What the data actually shows
The pivotal source is a comprehensive review by Baumeister, Campbell, Krueger and Vohs (2003), which examined the large literature on self-esteem and reached deflating conclusions. High self-esteem correlated with happiness and with reporting good things about oneself, but it did not reliably cause better academic or job performance, and the causal arrow more often ran the other way — success and good relationships tended to raise self-esteem rather than the reverse.
The same review found that efforts to boost self-esteem directly — a common feature of schools and self-help in that era — produced little benefit and occasionally backfired. Importantly, high self-esteem is also not uniformly benign: it spans both secure, well-grounded confidence and brittle, inflated, defensive forms, and the review cautioned against treating it as an unalloyed good to be maximised.
Against this, a body of research led by Kristin Neff makes the case for self-compassion as a healthier aim. Self-compassion — kindness toward yourself, recognising that suffering and failure are part of the shared human condition, and holding difficult feelings with balance — is associated with wellbeing comparable to or better than self-esteem, but without the same fragility, because it does not depend on outperforming others or on things going well.
Why this feels different from how it actually is
Self-esteem feels foundational because the movement that promoted it was enormous and culturally pervasive for decades, framing low self-esteem as the root of a wide range of problems. That framing was intuitive and hopeful, but the evidence that emerged was far more modest than the messaging, so the gap between cultural belief and data on this topic is unusually wide.
It also feels causal because of the direction we naturally read the correlation. We notice that people who are thriving often feel good about themselves and conclude the good feeling came first — but the review suggests it more often follows the thriving. Confusing a symptom for a cause is easy when the two reliably travel together.
And self-esteem's reliance on comparison is part of why it can feel so unstable. If your sense of worth is built on being competent, attractive, or successful relative to others, it rises and falls with each comparison and each setback — which is exactly the fragility self-compassion is meant to avoid by grounding worth in something other than coming out ahead.
What the research says to do about it
Pursue the things that actually raise self-esteem rather than trying to inflate the feeling directly. Since competence and good relationships tend to produce healthy self-regard as a byproduct, building real skills and investing in connection is more effective than affirmations aimed at the feeling itself.
Where the goal is wellbeing and resilience, the evidence points toward cultivating self-compassion. Neff's work and related research suggest that responding to your own failures with kindness, perspective, and a recognition of shared human imperfection is associated with lower anxiety and depression and steadier wellbeing — and, notably, does not reduce motivation, which is a common worry.
Practically, that means treating yourself in setbacks the way you would treat a friend in the same situation, rather than either harshly criticising yourself or scrambling to defend an inflated self-image. This is one of the better-supported alternatives to the self-esteem framing, and it does not depend on winning comparisons.
What the research says does not help
Generic self-esteem boosting — affirmations, indiscriminate praise, telling yourself you are great — has weak and short-lived effects and can backfire, especially for people who do not believe the message. The Baumeister review found little benefit from such efforts, and inflated praise can even undermine the people it is meant to help.
Building self-worth on being better than others is fragile by design, because it rises and falls with every comparison and collapses under failure. The very comparison-dependence that powers high self-esteem in good times is what makes it brittle in bad ones.
Treating low self-esteem as the explanation for most personal problems is not supported by the evidence and can misdirect effort. The data suggests self-esteem is more often an effect than a cause, so chasing it directly tends to be less productive than building competence, relationships, and self-compassion.
Real numbers in context
There is no single headline statistic here; the central finding is a direction of causation. The Baumeister, Campbell, Krueger and Vohs (2003) review concluded that high self-esteem is mostly a result of success and good relationships rather than a cause of them, does not reliably improve performance, and is not reliably raised — to any benefit — by programmes that target it directly. The honest number to carry is the size of the gap between what the self-esteem movement promised and what the evidence delivered.
On the alternative, self-compassion research (Neff and colleagues) consistently links it to wellbeing at least on par with self-esteem, with the added advantage that it is not contingent on outperforming others or on things going well. So the better-supported aim is less 'feel superior' and more 'be kind to yourself regardless of how you are performing' — a subtle but meaningful shift.