Dunning Kruger effect done with.

The Dunning Kruger effect is a lovely finding, which seems to suggest that the cognitively incompetent are too incompetent to realise that they are incompetent. They over-estimate their abilities. Brighter people are more aware of their short-comings, and under-estimate their abilities.

Once the Dunning Kruger effect was announced, much of the mess of life made sense. The dull are full of passionate intensity, while the best lack all conviction.

However, for a few years now it has been clear that the finding was a bit too lovely. Most of the apparent effect was regression to the mean. Also, most of the samples were drawn from university students, who are supposedly bright, or at the very least an upwardly biased sample of ability, but the restriction of range called into question the generalisation of the finding. Even on better non-university samples there are still other statistical problems with getting people to judge themselves against an average. For example, older and more experienced drivers are better than younger less-experienced ones, and so it is natural for older drivers to think they are above average, and insurers agree with them.  It is the old problem about the mean and the median when applied to non-normal distributions.

To give them their due, in their original article Dunning and Kruger had wittily admitted that they might have made a mistake in finding the effect, and were too dull to have noticed it.

Now a new study has dug deeper into a broader set of data, and comes to the conclusion that once you make a correction for regression to the mean there is something there, but that something doesn’t amount to much. The effect is too small to account for anything.

Reevaluating the Dunning-Kruger effect: A response to and replication of Gignac and Zajenkowski (2020)

Curtis S.Dunkel, Joseph Nedelec, Dimitri van der Linden

Intelligence Volume 96, January–February 2023, 101717

https://www.sciencedirect.com/science/article/abs/pii/S0160289622000988

There is always a sense of loss when a useful explanatory finding has to be dropped because the data don’t support it. Of course, we should be above such sentimentality, but empiricism is a brutal calling. You have to take down an explanatory scaffolding you had come to rely upon, and sally forth in ignorance until some other theory finds evidence in its favour.

So, it seems that we must drop this notion, and move on.