It's time to have more fun with some attempted responses to my paper attacking the hereditarian hypothesis! This time by the far right anthropologist Peter Frost. This may get technical so please feel free to ask for clarifications! https://twitter.com/frost61h/status/1382095112973484032?s=20

There are some minor things before the meat of the post. For example, the 1,271 DNA variants he mentions are not the ones that account for 11-13% of variation, rather that is when many thousands of variants are added together into a polygenic score

Additionally there's a case of the ecological fallacy in citing Piffer's work. The correlation observed at the population level is better explained as an artifact of polygenic score bias in non-European samples, like I explain in my article & has been argued by many researchers.

The first criticism is based on a misunderstanding on behalf of Frost. The method that used Fst is in fact a measure of selection. The idea is that if natural selection has acted on a trait, then DNA variants involved with the trait will be more different than those not involved

It is not simply about the magnitude of differentiation as measured by Fst but rather about whether trait associated SNPs are more different than control SNPs. The figure shows it well, the amount of genetic differentiation in education associated SNPs is not larger than controls

This method has been used in published work before https://www.nature.com/articles/s41467-018-04191-y. The categories from Wright are simply irrelevant, & are largely arbitrary to begin with as there is no real way to create a rule that applies to levels of genetic differentiation in all species.

Additionally, I don't think hereditarians are actually happy with a value like 0.11 even if it is "moderatelly differentiated" because it is roughly on par with Richard Lewontin's earlier work on genetic differentiation between races which has been constantly assailed.

Finally this simply isn't true and the paper is pretty clear. The Fst values are Fst between Europeans and Africans.

Next Frost discusses the other test for polygenic selection that compares evidence of selection from the biased between-family polygenic scores and the less biased within-family scores.

Here he engages in some circular reasoning. Less noise would only make group differences more easily identified if you assume they are there to begin with! They can just as easily better identify no group differences. Of course that possibility isn't considered.

While he is correct that smaller sample size may make the estimates less precise, the lack of bias makes them more accurate than between-family polygenic scores. The noise does make the test less powerful, but it is a more solid test than any performed by hereditarians.

Frost tries to object to the strong evidence that spurious associations & inflated effect sizes occur in GWAS studies & appear to be related to confounding from socioeconomic inequality, for example https://www.cell.com/ajhg/fulltext/S0002-9297(19)30231-9. These same processes can operate within & between groups

Then Frost again assumes that group differences should exist and therefore must be obscured for various reasons rather than assuming there are no group differences and more accurate methods can show that.

NB: Frost's claim about different genetic architecture is from misinterpreting the cause of lower polygenic score prediction in African Americans than Europeans. He incorrectly assumes PGS identify true causal associations & effects when it has been shown they are in fact biased.

Finally Frost's suggestions for better methods are odd considering methods like the ratio of synonymous & non-synonymous substitutions is not well suited for studies within species. They are better for comparisons at/above species level where there are more fixed substitutions

In fact the paper cited from Ohto is across the entire Mammalian phylogeny! The methods I used are among currently the best for cases of investigating polygenic selection within a species. The methods Frost suggests are good, but better suited to other questions.

Overall this criticism falls flat. Frost misunderstands core parts of the study and many arguments are circular, assuming the very conclusion that is being questioned in this search for genetic group differences.

Also Davide Piffer's paper trying to contradict my results have been "published" at an OpenPsych. The majority of the criticisms in this thread still stand, including the fact that when I correct for this mistakes the results are no longer significant. https://twitter.com/itsbirdemic/status/1250253881273135105?s=20