Hello.
I want to estimate a difference-in-differences-model for an article but I have a few questions:
Model with 2 time points:
- My model looks quite nice but I want wo bring in another difference: employment status (which is dichotomous and the same at both time points). Do I have to include the employment status as a group variable (bysort ewstatus: reg lifesatis treat_time treatment time2er, robust) or as a triple interaction (reg lifesatis treatment##time2er##vz_ne, robust). (I have Stata 16.)
Model with more time points:
I would like to include more time points (before and after treatment). Yet the treatment is not at the same time point for all respondents.
- How do I code the variable for time? t=0 for the year of treatment, -1=the year before 1=the year after the treatment? The range is -4 to 4 but I do not have 9 waves for the non-treatment group. How do I handle this?
- Or do I use dummy variables for the years and interactions with the dummy?
I have already looked in the popular literature but I cannot find a practical example. I am looking forward for your help!
Carolin
I want to estimate a difference-in-differences-model for an article but I have a few questions:
Model with 2 time points:
- My model looks quite nice but I want wo bring in another difference: employment status (which is dichotomous and the same at both time points). Do I have to include the employment status as a group variable (bysort ewstatus: reg lifesatis treat_time treatment time2er, robust) or as a triple interaction (reg lifesatis treatment##time2er##vz_ne, robust). (I have Stata 16.)
Model with more time points:
I would like to include more time points (before and after treatment). Yet the treatment is not at the same time point for all respondents.
- How do I code the variable for time? t=0 for the year of treatment, -1=the year before 1=the year after the treatment? The range is -4 to 4 but I do not have 9 waves for the non-treatment group. How do I handle this?
- Or do I use dummy variables for the years and interactions with the dummy?
I have already looked in the popular literature but I cannot find a practical example. I am looking forward for your help!
Carolin
