Thanks a lot for sharing your concerns, they are very much on spot.
I am the guardian of the personal data and that is why in the repo there will be only an assigned numerical ID per respondent. Never a link between that ID and the real personal data. The other data present is not usable to identify anyone. So, it is already anonimized.
Just in case I checked with my director and everything is perfectly compliant in this way. :)
Thanks again...