This section of AEM is designed for masters and PhD students who wish to polish their skills in linear regression analysis and gain a deeper understanding of the foundations of statistical inference, including some of the key competing theoretical perspectives and controversies that dominate current thought. The approach taken in this section is somewhat more conceptual than that found in a typical introductory course in econometrics, but is complemented throughout by an emphasis on applied statistical practice, especially in environments with "messy" data and/or in which substantive theory is weak–all of which are hallmarks of statistical work in public policy. We will work closely with a textbook, Introduction to Econometrics (2nd ed) by Stock & Waston. Major themes in the course include:
Review of the logic of descriptive, exploratory, and inferential statistics
Nonparametric versus parametric statistics
The principles quasi-experimental design and causal modeling
Likelihood inference, likelihood functions, and information theory
Bayesian inference: Explicitly accounting for uncertainty in underlying theory
Grappling with the "Specification Problem" in statistical inference
Qualitative response models (logit, probit) and other models for handling restrictions on the dependent variable
Random coefficient and hierarchical linear models
Students are assumed to have been exposed to linear regression at the graduate level and to be willing to learn to work with summation notation and matrix algebra. Most of the statistical work in the course will involve using the Stata statistical package.