Hi all,

Thanks for Limesurvey and his dynamic community !

Here is something I can't do with LS :
To avoid asking 15 questions to a large sample, I 'd like to ask only 3 out of the 15.
How can I do it with limesurvey ?
I checked in the manual, there is an example with relevance status. I can't make it works, I probably missed something.
Could anyone give me a hint ?

Best wishes,

Matthieu

Hi Mathieu,

A hint: Type the relevance equation in the question_text (between brackets) as a way of debugging.
That may help.

Not much we can do with the information given.
Maybe a screenshot?

Cheers,

Thank you Gabriel.
Here is some more information. I've got a survey made of 1 group of 15 questions.
But I don't want all the participants answer all the questions. For each participant, 3 questions (out of the 15) should be randomly sampled and submitted.
I tried to insert a formula like {sum(Q1.relevanceStatus, Q2.relevanceStatus, Q3.relevanceStatus, ...., Q15.relevanceStatus ) LT 3} as a question validation equation. But Limesurvey2.00 does not seem to understand.

Can you see where I'm wrong ?

Sincerely,

Matthieu

Did you follow any guide to use that approach?

I tried to insert a formula like {sum(Q1.relevanceStatus, Q2.relevanceStatus, Q3.relevanceStatus, ...., Q15.relevanceStatus ) LT 3} as a question validation equation.

That should be a relevance equation (without curly braces), not a question validation equation.

That works !
Thanks for the suggestions !

Just a summary of the solution to make it clear.

Here are the guidelines for an example of randomly selecting three questions out of 5 (Q1, Q2, Q3, Q4, Q5):
1. Put all potential (here 5) questions in the same group
2. Ensure questions will be asked in a random order
3. Define for each question a relevance equation based on the relevance status :
A question relevanceStatus as a value of 0 if it has not been asked yet and 1 if it has been asked
e.g. for Q1 sum(Q2.relevanceStatus, Q3.relevanceStatus, Q4.relevanceStatus, Q5.relevanceStatus) LT 3
In this example, summing the relevance status will ensure that maximum 2 questions have been asked before Q1.
Note in each relevance equation, the codes of all questions except the currently edited question appear.

I hope it will be helpfull for someone else (Thanks again for the help received).

Thanks for sharing your feedback. Can you attach your sample survey (LSS file) or grup (LSG file)?

Do you use question by question mode or group by group with all those 5 questions being within the same group?

Here is the saved group with the random selection.

Hello,

I know it has been a while since there was a post in here, but I was having the same problem. Thankfully, the solution given here works. However, my survey has an additional requirement: Every question needs to be shown on a single page (question by question). Sadly, the solution does not work with that mode. Which changes would have to be made to solve this?

dieterdubinsky wrote: However, my survey has an additional requirement: Every question needs to be shown on a single page (question by question). Sadly, the solution does not work with that mode. Which changes would have to be made to solve this?

Every question of this random selection part? Or every question of the whole survey?
You might try if the workaround here works with one question per page mode.
I don't see a way to display the random selected questions one per page. LimeSurvey isn't offering subgroups or separate page separators.
But that would be needed to distribute the random-selection on different pages.

It looks like the relevanceStatus is only generated after page load so my solution would be as follows...

1) Run the survey in group-by-group mode

2) Place each question in a separate group

3) In each group, precede the visible question with a hidden equation question (equation1, equation2...) containing something like this (varies by group):
4) Place relevance on the visible questions, referencing their corresponding equation questions something like this:

Sample survey attached:

Good idea.

Since "one question per page" was described as an "additional" requirement I have to point out that one requirement is no longer fulfilled with Tony's workaround: "Ensure questions will be asked in a random order". The order won't change.

Progress-bar should be corrected manually. Hopefully the progress bar routine will change behavior one day in the future. Counting groups is often misleading.

"Ensure questions will be asked in a random order". The order won't change.

Did you try my sample survey? The group order is randomized so, effectively, the question order is randomized.

I tried is a few times. Perhaps I didn't tried it often enough? Or I looked at the wrong indication number? Numbers were always rising.
Or there is a bug in demo.limesurvey.org.

I see that you used a randomizationgroup "randGroups1". So it should randomize the order of the groups as well.