The linked study uses ~1200 very high IQ individuals (4 sigma out), so it has the same population of very high IQ individuals as you'd have in a random sample of tens of millions.
There's also been large studies that have used years of educational attainment as a moderately-correlated proxy for IQ.
There's also been large studies that have used years of educational attainment as a moderately-correlated proxy for IQ.