Define causal effects with potential outcomes and causal diagrams. Estimate effects using randomization, adjustment, matching, weighting, instrumental variables, regression discontinuity, and difference-in-differences when their assumptions are defensible.