20.04: Difference-in-Difference

[malcolmbarrett]

• https://pubmed.ncbi.nlm.nih.gov/33791189/
• https://github.com/eleanormurray/DiD_Tutorial/blob/main/R/Code_DiD_tutorial.R
• https://diff.healthpolicydatascience.org/
• with marginaleffects: https://marginaleffects.com/bonus/hypothesis.html#difference-in-differences

[malcolmbarrett]

It occurs to me that the falsification example with universal also extends as a potential parallel trend. If we believe they are largely the same except for internal policies that might be a good DiD. But I don't have a specific change in mind yet and we still don't have the universal data for real