>Really tough problem to solve, but the data does not support the conclusion that the American maths education system is "behind" or "failing" as a whole.

Wait, what? That is exactly what it means. If the groups that are least likely to get external support (ie a tutor, prep, whatever) are under-performing that implies the _standalone_ system is broken. Sure, it is not so broken that it can't be patched with outside aid, but that is true of many things we consider "a failure."

This is akin to saying "Of course my car can get across the country. All I need is the occasional tow-truck, you know, for when it overheats."