It’s not nearly as clear as you make it out to be. Of the randomized controlled trials that I know of [1], one study has shown that reducing class size from 25 to 15 has a large positive effect, but another has shown that reducing class size from 40 to 30 has no effect. Also, assigning a teacher’s aide, so that a class of 50 is taught by two people also has no effect. The problem is that such studies are expensive and hard to generalize. Is it really that going from 25 to 15 always has an effect and that going from 40 to 30 never does, or was there something about the particular circumstances of those trials that made class size matter in one case and not the other?

Another problem is that even if reducing class size is effective is that, in the short run, there’s a fixed supply of good teachers. This isn’t a problem when you’re running an RCT on a few schools, but if you want to initiate state-wide (or nation-wide) reform, it’s not clear that throwing a bunch of unqualified teachers into small classrooms will help.

[1]Hacsi’s Children as Pawns: The Politics of Educational Reform has a good summary of evidence to date, as of 2003