lo

James Lo is an Assistant Professor at the University of Southern California. He earned his PhD from UCLA, and his publications have appeared in the American Political Science Review, American Journal of Political Science, Political Analysis, and various other journals. His research interests include political behaviour and legislative voting.

Course Content
Causal inference is fundamentally about learning about counterfactuals from factuals – that is, given what we observe among units treated or not treated with a particular intervention, how would that unit how looked like if it had been assigned a different treatment status? In this course we explore methods for conducting causal inference in the social sciences.

Course Objectives
Students should expect to learn a number of causal inference designs – such as experiments, regression discontinuity, matching, and difference-in-difference – that they are able to deploy into their own work. In addition, a significant portion of the course is devoted to discussing real examples in which researchers have deployed such research designs to answer real research problems. This in in part designed to help students recognize opportunities to exploit such designs in their own research.

Course Prerequisites
This course does not require any additional knowledge of statistics beyond OLS. That said, we will derive some proofs, particularly in the early classes, that require some mathematics. In general, good mathematical intuition is extremely helpful. For computer skills, students should have solid knowledge of R, and should have some familiarity with how to program Monte Carlo simulations in R. Some familiarity with logistic regression is helpful, but not essential.

Day 1: Introduction to Causal Inference. Introduction to Monte Carlo simulation.

Day 2: Using covariates in experiments. Blocking and matched pair designs.

Day 3: Sample selection and statistical power in experimental design.

Day 4: Dealing with non-compliance, instrumental variables.

Day 5: Experiments and causal mediation. Dealing with sample attrition.

Day 6: Introduction to Difference-in-Differences. Synthetic controls.

Day 7: Introduction to Regression Discontinuity. Sharp vs. Fuzzy designs.

Day 8: Introduction to Matching. Univariate matching, Mahalanobis distance, propensity scores.

Day 9: Matching II. Extended lab demonstration during lecture period.

Day 10: Course Conclusion. Extended discussion of applications.