Hello ,
Suppose I intend to use a difference-in-differences research design that exploits within-district variation in exposure to violent crime shock over time.
And to account for heterogeneous treatment timings I run
where x1 is the crimeshock indicator
x2=crimeshock*female
x3=lagged crimeshock
x4* laggedcrimeshock*female
Here I have included an indicator of lagged crime shock(indicator of crime shock in the previous year).
1)Is this okay to include a lagged explanatory variable in such a model ?
2) Is it alright to include the lagged interaction term as well?
Suppose I intend to use a difference-in-differences research design that exploits within-district variation in exposure to violent crime shock over time.
And to account for heterogeneous treatment timings I run
Code:
did_multiplegt outcome district year x1 , placebo(1) dynamic(0) cluster(village_id) breps(999) trends_lin(district) longdiff_placebo robust_dynamic controls( x2 x3 x4 girl_child) graphoptions( ylabel(-.3(.1).1) xlabel(-2(1)0) title() xtitle(Years from First Exposure) ytitle(Estimated Effects on outcome (SD)) legend(off))
where x1 is the crimeshock indicator
x2=crimeshock*female
x3=lagged crimeshock
x4* laggedcrimeshock*female
Here I have included an indicator of lagged crime shock(indicator of crime shock in the previous year).
1)Is this okay to include a lagged explanatory variable in such a model ?
2) Is it alright to include the lagged interaction term as well?
