This study looks highly flawed. They gave ets tests to random (categorized) people in the three countries. So far so ok (although the question is always what does the test measure, I mean what e.g. recursion is not big in china, because it is not used in production etc). But for the US the take ets data instead of testing people for this study. So it seems they compare it with people who took this exam for other reasons (and probably studied/prepaeed for this exam!); or maybe I understand this wrongly