How to count observations if cell contains string (but not a specific string)

[Claudia Gerber]

Hey,

I am working on a household-level panel data set, and we asked the name of each household member (string) and then the final household count (numeric). I want to create a new variable that counts the names of the household members and compare it to our household count variable. I want to check whether cells contain names so that I can kick out observations that have a mismatch between the reported household size and those that did not provide us with all the names of the household members. We counted up to 11 household members.

I could only find the regexm function, but that requires that I specify the string variable (e.g. 'a' in name) but I only care that the cell is not empty. It's really just like Excel's CountA(X1:X10) function.

Thanks so much for your help.

Best, Claudia

HH_member_1	HH_member_2	HH_member_3	… HH_member_11	HH_count	HH_count_check
Maria	Jon	Liz		3	Count of (Maria+Jon+Liz)

[Rich Goldstein]

look at the egen functions for rowmiss and rownonmiss; see "help egen"

[Nick Cox]

Sorry, no idea what that Excel function does, but it sounds as if you want to count non-empty strings in a loop over variables. That is a standard egen function

* Example generated by -dataex-. To install: ssc install dataex
clear
input str4 var1 str6 var2 str7 var3
"frog" "dragon" "phoenix"
"toad" ""       ""      
"newt" ""       ""      
end

egen present = rownonmiss(var*), strok

l

     +------------------------------------+
     | var1     var2      var3    present |
     |------------------------------------|
  1. | frog   dragon   phoenix          3 |
  2. | toad                             1 |
  3. | newt                             1 |
     +------------------------------------+

But watch out: a string with just some stray spaces counts as not missing. If that's a problem, there are easy solutions.

[Claudia Gerber]

Dear Nick and Richard!


Thanks so much for your help! It worked!

Best, Claudia