Duration variable

[Keith Preble]

Good evening:

Below is a sample of data from my data set. The last column, NoBust, is how I want the variable to appear:

year	partnercode	targetcode	AllBuster	NoBust
1972	UK	Cuba	0	0
1973	UK	Cuba	0	1
1972	UK	Cuba	1	2
1973	UK	Cuba	0	0
1974	UK	Cuba	0	1
1975	UK	Cuba	0	2
1976	UK	Cuba	0	3
1977	UK	Cuba	0	4

I tried to do this in Stata, but I can't get it to appear in the table above. Instead, it appears as you see in the table below -- when the event, AllBuster, is 1, it resets to zero rather than continuing the series and starting zero after the event (as in my table above):

year	partnercode	targetcode	AllBuster	NoBust
1972	UK	Cuba	0	0
1973	UK	Cuba	0	1
1972	UK	Cuba	1	0
1973	UK	Cuba	0	2
1974	UK	Cuba	0	3
1975	UK	Cuba	0	4
1976	UK	Cuba	0	5
1977	UK	Cuba	0	6

This is the code that I came up with below, but I can't for the life of me figure out how to get it to work the way I want. The data set is a panel data set (unbalanced). I have over 11,000 observations.

by partnercode targetcode (year), sort: gen nobust = (allbuster0 == 1)
by partnercode targetcode (year): replace nobust = sum(nobust)
by partnercode targetcode nobust (year), sort: gen nobust0 = _n-1 if nobust >=0
drop nobust

Any help would be greatly appreciated!

[Nick Cox]

You have duplicates on your variables country country_2 year but you seem to want them to be treated differently. Also, there is inconsistency between your code and your data example in what the variables are called.

To help anybody else, here is dataex code to reproduce your data (FAQ Advice #12). Otherwise I can't follow what you want and why.

* Example generated by -dataex-. To install: ssc install dataex
clear
input int year str2 country str4 country_2 byte(allbuster nobust)
1972 "UK" "Cuba" 0 0
1973 "UK" "Cuba" 0 1
1972 "UK" "Cuba" 1 2
1973 "UK" "Cuba" 0 0
1974 "UK" "Cuba" 0 1
1975 "UK" "Cuba" 0 2
1976 "UK" "Cuba" 0 3
1977 "UK" "Cuba" 0 4
end