There is a lot of extra rigor in a random controlled study than an observational study because it is harder to "cook up data" to support the scientists own beliefs. This may even happen unconsciously.
I believe the point was a meta one: when arguing in favor of studies with randomized controls, do you make the argument by setting up a randomized controlled study to test the superiority of randomized controlled studies? If the only things worth paying attention to are the results of randomized controlled studies, then that's the only possible way you could consistently argue in favor of them.
Instead, it seems more common to establish the "ground rules" of science by some other means, such as arguments about the pros/cons of various scientific methodologies, or even what counts as "scientific methodology" at all (and at this point, you're doing philosophy of science).
