Three conditions under which experiments and observational studies produce comparable causal estimates: New findings from within-study comparisons

• Published in: Journal of policy analysis and management Vol. 27; no. 4; pp. 724 - 750
• Main Authors: Cook, Thomas D., Shadish, William R., Wong, Vivian C.
• Format: Journal Article
• Language: English
• Published: Hoboken Wiley Subscription Services, Inc., A Wiley Company 01.10.2008; Wiley Subscription Services, Inc; John Wiley & Sons, Inc; John Wiley & Sons, Ltd; Wiley Periodicals Inc
• Series: Journal of Policy Analysis and Management
• Subjects: Academic Achievement; Analytical estimating; Bias; Causal Models; Causality; Class Size; Comparative analysis; Context Effect; Estimating techniques; Estimation; Estimation bias; Experiments; Group dynamics; Job Training; Methodology (Data Collection); Natural experiments; Observational studies; Policy analysis; Pretests; Pretests Posttests; Public policy; Random sampling; Regression analysis; Research Design; Sample size; Selection bias; Standardized tests; Statistical analysis; Studies; Training
• ISSN: 0276-8739; 1520-6688
• DOI: 10.1002/pam.20375

This paper analyzes 12 recent within-study comparisons contrasting causal estimates from a randomized experiment with those from an observational study sharing the same treatment group. The aim is to test whether different causal estimates result when a counterfactual group is formed, either with or without random assignment, and when statistical adjustments for selection are made in the group from which random assignment is absent. We identify three studies comparing experiments and regression-discontinuity (RD) studies. They produce quite comparable causal estimates at points around the RD cutoff. We identify three other studies where the quasi-experiment involves careful intact group matching on the pretest. Despite the logical possibility of hidden bias in this instance, all three cases also reproduce their experimental estimates, especially if the match is geographically local. We then identify two studies where the treatment and nonrandomized comparison groups manifestly differ at pretest but where the selection process into treatment is completely or very plausibly known. Here too, experimental results are recreated. Two of the remaining studies result in correspondent experimental and nonexperimental results under some circumstances but not others, while two others produce different experimental and nonexperimental estimates, though in each case the observational study was poorly designed and analyzed. Such evidence is more promising than what was achieved in past within-study comparisons, most involving job training. Reasons for this difference are discussed.