5 Comments
Sep 21, 2022Liked by Emil O. W. Kirkegaard

Would be hesitant to give too much weight to the GCSE data. While I have no doubt there is some causal benefit of intelligence to GCSE performance, these exams are generally not that hard and the people who get the best grades tend to be the ones who have parents who make them study a lot (hence why many black Africans and Chinese Asians are doing so well). In short, they have very high returns to effort and parental engagement.

I suspect A-level (UK 16-18 exams) results would be a much better proxy for intelligence, as these are often the hardest exams all students in the UK ever take and have much lower returns to effort. There is a bit more noise in the sense that some students choose easier subjects (e.g music or geography) to inflate their grades, so there's probably a clearer signal in just looking at STEM grades. Just don't use the data from the last three years, as there has been absurd amount of grade inflation post-Covid

Expand full comment
author

Yeah everything before 2020 suggests I am probably right, although I suspect the differences are even more drastic if you break it down by subjects taken.

Many black and white British students steer clear of the harder A-level exams (like maths or physics) to get better grades and get admitted into more prestigious universities, whereas almost all of the Indian and Chinese kids will take at least one STEM subject and/or one of the harder non-STEM subjects like history. Furthermore ,some of the subjects got considerably easier post 2018. In math, for instance, you had to take around 6 exams in order to get your grade, whereas now you only need to take 3.

That being said, would be really interested to know how much of the variance can be explained by intelligence?

On the one hand, I think a lot of Chinese and Indian Asian kids generally overperform because they are forced to spend almost every waking minute preparing for their exams. On the other hand, a lot of the black kids I knew at A-level probably underperformed as they were just not engaged at all and not in home or school environments conducive to doing well in anything, let alone A-levels.

Also, I think the whole "wordcel Vs shape-rotator" thing comes into it somewhat, since I often saw kids who were much smarter than me underperform in STEM subjects. For instance, my girlfriend scored at the 99th percentile on the Mensa IQ tests (Culture Fair and Cattell B iii), but she could not do any better than a B grade at A-level mathematics despite relentless studying.

Expand full comment
author

You could try using FOIA to get the data by subject. People have gotten a surprising amount of otherwise hidden data this way, crime stats, intelligence test results by ethnicity etc. I had a look with a search query, and here's some related FOIAs:

https://www.whatdotheyknow.com/request/a_level_results_by_ethnicity_wit

https://www.whatdotheyknow.com/request/a_level_results_for_specific_eth#incoming-1514347

Expand full comment

I think African Americans are like 20% English and 5% indigenous on average so the gap between them and Caribbean blacks makes sense.

Expand full comment