2
$\begingroup$

I am using data from a controlled experiment with different timing of treatment implementation across individuals in the treated group. I can date treated observations with time distance to the implementation dates; However, how can I make a trend plot for control group? In particular, how to date the trend for control individuals?

$\endgroup$
1
$\begingroup$

I found that the good way to do it is using "stacked cohort" diff-in-diff, where you create a dataset that covers a time window for each event, and then stack up the datasets into one. One example is in Gormley and Matsa (2011).

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ Actually, it is not correct to do this, because Post_t is not well-defined. We cannot test the parallel trend assumption in this case. $\endgroup$ – Hong You Jun 18 at 14:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.