According to scientist Richard Lynn, on a scale where British and White Americans average IQ 100, black Africans average 67.  However a scientist J.M. Wicherts & his colleagues say the actual average IQ in black Africa is 80.  Wicherts writes:

We attribute the difference of roughly 10 IQ points between his estimate and ours to (1) our use of systematic methods and a lack thereof in Lynn’s work; (2) our use of weighting by sample size to estimate the mean IQ across samples and Lynn’s indifference to sample sizes; (3) our decision not to include unhealthy samples, which Lynn admitted; (4) our exclusion of samples in which test administration had met with problems, which Lynn attributes to low cognitive ability of test-takers; (5) our exclusion of data from the Coloured Progressive Matrices (CPM) for ages above 11 because the conversion from CPM scores to adult and adolescent norms for the Standard Progressive Matrices (SPM) artificially lowers the IQ; (6) exclusion of a number of high-IQ samples that he deemed unrepresentative; and (7) Lynn’s ad hoc downward correction of mean IQs from primary and secondary school students by two and six IQ points, respectively. Below we provide new estimates for these groups on

(1) Wicherts has a valid point that Lynn should use a systematic method for including and excluding studies, but his other criticisms are less tenable.

(2) I don’t think samples should be weighted by sample size because some of the biggest samples might be the least representative, so weighting by sample size just compounds the error.

(3) With some exceptions, Lynn is correct to include unhealthy samples because Africa is a region afflicted with poor health (malnutrition, AIDS, etc) so excluding unhealthy people largely excludes a huge percentage of Africans, making Wicherts samples unrepresentative.

(4) Wicherts is also wrong, in my humble opinion, to exclude samples where there were problems with test administration, because low IQ samples by definition are going to have more problems being administered tests, so excluding “problem testing sessions” systematically biases the numbers upward.

(5) Wicherts also excludes data from the CPM for kids older than 11.  This is one area I’m not informed on, but excluding CPM scores sounds like a bad idea if you want to measure a low IQ population, because the CPM was specifically designed to have a very low floor, thus making it possible for people to score very low.  By contrast the SPM has a high floor, causing low IQ people to score artificially high.

So even though Wicherts does a good job exposing the contradictions in Lynn’s research, he creates a bunch of new criteria that seems to bias the estimated IQ of Africans way up, and they are already biased up just by virtue of the fact that the most disadvantaged Africans are not in schools, urban areas, or professions, where they can be easily tested.  They are isolated in poor rural areas, often don’t attend school, don’t know how to hold a pencil, can’t read, and are tragically orphaned by AIDs.

Commenter Lion of the Judah-Sphere asks:

I’m not sure if I understand the point all of this discussion over African IQ samples. I would ask: do elite Africans (average IQ 100) perform as well as elite whites (average IQ 130)? If they complete, say, Calculus 1 and 2 in high school, and rigorous engineering degree programs in university at the same rate as elite whites, then we know for sure that African IQ data is bullshit, because obviously 100 IQ in Africa means something totally different than 100 IQ in Europe or America. But if 100 IQ Africans are struggling to pass pre-calc in high school and they’re completing “engineering tech” degrees that are being called engineering, then it’s likely 100 IQ in Africa = 100 in the West. Voila! Problem solved…

The debate between Wicherts and Lynn largely concerns what the average IQ is, but that’s separate from the question Judah-Sphere is asking, which is, are the tests valid for Africans.  I would say that while tests like the SPM and CPM are not as culture reduced as they appear, they do seem to be valid for comparing Africans and non-Africans with equivalent schooling.  For example, if you scroll down to figure 2 of this study by J.P. Rushton, you find that the regression line predicting grades for African and non-African engineering students is roughly parallel, suggesting the test predicts very similar academic outcomes in both populations (at least in STEM).