Hi DCR – Three more sessions left! This week we get into the core issues of causal inference and confounding. These issues are conceptually challenging, but totally central to the whole discipline of epidemiology, so worth really digging into!
Deborah’s version of Lecture 5 is nicely polished, and provides beautiful and relatively simple examples from her long and illustrious career related to hormone replacement etc. My newer version of the lecture is recorded in class, is a big rougher overall (a bit too much detail on some points, in retrospect), but I think is a bit clearer on the main points about possible explanations for an association that I would like you to take away from this week. Also, from my own personal experience, confounding did not completely sink in conceptually for me until I saw an example with numbers...so I’ve tried to provide that for you in my lecture. Slides 49-53 go through a simple example of confounding, with numbers, that I hope will help you understand the concept as well.
Consider watching both!
Then, after watching the lectures and reading relevant chapters in the book, you can take a crack at describing issues with causal inference in your own study (with different instructions in Assignment 5 for observational studies, RCTs or studies of diagnostic or prognostic tests).
Hope you enjoy. I’ll be at Office Hours tomorrow as usual.
Mark