Researchers at the University of Bath and The London School of Economics and Political Science (LSE) suggest that the celebrated Dunning-Kruger effect, often taken to mean the least competent people are the most overconfident, may actually be the wrong way round.

For decades, the Dunning-Kruger effect has been understood to mean that people with low ability, those who know the least, are the most overconfident, while high performers are more realistic. Hundreds of studies have replicated the classic finding, illustrating that people who do well on multiple-choice tests overpredict their performance by less than those who do badly.

Backed by this mountain of evidence, the Dunning-Kruger effect has become the internet’s ultimate shorthand for calling out clueless confidence—explaining everything from bad bosses to Twitter trolls with a simple ‘truth’: low performers are ‘unskilled and unaware of it.’

However, this new study, published in Psychological Review, reveals how the Dunning-Kruger effect is an artefact of flawed statistical analysis rather than human nature.

When this flawed statistical analysis is corrected, the researchers found that high-ability people tend to be the most overconfident – reversing the Dunning-Kruger claim, widely understood as “stupid people don’t know they’re stupid”.

Professor Chris Dawson, from the University of Bath’s School of Management, said: “Our research demonstrates that the original methodology fell victim to a statistical illusion. The popular image of the incompetent person brimming with confidence tells us less about human psychology than it does about the hidden traps in statistical analysis.

"When we clear away the statistical problems, the apparent pattern disappears. Instead, we see that overconfidence is a universal human trait, but it is the most capable among us who exhibit it the most."

Professor David de Meza from LSE’s Department of Management, said: “Test performance is naturally volatile- impacted by factors such as luck or bad question phrasing-and sorting participants purely by their test scores mathematically distorts the data to manufacture a false psychological phenomenon.

“When a test score is exceptionally low, it’s often because the person just got unlucky with the questions—meaning their estimated performance naturally ends up looking way too high. The exact opposite happens to top performers, who benefit from a lucky break.”

To address it, the researchers applied improved statistical techniques to data from thousands of participants across massive replication datasets. This approach manages to handle randomness in both people’s test scores and their predictions about those scores. Once a statistical model respects this reality, the classic story flips. It now turns out that the more able tend to be the most overconfident, the opposite of Dunning-Kruger.

It is plausible that poor performers may sometimes lack the cognitive ability to recognise their own incompetence. There are, though, other reasons why overconfidence arises. The researchers go on to show that it can result from attempts to signal otherwise hidden ability, a mechanism that high-ability individuals have the most incentive to engage in.

Talking the Talk, Not Walking the Walk: The Coevolution of Overconfidence and Loss Aversion is published in Psychological Review at: https://doi.org/10.1037/rev0000644.