Choosing Between a Digital and In-Person Group: What the Evidence Actually Supports

A funder asks whether your loneliness programme should run on Zoom or in a church hall. There is no trial that answers this cleanly. What exists is a scattered literature on social prescribing outcomes, workplace remote-work data, and two small randomised trials of loneliness interventions that were not designed to compare format at all. You have to build your decision from adjacent evidence, not a direct answer. Here is how to do that without overselling what you know.

Start with what the format actually changes

Digital and in-person groups don’t just differ in venue. They differ in three things that matter to outcomes: the bandwidth of the interaction, the incidental contact around the edges of it, and who can get in the door.

Bandwidth is the obvious one. A video call transmits voice and a flattened face; a room transmits posture, timing, who sits next to whom, and the unplanned five minutes before the session starts. Workplace research on remote arrangements distinguishes loneliness from workplace isolation as separate constructs — isolation is about structural exclusion from informal networks, loneliness is the subjective ache, and the two don’t move in lockstep. A digital group can reduce isolation (contact happens on schedule) while doing little for loneliness, because the incidental, unscripted contact that seems to matter most is exactly what a screen doesn’t carry.

Incidental contact is the second difference, and it’s the one programme designers underrate. Ray Oldenburg’s account of the “third place” and Eric Klinenberg’s work on social infrastructure both argue that shared physical space does work that the scheduled activity inside it does not — the corridor conversation, the person who arrives early, the fact that you now recognise someone on the street. None of that survives a login link. Recent tracking of physical third places in the US found closures across every category between 2019 and 2021, concentrated in higher-vulnerability and rural areas — which is a separate problem, but it tells you the option to meet in a shared physical space is not evenly available to begin with.

Access is the third, and it cuts the other way. A digital group removes transport, mobility, childcare, and shift-pattern barriers that keep people out of a room. Gallup’s global workplace data found fully remote employees report loneliness at 25%, against 16% for fully on-site — but that same workforce chose or was assigned to remote work for reasons that had nothing to do with loneliness, and the causal arrow is not settled. A parallel study using a nationally representative sample of employed US adults examined the loneliness-remote work association directly, which is useful precisely because most of what circulates on this topic is anecdote dressed as evidence.

What the intervention trials actually tested

Two of the few controlled trials of loneliness interventions were run in person, and neither compared formats — but they are still the closest thing to hard evidence available, and worth reading for what they tested rather than for a verdict on digital-versus-physical.

The HEAL-HOA trial tested prosocial engagement and volunteering against a control among lonely older adults in Hong Kong, in person. A separate randomised trial of befriending in residential aged care found reductions in loneliness scores at both 8 and 16 weeks against control, also delivered face to face. Read together, they suggest structured psychological approaches outperform simple social contact — but both were physical-world interventions. Nobody has run the equivalent trial online. If a funder asks you to cite evidence that a digital group produces comparable reductions in loneliness, the honest answer is that the trial evidence for either format specifically compared to the other does not exist yet.

The wider social prescribing literature is not much more help on this question. Systematic reviews report consistent self-reported benefit — increased self-esteem, restored participation, a sense of purpose — but a 2025 review protocol notes that social prescribing’s effectiveness for older adults specifically remains unclear, with only one peer-reviewed randomised trial in the area. A qualitative synthesis found that structured, purposeful activity looks more effective than contact alone, which matters regardless of format: the group needs a task, not just a room or a call.

What this means in practice: don’t ask “does digital work.” Ask which of the three things your target group needs most — reduced friction to attend, or the incidental contact that only a shared room produces. If your population is mobility-limited, working shifts, or geographically scattered, choose digital and build in deliberate low-stakes chat time to substitute for what the format loses. If your population can get to a room, default to physical and treat the pre- and post-session slack time as part of the programme, not overhead to be trimmed.

The named failure mode: the substitution trap

Programmes fall into what’s worth calling the substitution trap: assuming a digital session that replicates the content of an in-person group — same agenda, same duration, same facilitator script — will replicate the effect. It usually doesn’t, because the effect was never coming from the agenda. It was coming from the fifteen minutes nobody scheduled. A digital group that copies the agenda but skips a deliberate “arrive ten minutes early, stay ten minutes late, cameras on” structure is optimising the part of the session that mattered least.

Evidence status, plainly

Claim	Evidence status
Structured group activity with a purpose beats unstructured contact	Reasonably supported, across in-person social prescribing and trial literature
In-person format produces more incidental contact than digital	Theoretically well-grounded (Oldenburg, Klinenberg); not directly tested against digital
Remote/digital arrangements are associated with higher reported loneliness	Associational, workplace data only; causal direction unclear
Digital groups reduce loneliness as effectively as in-person groups	No direct trial evidence either way
Access barriers matter more than format for reach	Plausible from third-place availability data; not tested as an intervention choice

What this does not solve

Nothing here tells you which format will work for your specific group, because no trial has run that comparison. It also does nothing for the people your programme won’t reach regardless of format — those who never sign up for a loneliness intervention, digital or physical, because reaching for help is itself the barrier. Format is a real design decision. It is not the biggest one.