Monday, August 15, 2022

Comparing an incomparable?


Stigmata Siciliana (1964), David McLure (Wikicommons)



What is the mean IQ of sub-Saharan Africans? There’s no clear answer. Current estimates come from an early stage of the Flynn effect and are also distorted by qualitative differences in cognition. Furthermore, mean IQ differs among African groups.




At present, there is little consensus on the mean IQ of sub-Saharan Africans. Estimates have ranged from a low of 66 to a high of 82 (Lynn 2010; Wicherts et al. 2010). Rindermann (2013) put forward a "best guess" of 75, which is inexplicably much lower than the estimated African American mean of 85. Yes, African Americans are about 20% European by ancestry, but that degree of admixture would not cause a 10-point difference. Malnutrition? That might depress IQ scores in some African countries but not most.


Noah Carl (2022) has reopened the debate by inferring mean IQ from harmonized test scores and GDP per capita. Sub-Saharan Africa looks somewhat better on the first measure and somewhat worse on the second. Both measures correlate roughly with mean IQ, but the correlation isn’t strong enough to tell us whether the mean is 62, 75, or 82. Moreover, the first measure suffers from the same problem that plagues IQ tests: Africa is just starting to experience the secular increase in mean IQ that the West experienced during the 20th century, i.e., the Flynn effect. By how much should we increase the estimate of mean African IQ to adjust for Africa being at an earlier stage of the Flynn effect?


As for the second measure, GDP per capita, the ability to create wealth is determined not only by cognitive ability but also by other mental traits: future time orientation (also known as time preference), willingness to follow rules and enforce them, feelings of guilt over breaking rules, reluctance to use violence to settle disputes, tendency toward individualism rather than nepotism and familialism, and so on.


In a reply to Carl’s article, Emil Kirkegaard (1922) infers mean IQ from the Social Progress Index. But that measure is no less problematic than GDP per capita. Social progress is driven by a basket of mental qualities, of which cognitive ability is only one. Emil himself makes that point:


One cannot just impute IQs reliably from non-IQ data in order to get some kind of unbiased estimates of a region's IQ because the regions themselves may under- or overperform on international rankings for whatever reason, [including] legacy of or current communism, nonWEIRDness, low individualism, or any other difference you can imagine.


Emil concludes: “There’s no avoiding having to collect more African IQ data.”


More data would be nice, but no amount of data will provide us with a mean African IQ that can be usefully compared with the mean IQs of other populations. There are several reasons:


·        Again, estimates of African IQ come from an early stage of the Flynn effect. They are not comparable with estimates of IQ that come from a later stage in other populations.

·        The genetic architecture of cognition is not the same. Sub-Saharan Africans seem to have alleles for cognitive ability that do not exist in other populations. To date, such alleles have been identified only in people of European descent.

·       Recent cognitive evolution, particularly in societies near the Niger, has created differences in mean cognitive ability among African groups. It is no more meaningful to talk about a single mean African IQ than it is to talk about a single mean European IQ.


Differences in the stage of the Flynn effect


IQ data from Western societies are not comparable with IQ data from African societies. The latter are just beginning to experience the rise in mean IQ that took place earlier in the West, specifically the increase of 13.8 points between 1932 and 1978 (Flynn 1984). The Flynn effect seems to be not so much an increase in cognitive ability as an increase in familiarity with the “test paradigm” at school and, more broadly, in society. Flynn (2013) situates the cause in the modernist paradigm: “We freed ourselves from fixation on the concrete and entered a world in which the mass of people began to use logic on abstractions and universalize their moral principles.”


Keep in mind that competitive exams began to appear in the West only in the late 19th century, first for entry into the civil service and then more generally for the educational system (Wikpedia 2022). Previously, people entered the civil service through patronage appointments, and education took the form of apprenticeship and imitation of role models. In those days, people were less inclined to formulate questions and look for the answers. The answers were already known, and you had to learn them. In fact, there was a stigma attached to asking too many questions, especially in rapid-fire succession.


