And, alternatively, a reply to that post:

http://investinginkids.net/2014/02/28/grading-the-pre-k-evidence/

An important point that gets lost a lot is that test score gains from pre-K often fade as children age and get subsumed into the inferior educational system. However, and fascinatingly, there still seem to be advantages that show up in adulthood.

RCTs are very, very difficult to put into practice with live children in large numbers, as anyone with social science background knows. While it's certainly important to try one's best to get close to that, we won't get much of anywhere if we insist on pure RCTs for anything to ever matter with human subject research.