Carrying Values Backward by Group

[Mark Weiss]

Hi all,

I have a panel of individual-level survey data with three variables of interest. id signifies a local geography tag, so id can repeat, as can year. Var is a signifier of a policy adopted by a municipality. It looks something like this:

id	year	var
1	2013	0
1	2013	0
1	2014	0
1	2015	1
2	2013	0
2	2014	0
2	2015	0
2	2015	0

I want to fill backward such that each locality that has adopted the policy has var = 1 in each year, regardless of the year of adoption. I want to tell the program to essentially carry backward any observation of 1 starting in 2015 back to 2013, sorted by id. Notably, I want nothing to change for id = 2, since that region has not adopted the policy.

I tried

tsset id

year, but the console tells me "repeated time values within panel." This makes sense, because there is no way to identify who is a unique respondent across years and each id / year can repeat multiple times. I also tried

bysort id (year): replace var = 1 if var[_n+1] == 1

but no changes were made. Any sense on how to do this?

[Nick Cox]

egen wanted = max(var) , by(id)

Otherwise this is an FAQ. https://www.stata.com/support/faqs/d...issing-values/

However

tsset id year

was what you need for other purposes, although it doesn't help here.

[Mark Weiss]

The first line of code worked, and what an elegant solution! Didn't even occur to me. Thanks for the help, as always.