Are Confident People Actually More Competent?

What the data actually shows

The best-known evidence is Kruger and Dunning's 1999 paper, 'Unskilled and Unaware of It.' Across tests of humour, grammar and logic, people who scored in the lowest band tended to overestimate their performance substantially, while top performers were more accurate and sometimes underestimated themselves. The proposed explanation is elegant: the knowledge required to be good at something is often the same knowledge required to recognise that you are not — so the least skilled lack the very tools needed to see their own gaps.

Importantly, the effect has been actively debated and refined since. Some researchers argue that part of the original pattern can be produced by statistical artefacts such as regression to the mean and the simple fact that everyone, skilled or not, estimates imperfectly. The mainstream view is that a real relationship between low skill and limited self-insight exists, but its size and exact mechanism are more contested and more modest than the popular version of 'Dunning-Kruger' suggests.

The broader, less disputed point stands: self-assessments of ability are noisy and only weakly correlated with measured performance. Confidence is shaped by personality, culture, gender, mood and context as much as by skill, which is why it cannot be read straight off as a competence score.

Why this feels different from how it actually is

Confidence is loud and visible; competence is often quiet and only shows up in results over time. In a meeting, a presentation, or a first impression, the confident voice fills the room while genuine expertise — which tends to be more cautious and more aware of caveats — can sound hesitant by comparison. So we mistake the more audible signal for the more reliable one.

We are also primed to equate certainty with credibility. Someone who speaks without hedging feels more authoritative, even when the hedging would actually be the more accurate response to a genuinely uncertain question. This makes confident people persuasive in a way that is largely independent of whether they are right.

And in the comparison that drives self-doubt, you are matching your own visible uncertainty against other people's visible confidence — your inside against their outside. Because the truly skilled often carry more doubt, not less, you may be comparing your honest hesitation to someone else's untested self-assurance and concluding, wrongly, that they are ahead.

What the research says to do about it

The practical takeaway is to treat confidence as weak evidence and look for the stronger signals instead: track record, specificity, willingness to say 'I don't know,' and how a person updates when shown they are wrong. Calibration — how well someone's confidence matches their actual accuracy — is far more informative than the raw volume of their confidence.

Applied to yourself, the research suggests that some self-doubt, especially in areas where you are genuinely skilled, is normal and even a mild marker of insight rather than inadequacy. The discomfort of seeing your own limitations is the same discomfort that less skilled, more confident performers are spared.

Seeking concrete, external feedback is the most direct corrective, because self-assessment alone is unreliable in both directions. Objective measures and honest outside views close the gap between how good you think you are and how good you are far better than introspection can.

What the research says does not help

Using other people's confidence as your benchmark is the core mistake this research warns against. Assuming the most assured person in the room is the most competent — and that your own uncertainty proves you are behind — reads a weak signal as a strong one and reliably distorts the comparison.

Manufacturing unearned confidence to match them tends not to help and can mislead you about your own ability, which is precisely the failure mode the Dunning-Kruger work describes. Confidence that is not anchored to evidence does not become competence; it just becomes harder to correct.

Treating 'Dunning-Kruger' as a settled, precise law — or as an insult to throw at people who disagree with you — also misuses the research. The honest version is a debated, refined finding about imperfect self-insight, not a clean formula proving that confident people are stupid or that doubters are smart.

Real numbers in context

There is no clean single statistic that captures how loosely confidence tracks competence, and the precise magnitude of the Dunning-Kruger effect is genuinely contested — different analyses, correcting for measurement noise and regression to the mean, arrive at different and often smaller estimates. Any confident-sounding exact figure here should be treated with suspicion, which is itself fitting.

What the evidence supports is qualitative and robust: self-rated ability and measured ability correlate only modestly, the least skilled tend to have the least accurate self-assessments, and outward confidence is therefore a noisy proxy for skill. The useful conclusion is not a number but a habit — discount confidence, and weigh evidence of actual performance instead.