I’m implementing a survey with a household roster from which one respondent will be randomly selected. I’m wondering if there is a way to prevent the interviewer from changing the selected respondent after selection has occurred? Currently, interviewers can simply add another person to the roster after selection has occurred. When this happens the selection is done anew, potentially changing the selected respondent. Is there a way we can program the instrument to prevent this from happening? Perhaps locking the roster after it is complete?
No. In part because the situation you’ve described is entirely plausible, and as a result of the detailed interview a new person may be mentioned who qualifies as a household member.
One tool you have is a screening question “Are you done with the list of members?”, which could be as simple as that, or be a dozen of probing questions to make sure the list is final.
Another is make randomization less transparent, scrambling and shuffling the individuals before picking.
In any case, the paradata will reveal such a behavior and you can warn your interviewers. Often, knowing you have tools to detect this is sufficient to deter them from doing such things.
There is nothing simple you can do here literally because the interviewer may discard the whole interview and restart from the beginning.
Instead of randomness, you could try a deterministic rule of selection, which would be easy to check.