Code:
eventdd series `xlist3' i.state##c.trend i.state i.year [pw=weight], noline timevar(timeToTreat) method(ols, cluster(state)) graph_op(title("(a) Total Population") ytitle("Coefficient") xtitle("Event Time") xlabel(-4 (1) 3) ylabel(-0.3 (0.1) 0.3) legend(off))
Array
The command eventdd is automatically setting event time 3 (i.e. lag 3) as the second ommitted category. I want to manually override the command to set event time -4 (i.e. lead 4) as the second omitted category.
I have attempted looking through Statalist, the Clarke and Schythe (2020) working paper found here https://ftp.iza.org/dp13524.pdf, the eventdd help file and its examples, related commands such as fvset, etc. and I cannot figure it out. I know that plotting the graph I am wanting must be possible as a nearly identical graph appears in Appendix Figure A.5 in this paper https://papers.carlyurban.com/PBE-an...nt_Jan2021.pdf . I can provide additional details on the data if it is helpful. Thanks!
0 Response to Problem with omitting two categories using eventdd with a state-specific linear trend control
Post a Comment