To make causal inferences outside the experimental realm, researchers attempt to control for bias sources by postprocessing observational data. Finding the subset of data most conducive to unbiased or ...