According to the questionnaire, respondents can answer open ended questions. Why are these variables not in the datasets?

By default, we do not provide variables with texts entered directly by the respondents because we have to check it separately for sensitive information that could enable re-identification. As this can be a very time-consuming process (depending on the variable), we only do this on request and, if possible, we then make the variables available to data users after an appropriate revision.