>>14256269>Fourth, another sample (average IQ of 96) was considered unrepresentative by Lynn and Meisenberg because the children were from a “fee-paying school in Nigeria [which] is evidently an elite sample”. Two other samples (average IQs of 59 and 72) were also drawn from fee-paying schools (Conant et al., 1999, Sternberg et al., 2002), but these samples were not excluded by Lynn and Meisenberg.Fifth, Lynn and Meisenberg rejected a sample (average IQ of 91) in Ohuche and Ohuche (1973) as problematic in part because “the ages of the children are unknown”. At the same time, Lynn and Meisenberg included samples (average IQs between 63 and 72) of which age information is lacking (Dent, 1937, Fahmy, 1964, Fahrmeier, 1975, Fick, 1929, Nissen et al., 1935).
Sixth, Lynn and Meisenberg excluded data (Kashala, Elgen, Sommerfeldt, Tylleskar & Lundervold, 2005) from the Digit Span test (average IQ of 94), because it “correlates poorly with the Wechsler full scale IQ”. Similar problems were established with respect to the Koh's blocks test in an African sample (Vernon, 1969), but Lynn and Meisenberg did not see this as a reason to exclude data from this particular test (Dent, 1937; average IQ 68). Likewise, they excluded data in one sample (average IQ of 91), because of a weak “correlation between IQs [with] tests of English, math and social science in grades 4–7, showing IQs have no validity for these ages.” However, in numerous studies which they did include, test scores failed to correlate with other tests or criteria (Richter et al., 1989, Sternberg et al., 2002) average IQs 75 and 72, respectively). In fact, we documented a rather weak validity of the Raven's tests in African samples (Wicherts et al., 2009). However, despite its poor validity in many African settings, Lynn and Meisenberg state explicitly that the Raven's tests need to be considered to arrive at a good estimate of the average IQ of Africans.
