... and ditch its poorer but apparently much cuter cousin, RCT. At least that is what I see as the takeaway from this article about external validity problems with RCT - which by the way is similar to the "how do you scale it up" and "how can we apply what you have found in [...]
[view whole blog post ]