,
12 tweets,
4 min read
Read on Twitter
Video #1 in a series on causality for my Econ 305 class. More info on the class at nickchk.com/econ305.html Video 1: What is Causality?
Video 2: The very basics of causal diagrams
Video 3: encoding what you know about the world in a model by drawing your own causal diagrams
Video 4: Using dagitty.net to create causal diagrams
Video 5: Paths from X to Y, and closing back doors
Video 6: Controlling for variables to close back doors
Video 7: Collider variables! It has come to this, and for more than one reason.
Video 8: The first tool in the toolbox, fixed effects! It is strange to me that FE tends to come later in the econometrics sequence. It's a direct and minor revision on just controlling for a variable. So introduce it as such, right after controls!
Video 9: Introducing the family of CI methods that contrast a treated group with a comparable untreated group, and our first stab at it, constructing our untreated group ourselves using coarsened exact matching (the easiest way to match by hand!)
Video 10: Difference-in-differences! If you squint you can see that I'm introducing it conceptually as what to do with an interrupted time series when you're worried about "time" being a confounder/on a back door path.
Video 11: Regression Discontinuity Designs. And yes, I *do* do them without regression! So really just discontinuity designs. This is the one where actual-practice and the-way-I-do-it differ most, uh, sharply.
And finally, Video 12: Instrumental Variables, introduced via imperfect random assignment. However well those statistical properties are holding up, can't beat the intuition for understanding causal inference! Hope you liked the videos.
