Once, questioning the commonly-accepted claim of European-level Mongolian IQ, I went looking for some international test scores to get an anchor point for my expectations about the subject. I will report those in a later post. What I found was the 2007 TIMSS reports, and right before the section reporting Mongolian scores, there was a section reporting scores by percentile for most tested countries. Among these, something struck me: the scores for Yemen (GDP per capita in 2007: $1210 nominal currency) and Qatar (GDP per capita in 2007: $84000 nominal currency -neither of those are typos, BTW). Have a gander:

Screenshot (20) Screenshot (21) Data here (xlsx).

What surprised me most was that Russia’s and Japan’s science scores were almost exactly the same, with Russia having a larger standard deviation. I also find it curious that while Yemen’s scores in 4th grade math clearly converged with those of Qatar at higher percentiles of achievement, as did Russia’s with Japan’s, this was not the case for Yemen’s and Qatar’s science scores, with there being a near-constant gap at every percentile of achievement in this subject, with only the mildest convergence at the 95th percentile. My guess is this is because while 4th grade mathematical knowledge could easily be gained from a tutor using a few old books, this is not the case for scientific knowledge, which requires a constant updating of sources of educational information and an immersion into a modern environment. Indeed, it is precisely questions like “how are dogs and rabbits similar”, with the answer being that they are both mammals, that have the largest Flynn Effect, with the Flynn effect for Arithmetic (the most important part of Fourth-grade Math) being almost non-existent. This makes me suspect that, consistent with the qualifier in my previous post, there might be a real Flynn-based IQ gap between Yemen and Qatar.

Note that in every case, non-income factors are more important at explaining the Arab test score gap with Japan (per capita GDP in 2007: $34000 nominal currency), and even Russia (per capita GDP in 2007: $9100 nominal currency) than national income per capita.

Update: I now also notice that while the Qatar-Russia Math test score gap remains of roughly the same size across percentiles, this is not the case for the Qatar-Russia science test score gap, where the gap narrows dramatically in the higher percentiles of achievement. This is the reverse of the pattern presented by the Qatar-Yemen test score gaps (see above). It suggests science skills are, on average, have a steeper learning curve than math skills, while science scores are also more affected by national income than math scores. This is all consistent with science scores being more Flynnable than math scores, while math scores being more accurately indicative of innate intellectual ability across space and time.