The advice is always the same sentence: stop comparing yourself to other people. It is offered as though comparison were a choice you keep making badly.
Three decades of research say it is not a choice at all. Comparison happens on contact, before you have decided anything. The effortful part — the part that takes concentration, and that falls apart when you are tired or busy — is undoing it afterwards.
The comparison is automatic. The correction is not
In 1995 Daniel Gilbert and colleagues showed the shape of it. People compare themselves with whoever is in front of them without meaning to, and then, if they have the mental room to spare, they adjust for the fact that the comparison was not a fair one.
So “stop comparing” is not a technique. Nobody has ever managed it. People who seem unbothered by other people’s lives are not skipping the first step. They are doing the second one quickly, and usually without noticing they have done it at all.
The underlying theory is older still. Festinger’s 1954 account starts from a problem you cannot get out of: for anything without an objective measure — how good a parent you are, whether your career is going anywhere, whether you are doing enough — there is no ruler in the drawer. Other people are the only instrument available. Comparison is not a flaw in the machinery. It is the machinery.
What comparing upward actually goes with
A 2026 review pooled 54 separate studies covering 36,583 people, all of them about one specific act: looking at someone doing better than you, on a screen.
Across all of it, the link with worse mental health was moderate — around 0.33, on a scale where 0 would mean no relationship whatsoever and 1 would mean a perfect one. That is a real finding. It is also the least interesting number in the paper, because the average flattens the thing worth seeing, which is the order.
Nearly twice
How much more strongly comparing upward is linked to the sting you feel in the minutes afterwards (0.458) than to your self-esteem (0.251) — the outcome every article on this subject is written about, and the weakest of the five measured. Anxiety fell between the two, then low mood, then general wellbeing.
Self-esteem is the thing everybody worries about. Across 25 separate measurements it had the smallest link of the five.
So this is not a picture of your sense of yourself quietly wearing away. It is a picture of something sharper and much faster: the specific sting of being outperformed where other people can see it. That feeling is tied to the behaviour roughly twice as tightly as your self-esteem is.
What this review can and cannot tell you
These are links, not proof of cause, and the loop is obvious once you say it out loud: people who feel worse compare more, and comparing more makes them feel worse. Nothing in this analysis can pull those two apart.
In other respects it is unusually clean. The pattern held whatever the age of the people studied, whether they came from Western or Eastern countries, and whichever year the data was collected. It also held whether researchers measured people once or followed the same people over months. That last one earns a little trust: if the link only appeared when everything was asked in a single sitting, you would suspect the questionnaire rather than the world.
The authors also ran the standard check for a particular kind of distortion — the pattern that shows up when small flattering studies get published and small unflattering ones quietly do not. It found nothing. They are careful to add that finding nothing is not the same as proving there was none, and they are right to be careful.
Nothing changed the pattern, which is itself a finding
Reviews like this usually spend most of their length explaining who the effect is bigger for. This one could not find anybody.
It did not differ by age — adolescents and adults alike. It did not differ between Western and Eastern cultural backgrounds. It did not differ by the year the data was collected, and it did not differ between studies that measured people once and studies that followed them over time.
Three of those quietly contradict the usual framing. This is not specifically a teenage problem. It is not a local one. And on the evidence available it is not getting worse — an awkward result for the version of the story in which feeds are making everything accelerate, given that these studies span the entire period in which feeds took over.
The only thing that changed the answer was which outcome the researchers chose to measure.
Comparing downward is not the antidote
The obvious counter-move is to go and find somebody doing worse. It has a long history in the research and a mixed record.
It does protect your self-evaluation, briefly. It also requires you to keep the ranking in place. Somebody who manages their self-esteem by locating people below them has not escaped the comparison frame; they have changed which end of it they are standing at. The instrument is still measuring them against other people rather than against anything they chose for themselves.
The correction step in the research is not “compare downward instead.” It is “check whether this is a dimension you were competing on.” Those are different operations, and only the second one produces an answer you can do something with.
The variable is not you. It is what you are given to compare against
If the comparison itself is automatic, willpower is the wrong place to apply pressure. The supply is the right place.
Festinger assumed you compared yourself with the people around you: colleagues, neighbours, a few dozen lives roughly like your own and seen whole, bad mornings included. A feed supplies something categorically different. An effectively unlimited number of people, filtered for being exceptional, edited down to the parts that went well.
Research on social media and self-esteem keeps landing on the same distinction: it is the frequency of comparison that tracks poorer self-evaluations, not the time spent. Not the hours. How many upward comparisons those hours happen to contain.
You cannot change what your attention does with what it is shown. You can change what it is shown.
- Stop trying not to compare. It is automatic, you will fail, and the failure will feel like a character flaw instead of what it is. Aim at the second step, the one you actually control.
- Change what is in front of you. Comparison uses whatever is available. An hour of a feed selected for exceptional outcomes supplies far more upward comparisons than an hour of almost anything else, and that supply is editable.
- Name the dimension. The automatic step does not check whether this is something you care about. “Am I actually trying to be that?” is the correction, and asking it deliberately is quicker than waiting for it to arrive on its own.
- Expect the hit to be emotional, and short. The strongest link in the data is with how you feel immediately afterwards, not with your self-esteem. Treating a bad ten minutes as evidence about your life is the mistake the numbers do not support.
- Watch when you do it, not just how much. The correction needs resources. Late, tired and alone is exactly when comparison turns up without the adjustment that normally follows it.
None of that is abstinence, and none of it asks you to feel differently about anybody’s holiday photographs. It is about how many comparisons your day contains, and how quickly the correction arrives.
Try this
For a week, when you notice the feeling — not the comparison, the feeling that arrives just after it — finish this sentence out loud or in a note: “I am comparing my ___ to their ___.”
Filling in both blanks does two things. It forces the dimension into view, which is exactly what the correction step needs. And it usually exposes that the two terms are not the same kind of thing at all: your ordinary Tuesday against their best week, your interior against their exterior.
The point is not to talk yourself out of the feeling. It is to make the correction deliberate instead of leaving it to depend on how tired you are.
The parts of this that are genuinely hard
The two-way problem is not a footnote. Depression involves a documented pull towards exactly this behaviour, so somebody in a low period will compare more, compare upward, and feel it more sharply. For them the supply-side advice is real but nowhere near sufficient, and the honest thing to say is that this is a symptom as much as a cause.
There is also a limit to what a link of that size can tell any one person. It describes a pattern across 36,583 people; it does not predict your Tuesday afternoon. And most of it was measured by asking people to report their comparing and their distress on the same questionnaire, in the same sitting — a design that reliably tightens the apparent connection between two things somebody happens to be feeling at once.
The hardest part is not psychological. Comparison is the only instrument available for judging the things that have no objective measure, which means it is doing real work even when it hurts. The aim was never to switch it off. It is to stop feeding it a sample that was selected for being unrepresentative.
The bottom line
The advice has been failing for as long as it has been given, and the reason is now reasonably clear: it aims at the step that runs on its own.
What is left is smaller and more useful. The injury the evidence actually finds is immediate and emotional rather than a slow erosion of how you value yourself, which means the thing to manage is a bad ten minutes and not your character. And the variable genuinely in your hands was never the comparing. It is how many exceptional lives your day puts in front of you.
Sources (4)
Every factual claim in this article comes from one of these sources.
Keep exploring
Continue reading
Deeper into what you just read.
Explore more
Adjacent ground, if this was useful.
One thing worth knowing
A surprising finding. The evidence behind it. And what it means for your life. BetterYou+ sends one useful idea at a time—nothing to keep up with.