Modern Causal Inference: Experiments, Matching, and Beyond (Boulder, CO)
Innovations in the realm of causal inference may be counted among the most exciting methodological developments in the social sciences over the past 25 years. But mastering the state of the art of causal research is not easy because the field has a language all to its own. Participants will learn the causal language of the potential outcomes framework, which dominates contemporary causal inference. Having learned this language, the course then walks participants through key approaches, starting with randomized experiments, continuing with selection on observables, and finishing with selection on unobservables. In the process, participants learn about - and actively work with - exact matching, propensity score matching, fixed effects panel designs, difference-in-differences, synthetic control, instrumental variables, and regression discontinuity designs.
The course will use the R statistical computing environment for computation.
Prerequisites: Participants should have knowledge of multiple regression using linear algebra and some familiarity with limited dependent variables.
Fee: Members = $1700; Non-members = $3200