This class will provide a practical, hands‐on approach to quantitative analysis of the impacts of public policy using the powerful (and relatively user‐friendly) Stata statistical programming system. The approach taken will blend a concise summary of the theory behind different approaches to estimating the causal effects of policy, with the practical challenges of coding and estimating actual models using real world data sets. Students will be required to estimate real models using real data sets on a weekly basis, and report on their results in class (20% of grade). In addition, there will be two in‐class project presentations (30% of final grade) using pre‐selected, common datasets, and a final paper and project presentation (50% of final grade) modeling a policy issue using a dataset of the student’s own choice. MPAff and MGPS students are encouraged to select a domestic or global policy issue, and dataset, related to their major field of interest.
Prerequisite: Students should already have a basic understanding of how to use ordinary least squares to estimate a linear regression model, from a prior introductory statistics or econometrics course.
Topics covered in this class will include:
Probability and Statistics Review, Intro to Stata Basics; Simulation Methods
Linear Regression Models: Classic Approach and Extensions
Linear Regression Models: Specification Issues and Causal Interpretations; Potential Outcome Framework and Treatment Effects
Coping with Endogeneity: Instrumental Variables Methods
Panel Data Models: Difference‐in‐Differences and Fixed Effects; Correlated Random Effects
Matching Estimators and Regression Discontinuity Designs
Censored Data and Selection Models
Categorical and Count Outcomes: Interpreting Logit, Probit, and Poisson Models
Big Data: Tools to Deal with Really Large Datasets Or Really Large Models