Regression, matching, control function and instrumental variables methods for recovering the impact of education on individual earnings are reviewed for single treatment and sequential multiple treatments with and without heterogeneous returns. The sensitivity of the estimates once applied to a...