Dunning-Kruger Effect
Dunning–Kruger Effect
What is the Dunning-Kruger effect?
- A bias where low skill breeds overconfidence
- The cause: weak metacognition
- Supplement self-assessment with data and feedback
Once you understand the Dunning-Kruger effect, the first thing to fix in an organization is not people but how you handle self-assessment data. In areas where many respondents answer a training needs survey with 'We don’t need this course,' the real gap is sometimes the largest. And the leadership scores that managers give themselves right after promotion are usually higher than the scores they give six months later. So the order of diagnosis changes. Instead of running a survey first, show the behavioral indicators that define 'doing well' first and then collect self-assessments. Respondents then at least measure themselves against the same yardstick.
This bias attaches not to a person’s character but to the domain and its feedback structure. In work where results come back immediately as numbers, such as sales performance or production yield, overconfidence doesn’t last long. But in areas where feedback is vague and delayed, such as leadership, communication, planning, and interviewing, it can go uncorrected for years. It also helps to tell it apart from similar concepts. The better-than-average effect is a whole group rating itself above average, while the Dunning-Kruger effect stresses that the perception gap is especially wide in the lowest-performing group. At the opposite tail, competent people rating themselves low is closer to a false-consensus assumption that 'everyone can do this much,' so its cause differs from impostor syndrome, where people feel like frauds.
Where this concept most often goes wrong in practice is when it is used not as a diagnostic tool but as a label to put someone down. Once someone says in a meeting, 'He’s such a Dunning-Kruger case,' less experienced team members stop sharing not only their confidence but also their questions. Telling a newcomer 'now is not the time to speak up' based on the internet’s 'Mount Stupid' graph is the same kind of misuse. The graph is an adaptation that isn’t in the original paper, and it keeps early overconfidence from surfacing, which only delays the problem by six months. One more point: pointing out overconfidence alone does not reduce it. Without specific standards and repeated, measured feedback, criticism only breeds defensiveness and cynicism, and the people who most need correcting go quiet in review sessions.
Sangmi LeeSpecialist instructor in customer service, leadership, and communication
John KimInstructor John Kim, author of B2B Marketing Design and CEO of Nanum ConsultingCornell University psychologists David Dunning and Justin Kruger first reported it in their 1999 paper “Unskilled and Unaware of It: How Difficulties in Recognizing One's Own Incompetence Lead to Inflated Self-Assessments.” Through experiments on humor, grammar, and logic tasks, they showed that low performers greatly overestimate their own results.
Consider a hypothetical case. Company A, an industrial parts maker (480 employees, KRW 120 billion (≈ US$89,780,000) in annual revenue), appointed 32 new team leaders last year. In the first 360-degree review, the new leaders rated themselves 4.2 out of 5 on feedback and coaching, while their team members rated them 2.8, revealing a 1.4-point perception gap. The HR development team redesigned its six-month new team leader program to close that gap.
- Instead of starting with a survey, Kim, a manager on the HR development team, recorded real 1:1 conversations of six of the company’s top team leaders, distilled them into 12 behavioral indicators, and showed those to the 32 new team leaders first.
- When the same measure was rated again, the average self-rating fell from 4.2 to 3.5, and 19 of the 32 leaders said, 'What I was doing wasn’t feedback, it was giving orders.'
- The 8 team leads with the biggest gaps said, 'The team members must have misunderstood.' Instead of arguing back, the HR head read aloud 3 written responses from the 360-degree review, word for word, and the debate stopped on the spot.
- In a role-play of a performance-review meeting with an underperformer, an observer timed the talking: of 10 minutes, the team lead spoke for 7 minutes 48 seconds. One team lead said, 'I thought I was listening.'
- The 5 high-performing team leads who clustered on the underestimating side became in-house instructors. They were asked to put know-how they had never been able to explain into the language of 12 metrics.
- Six months later, the gap between self-ratings and team ratings had shrunk from 1.4 points to 0.4. Engagement scores in the new-manager teams rose from 62 to 74, and voluntary turnover fell from 11% to 6%.