Much empirical research stands or falls on whether it provides a credible basis for causal inference: Does a change in the predictor variable x cause a change in an outcome variable y. This course introduces methods for credible causal inference, emphasizing intuition and hands-on work with real data. It follows Applied Econometrics I and II, but should be accessible to students with other reasonable prior methods training (e.g., non-Kellogg graduate students). Topics include: Rubin causal model (causal inference as missing data problem); inference in randomized experiments; difference-in-differences (including triple differences; event studies; synthetic controls); regression discontinuity; the logic behind instrumental variables (you will already know the math); matching, propensity score weighting, and subclassification methods for observational studies; assessing and achieving covariate balance; continuous treatments; causal inference with panel data; treatment effect heterogeneity and local treatment effects; when to use (and not use) regression methods; Rosenbaum and Manski bounds.
There are no sections matching your search criteria. Please search again with different criteria or contact the Kellogg School registrar for more information.