Mixtape University: Diff-in-diff with a checklist. Defining your target parameters

Greetings! Today I am posting two new 15-minute videos for Mixtape University. This is for my “Diff-in-diff with a checklist” series. It will be based on my workshop, Causal Inference 2. And at the moment, this is my checklist (although last week the way I taught this in a workshop, I found that Steps 4-6 have not yet fully gelled in my mind if that will be how I do it when I get to those steps for this series — but we have time to fix that).

Today’s material will fit under “Choosing your target parameter”. What I’m trying to do in these two videos (and in another one I will show you next week where I run a simulation) is that different aggregated datasets, based on the identical underlying counts, will have implicitly been weighted in different ways. I use as an example the idea of county level versus state level crime data to make my example. In the first video, I introduce the county level data example, and in the second, I introduce this notion “weighting treatment effects” in different ways according to geography.

This material is something that slowly percolated in my head after years of teaching this material, only for a then stray comment by Andrew Goodman-Bacon to make me realize it was largely what Gary Solon, Steven Haider and Jeffrey Wooldridge wrote about in this 2015 JHR article “What are we weighting for?”

Mixtape University is a growing collection videos are for my paying members. I want to thank them for their support and I hope that they find these helpful. These are theirs to do with whatever they want. They can use these for their classes, download them (if that is possible), whatever. No need to ask me.

And for everyone else who is not yet a paying subscriber, consider becoming one! Paying subscribers get:

1. Full access to the entire past articles I’ve written on here (pushing 500-600 substacks or something at this point).

2. Full access to all new substacks (instead of just around half, since I randomize the paywall).

3. Full access to the Mixtape Mailbag, a Q&A segment where people write in with their questions, and I try to take a stab at answering them.

And lifetime subscribers, if you want to do that, get annually a half hour of office hours if that’s what you want too!

I have not yet fully figured out just how exactly I will store these, except that they will be under the mixtape University tab at the top. As I mentioned, I think the plan is I’ll have a single newsletter that gets pinned the top, like a table of contents, linking to each one, but I’m not yet there. I may also sometimes post more than two of these every Sunday, just to speed things up a little, but right now it’s two videos, each approximately 15 minutes, every Sunday. But with that said, here is this week’s!