Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Parallel trend assumption when doing Diff in Diff with xtpoisson and multiple intervention dates

    Dear Stata users,

    I am studying the effect of an intervention in states on the number of contributions/donations in that specific state.

    My control group exist of: all states that do not experience an intervention + the time before the intervention in states that do experience an intervention.

    Since this is count panel data I want to use xtpoisson and do a difference in difference (before vs after intervention).

    However, for the parallel test assumptions I found until now you need the variable 'policy' which equals 1 from the moment of the intervention.

    Since I have different intervention dates for different states, I am not able to give such a dummy to the control group.

    Does anyone know how to test the parallel test assumption in my study?

    PS: This is the command I use now to study the effect: xtpoisson donationsnumber treated i.year, fe vce(robust)
    wherin treated==1 for treated companies from the moment of the intervention till the end of the sampe period

    Kind regards,
    Simone
Working...
X