How do we "control" for a large number of confounding variables when analyzing causal effects in an observational setting, and have very little control over which individuals receive which treatment?