Hi,
In a survey I recently did, I wanted to implement a conjoint experiment. For that I presented my respondents with a couple of words that were randomly selected from lists of words and let them make a choice that was dependent on which words were randomly selected. I implemented the random selection of the words by adding a hidden equation type questions and later displaying the variables from the equations.
Unfortunately, there is an issue in Limesuvey where Limesurvey runs the random assignment once, when the page on which the random variable is generated is displayed and again when the answers on this page are validated. Since I put the hidden equation questions and the question were randomly assigned words where displayed on the same page, the random variables that were displayed were overwritten when the answer was submiteted.
I know that there are workarounds for that (for example here [url] forums.limesurvey.org/forum/design-issue...ng-survey-using-rand [/url]). Unfortunately I was not aware of the problem before I activated the survey.
Do you know, if there is a way to retrieve the answers that the respondents were presented? The one thing I can think of, is to go through the Survey with the same random seeds that the respondents had, but I don’t know, if there is a way to manually set the random seed in Limesurvey.
I am currently using Limesurvey 3.28.5 and it is hosted on a cloud server. 

Thank you for your help and all best

Mathii wrote:

Do you know, if there is a way to retrieve the answers that the respondents were presented?
 

There is no workaround documented to regenerate the random numbers uses during the interview.
I have not seen any tool or routine to get the series of random numbers from the saved seed in the response set.

If you survey allowed the respondent to skip back and reanswer questions, the random number might change more than once and you will not know which random number in the static series was used.

As far as I know, the seed is used to randomize positions of groups questions etc. It is not for the randomizations generated by ExpressionScript.

Unfortunately I see no way for you to retroactively determine the ExpressionScript randomizations seen before page submission.

Oh man, that's too bad! But thank you both very much for your replies! I will make sure to use the workarounds to stop the double assignment in the future.

What kind of conjoint did you conduct with LimeSurvey? Looks like a lot of work with LimeSurvey (which offer no tools for doing e.g. CBC or MaxDiff experiments).
Even to get the look and feel is not that easy. Nor having more than one design of experiment in the survey.

I conducted (or rather tried to conduct) a forced choice conjoint, where respondents were presented two options and had to choose one. Both options had several attributes, each of which was randomly assigned independent of each other. Both options could therefore be made up of any combination of attributes.
Making it was a bit impractical but realativley sraigth forward. I yust had to add a lot of equation questions (Two per attribute per experiment). The look was not great, but I think it was okay. I was able to get the Radio List question working in a table layout that is adaptive to different page widths.
When I tested it I had it in question by question view and it worked fine. I only changed it to group by group relatively late and did unfortunately not test, if the displayed options matched the saved variables again after that change.