How to allow survey participants to contact me without breaching anonymity?

Question

I started a survey yesterday as a part of an empirical experiment, using an online questionnaire. It is among the employees of the institution where I work to pay the bills (that's not the university at which I'm doing my Ph.D.).

Soon I not noticed that there is a problem with the survey. 2/3 of the responses are empty from page 3 on. These are not people who chose to hit the "exit and clear" button, because then the responses won't land in the database. They either closed the browser window, or had a technical difficulty.

Totally stumped at what is happening, I set up a second "survey" with only one question: What problems are you experiencing with the original survey?. I sent it out asking people who experienced a problem to tell me about it. And now I have three attempted answers there - empty.

Now I don't know how to let people tell me about the problem if this isn't working either. If I ask them to tell me per email, I will see who is sending the mail, breaching anonymity. The survey software runs on a university server. If I tell them to write to the university admin, they won't trust him because they don't know him and cannot know he is not going to give out their names. I don't know who I could ask from our institution to assist - nobody here is part of my Ph.D. project, and nobody has any responsibility for the university server. If I ask somebody close to me to act as "problem relay" only as a personal favor, the trust of the participants will be eroded again, because people close enough to me to do me this favor won't be seen as impartial.

Any ideas how to get the problem reported? It doesn't matter if it is technical or the questionnaire is so unreasonable that 2/3 of the participants change their mind about completing it. I need to know what is wrong, as this survey is pivotal for my dissertation.

Answer 1

This answer may depend on how many participants your survey has in general - but I think one of the main objectives of conducting a survey "remotely", online, rather than personally, in a controlled environment, is to gather an exceptionally large amount of participants. Therefore, I'll assume a sufficiently large number.

Note that online surveys are quite an exception here in that you never get into contact with the participants. In general, you always know to some extent who participated in the survey (hardly anything else is possible when you have to meet in person with the participants). And even for online surveys, contact details are often required if there is any compensation for the effort of taking part, or only the offer to be notified once the results of the study are ready (which is generally the least that should be offered to participants in response to the time they reserved for you).

Therefore, the anonymity assured to participants generally does not mean that researchers never learn about any of the personal data of the participants; it means:

• Responses will not be connected to the personal data of participants.
• The personal data will never be published in detail, just as aggregate values (e.g. age minimum, maximum and median across all subjects).
• Any identifying data (e-mail addresses etc.) will be erased as soon as possible.

Unless there is just a small total number of participants, participants can feel reasonably safe that even though the researchers know both their personal data and their responses, those two won't be connected in the future.