Eppig claims that evidence from the U.S. backs up his international correlations:
Despite the strength of our findings, our study was not without its limitations. We did our best to control for the effects of education. But what we really needed was to repeat our analysis across regions within a single nation, preferably one with standardized, compulsory education. The nation we chose was the United States. Average IQ varies in the states. (Massachusetts, New Hampshire, and Vermont are at the high end, for example; California, Louisiana, and Mississippi are near the low end.) Again, infectious disease was an excellent predictor of average state IQ.
It is hard for me to believe that infectious disease is by far the most important cause of IQ variation. If this were so, I would expect to see significant numbers of very high IQs even in low-average-IQ regions (asymmetrically long high-end tails of IQ bell curves for these regions), because even in high-infectious-disease areas (I assume) the number of uninfected people is significant. In reality, the IQ bell curve is shifted to the left for low-IQ regions, but retains its symmetrical shape. Only if everyone in low-average-IQ regions had been infected with IQ-lowering disease does the infectious disease hypothesis predict that normal distribution curves will be shifted to the left without the shape being affected.
For example, Kenya has high rates of infectious disease, infant mortality, crippling accidents, and so forth. Yet, the Kenyans who avoid all that sometimes do extraordinarily well in Olympic running events.
Also, the hypothesis doesn't square with the fact that people have lower IQs in any region when their family origins and genetic pedigrees originate in low-IQ areas.