Hi!
I'm trying to do a Triple Diff in Diff event study with the feols function but I'm getting weird results. I wonder wether my formula is correctly specified.
The standard event study diff in diff takes the form:
feols(y~ i(week, treat, -1) | c, data=df)
And my intution is that for the triple diff in diff version, it takes the form:
feols(y~ i(week, treat * treat_other, -1) | c, data=df)
, where treat_other is the extra control group we add to the comparison. We followed examples of the following paper
For the static version (no event time) triple diff, we get the desired results using the treat_other variable. In fact, the standard triple diff with no event study is written as:
feols(y~ treat_week*treat_other*treat | c, df)
Am I doing this right? I wonder wether the second formula here is correctly specified.
Thanks!
Hi! I'm trying to do a Triple Diff in Diff event study with the
feols
function but I'm getting weird results. I wonder wether my formula is correctly specified. The standard event study diff in diff takes the form:feols(y~ i(week, treat, -1) | c, data=df)
And my intution is that for the triple diff in diff version, it takes the form:feols(y~ i(week, treat * treat_other, -1) | c, data=df)
, wheretreat_other
is the extra control group we add to the comparison. We followed examples of the following paper For the static version (no event time) triple diff, we get the desired results using thetreat_other
variable. In fact, the standard triple diff with no event study is written as:feols(y~ treat_week*treat_other*treat | c, df)
Am I doing this right? I wonder wether the second formula here is correctly specified. Thanks!