Kevin's Resources on Causal Inference Methods for Political Economy
This is a collection of resources on Causal econometric methods - with a specific focus on methods used in Political Science and Political Economy (PSPE). These resources are a collection of the content I learned during my postgraduate degree at the London School of Economics.
For implementation of difference-in-differences in R and Python, see my repository on DiD. For more about me, see my github page.
Topics: confounders in causal identification, and how it relates to randomness/exogeneity.
Topics: potential outcomes framework for causal inference, and introduce the main estimands: ATE, ATT, and LATE.
Topics: introduce the classical 2-period Difference-in-Differences design used for causal identification, and the parallel trends assumption.
Topics: expand DiD to multiple pre-and-post periods, introduce the two-way fixed effects estimator, and discuss conditioning for parallel trends.
Topics: expand DiD to staggered treatment settings, discuss the issues with TWFE in staggered settings, and discuss modern advances in DiD.
Topics: extending Difference-in-differences to scenarios where parallel trends does not hold, including IFEct, MC, and synthetic controls.
Topics: the theory behind instrumental variables and why they can find causal effects, as well as the 2-stage least squares estimator.
Topics: the regression discontinuity design and its sharp variant, as well as the key identification assumptions and design of RDD.
Topics: extensions of the RDD design, including fuzzy RDD for non-compliance with cutoffs, and regression kink design.
Topics: design of randomised experiments, including blocked experiments, non-compliance, and survey experiments.
Topics: selection on observables. I discuss regression and matching strategies.