I have a questionnaire with sections A to F. Section A should be answered for by all respondents, but the rest of the sections should be randomly allocated at one section per respondent. I wound like to assign a section at random to a given enumerator. For example, interviewer1 is randomly allocated section B while interviewer2 is randomly allocated section E.
However, I don’t know how to randomly allocate questionnaire sections in the design. Kindly assist with the appropriate syntax.
You are saying that interviewer1 must be allocated section B.
I don’t see any randomness here.
Randomness implies at least two possibilities (given all other circumstances).
What is the other possibility for interviewer1?
This looks like a contradiction to me:
the rest of the sections should be randomly allocated at one section per respondent. I wound like to assign a section at random to a given enumerator.
Let me try and make it clearer. For the sections B to F, only one section should be administered for each given respondent. However, the section to administer must be selected by randomization.
How can I do that?
If I understand you correctly, the section assignment should be per enumerator (for all of his interviews), not per interview?
In this case:
Since the interviewer is not identifiable by the interview, but is usually known at assignment uploading time, you could preload a random number between 1 and 5 (corresponding to sections B to F) to all assignments for each interviewer. That is you perform the randomization at assignment uploading time, not at interview time.
Exactly. This is if you interpret the contradiction as the Interviewer-based strategy should prevail.
If you interpret the contradiction the other way around, then this is a simple random selection of which section to show:
Hence, for section number # the condition should be:
It would be good if you mark which solutions you’ve tried and which one you preferred in the end, and why.
Klaus’ suggestion is also good, and is what has been practiced in Survey Solutions before the function to access the random numbers was introduced in v5.1. (October 2015). It corresponds more closely to what you described as some module for some interviewers, but clearly is not going to be responding to reassignment from one interviewer to another (should such reassignments subsequently take place).