I am working on a dataset that you can find attached, which displays as dependent variable the votes of populist parties in the years 1979-2020 in Europe (Panel Variable: country, Time variable: Year). My task is to see if, as LGBT rights increase (i.e. dummy variables Union, Marriage, Stepchild_adopt and Joint_adopt) there is an increase in the total votes of populist parties (gen PopulistTotal = PopulistLeft + PopulistRight); so I would have 4 different treatments. The problem is that the parallel trend assumption for each treatment does not hold (it is very ambiguous, in some cases even the increase in populist votes increases even before the introduction of the LGBT law in question). The problem is the following: I have different treatments (four: Union, Marriage, Stepchild_adopt and Joint_adopt), and countries that introduce these treatments at very different times (some in the 1980s, others in 2020).
What do you think is the best method to follow?
I was thinking of using xtreg, with i.country and i.Year:
xtreg PopulistTotal Union Marriage Stepchild_adopt Joint_adopt i.country i.Year. How can I correct the fact that the parallel trend assumption does not hold?