Do Positive Affirmations Work? Who They Help, Who They Hurt

Key takeaways

In the experiment, low-self-esteem students who repeated “I am a lovable person” felt worse than those who said nothing; high-self-esteem students felt slightly better, and the harm was more pronounced than the benefit.
The instruction did not replace the negative thought: low-self-esteem students were as likely to write that they were unlovable as that they were lovable.
It was the ban on doubt that did the damage. Told to consider ways the statement might be true and might not, low-self-esteem participants ended up better off.
Research “self-affirmation” means writing about a value you hold. Across 58 classroom studies it had a small effect under identity threat, essentially none otherwise; large-scale replications found it in one cohort and not the next.
For self-criticism, self-compassion interventions have better support across 27 randomised trials, though mostly against waiting lists rather than active alternatives, and only one was preregistered.

Sixty-eight students sat down to write whatever came into their heads for four minutes. Half of them had one extra instruction: every time they heard a sound like a doorbell, they were to repeat to themselves, I am a lovable person. The doorbell went every fifteen seconds. Sixteen repetitions.

Then everyone’s mood was measured, along with how they felt about themselves at that moment.

The students who already had high self-esteem felt slightly better for having said it. The students with low self-esteem — the ones the sentence was designed for — felt worse than the students with low self-esteem who had said nothing at all.

The sentence does not land the same way on everyone

That experiment is the middle study in a 2009 paper by Joanne Wood and colleagues, and its title says the finding out loud: Positive Self-Statements: Power for Some, Peril for Others.

They started by asking whether anyone actually does this. In a survey of 249 undergraduates, only three per cent said they never used positive self-statements. People also rated them as helpful — and the higher a person’s self-esteem, the more helpful they thought they were. Which is the first clue: the technique is most trusted by the people who need it least.

The experiment then tested it. And the useful part is not just the direction of the result but the size of the asymmetry.

Backwards

Repeating “I am a lovable person” made low-self-esteem participants feel worse than saying nothing, and high-self-esteem participants feel slightly better. The harm to the first group was more pronounced than the benefit to the second — the same on mood, on desire to do enjoyable things, and on how they felt about themselves.

The researchers also had two coders read what people had written during those four minutes. Among the high-self-esteem students, thoughts affirming their lovability outnumbered contradicting ones heavily. Among the low-self-esteem students, they were as likely to write that they were unlovable as that they were lovable.

So the instruction did not replace the negative thought. It summoned it, and then sat next to it.

What this study can and cannot say

These are small laboratory experiments on undergraduates — 68 in the second study, 116 in the third — measured immediately afterwards. They do not tell you what months of a daily affirmation practice do, and they are not a clinical trial.

What they do establish is that the effect of a positive self-statement depends on who is saying it, and that for one group the direction is negative rather than merely absent. A technique sold as universally harmless is not.

Why a true-sounding sentence can make things worse

The explanation the authors offer is not complicated, and once you have read it you cannot unsee it.

People defend their view of themselves, including an unflattering one. A statement that lands far outside what you currently believe about yourself does not get absorbed; it gets argued with. Say I am a lovable person to someone who is not sure they are, and the mind supplies the counter-evidence, because that is what minds do with claims they doubt.

The third study tested a sharper version of this. Instead of repeating the sentence, 116 people were asked to think about it — but half were told to focus only on the ways it was true, and half were told to consider ways it might be true and ways it might not.

For people low in self-esteem, being allowed to hold both sides produced better mood, higher state self-esteem, and more happiness with themselves than being told to focus only on the positive. On the last of those, the predicted scores were 4.98 with permission to doubt against 4.16 without it. For people high in self-esteem, it made no difference either way.

That is a strange and specific result. It was not the negative thoughts that did the damage. It was the instruction to have none — because when they arrived anyway, as they were always going to, their arrival became evidence.

What researchers call self-affirmation is a different practice entirely

Here is where most articles on this subject go wrong, and it is worth being precise, because the two things share a word and almost nothing else.

“Self-affirmation” in the research literature does not mean asserting a trait you doubt. It means spending ten or fifteen minutes writing about a value that genuinely matters to you — family, music, faith, fairness — and why. The point is not to argue that you are good at something. It is to stand somewhere stable while a threat to some other part of your self-image is happening.

That practice has real evidence, and it also has real limits, and both belong in the same paragraph.

