Your first two links as well as the "Teaching Math to the Talented" link are about high and low performing cohorts. The author is slicing the data along nationa

Your first two links as well as the "Teaching Math to the Talented" link are about high and low performing cohorts. The author is slicing the data along national origin lines, which includes the entire ethnic population, not a slice of students that crosses national origin boundaries but differentiates based on relative performance. You can define "demography" in multiple ways, as aptitude (or environmental factors thereof) or national origin and associated culture. Both are valid demographic distinctions. However, you can't compare population slices based on aptitude with population slices based on national origin. Comparing by national origin catches all percentiles of the population, while only comparing the top and bottom percentiles to each other misses the comparison made between the middle cohorts, which are most students. I did not see anything in that blog post which suggested ethnic or national IQ differences. The author was more concerned with learning cultures between national groups, and how those cultures are carried over to new educational regimes, and how those shared cultures can be used to compare different educational regimes.