If treatment assignment is strongly ignorable, then adjustment for observed covariates is sufficient to produce consistent estimates of treatment effects in observational studies. A general approach ...