A 2021 meta-analysis pooled 58 studies of values affirmation in actual classrooms. Among students under identity threat it found an effect on academic achievement of Hedges’ g = 0.15 — small. Among students not under that threat it found essentially nothing: g = 0.01.

And it does not travel reliably. When the original school intervention was replicated at scale by other teams, results varied a great deal: one large replication found the effect in one cohort of students and not in the next, and effects in university courses have appeared in one semester and vanished in the following one. The authors of that replication work were blunt that this implies practical limits on self-affirmation as a tool.

So the honest summary is narrow: a brief writing exercise about your values produces a small benefit for people in a specific situation — under threat, in a context that matters to them — and is unreliable outside it. That is a real finding. It is not the thing on the mirror.

What has better evidence for the people affirmations hurt

If the problem is that you cannot say a kind thing to yourself without your mind objecting, the research points somewhere other than a louder claim.

A meta-analysis of 27 randomised controlled trials of self-compassion interventions — which train something closer to how you would speak to a friend having a bad week than to a slogan — found moderate improvements across a range of outcomes.

Notice which group you are in. If positive self-statements make you feel steadier, they are working and there is no reason to stop. If saying them produces an internal rebuttal, that reaction is the documented result, not a personal failure of technique.
Stop banning the counter-thought. The measured harm came from the instruction to think only positively. “This is true some of the time, and not other times” outperformed “this is true” for exactly the people who struggle with it.
Aim at a value, not a trait. The exercise with evidence behind it is writing for ten minutes about something that matters to you and why — not asserting a quality you are not sure you have.
Make the statement one you can actually verify. “I answered that email” survives scrutiny. “I am a lovable person” invites it. A claim your mind cannot argue with cannot backfire.
If you are talking to yourself the way you would never talk to a friend, that is the target. Self-criticism is the thing with a measured intervention behind it, and it is a different job from manufacturing praise.

None of that is an argument for pessimism, and none of it says encouragement is bad for you. It says that a sentence has to get past the person hearing it, and that the people most often handed this advice are the people best equipped to refute it.

How strong is the self-compassion evidence

Across 27 randomised trials, self-compassion interventions produced moderate effects on self-criticism (g = 0.56), depression (g = 0.66), anxiety (g = 0.57) and stress (g = 0.67), and larger ones on rumination (g = 1.37).

The caveat is structural and the authors state it themselves: most of those trials compared the intervention against a waiting list rather than against another active treatment, and only one of the 27 was registered before data collection began. A waitlist comparison shows that something helped more than nothing. It does not show it beat the alternatives.

Set against each other, then: the practice with the weakest case is the one you have been told to do in a mirror, and the two with better support are both quieter — a piece of writing about what you care about, and a change in the register you use on yourself. Neither promises anything. Both are available this afternoon.

If you want to test the difference on yourself, the values exercise is the one that takes ten minutes and needs nothing.

Try this

Take ten minutes and write about one thing you genuinely value — not something you are good at, something you care about. Why it matters to you, and one specific time it did.

No claims about yourself. No adjectives you would have to defend. The exercise is not to feel better about who you are; it is to be reminded that you are standing on something while a difficult thing is going on.

If a critical thought shows up while you write, let it. The instruction to keep it out is the part the evidence says does the harm.

The parts of this that are genuinely hard

The uncomfortable implication is that you cannot pick your intervention by which one sounds most encouraging. The technique that felt most supportive to the researchers’ survey respondents was the one that measurably hurt a third of them, and there is no way to feel your way to that conclusion from the inside.

The evidence is also thinner than the confidence around it, on both sides of the argument. The Wood experiments are small, laboratory-based and immediate. The values-affirmation literature is large but heterogeneous, with a small pooled effect and replications that come and go. The self-compassion trials are mostly against waiting lists. Anyone telling you this is settled — in either direction — has stopped reading early.

And none of this substitutes for treatment. Persistent, grinding self-criticism that is affecting how you sleep, work or relate to people is not a technique problem, and the honest thing an article can do at that point is say so and stop.

The bottom line

Repeating something you do not believe is not a neutral act. For people whose self-esteem is already low, the evidence says it produces the counter-argument rather than the belief, and they end up further from the sentence than when they started.

The version with support behind it is quieter and less quotable: write about what you value rather than what you are, let the doubt be in the room instead of banning it, and treat how you speak to yourself as the thing to change — not the volume at which you insist.

Sources (5)

Every factual claim in this article comes from one of these sources.