Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.

# st: RE: Computing difference of counts within groups

 From Nick Cox To "'statalist@hsphsun2.harvard.edu'" Subject st: RE: Computing difference of counts within groups Date Mon, 1 Nov 2010 16:14:04 +0000

```No looping needed at all. This yields to -by-.

sysuse auto, clear
egen nforeign = total(foreign == 1), by(rep78)
egen ndomestic = total(foreign == 0), by(rep78)
gen diff = nforeign - nodomestic

etc.

-egen- here is optional, as

bysort rep78: gen nforeign = sum(foreign == 1)
by rep78: replace nforeign = nforeign[_N]
by rep78: gen ndomestic = sum(foreign == 0)
by rep78: replace ndomestic = ndomestic[_N]

gets you to the same place.

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

Grealy, Patrick J

I thought there was a simple way to do what I am attempting using the egen command but I have not gotten that to work right. Can someone make another suggestion? I am trying to check if the difference between two groups within another group exceeds a certain value. For example, using the auto dataset, within the five rep78 groups I would like to calculate the difference between Domestic and Foreign and check if it exceeds some limit, say 7.

This code below does what I want but I was hoping to find something simpler:

sysuse auto,clear
gen fdiff = .
foreach ix_rep78 of numlist 1/5 {
quietly count if foreign==0 & rep78==`ix_rep78'
local foreign0 `r(N)'
quietly count if foreign==1 & rep78==`ix_rep78'
local foreign1 `r(N)'
replace fdiff = abs(`foreign0' - `foreign1') if rep78==`ix_rep78'
}
tab rep78 foreign if fdiff > 7

*
*   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/
```