From Nassim Taleb
From Stephen Wolfram
Statistics
Society
Disqualified
-
Dunning–Kruger effect: a case of autocorrelation.
If people had random judgments of their performance, plotting the difference between these and their actual performance would still surface the reported effect, simply due to regression to the mean.
In a way the concept is still valid, but its only conclusion should be that people tend towards the mean when judging their performance.
Most of all, this paper is a lesson in poor graph design and the power of memes and herd mentality.