First let me agree with the major finding of all of these. The core of D-K is mostly debunked, there is only a small bias towards overconfidence in the less-skilled or less-knowledgeable which turns into a very small bias towards underconfidence in experts.
But I am going to quibble a bit with the following, especially the second part.
There also doesn't appear to be any valid rationale for extending the acceptable scoring criteria from their previous work (the "good" scores) through their new categories of "adequate" and "marginal" and "inadequate, but not extreme". However, this did allow them to make the headline claim that only 5.3% of those tested fell into the D-K category of unskilled and overconfident versus the nearly 30% that fall into this category if we just consider the novices and their original boundary of "good".
The other interesting thing from the paper was the statistically significant finding of the difference in self-estimation accuracy between men and women. Women slightly underestimate their abilities while men significantly overestimate theirs.
But I am going to quibble a bit with the following, especially the second part.
While the self-confidence errors do appear to be fairly evenly spread between overconfidence and underconfidence regardless of actual knowledge, the range definitely tightens as knowledge increases. Experts do a far better job of estimating their abilities than novices. The authors in #3 appear to be deliberately hiding this by lumping together all of the results, but by careful examination of the results from Figure 4, it appears that only around 44% of the "novices" (freshman and sophomores) had a "good" score on the self-estimation task compared to 74% of the professors.Instead, less skilled (or at least less-experienced) individuals tend to over AND under-estimate their competence level by a wider range than more experienced individuals, but most folks generally have a "good" or "adequate"(their terminology - see #3) self estimation with outliers (basically a Gaussian distribution).
There also doesn't appear to be any valid rationale for extending the acceptable scoring criteria from their previous work (the "good" scores) through their new categories of "adequate" and "marginal" and "inadequate, but not extreme". However, this did allow them to make the headline claim that only 5.3% of those tested fell into the D-K category of unskilled and overconfident versus the nearly 30% that fall into this category if we just consider the novices and their original boundary of "good".
The other interesting thing from the paper was the statistically significant finding of the difference in self-estimation accuracy between men and women. Women slightly underestimate their abilities while men significantly overestimate theirs.