Statistical criteria for a valid comparison group

[mariano-recurve]

So far we are computing t-test and KS-test on a set of features in order to determine if a treatment and control group "match". What other metrics should we be considering?

(per Ken Agnew's question)