Friends,

I am using panel data set for individuals * years.

Individuals are treated at different years, ranging from 2000 to 2015.

I want to show the effect of being in the treatmentgroup (vs control) using all observations and a normalized time scale (t=-1 , t=0, t=+1).

The controll group has no treatmentperiod obviously, so it cannot have any values of t=-1, t=0 etc.



So is there a neat way of comparing those with treatment in 2007, to the controll in 2007 (but for every year), etc via regression?


using only the ones who have treatment in 2007:

reg Y i.treated##i.year
margins, dydx(treated) at(year = 2000 (1) 2015)) plot (xline(2007))

This produces pretty much what I want, but only for the group with treatment in 2007.

So I need some way of making one of each of these (for every treatment year) and combining them into one regression, to reduce errors and plots.


Thank you very much.