Hello stata-users,
When using -cem- matching approach, I get 3 variables cem_strata, cem_matched, and cem_weights. The original stata paper doesn't mention anything about how to divide the dataset into treated vs untreated based on your chosen treatment variable.

If not, then I guess -cem- is only used for ATE on the population and doesn't quite help with how the two groups are different, correct?