Researchers should be aware of the threat of regression to the mean when constructing matched samples for difference-in-differences. We provide guidance on when to incorporate matching in this study design.
No takes yet. Share an insight, caveat, or question.
Daw et al. (2018) studied this question.