I am creating a questionnaire with selection of people according to specific criteria. For that I am using the public demo hosted in this test questionnaire
This questionnaire performs what I need, but there is one issue that concerns me. The internal variables change every time I modify the list of members and enter the information of the people (necessary to perform the selection).
This is an issue for me because it can be used as an element of survey spoofing. Where the surveyor does the draw, and the selected person is absent or does not want to participate, and can change the elements at convenience. So, my question is can I “lock” the internal variables that do the selection with the “first thing that enters into them”?
In summary the flow would be: I enter the participants, the draw is performed, the internal variables are filled in this process, block the answers of these internal variables that are already “filled in”. In this way, even if you change the previous elements (the internal variables) the first selection would be maintained.
Is there a way to do this, perhaps by using more complex expressions in the selection variables?
To my knowledge, there’s no way to “lock” variables, per se.
However, there might be a way to defer their evaluation. For example, imagine that an interviewer lists members of the household and only see the selection(s) after pressing a “Done” button.
Random selection is one of my many weak suits. Undoubtedly others will have better solutions.
To avoid this, require the members to be listed in a particular order, e.g. from oldest to youngest, so that if they are not in that order, you know it has been manipulated.
(This is not my solution, of course, see Leslie Kish). There is a worked out example here. And of course you don’t have to use the Kish grid for selection in Survey Solutions. It was developed in the times when interviewing was done on paper without any electronic tools to generate random numbers: see L. Kish, ‘A Procedure for Objective Respondent Selection within the Household’, Journal of the American Statistical Association, 1949.
Asking a separate “DONE” question is not a guarantee against spoofing. After the first entry and pressing “DONE” the interviewer will know the position of the selection (for example, 4th person out of 6) and still will have a possibility to go back to doctor the list of the members, and press “DONE” again.
Other mechanisms must be used to detect and address such behaviors.