Ben Weidmann and Luke Miratrix have published "Lurking inferential monsters? Quantifying selection bias in evaluations of school programs." Ben has made some great strides in understanding how well we might use matching to conduct observational studies on national student datasets for interventions taken up at the school level. In this work, he conducts a within study comparison, seeing if he can replicate a series of actual RCTs using observational study methods. He finds that he can, in this context. Along the way he clarifies a lot of aspects of what a within study comparison actually is, and provides some nice statistical tools for doing this sort of thing more generally.
Congratulations to Ben and Luke!