Hypothetical potential outcomes for a tailored-treatment design based on the Chong, De La O, Karlan and Wantchekon (2011) municipal audit experiment. The content of treatment depends on the subject's type, so some cells of the schedule of potential outcomes are never observable.