The friendship recession, and why it hit men hardest

The headline figures come from the May 2021 American Perspectives Survey, benchmarked against Gallup data from 1990. The men’s series moves roughly twice as far as the overall one, which is why this survey gets cited so heavily in coverage of a "friendship recession".

It is worth stating the caveat plainly: these are two surveys thirty years apart, run by different organisations with different instruments, and some of the gap could be method rather than change. "Close friend" is also self-defined, and the meaning may itself have shifted.

The time-use data is independent of the survey and points the same way, which is the main reason to take the trend seriously even if you discount the exact percentages.

The structural account is the most defensible one available. Male friendship in many cultures has been organised around shared activity rather than around disclosure — you were friends with the people you did something with. When the activity ends, so does the mechanism, and there is no habit of maintenance-by-contact to fall back on.

That fits the formation research: proximity plus repeated unplanned interaction plus a low-guard setting. An activity supplies all three at once, which makes it efficient and also fragile. Lose the five-a-side game and you lose the friendship group attached to it.

We are flagging this as a plausible reading rather than a finding. The survey documents the pattern; it does not establish the cause, and we have not found a study that does.

That social media caused it. A meta-analysis of 141 studies, roughly 145,000 participants and 897 effect sizes found that most associations between social media use and wellbeing were negligible. Panel work found little support for the idea that it displaces direct contact with close others — it most likely displaces other media.

That the elderly are the loneliest group. The BBC Loneliness Experiment, with 55,000 respondents, found 16–24 year-olds were the loneliest bracket: 40% felt lonely often or very often, against 27% of the over-75s.

We publish both of those because they cut against convenient narratives, including ones that would suit us. A page that only carries the findings supporting its own product is a page that selected them.

Rebuild the activity, not the friendship. Since the mechanism was shared activity, the repair is a recurring thing in a calendar rather than a resolution to be better at keeping in touch. A standing fortnightly game does more than a hundred good intentions.

Lower the bar on what counts as contact. Brief exchanges with people at the edge of your network are associated with more positive affect and greater belonging, and one-line messages are contact.

Send the message you are talking yourself out of. Recipients appreciate being reached out to more than senders predict, and the gap widens the longer the silence.