It's a very old story: But if compare students in Finland with students in the US with ancestors from Finland, then might have something. Similarly compare st

It's a very old story: But if compare students in Finland with students in the US with ancestors from Finland, then might have something. Similarly compare students in Nigeria with students in the US with ancestors from Nigeria. But do NOT compare all students in the US with all students in Finland. To heck with Nigeria and Finland. Instead, there is a general result: For some positive integer n of, say, a few dozen, take n relatively homogeneous populations. Give the tests. Take the best, say, 5 populations on that test. Now, for population n + 1, take a 'mixture' of the populations of the n other populations, give the test, and compare with the top 5. Then very likely population n + 1 loses. This is true just by a simple convexity argument that never mentions Finland, Nigeria, or the US but still does explain why the US has a hard time beating the best of the homogeneous countries. The NYT, the McKinsey study, etc. won't tell you any of this.