What actually helps you feel less lonely, according to 58 trials

A 2022 systematic review and meta-analysis gathered 58 randomised controlled trials of interventions for social isolation, loneliness, social participation or social connectedness in adults. Its contribution was to stop treating those as one outcome. It measured separately how much contact people had and how good their connections felt, two things earlier reviews had merged and then found mixed results for.

Kept apart, both moved. Across the trials, interventions raised objective social contact by a standardised effect of g = 0.43 and the perceived quality of connection by g = 0.33. Those are moderate effects, not dramatic ones, averaged across very different programmes. What produced them differed. Giving people access to other people was the most effective way to raise contact (g = 0.67). Teaching skills for the biased beliefs, fear-driven avoidance and practical barriers that get in the way was the most effective way to improve how connected people felt (g = 0.53).

The limits matter here. The mean age in the trials was 62, so the results apply most directly to older adults. The full paper is behind a paywall and we have read only the abstract, so follow-up lengths, risk of bias and the comparison between theoretical approaches are not covered here. None of the trials tested an app or a journal.

The split puts numbers on something many people know from experience: being busy with people is not the same as feeling close to any of them. In these trials, putting people in contact with others reliably produced more contact. It did not follow that they felt more connected. The intervention that did most for that was a different one, aimed at what people believed about social situations and at the avoidance those beliefs produce.

That has a practical consequence. If the only response to feeling lonely is to add more, more events, more groups, more plans, it works on the outcome that access moved best and leaves the other one to chance. Both matter. Contact is the raw material, and without it there is nothing to feel connected through. But how connected you feel seems to depend on what you think is happening in those contacts, and that part responded to a different kind of help.

An earlier piece here cited a similar ranking from three older meta-analyses, with the caveat that it rested on a small number of studies. This review points the same way across 58 trials, with caveats of its own: older samples, moderate effects, and an abstract rather than a full text behind our reading of it.

The review’s abstract names biased beliefs and fear-driven avoidance without listing which beliefs. The best-measured examples come from a separate set of experiments, and connecting the two is our reading rather than the review’s. In those experiments, people who sent an unprompted message predicted how the recipient would feel, and recipients appreciated it more than senders expected. The gap widened the more surprising the contact was, so the longer the silence, the bigger the underestimate. These were experiments about single moments of contact, not about friendships over time.

The liking gap runs the same way. Across studies of strangers paired in a lab, students sharing university housing over an academic year and adults in workshops, people underestimated how much their conversation partners liked them, and the gap survived months of living alongside each other. The samples were mostly students and workshop groups, liking was self-reported on both sides, and the size of the gap varied between studies. The direction did not.

Two more findings lean the same way. People who wrote gratitude letters underestimated how positive the recipient would feel and overestimated how awkward it would be. People assigned to deep rather than shallow conversations found them less awkward and more connecting than they had predicted, most of all for the deeper topics; those conversations were assigned, often between strangers, so the stakes were lower than in life. Together they describe a pessimistic model of how you land on others, and a pessimistic model is exactly what produces avoidance.

Not all of it is about close friends. In daily diary studies, people were happier and felt a greater sense of belonging on days when they had more interactions with people they barely knew. In a field experiment, a brief genuine exchange with a barista produced more positive feeling than a purely efficient one, and the effect ran through belonging.

These are small effects on momentary mood, mostly from student samples, and the barista study is a single brief manipulation. Nothing in them says acquaintances substitute for close friends. They say a short real exchange is not nothing, and that it is available on days when nothing bigger is.

It is also the lowest-stakes place to practise what the trials were teaching: saying one sentence more than the moment needs, when the forecast says not to bother. That framing is ours. The diary studies measured mood, not forecasts.

Work out which half you are short of. If you see almost nobody, the trials say the most effective step for contact is access: put yourself somewhere other people regularly are, and keep going back. If you see people and still feel alone, more access is not the main lever. The place to look is the forecast you make before each contact.

When you hesitate before a message, treat the hesitation as a forecast with a known bias, not as information about the other person. The expectation that you will be unwelcome, unliked or awkward has been measured and, on average, found to be too pessimistic. That does not make every message land. It makes the hesitation a poor guide. Put one specific thing in the message, something they told you last time, because a remembered detail gives it a reason to exist and gives them an obvious thing to reply to.

None of this needs an app. A short list of the people you mean to contact, one thing you know about each of them, and the decision to send the message when the forecast says not to, is the whole method. What does not work is waiting until you feel sure it will be welcome, because that forecast is the part the research says is wrong.