st: RE: creating new variable under certain conditions

[Nick Cox <n.j.cox@durham.ac.uk>]

Look at -egen-. You want to calculate in groups, something like the following 

bysort gvkey appyear : egen NEW1 = total(n_int_cites)
by gvkey appyear : egen NEW2 = mean(n_cites) 

Nick 
n.j.cox@durham.ac.uk 

S.H. Former

I have a problem concerning the composition of my database. The database looks as follows:

gvkey	appyear	patent		n_int_cites	n_cites	n_patentsyr
1004	1985		4751021	1		6	1
1004	2001		6622640	1		2	1
1010	1974		4178953	1		2	1
1010	1975		4017058	1		2	1
1010	1977		4113233	1		3	7
1010	1977		4121614	4		8	7
1010	1977		4137937	1		4	7
1010	1977		4158453	3		3	7
1010	1977		4160460	3		5	7
1010	1977		4178755	3		6	7
1010	1977		4184663	2		11	7

I want to create a new variable where the n_int_cites is added up but only when gvkey is similar and appyear is similar. Another variable should be created where the average is taken from n_cites but again only when gvkey is the same and appyear is the same. So it should look as follows:

gvkey	appyear	patent		n_int_cites	n_cites	n_patentsyr	NEW1	NEW2
1004	2001		4751021	3		6	2		4	5	
1004	2001		6622640	1		4	1		4	5
1010	1974		4178953	6		2	1		6	2
1010	1975		4017058	1		2	1		1	2
1010	1977		4113233	1		3	7		17	5.71
1010	1977		4121614	4		8	7		17	5.71
1010	1977		4137937	1		4	7		17	5.71	
1010	1977		4158453	3		3	7		17	5.71
1010	1977		4160460	3		5	7		17	5.71
1010	1977		4178755	3		6	7		17	5.71
1010	1977		4184663	2		11	7		17	5.71

So the important thing is that only observations should be considered in a calculation when gvkey and appyear are the same!

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/