Variable for _n per distinct id

[Hanna Larsson]

Hi!
I have a problem I would appreciate your help with. My data (>700 000 rows) consists of the variables id, id2 and var and I want to generate newvar (se below).
Newvar should be based on the number of distinct id2 per id, starting with 1 and counting upwards - order defined by the quantity of var.


data

id	id2	var	newvar
1	1	1000	1
1	2	2000	2
2	3	1200	1
2	3	8000	1

If I use _n it counts each row per id2. I want newvar to be the same for each id2.

Thank you so much!

[Andrew Musau]

order defined by the quantity of var

It is not clear what this means if "id2" does not prescribe the order. You need to expand on this point.

bys id (id2): gen wanted= sum(id2!=id2[_n-1])

[Hanna Larsson]

It is not clear what this means if "id2" does not prescribe the order. You need to expand on this point.

bys id (id2): gen wanted= sum(id2!=id2[_n-1])

Thank you!!!