As a Christmas gift to my readers, Santa Pumpkin is going to answer some questions his readers have been asking. I’ll start with Thinking Mouse, since he asked his question in the comment section:
What I want to figure out is how large of an percentage of all genomically high iq people are in the upper class, within the third world. Make that into a ratio to see the correlation between genomic iq and wealth within the third world. From that we could also guess the correlation between wealth and iq by comparing it to the west on the same traits.
The best we can do is try to estimate this correlation is the First World and assume it’s the same in the Third World.
The correlation between IQ and self-made lifetime income in the U.S. is about 0.48, and the correlation between IQ and inherited wealth would be about 0.22. So lumping the self-made rich and the trust-fund babies together, the correlation between wealth and IQ in the U.S. would be about 0.35.
Now to find the correlation between wealth and genomic IQ, we need to divide the correlation between IQ wealth by the correlation between IQ and genomic IQ. If one takes twin studies at face value, IQ has a heritability of about 0.75, implying a genotype-phenotype correlation of 0.87 (at least among the adopted).
But as Mug of Pee and Swank have noted, a major problem with twin studies all done in the same or similar countries is reaction norms. In other words, genome A may score higher than genome B in all U.S. environments, not because genome A causes higher IQ per se, but causes some other IQ enhancing behavior (i.e. eating brain food like fish), but if both genomes were reared in the Third World (where fish is too expensive) genome A might have a lower IQ phenotype than genome B.
One way to determine the actual causal correlation between IQ and DNA is to create a single test that is highly g loaded in the worldwide population and have an international genotyped sample of a million people take that test, and then have machine learning identify the genetic variants that best predict IQ among these culturally diverse people, holding the Human Development Index of each genome’s country constant.
Of course the genomic predictor the computer spits out will underestimate the true within-country causal correlation between IQ and DNA because perhaps only 75% of IQ is caused by common additive genetic variants while the remaining is caused by rare and/or non-additive genomic effects which are hard to identify. But mathematically adjusting for this fact might give a good estimate (assuming IQ has the same genomic architecture in all countries).