Is Teenage Loneliness Actually Getting Worse?

The short version, with the numbers behind it

There are three bodies of evidence worth knowing about, and they broadly agree.

International school surveys. Twenge and colleagues, publishing in the Journal of Adolescence in 2021, used PISA data covering 1,049,784 students aged 15 and 16. School loneliness increased between 2012 and 2018 in 36 out of the 37 countries measured. The effect size was d equals 0.35, which is small to moderate. The authors state plainly that causation cannot be proven.

A longer run of the same question. Freije and colleagues, in the Journal of Adolescent Health in 2025, covered 38 countries from 2000 to 2022 and more than a million adolescents. The rise did not stop in 2018. It continued.

How much people are together. The Bureau of Labor Statistics measures how Americans spend their days. In its 2025 survey, released on 25 June 2026, 30 percent of people socialised on an average day, against 38 percent in 2015. The time spent on it was 35 minutes, against 41.

A longer run of the same survey exists, and it needs a label. It is an independent analysis of the survey's raw data, not a figure the Bureau published. It covers ages 15 to 24. On that analysis, average daily time socialising fell from 61 minutes in 2003 to 39 minutes by 2019.

That age band is the same objection this page raises further down against the figure everybody else prints. It applies here too. What is different is that this one stops before the pandemic, and it says what it counted. So it is used here with the limit stated.

Put together: teenagers report more loneliness at school than they did ten years ago, in most countries anyone has measured. People in general are spending less time together. That second part was well under way before the pandemic and did not come back after it.

The number this page does not print, and why

You have almost certainly seen a much more dramatic figure than any of the above. A collapse in time spent with friends, quoted as a single arresting pair of numbers, usually from a national health advisory.

It is not used here. Four reasons:

1. The age band it covers runs well past the teenage years, so it is mostly not describing teenagers. The 2003 to 2019 figure above has the same flaw, which is why it is flagged there.
2. Its end point is 2020, a year in which people were legally prevented from seeing each other. Ending a trend line in a lockdown produces a trend line about lockdown.
3. The measure is never defined in the document that made it famous. It is not stated what counted as time with friends.
4. It disagrees with the federal time use survey by roughly a factor of two and a half on the same baseline. Both cannot be right, and only one of them publishes its method.

The underlying paper could not be retrieved. That leaves a widely repeated figure whose source cannot be checked and whose value conflicts with the government's own measurement.

The direction it points is correct. Time together has fallen. That is why the honest thing is not to bin the claim but to use a source that shows its working, which is what the time use survey above does.

Two other much quoted figures in this area are also absent from this page. One is drawn from a survey of adults rather than teenagers, and compares a telephone survey with an online one, which is not a clean comparison. The other has been reported by its own author at two different values in a single piece of writing. Neither is dishonest at source. Neither is solid enough to hand to a frightened parent as a fact.

What nobody has shown

The cause. This is the part that gets skipped, and it is the most interesting finding of the lot.

The 38 country study tested six likely explanations against the rise. None was directly linked to it. More striking still: where internet access rose, loneliness was lower rather than higher. That is a correlation and not a cause, and it is the opposite of the standard story. Both of those should make everybody slower.

So the state of knowledge is this. The rise is real. It shows up across dozens of countries. Nobody has established what drives it. Anyone offering you one confident cause is ahead of the evidence.

What a parent should take from it

Very little about their own teenager, and that is the point.

An increase of d equals 0.35 across a million students is a shift in an average. It does not mean your fifteen year old is lonelier than you were at fifteen, and it does not mean they are not. National trends do not diagnose households.

What it does give you is context. If your teenager says school is the lonely part, they are saying something a great many people their age also say. That matters for one reason. It removes an explanation you might otherwise reach for, which is that this is rare and therefore a sign that something is wrong with them.

What this page cannot tell you

Why it is happening. Whether it continues. And whether the trend has anything to do with your teenager, which national averages are not built to answer.

What it can tell you is which numbers in this area hold up. The international school surveys and the federal time use data do. Several of the figures that circulate most widely do not, and they are the ones most likely to arrive on your phone at eleven at night.