Differences in the genetic architecture of cognition


As a means to estimate cognitive ability, the IQ test is becoming superseded by the educational polygenic score. This measure is based on SNPs that have been shown to be associated with educational attainment. Your polygenic score is higher to the extent that the alleles at those SNPs are associated with higher educational attainment. It is thus a measure of innate cognitive ability. At present, we have identified 1,271 SNPs that are associated with educational attainment and which, together, explain 11-13% of the variance in educational attainment among individuals (Lee et al. 2018). The educational polygenic score has shown good reliability in predicting the IQ of individuals and even better reliability in predicting the mean IQ of populations.


Again, we have identified alleles associated with educational attainment only in people of European descent. For this reason, the educational polygenic score is five times worse at predicting the cognitive ability of African Americans (Lasker et al. 2019). The loss of predictive power seems greatest in the domain of language ability, according to two studies:


·        Guo et al. (2019, p. 27) found that the educational polygenic score is ten to eighteen times worse at predicting the verbal ability of African Americans, in comparison to White, Asian, and Hispanic White Americans. They attributed this difference to the smaller size of the African American sample, to gene-environment interactions, and to “significantly less than full coverage of African genetic variants related to cognitive ability.”

·        With a sample of school-age African Americans, Rabinowitz et al. (2019) found that the educational polygenic score fails to predict performance on a standardized reading test but does predict pursuit of postsecondary education, getting a criminal record (only among boys), and performance on a standardized math test (only for one of the three cohorts).


When modern humans began to spread out of Africa some 60,000 years ago, those left behind began to pursue their own trajectory of cognitive evolution. The evolutionary change seems to have been greatest in the domain of language, i.e., the ability to express oneself in speech and writing. Polygenic scores cannot predict innate reading ability because too many of the relevant alleles are exclusive to the African gene pool and remain unidentified. Other relevant alleles may simply be more important or less important in other gene pools.


Although the educational polygenic score is based on alleles identified in Europeans, it can still be used for rough predictions of cognitive ability among people of African descent. Lasker et al. (2019, pp. 444-445) were able to increase its predictive power for African Americans by almost a factor of three, i.e., an increase from 20% to 54% of its predictive power for European Americans. They achieved this improvement by using alleles from a much smaller subset of SNPs that are less sensitive to decay of linkage disequilibrium.


Differences among African groups in the trajectory of cognitive evolution


Within the larger African trajectory of cognitive evolution, various African populations have pursued their own sub-trajectories. This has been especially true for populations in West Africa over the past millennium and a half. Their educational polygenic scores vary as you go from west to east, being lowest among the Mende (Sierra Leone) and progressively higher among Gambians, the Esan (Nigeria), and the Yoruba (Nigeria). The Yoruba have almost the same educational polygenic score as that of African Americans, who nonetheless are about 20% admixed with Europeans (Piffer 2021, see Figure 7).


Before European contact, West African societies were more complex in the north and the east, i.e., in the Sahel and the Nigerian forest. Those areas saw the creation of towns, the formation of states, and an increasing use of metallurgy and luxury goods from the fourth century onward. The increase in social complexity seems to have been driven by the development of trade along the Niger, which served as the main trading route between the coast and the interior (Frost 2022).


In West Africa, cognitive evolution seems to have gone the farthest among the Igbo of the Niger delta. We have no educational polygenic data on them, but their record of academic achievement in Nigeria, the UK, and elsewhere indicates an unusually high level of cognitive ability (Chisala 2015).




We should get more data, while recognizing the limits of what the data may tell us. IQ tests will always be problematic, and future research should focus on educational polygenic scores. In particular, we need to identify relevant alleles in non-European populations. Some of those alleles may be population-specific, and others may be universal but more important in some populations than in others. Finally, Africa is not a monolith. Different African populations have pursued different trajectories of cognitive evolution.





Carl, N. (2022). How useful are national IQs? Noah’s Newsletter, July 13.  


Chisala, C. (2015). The IQ gap is no longer a black and white issue. The Unz Review, June 25.   


Flynn, J.R. (1984). The mean IQ of Americans: Massive gains 1932–1978. Psychological Bulletin 95(1):29–51.  


Flynn, J.R. (2013). The “Flynn Effect” and Flynn’s paradox. Intelligence 41: 851-857.   


Frost, P. (2021). Polygenic scores and Black Americans. Evo and Proud, April 27.   


