I would like to administer an anonymous survey where I ask a respondent to answer the same questions before and after an intervention.

How do I design my survey so that I can link the pre and post data for each respondent while still remaining anonymous?

That's not possible if you want to be truly anonymous. In anonymous surveys there is no connection between the token and data tables and time-stamps are not recorded.

I agree with tpartner. Truely anonymous is not possible if you need to compare responses from one wave to another for the same person. Somehow you need to identify this person.

However, how do those things happen in market research? We guarantee respondents that we do not use their personal data to analyse (and of course we do not pass this information on to the end client), which means their responses are basically anonymous as we will only use some information to connect both data sets (e.g. an ID), then the personal data is seperated from the actual responses. Could we make connections? Yes. Do we do it: NO! Do respondents have to trust us: Yes.

Within the market research companies, there are usually processes in place, that guarantee a maximum of anonymity. As soon as possible the personal data will be deleted from the data set. we are following a rigid code of conduct and thus do not attribute any personal data to the results.

One thing: Of course you have to be clear upfront. You have to tell them that their responses will be treated anonymously. You can't say that the survey is completely anonymous, because it isn't.