Causal Inference without Control Units

Randomized experiments are the gold standard for causal claims, yet randomization is not feasible or ethical for many questions in the social sciences. Researchers have thus devised methods that approximate experiments with observational data, usually via regression or matching. These techniques use nonexperimental control units to estimate counterfactuals. However, control units may be costly to obtain, incomparable to the treated units, or completely unavailable when all units are treated. We explore the possibility of making causal inferences without control units by using front-door and front-door difference-in-differences estimators that make use of post-treatment information. We illustrate the viability of this approach using multiple applications from the social sciences, including the evaluation of a job training program and the evaluation of an early voting program. Our statistical contributions have important implications for both experimental and observational research design and analysis.

Speaker Details

Konstantin Kashin is a PhD Candidate in the Department of Government and an affiliate of the Institute for Quantitative Social Science at Harvard University. His primary research interests include quantitative political methodology and computational social science, with a focus on text as data and causal inference. He is interested in applying these methods to the study of policy diffusion and interest group politics.

Date:
Speakers:
Konstantin Kashin
Affiliation:
Harvard University
    • Portrait of Jeff Running

      Jeff Running