Dunning & Kruger (1999) [1] studies individuals' perceived and actual performances in humour, grammar, and logic. It observed that the gap between expectation and reality was inversely predicted by competence. The authors conclude that "people tend to hold overly favourable views of their abilities in many social and intellectual domains". C'est tout.

The implied mis-interpretation of Dunning & Kruger, that inability and confidence are correlated, is as commonplace as it is untrue. The Dunning & Kruger cohort which estimated its humour at the top quartile did perform in the top-quartile. And the cohort which estimated its humour in the bottom quartile did, too, perform the worst. It's just that the top quartile estimated itself at 75 when it came out at 90 while the bottom estimated itself at 60 when it performed at 10. We are good at relative evaluation. What we suck at is pinning that to an absolute scale.

"Several of my teachers [said] I’d gotten far too arrogant...this caused me to once again return to my previous mode of thinking. I wasn't actually a good programmer."

One can be a good programmer and arrogant. The former refers to technical competence. The latter to social competence.

[1] http://www.psoriasissociety.ttsg.org/pdfs/Dunning-Kruger%20E...