Maximum Likelihood

This seminar is a survey of maximum likelihood (ML) methods and their applications to empirical political questions. It was the third course in the University of New Orleans graduate research methods course sequence when I was at UNO. This course focuses on understanding the conditions when the assumptions of ordinary least squares (OLS) regression are violated, the principles of maximum likelihood estimation, and what models are appropriate given observed data. This seminar centers on the use and interpretation of ML and on linking theory to statistical models.

The models covered in this course are widely used in political science today. To engage with other researchers’ quantitative empirical work it is necessary to be able to understand and evaluate it. This course covers a number of different models—some of which will be more of use to you than to others. This course enables students to explore models suited to the nature of their data in detail and use these models to replicate and extend current research.

Syllabus (pdf)
Course overview
Week 1: Introduction and review of linear models (Week 1 slides)
Week 2: No class (APSA)
Week 3: OLS & time series review; intro. to likelihood inference (Week 3 slides)
Week 4: Likelihood inference (Week 4 slides)
Week 5: Binary dependent variables I (Week 5 slides)
Week 6: Binary dependent variables II; heteroskedastic models (Week 6 slides)
Week 7: Ordered dependent variables (Week 7 slides)
Week 8: No class (mid-semester break)
Week 9: Unordered/choice models (Week 9 slides)
Week 10: Event count I: Poisson (Week 10 slides)
Week 11: Event count II: Negative binomial, zero-altered (Week 11 slides)
Week 12: Hazard models I: Discrete/continuous time, semi-parametric (Week 12 slides)
Week 13: Hazard models II: Parametric, special topics (Week 13 slides)
Week 14: No class (Thanksgiving)
Week 15: Censored/truncated variables (Week 15 slides)
Week 16: Multiple equations (Week 16 slides)
%d bloggers like this: