Hi @seglea, There might be a case that the incorrect option answers were later on changed to correct ones but weren’t published so the latest changes won’t be visible to the respondents taking the live survey. Please make sure that the current survey version is being published.
I’m not sure if there is any way to guess the answers that the respondents were trying to give (in case of incorrect answer choices) but you can manually edit the collected responses, but please do it under the guidance and agreement of your team member/colleague working on this project/questionnaire, as editing the responses manually can be risky.

Thanks for your suggestion, omakrewat.  However I don’t think that’s the case, because we trialled the survey before distribution (and I have just trialled it again) and the correct response options are offered.

@seglea,

I’m guessing your turned on Variable Naming (under Recode) on the questions you are having an issue with, then changed the scale after you did that. “Click to write answer 4” is probably one of the “Strongly...” answers.  If that’s the case, the good news is you just need to edit your Variable Names and republish and all your data will be fixed. 

@TomG, many thanks, you are exactly right about the Recode (I’m baffled how that happened, but that’s another problem).  And republishing the survey fixes the data within Qualtrics.  But then when we export them to .csv, the old responses are found in the .csv file.  Is there some kind of refresh I have to do to get them to Export?

@TomG … I have now resolved this, by making and then reverting a nonsense edit in a text field so we could republish a second time; and the Export is now coming through correctly.  Many thanks for your help - we would not have managed to resolve this issue without it.

Alas, I spoke too soon.  Two days later, the survey has reverted to spitting out the same incorrect data on Export.  I’m going to raise this with Support