Advanced Regression Analysis

Regression is perhaps the most ubiquitous tool for quantitative analysis in the social sciences. Why? What can it be used for, how can we interpret results, when do its assumptions fail, and what can analysts do if the core assumptions no longer hold? This course will answer all of these questions by providing building blocks for causal modeling (potential outcomes and structural causal models), the multiple motivations of linear regression (plug-in estimator, geometric, linear algebra, FWL), modifications required for limited dependent variables (logistic regression), how it operates in the presence of covariates (moderation, mediation), examine its various use cases for drawing inferences from data (selection on observables, instrumental variables, panel data and differencing, regression discontinuity) while providing students with practical lessons on statistical programming.