Impostor Syndrome or Dunning-Kruger: Two Edges of the Same Axis

Two notions circulate everywhere, and almost always against each other. On one side, impostor feelings, attributed to competent people who doubt themselves. On the other, the Dunning-Kruger effect, wheeled out to describe the self-assured colleague who masters nothing. The implicit conclusion is flattering and widely held: if I doubt myself, I'm probably worth something.

That conclusion is comfortable, and it is wrong — or at least, it goes far beyond what the data allow.

Because the two phenomena are not opposites. They are the same one, observed at the two ends of a single axis: the gap between actual level and estimated level. In one case the estimate is too low, in the other too high. The faulty skill is identical in both, and it isn't technical competence: it is the ability to evaluate your own output.

In the article that follows, I explore professional self-assessment and what it can and cannot teach us about ourselves. If you recognise yourself in this theme, the tests on the platform help you take stock.

What Kruger and Dunning actually published

In 1999, Justin Kruger and David Dunning published a study that became one of the most cited and most distorted in social psychology. They gave students tests of humour, logical reasoning and grammar, then asked them to estimate their ranking relative to others.

The main result fits in a sentence: the lowest performers markedly overestimated their ranking, while the highest performers slightly underestimated theirs.

The explanation the authors proposed is more interesting than the result itself: the skills needed to do a task well are largely the same as those needed to judge whether you have done it well. Someone who doesn't know the rules of grammar cannot spot their own mistakes — they lack precisely the instrument that would tell them they are making any. This is what is called a metacognitive deficit, and it is the paper's real contribution.

Note the second finding, systematically forgotten: the best performers underestimate themselves too. The phenomenon does not concern only the bottom of the distribution. The authors explain it by a projection bias: when something comes easily to you, you assume it comes easily to everyone, and you therefore place yourself around average.

Three things you read everywhere that are false

The graph does not exist. The roller-coaster curve — "peak of Mount Stupid", "valley of despair", "slope of enlightenment" — appears in no publication by Kruger and Dunning. It was invented afterwards, online, with axes one of which (confidence) was never measured in the original study, which was about estimating a ranking. It has become more famous than the research it claims to summarise. It is not a diagnosis, and it is nobody's name. "He's peak Dunning-Kruger" describes nothing measurable. The original effect is an aggregate statistical result across groups, not a label applicable to an individual — still less to a colleague you find irritating. The effect itself is contested. Several papers have shown that a substantial part of the observed pattern can be explained by statistical phenomena, notably regression to the mean and the fact that almost everyone rates themselves slightly above average regardless of any metacognitive deficit. The debate is not closed. What remains solid is the basic finding: self-assessment is an unreliable instrument, in both directions.

The two edges, side by side

Look at the two middle rows: they are exactly symmetrical. The attribution mechanism that protects impostor feelings is the same one that protects overconfidence, with the signs reversed. It is the engine common to both.

And above all: the bottom row is identical. What is missing on both sides is not lucidity, it is outside information.

Is doubt a sign of competence?

That is the question everyone asks, and the honest answer is: statistically often, individually never.

At scale, yes: more competent people tend to perceive the extent of what they haven't mastered, because they hold the map of the field. The more you know a subject, the more you see its deep water. Doubt therefore often rises with competence.

But a group-level result does not convert into an individual verdict. Doubt can equally come from a belittling environment, a period of exhaustion, a personal history, or an internal standard set far too high. None of those causes says anything about your actual level.

Concluding "I doubt, therefore I am competent" would be exactly the error we are trying to avoid: replacing one verdict about yourself with another verdict about yourself, on no better evidence. It is comfortable for a day. It does not survive the first setback.

The only workable way out

Since self-assessment is the faulty instrument, calibrating it more finely achieves nothing. What is needed is a different instrument.

In practice that means three things.

Ask for feedback about facts, not about people. "What was missing from that document?" produces usable information. "Do you think I'm doing all right?" produces reassurance, which relieves for two hours and teaches nothing. Look for decisions rather than compliments. A compliment is an opinion, reattributable to politeness within a second. A decision — giving you that file, that day, rather than someone else — commits the person who took it, and it has a date. Build a series, not a memory. Memory keeps only the peaks: the two striking failures and nothing in between. A dated record restores proportion, and it is often the only genuinely new element in the file.

That is the logic behind ScanMyJob: from written work exchanges, producing a dated record of facts that surfaces what contradicts your reading as much as what supports it. On this particular subject, that is the point that counts. Someone convinced they are merely tolerated in their role sometimes finds the opposite trace — difficult files systematically routed to them, their wording reused in committee, requests to arbitrate. And the reverse also happens: the record can confirm that the negative feedback is real, repeated and documented. An instrument that could only ever say one of those two things would not be an instrument. The gallery of examples shows four real cases, including some where the analysis contradicted the person's initial reading.

What this article does not do

It places you nowhere on this axis. That would be a verdict, and a verdict is precisely what is at issue here: nobody, and certainly no text, can estimate your level at a distance. If you want a structured reference point on impostor feelings, the platform's questionnaire draws on the Clance scale and is available here — it describes an experience, it measures no professional competence. It poses no diagnosis, and there is none to pose. The impostor experience, described by Clance and Imes in 1978, is neither an illness nor a disorder and appears in no medical classification. It does not replace professional advice. If this doubt weighs on your choices, your sleep or your health, it is something to work on with someone. And it leaves the question of the environment open. Doubt can be sustained by how a team operates — criteria never stated, feedback only ever corrective, public comparisons. In that case your metacognition is not what is at fault, and that is a different investigation.