Open s3alfisc opened 3 weeks ago
looks good; can also add as a method attached to the DiD estimator classes but I like the idea of having it free-standing because ideally you make this plot before running any regressions since it should inform your choice of estimator (e.g. if all treatment is in one cohort, 2WFE doesn't have negative weighting problems).
I will clone this and develop it a bit further (e.g. collapse the data by some level, e.g. state) before plotting
I like the idea of having it free-standing because ideally you make this plot before running any regressions
I agree 👍 to clarify my suggestion, it would have been to keep it as a standalone function, but to also add it as a post-estimation method, purely as convenience for users who might want to check post hoc. But maybe we shouldn't incentivize poor research practices so happy to drop the post-estimation method idea.
On the point of setting a good example - it would be to add an example of the function at the beginning of the DiD notebook =)
Code by @apoorvalal .
I have slightly renamed function arguments, added docstrings & type hinting, and a very basic API test.
This also took me twice because I forgot to check out the dev branch 😅
@apoorvalal , do you think there would be value in adding this function as a post-estimation method to the DiD estimators?