Say It to Yourself a Hundred TimesWhy affirmations are almost never what the studies actually looked at

You have probably tried it yourself. One sentence, in the morning, in front of the mirror: *I am enough. I can do this. I am a lovable person.* Perhaps it felt good. Perhaps it also felt hollow — as though you were telling something to someone who doesn't buy it.

Both experiences now have their own camp. One says: affirmations work, science has proved it. The other says: affirmations are toxic positivity, they only make things worse. Both cite studies.

I went and read those studies. The result is uncomfortable for both sides — and considerably more interesting than the argument.

The largest study of the subject contains no studies of the subject

In October 2025 the *American Psychologist* published the most extensive review of affirmations to date: 129 individual tests from 67 journal articles, 17,748 participants in total. The press reported: affirmations work.

If you look up which exercises are actually in there, the breakdown reads like this: 114 tests of a writing exercise about personal values. 15 of an exercise recalling one's own kind actions. And zero on repeating positive statements about oneself.

Zero. In the largest review of affirmations, the thing that virtually everyone means by an affirmation does not appear a single time.

Two different things share one name

This is neither a coincidence nor a scandal. It is a confusion of terms that has been bedding in for forty years.

• What you know: repeating a positive statement about yourself. “I am a lovable person.” The sentence is about you.
• What was studied: picking, from a list, the value that matters most to you personally — family, music, honesty, craftsmanship, faith — and writing for ten to fifteen minutes about why it matters to you. Not one sentence in it is about yourself.

The second approach goes back to the social psychologist Claude Steele, who described it in 1988. His reasoning: people want to see themselves as competent, decent and capable of acting. When that picture is dented in one place, you do not have to repair it in that very place. It is enough to widen it somewhere else — the sore spot then takes up proportionally less room.

That is almost the opposite of the mirror exercise. One goes straight for the place that hurts. The other explicitly does not.

The study the critics rely on

Anyone reading online that affirmations might do harm ends up at the same paper almost every time: Joanne Wood and colleagues, 2009, *Psychological Science*. The setup is odd enough to be worth knowing.

Sixty-eight students wrote down whatever went through their minds for four minutes. Half of them heard a doorbell chime every fifteen seconds — the signal to tell themselves inwardly, “I am a lovable person.” Sixteen repetitions. Then their mood was measured.

The result: participants who already had low self-esteem felt worse afterwards than those who had said nothing at all. Among those with high self-esteem, mood rose slightly. The finding travelled the world, usually in short form: affirmations harm the very people who need them most.

Except: 68 people, and of those only the top and bottom third of the self-esteem distribution entered the analysis. The authors note in a footnote themselves that such an extreme-groups comparison can overestimate effects.

In 2020 another research group repeated the study — with roughly 225 and 237 people, more than three times as many. They found nothing. No difference in mood, none in self-esteem, none in goal attainment, and no interaction with trait self-esteem either. Neither benefit nor harm.

That is the honest position, and it is unspectacular: for the sentence in front of the mirror there is to this day no clean evidence of benefit — and none of harm either. Anyone writing “affirmations are proven to be harmful” is overselling exactly as much as the other side.

The finding both sides overlook

The same 2009 paper contains a third study, and I consider it the most useful of all. 116 people were again given the sentence “I am a lovable person.” One group was to focus exclusively on how it was true. The other was also allowed to consider how it was not true.

Among people with little self-confidence, the second group came off better. The sentence was not the problem. Banning the contradiction was.

Which is, incidentally, precisely the charge behind the phrase “toxic positivity” — except that here it comes from the study itself rather than from an Instagram post.

The other branch — and what became of it

That leaves the writing exercise about one's own values. For that there is a famous finding. In 2006 *Science* published a field study from an American middle school: seventh-graders completed an unremarkable fifteen-minute writing task in class, with teachers unaware of who was in which group. The grades of the African American students rose by 0.26 and 0.34 points respectively — arithmetically around 40 per cent of the achievement gap.

Those 40 per cent have appeared in countless articles ever since. What appears less often: it was one school with 119 African American children.

In 2017 the same exercise, with the same materials, was repeated across eleven schools in one Midwestern district. First cohort: 939 children, a small positive effect. Second cohort, same district, same materials, one year later: 1,170 children — and a slightly negative value. The difference between the two cohorts was statistically meaningful. In the end the researchers could rule out effects above a very small threshold.

In 2021 a meta-analysis then pooled everything available — co-authored by Geoffrey Cohen, one of the originators of the method. The result is remarkably sober: a small effect for people currently under pressure. And for everyone else, essentially zero.

“Any specific future study could plausibly yield a null effect.”— Wu, Spreckelsen & Cohen, Journal of Social Issues 2021

That is not a demolition of the method. It is something more important: you cannot predict for an individual case whether it will help. And yet that is exactly what every affirmation app promises.

What the research does not say

This is where it gets uncomfortable, but without this section the article would be dishonest — and would do exactly what I accused both camps of above.

• It does not say that affirmations are harmful. The only serious indication of that failed to survive a repetition three times its size.
• It does not say that values writing reliably helps. The effect is small, and the spread between studies is so wide that a null result for any new case is well within the expected range.
• It says nothing about dosage and duration. Of 129 tests, only 27 had any follow-up measurement at all, on average barely two weeks later. On “how often” and “for how long” there is simply no data.
• It says almost nothing about us. Practically all of this research comes from the United States. I found no German-language studies on it — and with this topic that is not a detail, as the next section shows.
• It says nothing about people in real crisis. The large meta-analysis explicitly excluded clinical groups. Anyone in a genuinely distressing state of mind will find no answer here — and should not be looking for one in a blog article.

