Identifying differences in a variable within a dataset

[Chris Rooney]

I have a dataset that has data that is pooled across different years.

At the moment, I would to like be able to identify the occupational codes which are common across two years (e.g. 2004 and 2005).

So for example, say in the year 2004 the following occupational codes are present:

0101
0102
0103
0104
0105

And in the year 2005, the following occupational codes are present:

0101
0102
1111
1112
1113

Is there a command/function in Stata which can tell me that only occupational codes "0101" and "0102" are present in both? Note that I can't use merge as I'm not using two different .DTA files - it's within the dataset.

Thanks

[Nick Cox]

https://www.stata.com/support/faqs/d...ions-in-group/ applies.

bysort code (year) ; gen diff = year[_N] != year[1] 

tab code if diff

If there were more years than two in the dataset the code could be extended

gen relevant = inlist(year, 2004, 2005) 

bysort relevant code (year) : replace diff = year[1] != year[_N] 

tab code if diff & relevant

[Chris Rooney]

https://www.stata.com/support/faqs/d...ions-in-group/ applies.

bysort code (year) ; gen diff = year[_N] != year[1]

tab code if diff

If there were more years than two in the dataset the code could be extended

gen relevant = inlist(year, 2004, 2005)

bysort relevant code (year) : replace diff = year[1] != year[_N]

tab code if diff & relevant

Hi Nick,

Thanks for that. Excellent. I've tried it for two years and it works! How does one extend it across say, 10 years, as that is the challenge I'm facing?

Thanks.

[Nick Cox]

How you would want to extend it? You could work out e.g. the number of years in which each code appears. Indeed why don't you just cross-tabulate code and year?

[Chris Rooney]

How you would want to extend it? You could work out e.g. the number of years in which each code appears. Indeed why don't you just cross-tabulate code and year?

I'd like to extend such that the same occupational code must occur in all the years (e.g. 2004, 2005, 2006, 2007, 2008 etc.), not just the first two years.

[Nick Cox]

egen tag = tag(code year) 
egen wanted = total(tag), by(code) 
list code if wanted == 10 & tag, noobs