label define but it's not 1-to-1

[Yao Zhao]

I don't think I make a clear question, so let me show my data.

* Example generated by -dataex-. To install: ssc install dataex
clear
input str31 q14_marriage
"Single; never married"          
"Single; never married"          
"Single; never married"          
"Married or domestic partnership"
"Married or domestic partnership"
"Divorced"                       
"Married or domestic partnership"
"Married or domestic partnership"
"Married or domestic partnership"
"Married or domestic partnership"
end

For this string variable, it has multiple choices and I want them to be 0,1,1,0,0: (my reasoning here is whether people are in a relationship or not. I'm open if you have different opinions.)

Divorced
Engaged; to be married
Married or domestic partnership
Separated
Single; never married

In my case, three string values are 0 and two string values should be 1. How to do it?

[Rich Goldstein]

there are lots of ways to do this (but note that your example does not include "Engaged; to be married"); here is one:

gen byte partner=substr(q14,1,3)=="Mar" | substr(q14,1,3)=="Eng"

at this point you can add labels to the new variable; see

help label

[Nick Cox]

gen wanted = strpos(Q14, "Single") | strpos(Q14, "Separated")  | strpos(Q14, "Divorced")

will assign Single and Separated and Divorced to 1 and the others to 0. Or use negation to flip it round.

That may not be your distinction but it may help with technique. I don't follow your 0 1 1 0 0 or your ordering as being anything but alphabetical.