And what about toxic positivity?

The term sounds like technical language, but it is not. It was popularised in 2019 by an American therapist through Instagram graphics; her book followed in 2022. To this day there is no agreed definition, no accepted measurement instrument, and not a single study testing whether having less of it improves anything. The first academic review that even attempts to order the concept appears in 2026 — seventeen years after it was coined.

That does not mean there is nothing behind the feeling. It only means that what is documented goes by a different name.

In a study of 7,443 people across 40 countries, participants were asked how strongly they experienced societal pressure to be happy — and to keep negative feelings to themselves. The higher that pressure was felt to be, the lower their well-being. And the real punchline: the association was strongest in the countries where people are objectively best off.

A second finding fits with this: it is not one's own standard of being in good spirits that goes with feeling worse, but the assumption that others expect it. You then feel bad because you feel bad. It is that second layer that weighs.

And what helps against it? In one of the most careful studies on the question — over 1,300 people, a laboratory experiment, a diary phase and six months of follow-up — this emerged: people who allow their own difficult feelings rather than judging them report more stable well-being over months.

Two subtleties from the same work matter to me, because they are almost always left out. First: this does not mean accepting whatever happens. Accepting *situations* had nothing to do with the outcome — it is exclusively about how you handle your own inner life. Second: this stance lowered negative feelings but did not increase positive ones. Acceptance is not a recipe for happiness. It is a relief.

The other direction deserves the same honesty. The famous line “seeking happiness makes you unhappy” comes from two small studies with 59 and 70 female participants. In 2024 a Dutch longitudinal study of 8,331 adults over four years was published. There, people who valued happiness had higher well-being on average. And a comparison across four world regions found the feared disadvantage in the United States — in Germany, not at all.

So the charge of forced positivity is both justified and overstated. Justified where people are made to feel that their low mood is a failing. Overstated where it turns into the idea that one may no longer strive for anything good.

What remains — and what you can do with it

After all that clearing away, less remains than the self-help books promise, but more than nothing. And what remains is pleasingly concrete.

1. 01The sentence is not the tool. The moment is.By far the best-studied form of influencing yourself is not a declaration but an if-then: “If X happens, then I do or think Y.” More than 600 individual comparisons show that this form changes not only actions but explicitly also thoughts and feelings — the “then” part may be a sentence to yourself. What matters is the “if” part. Not: find the finest sentence. But: find out in which situation the old sentence actually turns up.
2. 02Write rather than speak. About a value rather than about yourself.The one variant with solid research behind it is the least known: ten minutes writing down why something matters to you. Not “I am valuable”, but “Why does reliability actually mean so much to me?” Self-worth comes in through the back door rather than by assertion. And yes: the effect is small and not guaranteed. But it has at least been studied.
3. 03When things get tight — not every morning.In the meta-analysis the exercise helped people who were currently under pressure. For everyone else the effect was essentially zero. Which makes the standard advice — every morning in front of the mirror — the application with the least going for it. After a setback, before a difficult encounter, on a day full of self-doubt: that is where it belongs.
4. 04Allow it not to be entirely true.The finding from Wood's third study is the most practical of the whole topic. Repeating a sentence against your own inner contradiction works against you. “Sometimes I am enough, and this morning it doesn't feel that way” is the sentence that stays standing — because you actually buy it.
5. 05Instruction rather than encouragement, path rather than goal.Three separate lines of research point the same way. In sport psychology, instructional self-talk (“this is how I do it”) beats motivational self-talk (“I can do this”). People who picture the process of studying before an exam study more than those who picture the good grade. And if-then plans work better than general resolutions. All three shift attention away from *what I want to have* and towards *what I do next*.

And the 66 days?

Anyone who read the last article will be asking: if a bottle of water with lunch takes around two months on average to become second nature — how long does a sentence in your head take?

There is no answer to that, and I am not going to invent one. Nobody has ever studied how long a newly introduced habit of thought takes. The 66 days come from studies of fruit, water and going for walks. Transferring them to thoughts would be exactly the mistake I accused the self-help literature of in the last article.

Two things can be said nonetheless. First, the time scale: if even a bottle of water takes two months, then a pattern of thought that has been running for twenty years does not give way in a fortnight. That is not bad news, it is a different expectation.

Second, the anchor. That thoughts can run automatically is measurable — there is a questionnaire for exactly that, and what it captures predicts how people feel nine months later. And if-then plans demonstrably change not only what people do but also what they think and feel. The number does not carry across the border. The fixed point in the day does. Why we notice the beginning of a pattern so poorly I have described elsewhere.

One last detail I came across while reading, which I find too honest to leave out. In the review of one of the few genuinely studied exercises, exactly one difference mattered in the end: whether someone worked through it alone or was guided. Guided, the effect was almost twice as large.

Not because the other person knows more. But because we rarely see such things through on our own — and because someone who isn't standing inside the habit notices the moment sooner than you do.

It is not the fine sentence that changes anything. It is the moment when the old sentence arrives — and what you have prepared for exactly that moment.

This article reflects the state of psychological research and is intended as everyday orientation. The coaching methods described support general wellbeing and holistic balance. They do not constitute a medical diagnosis, promise no cure, and do not replace medical or therapeutic treatment.