And of course, inevitably you move on to pretending that the huge and stable variation in measured test performance within a population is of lesser import than relatively small and unstable difference in average test performance between populations. Even if we grant the rather silly proposition that tests so unstable that the Dutch population improved by more than a standard deviation in thirty years are actually a good measure of innate intellectual differences unaffected by non-genetic factors, you've got the problem that it's impossible to predict with any degree of certainty what the person who took the test's genetic background actually is. I mean why would anyone interested in investigating innate intelligence choose to focus on genetic markers that predict my skin tone pretty accurately but can't even rule out the possibility of me being in the very top or very bottom percentile for results in any cognitive ability test?
I'm going to go out on a limb and suggest that if a test indicates that me and my fellow white English thirtysomethings produce average scores a standard deviation ahead of (i) thirtysomething black people living in my country and (ii) my white grandparents' generation when they were my age, what's actually measured by the observed one standard deviation average differences across cohorts probably isn't genetic differences....
We can reliably identify haplogroups associated with certain phenotypes popularly categorised as "races", but we can also [more] reliably identify genetic markers associated with other phenotypical differences which have little or no correspondence with haplogroups. The presumption of greater significance of haplogroup-associated differences is the social construct here.