Correlation is not causation. Counterfactual causal inference is undoubtedly one of the most important inventions in statistics and social science research methods. Based on the potential outcomes framework, this course presents the state-of-art of causal inference methods. In-depth topics include the concept of potential outcomes, experiments (covariate-balancing test and randomization test), matching, propensity score methods, sensitivity analysis, instrumental variables, regression discontinuity, difference-in-difference, synthetic control, and causal mediation analysis. Materials (recordings, slides, and lab code) for optional topics including repeated treatment (marginal structural models), nonbinary treatment (dose-response functions), and interference (contagion and spillover) may also be acquired for self-study. The seminar will provide a delicate balance of major methods and diverse applications as well as readily useful computer code. Knowledge of Stata (preferred) or R and logistic regression is required.
Day 1: Potential Outcomes, Experiments, Matching, and Propensity Score Methods
Day 2: Sensitivity Analysis, Instrumental Variables, and Regression Discontinuity
Day 3: Difference-in-Difference, Synthetic Control, and Causal Mediation Analysis
Optional Topics: Repeated Treatment, Nonbinary Treatment, and Interference
Dr. Weihua An is Associate Professor (tenured) of Sociology & Quantitative Theory and Methods and associated faculty of the East Asian Studies Program, the Goizueta Business School, and the Rollins School of Public Health at Emory University. He received a Ph.D. in Sociology and an A.M. in Statistics from Harvard University and was a doctoral fellow and a postdoc fellow at Harvard Kennedy School. His research advances theories and methods for network analysis and causal inference with applications to studying inequality and social policy (e.g., immigration, housing, education, and redistributive policy), health (e.g., peer effects, perceived risk, social network-based interventions, and pandemics), and organizations. He has published widely in both methodological and substantive journals and has also authored multiple statistical software including "fglsnet", "LARF", and "keyplayer" in R and "DIDMatch' in Stata, which have received over 130K downloads in total. He has served or is serving on the editorial boards of American Sociological Review, Journal of Machine Learning Research, Social Science Research, Sociological Methodology, and Sociological Methods and Research and has edited several special issues at top journals including the latest Methodological Advances in Quantitative Social Science (Social Science Research 50th Anniversary Series). He is a recipient of the Faculty Teaching Award from Emory Sociology and has advised over 20 dissertations and multiple honors theses. He is an instructor for the NIDA-funded program “Training in Advanced Data Analytics and Computational Sciences” at Emory and also has extensive experience in leading methods training programs for researchers and working professionals. Dr. An studied causal inference at Harvard with Prof. Guido Imbens (A 2021 Nobel Laureate in Economics), Prof. Donald Rubin (a founder of modern causal inference), and Prof. Christopher Winship (A co-author of the popular book Counterfactuals and Causal Inference: Methods and Principles for Social Research). He has researched causal inference for over 16 years and has published on many topics including Bayesian propensity score estimators, combining difference-in-difference with matching, assessing covariate importance through treatment effect deviation, analyzing outcomes with network dependence, instrumental variable methods, causal inference with treatment diffusion, causal network analysis, etc.
June 26-28 (8-10:30 pm, US ET), 2023 and on-demand. Zoom class with recordings available.