Purely-observational data

The data is purely-observational, and it is used to test the effect of treatment on the response. The data is sampled from a uniform distribution over all the subsets of N of size s, and the set of treated people contains a higher proportion from A than the set of untreated people.