What is the Dunning-Kruger effect?
The Dunning-Kruger effect began as a finding about self-estimates on short tests. Researchers still disagree about how much of it is statistical, and the wider evidence on self-ratings of ability shows moderate insight.
Answer to the question
What is the Dunning-Kruger effect?
The Dunning-Kruger effect is the finding that people who score low on a test tend to estimate their performance well above their score. Justin Kruger and David Dunning reported it in 1999. Later studies dispute how much of the pattern is psychological and how much statistical, and the size of any remaining effect differs across samples.
Key takeaways
- Kruger and Dunning (1999): participants in the bottom quarter on tests of humour, grammar and logic put themselves at about the 62nd percentile when their scores sat at the 12th.
- Zell and Krizan (2014) combined 22 meta-analyses of self-evaluated ability and objective performance, including supervisor evaluations: the mean correlation was .29, described as moderate insight.
- Krueger and Mueller (2002) and Gignac and Zajenkowski (2020) argued that much of the familiar pattern is statistical, from regression to the mean and the better-than-average effect.
- Dunkel and colleagues (2023), using 13,977 Americans, found a statistically significant but very small effect, and Jansen and colleagues (2021) found support for low performers being less able to tell whether they were correct.
- Mazor and Fleming (2021) showed how one rational model can produce a group-level effect with identical insight at every skill level.
The Dunning-Kruger effect is the finding, first reported by Justin Kruger and David Dunning in 1999, that people who score low on a test tend to rate their own performance well above where it landed.
What did the 1999 studies measure?
Kruger and Dunning reported four studies in the Journal of Personality and Social Psychology, built on tests of humour, grammar and logic. Participants took a test, then estimated how well they had done and where they stood against other people. Those in the bottom quarter by score put themselves well above their true standing.
The authors tied the gap to metacognitive skill, the ability to tell accurate answers from mistakes. The Dunning-Kruger effect, as they framed it, comes from that shortfall: the knowledge needed to do a task well is much the same knowledge needed to judge it. Furnham, in his book The New Psychology, stresses the consequence that the claim concerns skill in a particular field, so the same person can sit near the bottom in one area and near the top in another.
The gap was wide. Bottom-quarter participants placed themselves at about the 62nd percentile when their scores sat at the 12th. In the fourth study a short training in logical reasoning improved both skill and the accuracy of participants' judgements of their own performance, which the authors took as support for their account.
Self-ratings and results in many fields
The 1999 participants were students on short tests. Working life enters the story of the Dunning-Kruger effect through a wider body of research. Ethan Zell and Zlatan Krizan gathered 22 meta-analyses in Perspectives on Psychological Science that compared self-evaluations of ability with objective performance, across academic ability, intelligence, language competence, medical skills, sports ability and vocational skills. Their performance measures included standardised tests, grades and supervisor evaluations.
In the Zell and Krizan data the correlations ran from .09 to .63, with a mean of .29, and the authors described the result as moderate insight into one's own abilities. The link was stronger for domain-specific self-evaluations and for tasks that were objective, familiar or low in complexity. Applied to a self-rating in an appraisal, that makes it an input with real information in it, strongest when the task is concrete and well known.
Across 22 meta-analyses, self-evaluations of ability tracked objective performance with a mean correlation of .29.
Difficulty adds a further point. Katherine Burson and colleagues ran twelve tasks across three studies in 2006. On moderately difficult tasks the best and worst performers judged their standing about equally well, and on hard tasks the best performers judged it less accurately than the worst. They proposed that perceived difficulty, rather than skill alone, drives how far people miss when they compare themselves with others. How skill is built in the first place is a separate question, taken up in Does Deliberate Practice Make an Expert, and a related misjudgement of one's own future work is covered in Why the Planning Fallacy Persists.
Two readings of one quartile chart
The familiar picture of the Dunning-Kruger effect is a chart of four groups ordered by score, each with its average self-estimate. One reading holds that much of that picture comes from the statistics of such charts. Joachim Krueger and Ross Mueller proposed it in 2002. In a replication they found no mediation by the metacognitive variables Kruger and Dunning had pointed to, and when either regression to the mean or the better-than-average effect was removed statistically, the errors were no longer asymmetric. They also reported that high performers were the more error prone, since they more often neglected their estimates of other people's performance.
Regression to the mean is the tendency of a group chosen for an extreme score on one noisy measure to land nearer the average on the next. The better-than-average effect is the habit of placing oneself above the typical person. Because most people do so, the lowest scorers have the most room to overestimate, and a quartile chart shows that room.
A 2020 paper in Intelligence by Gilles Gignac and Marcin Zajenkowski carried the argument into the method. In simulated data, the quartile method produced the familiar pattern even when none was built in. In a community sample of 929 adults, the quartile method showed the pattern, while two alternative tests, the Glejser test and quadratic regression, did not. They concluded that the effect is mostly, though not entirely, statistical in origin.
A national sample gave Curtis Dunkel and colleagues a different result in 2023. Across 13,977 Americans the quartile method again showed the pattern, and the alternative tests also found an effect, statistically significant but very small, with the quadratic term adding about 1 per cent to the variation explained. They pointed out that more than half of the earlier sample were undergraduates, which may have narrowed the range of intelligence, and concluded that it was premature to call the Dunning-Kruger effect mostly an artefact.
