In longitudinal data, coding with binary variable of 0/1 whenever person ID changes (but keeping 0 for within person ID)

[Claire McKenna]

I'm using longitudinal data, that includes up to four year-month (YEAR_MONTH) observations for each person (PERSON_ID). So each PERSON_ID has up to four records. Assuming the data are sorted by PERSON_ID and YEAR_MONTH, I want to generate a binary variable that flags with a "1" whenever PERSON_ID changes. The variable would take a value of "0" whenever current PERSON_ID and the prior record's PERSON_ID are the same. So I would like data that looks something like the following. Would appreciate any help with this. I'm sure it's more straightforward than I realize. Thank you very much.

PERSON_ID	YEAR_MONTH	NEW_PERSON
1	2020_1	1
1	2020_2	0
1	2020_3	0
1	2020_4	0
2	2020_1	1
2	2020_2	0
2	2020_3	0
2	2020_4	0
3	2020_1	1
3	2020_2	0
3	2020_3	0
3	2020_4	0
4	2020_1	1
5	2020_1	1
6	2020_1	1
7	2020_1	1
8	2020_1	1

[Andrew Musau]

Seems like you just want to tag the first observation of each person, in which case, you can

bys PERSON_ID (YEAR_MONTH): g wanted=_n==1

[Claire McKenna]

Thanks very much. That seems to work. But now I'm having trouble with my grouping function (how I generate PERSON_ID). I've concatenated a couple ID variables, and I'm unsure if the resulting string variable is causing Stata to be unable to distinguish between unique IDs. Does the "group" function work on string variables?

[Rich Goldstein]

I assume you mean the group function to the -egen- command; here is what it says in the help file: "varlist may contain numeric variables, string variables, or a combination of the two."

[Claire McKenna]

I'm all set now, thank you, Rich