What Dunning and Kruger Actually Found in 1999

In 1999, two Cornell psychologists ran a small experiment that produced one of the most misquoted findings in pop psychology. David Dunning and Justin Kruger gave undergraduates tests on grammar, logic, and humor, then asked each student to estimate how well they had done compared to peers.

The students in the bottom quartile averaged the 12th percentile on logic. They guessed they had landed near the 68th. The high scorers, meanwhile, slightly underestimated themselves — they assumed the test had been easy for everyone.

The paper's actual claim is narrow: the same skills you need to perform a task are the skills you need to evaluate your performance on it. If you don't know what good logic looks like, you can't tell that yours is bad. Competence and the ability to judge competence are the same muscle.

That is not the cartoon "Mount Stupid" graph that floats around the internet, with confident idiots towering over humble experts. The original chart shows poor performers overestimating themselves by a wide margin and top performers underestimating themselves by a small one. Everyone, on average, thought they were a little above average.

Later statisticians have argued the effect is partly a regression-to-the-mean artifact: people on the extremes tend to guess closer to the middle. Dunning has agreed the visualization is often abused but stands by the underlying point. Beginners don't know what they don't know — and that gap, not arrogance, is the whole story.