LimeSurvey Cloud Version 6.8.2
Hi,
I would like to create a questionnaire consisting of 10 questions, with the first two questions being the same for all respondents. Starting from the third question, I would like to introduce randomization:

• Respondents would be randomly assigned to receive one of two versions of the questionnaire (50% Version 1, 50% Version 2).

However, I would like this randomization to be stratified based on the responses to the first two questions (e.g., gender, age, or socio-professional category) to ensure that the groups receiving Version 1 and Version 2 are balanced according to these characteristics.Could you please confirm if LimeSurvey allows such stratified randomization? If yes, could you provide detailed instructions on how to set it up?Thank you very much for your assistance.Best regards,

Hi,

Could you please confirm if LimeSurvey allows such stratified randomization? If yes, could you provide detailed instructions on how to set it up?

Before I try to find a solution, please provide the lss export of this part of your survey.
How many categories of gender, age, ...?
And please explain what exactly you mean by "balanced"?

What comes to my mind at first, is "statFunctions".

Let's see

Joffm
 

Let's elaborate my idea.
Assuming
2 gender groups, coded 1,2
3 age groups, coded 1,2,3
3 socio-professional groups, coded 1,2,3
there 18 possible combinations.

Calculate it in an equation (combi)
{sum((gender-1)*9,(age-1)*3,socio)}

With the function "statCountIf"
{statCountIf(combi.sgqa, combi)}
count the answers with this calculated combination.
Now: If it is odd dislay "version 1", if it is even display "version 2"

Joffm

Hi

Thank you for your reply.
I can't send you the LSS file (confidentiality issue).
When I say “balanced”, I mean that each of versions 1 and 2 must be proportioned. In other words, I want to ensure that at the end of my survey, there is no over-representation of people aged 18 and over compared to people aged 35 and over. I want to know if I should set quotas and use your randomization method.

Hi,

I can't send you the LSS file (confidentiality issue).

We never want the "rreal" survey.
But you'd create a sample to show the secenario.

But it's still not clear.
What I understood is that you want to get (example)
50% version 1 (young, male, socio 1)
50% version 2 (young, male, socio 1)

50% version 1 (old, female, socio 2)
50% version 2 (old, female, socio 2)

50% version 1 (mid, male, socio 3)
50% version 2 (mid, male, socio 3)

for all combinations of your three attributes.
This is what I explained.
It's only some kind of equal distributiuon of version 1 and version 2.

But there is no guarantee that you won't get 70% young people, 15 mid and 10 old.
Here you have to use quotas for all possible combinations.

Joffm