>>16779860
>>16779866
All IQ test issuers explicitly say that the tests are not valid for comparing groups. For example, when calculating the covariance between the tests and g (="fluid intelligence"), all adversity cases from genetical clinical retards to extreme poverty meth addicts, are excluded up-front, because the tests are calibrated only against "normal" cases.

When you put the "adversity cases" back in, which you do if you're comparing groups, e.g. africans against europeans, US whites against US blacks etc, then you have - even by IQ issuers own measure - a much much lower, and unknown, covariance.