The lectures are over, the problem sets are submitted -- all that's left for the fall semester are finals in a couple weeks. One of the courses I'm taking is Germán Rodríguez's "Generalized Linear Statistical Models" and it occurred to me that I should highlight the course website for blog readers. Princeton does not have a school of public health (nor a medical school, business school, or law school, amongst other things) but it does have a program in demography and population research, and Professor Rodríguez teaches in that program.
The course website includes Stata logs, exams, datasets, and problem sets based on those data sets. The lectures have closely followed the lecture notes on the website, covering the following models: linear models (continuous data), logit models (binary data), Poisson models (count data), overdispersed count data, log-linear models (contingency tables), multinomial responses, survival analysis, and panel data, along with some appendices on likelihood and GLM theory. Enjoy.