Estimating causal effects of treatments in randomized and nonrandomized studies.
Donald B. Rubin
Journal of Educational Psychology · 1974
10.1037/h0037350