This is a very simple one. If the question is wrong, change it. There is no point continuing to collect rubbish data just for the sake of 'tracking'. It will not assist you in any way, in fact it will feed misinformation.
Just make sure that when reporting changes over time, the alteration to the question is noted. Indeed, some question changes will result in not being able to compare to previous findings at all. When analysing this information, I would generally recommend providing the previous tracking information, comment on why the question has changed, and then present the new data with a discussion on how the re-wording has enhanced the analysis.
If you must include the existing incorrect question, find a way to also include a revised question asked in the correct way. However, be sure to consider how ordering and placement of the question will impact on findings.
Just make sure that when reporting changes over time, the alteration to the question is noted. Indeed, some question changes will result in not being able to compare to previous findings at all. When analysing this information, I would generally recommend providing the previous tracking information, comment on why the question has changed, and then present the new data with a discussion on how the re-wording has enhanced the analysis.
If you must include the existing incorrect question, find a way to also include a revised question asked in the correct way. However, be sure to consider how ordering and placement of the question will impact on findings.
