Announcement

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

  • -didregress- with repeated cross-section


    I am running DID on a repeated cross-section. The survey is household level and the treatment is given at the district-level, all at once in the year 2018. The same districts are surveyed each other but the households change. All the households in a treated district are considered treated. I have created the treat-post interaction term as gen treatpost = treated*(year>=2018)
    The tabulation of year and treatpost is given below.



    Click image for larger version

Name:	Screenshot 2025-03-15 114908.png
Views:	1
Size:	13.2 KB
ID:	1774400


    My concern:
    There are more treated units in 2022, than there were in 2018. Does this create a problem? I am unable to run -estat ptrends- after -didregress- and get an error "treatment assignment times vary". Could it be due to this?

    Further, this is what I get as the first table after running didregress. Should the minimum and maximum in the treated column be the same? What I currently have is 2018 and 2022---is this what creates a problem for estat ptrends? If so, what are the potential solutions?

    Click image for larger version

Name:	Screenshot 2025-03-15 114759.png
Views:	1
Size:	6.7 KB
ID:	1774401

  • #2
    Was your treatment intervention staggered? If yes, then you should consider using staggered DiD methods.

    Comment


    • #3
      It is not staggered. All treatment is given all at once in 2018, as mentioned in #1.

      Comment

      Working...
      X