Diff-in-diff as a causal design
Causal inference
Research methods
Difference-in-differences
Originally shared on LinkedIn.
According to The Mixtape, “diff-in-diff is a causal design, not simply four numbers subtracted in a row.”
“Design” implies some assignment to treatment by the researcher, otherwise it’s “simply four numbers subtracted in a row” plus “assume a can opener” (parallel trends). To get parallel trends by design probably means some kind of random assignment. But then you probably don’t want to do diff-in-diff:
Evaluating natural experiments, in Empirical Research in Accounting: Tools and Methods.