A case like that depends mostly on your reporting needs. By reporting needs I mean: 1) how will the outcome of this question show across the UX and 2) how do data analysts intend to exploit answers to the question.
Because it’s OHRI, I suspect that 2) is the key factor.
In which case why not literally having 5 boolean questions?
- Belongs to general population (yes/no).
- Is a MSM (yes/no).
- Is a FSW (yes/no).
- Is a MSW (yes/no).
- Is a PWID (yes/no).
That will be very easy to flatten.
I’m a little confused about the first question. I would assume that being part of the “general population” means being part of the general population of HIV positives right? In which case the question might be either irrelevant (because there should be a diagnosis or some other findings somewhere already that lead to this questionnaire) or the question should become
- Is HIV positive (yes/no).