Nice piece in ‘Science’ with the title: ‘No easy answers: What does it mean to ask whether a prekindergarten math program “works”?’ Geoff Norman, many years ago, used the term RCT in the context of medical education to stand for Randomised, Confounded and Trivial. Research into what works and what does not work in education is hard, and most studies (IMHO) fail to inform. Education isn’t a product like a drugs is, and gee it is hard to demonstrate when and where most drugs will work if you do not have an understanding of the biology and large effects to play with and outcomes that need to be measured over the long term.
I think about this a lot, but have no easy rules to guide action. Which is, of course, exactly the problem.