6 Comments
Apr 14, 2022Liked by Emil O. W. Kirkegaard

A better way to estimate iq is to use percentiles: In a group of 100 random persons, how many of them are you smarter than?

Expand full comment

This, but also find benchmark tasks of each IQ level ala NALS and PIAAC (see previous posts).

Expand full comment

Jesus. How does this pass peer review? It took you, what, all of 2 minutes to see their flawed methodology?

Expand full comment
author

Peer review is largely random and with a left-leaning bias. This paper isn't political, so it comes down to just the random luck part I think. I know one of the authors, he's a nice guy, and has some other good work.

Expand full comment

That is a very bad error to report a result that is a function of some arbitrary feature of your transformation. You are anchored to average being 100 for IQ but it's not clear what should be regarded as "very high" or "very low." I suppose you could use increments anywhere between 1 and 10. Someone could reasonably see the far right box as the smartest people alive. I think that an IQ of 120 - 130 could arguably be regarded as "very high" too.

Expand full comment

Also the metropolitan vs national average as the anchor is also concerning.

Expand full comment