[gwern]

j15t gives you the money quote of the results in https://news.ycombinator.com/item?id=12025219

If you just want to know what percentage of variance in family SES is caused by the measured SNP differences, this study puts it at ~20%. This isn't too interesting since we already know that as most SNP GCTAs turn in results similar to that and there are several other studies establishing SES SNP heritability similar to that ( https://en.wikipedia.org/wiki/Genome-wide_complex_trait_anal... ).

This heritability doesn't tell you how the genetics is causing the higher/lower SES, though. It might be because they are increasing intelligence (most plausible); changing personality to more extroverted and Conscientious (possible); or increasing height (possible); or changing preferences to more rewarding fields like STEM (plausible given other results on subject area interest being genetically influenced); or maybe because they are yielding blue eyes/blond hair (highly unlikely). In this case, the researchers look at intelligence, since that is a trait closely linked to SES and also highly genetic.

So what you do is do the GCTA on both SES and intelligence, getting the usual 0.2-0.3 heritability estimates, then you look for overlap. This paper is interesting because it finds a great deal of overlap, indeed, almost total overlap - so you can interpret this as showing that at least part of SES is heritable and passed on in families, not because rich families are better at finding summer jobs for their kids or they leave their kids fortunes in their wills or live in better neighborhoods, but because they give their kids more intelligence genes and intelligence is important for success in modern society.