0E Maximum Likelihood
Luke Keele, Penn State University
17 -28 June (two week course / 35 hrs)
Course Content
This course presents an overview of maximum likelihood estimation (MLE) for models where the traditional assumptions of ordinary least-squares regression are violated often because the dependent variable is discrete. The course starts with a review of estimation via the maximum likelihood principle and its roots in probability distributions. The rest of the course will be devoted to the estimation and interpretation of the most commonly used models. Specific models that we will cover include binary logit and probit, multinomial logit, ordered logit and probit, Poisson regression models and other models for event counts, and basic event history models. The course will feature hands on data analysis with daily lab assignments that use real data..
Course Outline
Day 1 – Introduction to R and Stata
Day 2 – Introduction to the Theory of MLE
Lab Day 2 – Basic MLE Assignment with simple optimization of single parameter models.
Day 3 – Estimation of the linear model via MLE. Revising the linear model to incorporate heteroskedasticity.
Lab Day 3 – Basic Regression via MLE – Program the linear model via MLE, estimate the heteroskedastic regression model in Stata.
Day 4 – Introduction to the logit and probit model
Lab Day 4 – Estimation and interpretation of logit and probit models.
Day 5 – Uncertainty estimates, models for grouped binary data.
Lab Day 5 – Logit and Probit II – estimates confidence intervals for predicted probabilities and estimation and interpretation of grouped binary data.
Day 6 – The ordered probit and logit model
Lab Day 6 – Estimation and interpretation of the ordered probit model.
Day 7 – Models for unordered outcomes
Lab Day 7 – Estimation and interpretation of the multinomial and conditional logit model.
Day 8 – Models for counts
Lab Day 8 - Estimation of interpretation of poission and negative binomial regression.
Day 9 – Introduction to Survival Models
Lab Day 9 - Estimation and interpretation of survival models
Lab Day 10 - Projects
.