Links!
The past three class sessions we (by which I mean “I”) have fumbled around trying to describe instrumental variables — the basic strategy, the details of how and when this strategy will work, and some basic formal derivations. I’ve finally settled on a presentation that I like reasonably well (linked above). I try to tie together the potential outcomes approach, the causal graphs approach and a causal (“structural”) equations approach. You can let me know how successful I was.
Perhaps more convincingly, the provided R script generates data from various sorts of data generating processes and confirms what the theory tells us — that what the traditional IV estimator is actually estimating will differing depending on the various assumptions that are satisfied. Please take the time to explore the script and make sure that you understand what is going on. If you don’t track something, ask me in class!
As always, thanks for your patience.
Stat Ranch 