2

How is there a parallel trends assumption if there is no formally defined treatment group?

Tushar Mehra
  • 21
  • 1
    Have you seen this question? https://stats.stackexchange.com/questions/465167/how-does-the-parallel-trends-assumption-operate-when-there-are-multiple-interven – RobertF Mar 02 '22 at 13:37
  • What do you mean by no “formally defined” treatment group? Do you mean different units start treatment at different times? – Thomas Bilach Mar 03 '22 at 18:12
  • @ThomasBilach What I mean is that the since the treatment is not binary, there is no formal "treatment group" in the traditional sense of the term – Tushar Mehra Mar 04 '22 at 23:04
  • @RobertF --> thank you for the link; I think it is addressing a slightly different question to the one that I am asking given that it is looking at multiple treatments as opposed to a single continuous treatment – Tushar Mehra Mar 04 '22 at 23:05
  • You could still have a “group” in some sense. Do a subset of units start treatment at a particular point in time? Maybe some policy is introduced, or a new law is passed? – Thomas Bilach Mar 06 '22 at 21:03
  • @ThomasBilach I am referring to the kind of treatment measured in the first answer to the question here: https://stats.stackexchange.com/questions/152684/how-do-i-interpret-a-difference-in-differences-model-with-continuous-treatment – Tushar Mehra Mar 07 '22 at 17:28
  • @ThomasBilach I mean not necessarily. There could be a policy that affects everyone but the effect is doses by an exogneous characteristic, which is a continuous variable. In that case you have everyone somehow treated by with different intensity, no group is a pure control. – PhDing Oct 21 '22 at 20:12

0 Answers0