Correlate responses after anonymous setting?
A survey I created was originally set to anonymous. However, following the responses received, a colleague wanted the results correlated by attendees. Is there anyway to re-identify without creating a new poll?
Unfortunately, there is no way to correlate the responses to anonymous questions and we do this intentionally.
When we say “Anonymous” we mean anonymous. If you have a question earlier in the presentation/survey that is identifiable such as “what is your name” and we correlate an anonymous response with this, then we are no longer anonymous. There’s just too many dangers and ways it could be abused and we owe our company’s success on both the Presenter and the Participant. Both must be protected. It’s the American Way.
So, while we understand where you and your colleague are coming from and you probably have only the best intentions, we just can’t rely on it.
Let us know if there’s anything else though! We appreciate the feedback and ideas
@Brian - I totally understand your concerns, and I agree that the Participant's Anonymity must be protected. At the same time, the ability to correlate data between polls would open up a whole new dimension of value from your service, and I believe there is a creative solution to the problem of protecting anonymity while empowering correlation between anonymous polls.
First, an example business case: I ask the audience what their state of residence is, and then I ask them if they believe racism is a problem in America. By being able to correlate data, I can now tell how state residency influences perspectives on racism in America. Of course, if I only have twelve people responding to the poll, then I'll be able to identify respondents by their response, which is why you guys want to be so careful about this.
Possible solutions: When correlation is turned on, include a warning message to pollees that this is correlated anonymously, but if they reveal any information that would allow them to be identified, their anonymity could be compromised, or even a warning that says, "While individual responses are anonymous, your identity could be determined through correlation of responses." Additionally, you could allow the pollee to opt out of correlation and then the admin can see the % of respondents that allowed correlation so they know whether their sample set is large enough to be statistically relevant. And I'm sure there are more ideas. I think this one is worth developing a solution for. I would pay more for this feature.