How to use collapse (sum) + missing values

[Tilo Gashi]

Hi,

I have a question regarding collapse(sum) command in STATA. Is it okay if I have three observations falling in the same category say "a" such as: 6, . , 3 and collapse(sum) gives as a result 9? Hence, it treats the missing value as 0 in adding them up.

I was thinking would it more appropriate if it adds the 6 and 3 but it also leaves the missing value not added like:
a 9
a .

If yes, do you know how could this be done?

Thanks and sorry in advance if my question is not structured properly and is not clear enough

[Clyde Schechter]

Aggregate sum() functions in Stata treat missing values as zero. This is typically the most useful approach to missing values in situations where grups of numbers are being added. However, if you want to treat missing values separately, you can do so. For example:

* Example generated by -dataex-. To install: ssc install dataex
clear
input str1 category float summand
"a" .
"a" 2
"a" .
"a" 1
"b" 3
"b" .
"b" 7
end

gen byte summand_missing = missing(summand)
collapse (sum) summand, by(category summand_missing)
replace summand = . if summand_missing

list, sepby(category)

[Tilo Gashi]

Thanks a lot Clyde Schechter!