This is hard to test. What we need here are studies that test different methods of teaching math on randomly selected people.

Of course people self-selecting to participate in the study would ruin it, and most people hate math after the experience and wouldn't participate unless paid large sums.

On the other hand, a study of highschool students who are forced to participate also isn't very useful because the fact of forcing students to study may well be the major reason why they find it a not fun experience and don't study well.