Universal treatments without clean controls, continuous policies without untreated groups, reversible reforms over time and statistical inference with handful of cases