Hello,

I know there's an LS plugin called "updateTokenByResponse", authored by Chenu Denis. With this plugin you can save answers for specific questions in token attributes.

I'm looking for a solution for the reverse case, so to speak an "update Response By Token". In a survey there should be a (hidden) question, for which the value of a token-attribute should be saved as answer, after submitting.

To be precise, it is about seeing the "group membership" of the participants (stored in a token attribute) in the answers.

This is not difficult to achieve for non-anonymous surveys.
But in this case, apart from group membership, the survey must be anonymous.

I have already tried a solution using additional parameters in the survey URL, but it was not satisfactory.

Is there a solution for this problem, does anyone have an idea?

Kind regards
Roland

Of course not.
This way you would circumvent the anonymity.

You'd connect data of the participants table with the answer table.

So your solution with survey url should be fine.

Joffm

As Joffm says, if this would be possible in "anonymous mode" Limesurvey can not guarantee anonymity anymore, thus it could NOT claim to be technically anonymous.

However, there are different levels of anonymous and over years in Market Research we have analyzed data anonymously, despite in theory being able to connect personal data with response data.

So technically the data is not anonymous, because people could connect them if they want to, but in practices the personal data was treated separately and thus analysis was anonymous. However, with this approach you can NOT claim a 100% technical anonymity. You need your respondents to trust you.

holch wrote: …
So technically the data is not anonymous, because people could connect them if they want to, but in practices the personal data was treated separately and thus analysis was anonymous. However, with this approach you can claim a 100% technical anonymity. You need your respondents to trust you.

Totally right !

anonymous is a contract , not a technical system

Thank you all for your comments!
I should have mentioned that each electoral group always includes several people, so that the anonymity of the election is not compromised by indicating the group in the responses.
Furthermore, the participants will be informed in advance that there is a division into groups and to which group they will be assigned, and that the group property will be shown in the answers.

My solution with a modified URL basically does exactly what you rejected: information from the participant table (from a token attribute "Group") is built into the URL and then read into a question in the survey, and in this way transferred to the responses.

One could argue: the participants see the data ("Group") in the URL and can decide whether they want to participate. But participants could also change the group information in the URL, and that is what I would like to prevent.

Roland

We are not rejecting anything. We are just saying, if you could pass on data from the token table of an anonymous survey to the response table, Limesurvey could not claim that it is "technically anonymous". Because you could pass on an ID that clearly identifies a person, their name, their email address, etc. Thus the text that LS displays for anonymous surveys would be misleading.

What you are doing is basically to bypass the measures of an anonymous survey in Limesurvey and while the information that you are passing on might not be "dangerous" itself, it could be. Thus I find the impression of technical anonymity that you present to your respondents through the fact that your survey is labeled as "anonymous" rather misleading, because with the same bypass, you could also send a personal ID to your survey.

Actually, I think I will propose that anonymous surveys in LS should not accept external URL parameters, because it could lead to people saying that anonymous surveys in LS are not really anonymous.

Now, don't get me wrong. This is more a technical thing and something that I think Limesurvey needs to check, because it is not good if we give the impression that the survey is technically anonymous and you can bypass it that easily, if you are malicious.

What you are basically doing is what I described here, and which in my opinion is perfectly fine, if you explain it in the right way and don't say that there is no technical way to connect personal data with responses, because obviously there is:

So technically the data is not anonymous, because people could connect them if they want to, but in practices the personal data was treated separately and thus analysis was anonymous. However, with this approach you can NOT claim a 100% technical anonymity. You need your respondents to trust you.

So here my "feature request": bugs.limesurvey.org/view.php?id=16228

But thinking about it, I am wondering it is not actually a bug. I think a survey that is labeled "anonymous" by Limesurvey should not accept URL parameters.

With a "normal" survey, one could still pass on parameters and then define their "own" definition of anonymity. Which is totally fine, as long as things are explained clearly to the respondent.

But as Limesurvey, I think we should not label a survey as "anonymous" if it will accept URL parameters.

I think that it is really hard to conduct an electronic survey in a truly anonymous way.

With LimeSurvey you always have to trust someone, the survey administrator. He can do whatever he wants with the responses.

But if you truthfully explain to the participants of a survey with LS which of their data will be stored for a survey in LimeSurvey (in the participant table) and which of these data will be used in the answers, then you have provided the necessary transparency as a basis for trust.

Roland

I agree with you, Roland. In market research we have done this for years. We HAD the personal data to invite them, but we would guarantee that we will not link personal data with the response data (we would only check who has not responded yet and then send reminders, but the personal data would ever be connected with the responses for analysis purposes). So respondents needed to trust us (and of course they could).

But we could not claim that it is technically impossible to connect personal data with response data, which in someway the anonymous mode in LS does, I think.

If you explain fair and clear, what is possible and what is done and what not and people are OK with that, I don't see a problem. I need to read the text again that is shown for anonymous surveys, but I guess it would make false claims if this would be possible.