, a shared project — rather than conversation as the entire content of the session.
3. A visible responsibility that depends on the person showing up — if someone else has to cover for them, absence has a cost the group notices immediately, not three weeks later.
4. A session structure with a beginning, middle and end — not open-ended socialising, but a shape that gives people something to arrive for and something to have completed by leaving.

None of these require large budgets. They require deciding, before the first session, who does what — and building that decision into the programme design rather than leaving it to emerge.

Evidence status: what you can and cannot claim

Claim	Evidence status
Unstructured befriending/social contact reduces loneliness	Supported by randomised evidence, though effect sizes are modest and time-limited
Structured, role-based prosocial engagement outperforms befriending on loneliness outcomes	Supported by one randomised trial (HEAL-HOA volunteering arm); not yet widely replicated
Structured behavioural/psychological content delivered in a group format outperforms befriending	Supported by one large randomised trial in a specific, disadvantaged population; generalisability to other populations untested
Participants perceive structured, purposeful groups as more valuable than unstructured contact	Supported by qualitative synthesis; self-report, not an outcome measure
Social prescribing generally reduces loneliness	Systematic reviews report positive outcomes across included studies, but trial quality is described as heterogeneous and the number of studies is small
Social prescribing’s effectiveness for older adults specifically	Described in a 2025 systematic review protocol as still unclear, with only one peer-reviewed randomised trial existing in this specific area

The pattern across the row is consistent even where the underlying evidence base is thin: role and structure appear to add something contact does not, but the number of trials testing this directly is small, and most are in older-adult populations. Do not extend this claim to workplace groups, youth programmes, or general community groups without saying so.

What to put in the evaluation plan

If you are commissioning or funding a group programme, ask for these measures specifically, rather than accepting attendance and satisfaction alone.

• Role assignment log. Which member held which role, for how long, and whether the role was reassigned or lapsed. This is the operational proxy for “structure” and it is cheap to collect.
• Individual attendance trajectory, not aggregate attendance. A programme with 80% average attendance can still be losing a third of participants by week six if the other two-thirds are near-perfect attenders. Track individuals.
• A loneliness measure taken at baseline and at a fixed follow-up point, using a validated instrument rather than a bespoke satisfaction question, so results are comparable across cohorts and, ideally, across programmes.
• A separate measure of role engagement or task completion, distinct from the loneliness measure, so you can test whether the two move together — which is the actual mechanism claim you are trying to fund.

What this means in practice: if two group programmes cost the same and one assigns members defined roles or joint tasks while the other is an open drop-in, ask the structured programme for role-assignment data and individual attendance trajectories, not just headline attendance and satisfaction scores. If it cannot produce them, it has not actually tested the mechanism it is claiming credit for.

A caution about causal claims

None of the trials cited here were run on general community groups, workplace teams, or younger adults. They were run on older adults, often in conditions of poverty or isolation specific to the study population. A funder should not accept “roles improve outcomes” as a settled, portable finding. It is a finding replicated twice, in a narrow population, against the same comparator. That is a real pattern worth designing around — it is not yet a rule.

It is also worth separating two different funder questions that get conflated. “Does this programme reduce loneliness?” is an outcome question. “Does giving members roles improve on contact alone?” is a mechanism question. A programme can answer the first with weak evidence and still be worth funding for other reasons — cost, reach, community buy-in. But if a proposal is asking for a premium specifically because of its structured, role-based design, it should be evaluated against the mechanism question, not the general outcome question, and held to the higher evidentiary bar that claim requires.

What this does not solve

This guide addresses design and measurement inside a programme that people have already joined. It says nothing about who joins in the first place. Structured, role-based groups likely demand more of a participant up front — a role is a commitment, not just an invitation — which may screen out exactly the most isolated people the programme is meant to reach. Reach and role structure may trade off against each other, and no source here tests that trade-off directly. Any evaluation plan built on this guide should track who enrols and who declines, not only what happens to the people who stay.