Frost, P. (2022). Recent cognitive evolution in West Africa: the Niger’s role. Evo and Proud, April 30.  


Guo, G., Lin, M.J., and K.M. Harris. (2019). Socioeconomic and Genomic Roots of Verbal Ability. bioRxiv, 544411.  


Kirkegaard, E.O.W. (2022). African IQs without African IQs: it’s complicated. Just Emil Kirkegaard Things. August 7.  


Lasker, J., B.J. Pesta, J.G.R. Fuerst, and E.O.W. Kirkegaard. (2019). Global ancestry and cognitive ability. Psych 1(1).  


Lee, J. J., Wedow, R., Okbay, A., Kong, E., Maghzian, O., Zacher, et al. (2018). Gene discovery and polygenic prediction from a genome-wide association study of educational attainment in 1.1 million individuals. Nature Genetics 50(8): 1112-1121.


Lynn, R. (2010). The average IQ of sub-Saharan Africans assessed by the Progressive Matrices: A reply to Wicherts, Dolan, Carlson & van der Maas. Learning and Individual Differences 20(3): 152-154.   


Piffer, D. (2021). Divergent selection on height and cognitive ability: evidence from Fst and polygenic scores. OpenPsych.     


Rabinowitz, J.A., S.I.C. Kuo, W. Felder, R.J. Musci, A. Bettencourt, K. Benke, ... and A. Kouzis. (2019). Associations between an educational attainment polygenic score with educational attainment in an African American sample. Genes, Brain and Behavior, e12558.   


Rindermann, H. (2013). African cognitive ability: Research, results, divergences and recommendations. Personality and Individual Differences 55: 229-233.   


Wicherts, J.M., C.V. Dolan, and H.L.J. van der Maas. (2010). A systematic literature review of the average IQ of sub-Saharan Africans. Intelligence 38: 1-20.   


Wikipedia. (2022). Imperial examination – Influence - West.  


Anonymous said...

As a means to estimate cognitive ability, the IQ test is becoming superseded by the educational polygenic score.


At present, we have identified 1,271 SNPs that are associated with educational attainment and which, together, explain 11-13% of the variance in educational attainment among individuals (Lee et al. 2018). The educational polygenic score has shown good reliability in predicting the IQ of individuals and even better reliability in predicting the mean IQ of populations.

Reliability is useless without validity. How much of the variance in cognitive ability does the educational polygenic score explain? How much of the variance in cognitive ability does the IQ score explain?

Anonymous said...

I am here before Santo starts spouting stupid shit like "IQ does not exist."

on another note dont take credit for what i say when instead the white man decides to exploit credit for what i do instead.

and yes we know africans are dumb as rocks but at least they outsmarted the malevolency of the white man did they not?

Anonymous said...

It will be very interesting with the Christianization of Sub-Saharan Africa and of the spread of Agrarian Civilization continent wide. If Clarkian Selection for Higher IQ would spread and become more effective with a larger population.

Santocool said...

''and yes we know africans are dumb as rocks''

Look yourself in the mirror.

Literally speaking, IQ does not exist.

Peter Frost said...


That would be a logical impossibility. You can validate a measure of cognitive ability (like the educational polygenic score) with respect to another measure of cognitive ability (like IQ) but not with respect to the actual object of your measurement (cognitive ability itself).


Africans are not as dumb as rocks. Please think before you write.


Literally speaking, words don't mean what we think they mean. Therefore, communication is impossible. And yet people communicate. Can you explain that paradox?

Santocool said...

Your name doesn't exist as you do. Words are symbols invented to be associated with "things", like legends on maps.

IQ belong to the realm of object-symbol languages, mathematics too.

People learn this very basic trick to communicate one another, but symbols are not equivalent to real things thought. They/we understand by association.

Primary function of words is not give a meaning of things but a particular or differentiable identity.

Anonymous said...

Even when full scale IQ scores are the same, there are differences in the specifics. Black IQ is weighted towards memory and coding tasks and depressed on perceptual tasks, and black verbal reasoning probably would be average if cultural differences were accounted for.

If there are differences in heritability, on what cognitive functions and to what extent, it's possible for the mean black IQ to be around 100 and for the average black person to have subpar abstract and spatial reasoning.