Trap/Cognitive Bias/No. 0295
Dunning-Kruger Effect
The Dunning-Kruger effect is a pattern in which low scorers on some tasks overestimate their performance more than high scorers do. Described by Justin Kruger and David Dunning in 1999, it may reflect limits in self-monitoring, statistical regression and measurement effects.
- Evidence
- Mixed evidence
- Read
- 6 min
- Links
- 8 connections
01You've seen this when…
- in life
You expect to place near the top on a language practice test. Your score lands near the bottom, including errors in sentences you felt sure were correct.
- at work
A new interviewer rates their hiring judgment highly. In a scored interview exercise, they miss several job-relevant clues and mistake fluent answers for evidence of skill.
- out in the world
At a neighborhood emergency-preparedness workshop, a volunteer rates their knowledge highly before a quiz. Their missed answers include situations they hadn’t realized were dangerous.
02The idea
The answer key reveals your mistakes. It also shows that your way of judging whether you were right was unreliable.
The Dunning-Kruger effect describes a pattern found on some tasks: people with lower measured performance show larger gaps between their performance and their self-assessment. Someone near the bottom of a test’s rankings may think they performed above average.
One proposed explanation is a double difficulty. The knowledge needed to answer correctly can also be needed to recognize an incorrect answer. Without it, you may lack both the skill and the means to check your skill. That checking ability is part of metacognition: monitoring your own thinking.
The familiar graph by itself leaves the cause unresolved. Noisy tests, ranking scales and people’s tendency to place themselves above average can also create the pattern.
The key distinction is larger error, not necessarily greater confidence. A low scorer can rate themselves below a high scorer and still overestimate by much more. The effect does not mean that every beginner feels brilliant or every expert feels uncertain.
03Why it happens
- Missing skills can hide missing skills. If you don’t understand what makes an argument valid, an invalid argument may look convincing. Feedback becomes more useful when you learn the distinction you previously couldn’t see.
- People often begin with a favorable guess. When evidence is thin, people may assume they are reasonably capable. Illusory superiority, the tendency to rate yourself above others, can leave low performers with especially large errors without requiring a special deficit in self-awareness.
- Test scores contain noise. A low score can reflect both limited skill and an unusually bad performance. An estimate closer to someone’s usual performance will then look too high. Sorting people by noisy scores can produce a pattern related to regression to the mean, even without differences in their ability to monitor themselves.
- Rankings leave unequal room for error. Someone at the 10th percentile has much more room to overestimate than to underestimate. Someone at the 90th has the reverse. Estimates clustered toward the middle will therefore produce upward errors at the bottom and downward errors at the top.
These explanations can operate together. The research question is how much each contributes on a particular task, not whether one explanation must account for every case.
04A worked example
In Justin Kruger and David Dunning’s original experiments, college students completed tasks and then assessed their performance. The students judged humor on some tasks and applied logical reasoning or grammar skills on others. The paper’s summary across four studies reported that participants in the bottom quarter scored around the 12th percentile but estimated themselves around the 62nd.
What it looks like The weakest performers believe they are better than most of their peers, despite results placing them well below most.
What’s actually going on There is a large mismatch between measured rank and estimated rank. That establishes poor self-assessment against the test while leaving the cause unresolved. The authors also investigated recognition of good performance and the effects of training to examine their proposed skill-based explanation.
What would have helped Learners could use clear standards to evaluate their answers and explanations of mistakes to guide practice before taking another assessment. Simply telling someone their rank gives them a result; teaching them why their answers fail gives them a way to judge future answers. For researchers, repeated assessments and checks for measurement error help distinguish limited insight from the statistical effects of grouping people by scores.
05How to spot it
Treat these as reasons to check your judgment. Assessing someone else’s self-assessment requires evidence beyond a conversation:
Evaluating a confident speaker requires evidence of both performance and self-assessment. The same requirement applies to beginners and people who disagree with you.
06What to do instead
- Predict something measurable. Before a task, estimate how many answers you’ll get right or which requirements your work will meet. Turning a general feeling of competence into a specific prediction makes calibration possible.
- Use a standard outside your own impression. Try a scored exercise, a clear rubric or an independent review. Prefer evidence of correctness to praise or reassurance.
- Learn what separates good work from bad work. Compare worked solutions and explain the difference. In deliberate practice, feedback that identifies an error you can work on gives you more guidance than a score alone.
- Check again on fresh tasks. After understanding yesterday’s answer key, test whether you can recognize today’s mistake. Compare predictions with results over several attempts.
- Keep confidence attached to a task. Skill at investing needs its own evidence, even when you’re good at spreadsheets. Intellectual humility means leaving room for correction while acknowledging the skills you have.
The goal is accurate confidence. Lowering confidence helps only when it brings confidence closer to performance.
07When it isn’t a Dunning-Kruger case
The Dunning-Kruger effect describes a specific pattern in measured performance and self-assessment. Calling arrogance, stupidity or bad judgment a Dunning-Kruger case requires evidence of that pattern. Overconfidence is broader: capable people can overestimate themselves too. A person can also perform poorly while knowing perfectly well that they need help.
The familiar confidence-versus-experience sketch—a beginner’s peak, a collapse, then a gradual recovery—depicts a person’s journey from novice to expert. The original studies compared participants with one another on particular tasks.
A test score is an imperfect measure of competence. An unfamiliar format, ambiguous questions or a narrow sample of skills can distort the comparison. Test self-assessment against a credible measure. When it disagrees with one score, check that score’s limits before deciding whether the self-assessment is wrong.
Using the effect to dismiss an opponent can recreate the problem: you become certain that your own judgment needs no checking.
08Roots
At Cornell in the late 1990s, psychologist David Dunning and graduate student Justin Kruger turned an everyday puzzle into a set of tests. How could people recognize their own poor performance if they lacked the knowledge needed to recognize a good answer? For one task, students rated jokes, and their judgments were compared with professional comedians’ ratings.
Their 1999 paper proposed that performing a task and evaluating performance could depend on overlapping skills. In a logic experiment, training helped low performers recognize limitations in their earlier performance. The argument was more specific than people flattering themselves: learning could improve the ability to notice what had previously gone wrong.
The name spread into classrooms, management advice and online arguments. Meanwhile, researchers challenged its interpretation. A 2002 critique emphasized above-average self-ratings and statistical regression; later work examined how similar graphs could arise from measurement alone. Follow-up experiments also supplied evidence for a role for self-monitoring. The result is a recognizable pattern with a continuing debate about what produces it—not a universal law about beginners.
09How solid is this?
The low-score/large-overestimate pattern recurs across studied tasks. Experiments support a role for skills and self-monitoring, but statistical and measurement effects can produce much of the same pattern. The claim that poorer performers uniquely lack insight remains disputed and task-dependent.
10Connections
- Often confused with Illusory Superiority, Overconfidence Effect
- Countered by Deliberate Practice, Intellectual Humility
- Part of Calibration, Metacognition
- See also Regression to the Mean, Measurement Error
11Origin and sources
Justin Kruger and David Dunning described the effect in 1999. Subsequent research debated the contributions of metacognitive limitations, statistical regression and measurement.
- [1]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.
- [2]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.
- [3]Ehrlinger, J., Johnson, K., Banner, M., Dunning, D., & Kruger, J. (2008). Why the unskilled are unaware: Further explorations of (absent) self-insight among the incompetent. Organizational Behavior and Human Decision Processes, 105(1), 98–121.
- [4]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.
Suggest an edit· Updated 2026-10-02