How to plot parallel trend*graph with event study

[Berit Alsson]

Hi,

What is the best way to visually inspect if the parallel trend holds or not when working with event study with variation in treatment timing?
im currently trying to replicate Pedro H.C. Sant´Anna graph:

with 4 different groups based on when treatment starts, and before year 1986 no group is treated. In this example plot there is no-never treated group which I have in my data, so mine will look like this but with an additional group that is never treated.

I do not know how to do this in another way, I cannot use event study graph with normalized time because then I cannot plot my control group because all my treated groups are treated in different times and then normalized, so I cannot define a time for my control group obv.


My code:
collapse(mean) outcome, by(year gvar)
reshape wide outcome, i(year) j(gvar)

graph twoway connect outcome* year


is this the right approach for this design?






Btw. I cannot provide an example on my data due to confidentiality

Thanks!