Rachel Jansen and colleagues approached the question from a model. In Nature Human Behaviour in 2021 they showed that the pattern could arise from prior beliefs about ability, from a link between performance and skill at telling whether each answer is correct, or from both. In two online replications of about 4,000 participants each, the comparison of two versions of the model favoured the second: low performers were less able to tell whether they were correct, in grammar and logical reasoning.
Is the Dunning-Kruger Effect Real sets the same papers out in date order.
Does the pattern reflect one burden or two?
The measurement was argued over as well. Avram Hiller wrote in Intelligence in 2023 that Gignac and Zajenkowski's finding of equal prediction error across levels of intelligence probably came from converting a comparative self-rating onto a linear scale, and he cautioned that ordinal ratings treated as interval data can mislead. Gignac and Zajenkowski replied that year that they had re-run the analysis with the transformation he suggested and again found no support for the hypothesis.
Matan Mazor and Stephen Fleming offered a way of holding both readings in view. They explained that regression to the mean can arise inside a single rational observer who pulls surprising results toward a prior expectation, so a model with identical insight at every skill level can still produce a group-level effect. A second model, in which weaker scores go with noisier self-estimates, fitted the online data better than shrinkage alone.
Standard models of decision-making, Mazor and Fleming added, predict more uncertainty in low performers without any extra metacognitive deficit, so the pattern may reflect one burden where Kruger and Dunning proposed two. They suggested trial-by-trial confidence data as the next test, since such data would show how certain people feel about each answer.
Common questions
What is the Dunning-Kruger effect?
The Dunning-Kruger effect is the finding that people who score low on a test tend to estimate their performance well above their score. Justin Kruger and David Dunning reported it in 1999, in studies of humour, grammar and logic. Later work disputes how much of the pattern is psychological and how much statistical.
Does the Dunning-Kruger effect mean low performers are the most confident?
Not in the studies reviewed here, which treat the Dunning-Kruger effect as a gap between estimate and score. Kruger and Dunning compared estimated and actual percentile by quarter of the score distribution. Burson and colleagues found that on hard tasks the best performers judged their standing less accurately than the worst, and Krueger and Mueller reported more error among high performers.
How well do people judge their own work performance?
Zell and Krizan combined 22 meta-analyses of self-evaluated ability and objective performance, including supervisor evaluations. The mean correlation was .29, which they described as moderate insight.
Sources
Furnham, A. (2024). The New Psychology. Seed book; vault hub Books/The New Psychology.
Kruger, J., & Dunning, D. (1999). “Unskilled and unaware of it: How difficulties in recognizing one’s own incompetence lead to inflated self-assessments.” Journal of Personality and Social Psychology, 77(6), 1121-1134. doi:10.1037/0022-3514.77.6.1121 https://doi.org/10.1037/0022-3514.77.6.1121
Zell, E., & Krizan, Z. (2014). “Do people have insight into their abilities? A metasynthesis.” Perspectives on Psychological Science, 9(2), 111-125. doi:10.1177/1745691613518075 https://doi.org/10.1177/1745691613518075
Burson, K. A., Larrick, R. P., & Klayman, J. (2006). “Skilled or unskilled, but still unaware of it: How perceptions of difficulty drive miscalibration in relative comparisons.” Journal of Personality and Social Psychology, 90(1), 60-77. doi:10.1037/0022-3514.90.1.60 https://doi.org/10.1037/0022-3514.90.1.60
Krueger, J., & Mueller, R. A. (2002). “Unskilled, unaware, or both? The better-than-average heuristic and statistical regression predict errors in estimates of own performance.” Journal of Personality and Social Psychology, 82(2), 180-188. doi:10.1037/0022-3514.82.2.180 https://doi.org/10.1037/0022-3514.82.2.180
Gignac, G. E., & Zajenkowski, M. (2020). “The Dunning-Kruger effect is (mostly) a statistical artefact: Valid approaches to testing the hypothesis with individual differences data.” Intelligence, 80, 101449. doi:10.1016/j.intell.2020.101449 https://doi.org/10.1016/j.intell.2020.101449
Dunkel, C. S., Nedelec, J., & van der Linden, D. (2023). “Reevaluating the Dunning-Kruger effect: A response to and replication of Gignac and Zajenkowski (2020).” Intelligence, 96, 101717. doi:10.1016/j.intell.2022.101717 https://doi.org/10.1016/j.intell.2022.101717
Jansen, R. A., Rafferty, A. N., & Griffiths, T. L. (2021). “A rational model of the Dunning-Kruger effect supports insensitivity to evidence in low performers.” Nature Human Behaviour, 5(6), 756-763. doi:10.1038/s41562-021-01057-0 https://doi.org/10.1038/s41562-021-01057-0
Hiller, A. (2023). “Comment on Gignac and Zajenkowski, The Dunning-Kruger effect is (mostly) a statistical artefact: Valid approaches to testing the hypothesis with individual differences data.” Intelligence, 97, 101732. doi:10.1016/j.intell.2023.101732 https://doi.org/10.1016/j.intell.2023.101732
Gignac, G. E., & Zajenkowski, M. (2023). “Still no Dunning-Kruger effect: A reply to Hiller.” Intelligence, 97, 101733. doi:10.1016/j.intell.2023.101733 https://doi.org/10.1016/j.intell.2023.101733
Mazor, M., & Fleming, S. M. (2021). “The Dunning-Kruger effect revisited.” Nature Human Behaviour, 5(6), 677-678. doi:10.1038/s41562-021-01101-z https://doi.org/10.1038/s41562-021-01101-z