5 ms·
The author states, as regards the interpretation of the Dunning-Kruger diagrams, that [i]n two of the four cases, there’s an obvious positive correlation bet
by hf 11y ago
The author states, as regards the interpretation of the
Dunning-Kruger diagrams, that
[i]n two of the four cases, there’s an obvious positive correlation between
perceived skill and actual skill, which is the opposite of the pop-sci
conception of Dunning-Kruger.
In my corner of the universe, you don't get to cherry-pick which pieces
of data (ie "what instances of two sets of random variables") you bestow
the golden twig of correlation upon. If I'm not entirely mistaken,
correlation is very much a global feature, not a measure of proximity of
two points on a chart.
So, yes, Dunning-Kruger (as evinced from the diagrams sported here) indeed seems to make a weaker claim: that there's no
correlation between “perceived ability” and “actual ability”. As such,
this claim is as far from the "pop-sci conception" of Dunning-Kruger as
it is from the author's.
- maxerickson 11y agoYeah, the valuable conclusion to make from the Dunning-Kruger paper is that self assessment isn't worth a lot.
- mreiland 11y agoself assessment is worth plenty.
- Tenoke 11y agoErr, no. There is clearly a pattern to the data, and drawing the conclusion that self-assessment and actual skill are uncorrelated is not what you should do (the simplified pattern being that unskilled individuals overestimate their skill, while skilled individuals underestimate theirs, or a regression to the average if you will). At any rate, even if we don't take that into account self-assessment is worth plenty as demonstrated by many studies which manage to get coherent data from self-assessments. Sure, you should take them with a grain of salt, and you can expect biases, but no need to throw them out.
- maxerickson 11y agoI apologize for not writing an essay carefully specifying what "worth a lot" means to me.
- stdbrouw 11y agoThe referenced graphs measure performance and perceived ability on 4 different tasks. You're right that ideally you'd pool this data to get to an overall correlation, but for the point the author makes, eyeballing it and taking a mental average does the trick, no? Also, what corner of the universe are you from? Loess regression, hierarchical modeling, conditional analyses... methods for finding "non-global" correlations aplenty.