Does social media make you lonely?

Which way does the arrow point?

Both ways, and weakly in each. Fam and Männikkö pooled 26 longitudinal studies covering 24,798 people and separated the two directions using cross-lagged coefficients, which account for where someone already was at baseline.

Loneliness predicting later problematic media use came out at 0.10. Problematic media use predicting later loneliness came out at 0.09. The two paths are effectively the same size, and both are small.

That symmetry is the finding. There is a loop here, and the popular version of it, in which the feed does something to you, is only half the picture. Lonely people reach for it, and reaching for it leaves them slightly lonelier.

Does cutting back help?

A little, and the detail matters more than the headline. May, Malouff and Meynadier pooled ten randomised controlled trials covering 1,491 people and found reducing social media use improved depression symptoms at an effect size of 0.28, falling to 0.25 after adjusting for publication bias.

Then the moderator analysis splits it open. Trials that asked people to limit their use produced a significant effect of 0.33. Trials that asked for full abstinence produced 0.15, which was not significant.

Limiting worked. Quitting did not. That is the opposite of the advice most people are given.

Why would quitting do nothing?

A second meta-analysis says the same thing from a different angle. Lemahieu and colleagues pooled ten abstinence studies covering 4,674 participants and 38 effect sizes, and found nothing on any outcome: positive affect 0.03, negative affect minus 0.01, life satisfaction 0.03. None approached significance.

One plausible reading is that abstinence removes the connection along with the comparison. If a platform is where your friends actually are, deleting it costs you something real, and that cost may cancel the benefit.

Limiting keeps the connection and cuts the scrolling. That is a different intervention, and it is the one with a result behind it.

How much should you trust these numbers?

Less than you would like, and the authors are candid about why. The reduction trials ran three weeks at most with no long-term follow-up. Only four of ten included people who failed to comply. Half did not report attrition.

The abstinence studies were mostly Western students, mostly underpowered, with no power analyses and inconsistent definitions of what counts as social media.

The longitudinal work carries its own caveat: the funnel plots were asymmetrical, heterogeneity was moderate to high, and only English-language studies were included.

What is worth doing?

Cap it rather than quit it. The evidence supports limiting time, and the trials that tested going cold turkey found no benefit at all.

Watch what you use it for. Problematic use, meaning compulsive checking and distress when you cannot, is what these studies measured. Messaging a friend and doom-scrolling a feed are both social media, and only one of them looks like the problem.

Treat it as one input among many. An effect size of 0.10 in either direction means your feed explains a small fraction of how connected you feel, and the evidence on social connection and lifespan points at relationships that exist offline. If loneliness is the actual problem, the research on building friendships as an adult is the more direct route.