A footnote (or headnote):
-egen, rsum()- will work in modern Statas but is undocumented: the
modern name is -rowtotal()-.
Nick
[email protected]
Tim Wade
Michelle, maybe a different approach, something like this, will work.
Basically create a set of temporary variables equal to the original
set of variables which you recode to 1 and 0 for each race category,
sum them up, and drop them then do it again for the next race
category.
forvalues i=1/[number of race categories] {
forvalues j=1/28 {
gen temp`j'=sprace`j'
}
forvalues k=1/28{
replace temp`k'=1 if sprace`k'==`i'
replace temp`k'=0 if sprace`k'~=`i'
}
egen count`i'=rsum(temp1-temp28)
drop temp1-temp28
}
hope this helps, Tim
On Fri, Oct 31, 2008 at 2:53 PM, Michelle Debbink
<[email protected]> wrote:
> Hi All - New to the listserv and seem to be running into an issue with
a forvalues loop. We have a categorical variable for the race/ethnicity
of a list of contacts for each respondent to a survey up to 28 contacts.
Some respondents have fewer than 28 contacts and so have missing data
for some contact entries. We would like to sum the number of contacts
of each race for each respondent (so as to have a variable for number of
white contacts, number of black contacts, etc). In order to sum the
contacts, we need to make a dummy variable for each contact regarding
their race, and missing values must be replaced with zero so that the
sum isn't returned as missing. We have tried each of the following
forvalues loops but it seems to not be working (all return "_=invalid
name") and we would appreciate any guidance you could give.
>
> // generate dummy variables for ethnicity of all sex partners in last
12mo and replace missing with 0
> forvalues i = 1/28 {
> tab sprace`i', gen(sprace`i'cat_)
> replace sprace`i'cat_* =0 if sprace`i'==.
> }
>
> //generate an identical variable that has no missing values (which we
could then use to make the dummy variables to sum)
> forvalues `i'=1/28{
> if sprace`i'!=. {
> gen spracem`i'=sprace`i'
> }
> else{
> gen spracem`i'=0
> }
> }
>
> //sum across all the new dummy variables to get number of contacts of
each race
> forvalues i=1/28{
> gen numwhitepart = sum(sprace`i'cat_1)
> gen numblkpart=sum(sprace`i'cat_2)
> gen numhispart=sum(sprace`i'cat_3)
> }
>
*
* 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/
