Module 1, Part 2: Matching, Weighting, and Selection on Observables
Part 2 begins our first attempt to estimate a causal treatment effect. We’ll do so with a few approaches, all of which rely on the combination of a “selection on observables” or “conditional independence” assumption, along with a “common support” assumption.
Objectives
- Understand and explain the differences between matching and re-weighting
- Explain the implicit weighting in OLS regression and how to avoid such weighting
- Calculate an ATE or ATT using regression, matching, and weighting
Activities
- Continue working with the Hospital Cost Report Information System (HCRIS) repository.
- Estimate treatment effects with simulated data, using matching, regression, and re-weighting. The code file for in-class simulated data is available here
Slides
Use the buttons below to open the slides either as an interactive website or as a static PDF (for printing or storing for later). You can also click in the embedded slides and navigate through them with your left and right arrow keys.
Fun fact: If you type ? (or shift + /) while going through the slides, you can see a list of special slide-specific commands.