Hello,

I hope everybody is fine.

I am in a dilemma, I am a beginner in this wonderful world of LimeSurvey and I hope you will be very patient with me.

I am conducting a survey where they ask me to make rotations, that is, that THERE ARE 21 ROTATIONS AND I MUST HAVE 4 OF EACH ROTATION AMONG USERS OF EACH BRAND.

Below I leave the question itself and how they ask me to be the rotations.

I appreciate your kind help.

Please, explain two more things:
1. I understood that there are these 21 "rotations", meaning the objects are displayed in exactly this order.
But I do not find the order of the first image in your "rotations".

2.

I MUST HAVE 4 OF EACH ROTATION AMONG USERS OF EACH BRAND.

As each "rotation" contains all brands it is not more than "Show 4 different "rotations"?
But this I would not understand. Why shall I answer the same question 4 times? Only with objects in different order.

If I read this question (P11) it seems to be just one question, where the objects are displayed in one of the 21 possible "rotations",
Please, clarify.

If it is really only one question with a randomly selected "rotation" you may use a random number,
or - if there are really more - the method "Show x questions out of y".
This is:
Create two groups:
1st group contains x questions
2nd group contains the rest and is hidden
All question get the same randomization name.

So, please clarify. And we can tell you more.

Joffm

P.S.
You see that I always wrote "rotation" in quotes.
This, because it is not a rotation.
Rotation means:
1. A B C D E F
2. B C D E F A
3. C D E F A B
4. D E F A B C
You see the order is always the same.

What you show is a selection (21) of all possible (5040) randomized sequences of the seven objects.

But don't care: It's just wording.

I really don't think this rotation approach is necessary. I can understand that you don't want to present the ingravings always in the same order, but having such a very exact distribution? Maybe the colleagues are not aware that you can present the subquestions in aleatory order? In my opinion this should already solve the whole problem. Maybe they have a very specific methdological reason for it, then ok. But generally I have the feeling that they are applying offline research methods to online research methods. Offline you need to make very specific rotations, you can't let the interviewer choose, because they will generally use some kind of pattern. Now with a programmed questionnaire and randomization, I think they fixed rotation plan is not necessary.

Also, what the at least 4x the same rotation of users among the same brand, that is basically impossible to control.

I think what you are asked to do is not possible to control with Limesurvey, besides creating separate questions for each rotation and then randomly apply the questions. This still does not solve the quota of min n=4 for each rotation. But if you have a high enough number of participants the random number should help you to get enough participants for each rotation question.

But as I said, I would discuss this approach with the one designing it. It sounds to me, that paper & pencil approaches are applied to a programmed questionnaire and I don't think it is necessary.

Hello Holch and Joffm

I really appreciate your answer, I explain a little better.

I have 4 brands to evaluate.

Mark 1
Mark 2
Mark 3
Mark 4

And I have 21 rotations.

What I am trying to do is that:

4 people answer the rotation 1
4 people answer the rotation 2
4 people answer the rotation 3
4 people answer the rotation 4
4 people answer the rotation 5
4 people answer the rotation 6
4 people answer the rotation 7
4 people answer the rotation 8
4 people answer the rotation 9
4 people answer the rotation 10
4 people answer the rotation 11
4 people answer the rotation 12
4 people answer the rotation 13
4 people answer the rotation 14
4 people answer the rotation 15
4 people answer the rotation 16
4 people answer the rotation 17
4 people answer the rotation 18
4 people answer the rotation 19
4 people answer the rotation 20
4 people answer the rotation 21

And this for each brand to evaluate. that is to say:
84 people will evaluate me to mark 1
84 to mark 2
84 to mark 3
84 to mark 4

Hopefully this time it has been clearer, (sometimes that is difficult for me)

Hello,

Reading some parts of the forum I found this to be able to randomize 2 groups of questions

if (is_empty (randnumber.NAOK), rand (1, 2,), randnumber.NAOK)

But I would like to know if with this type of equation I can perform the type of rotation I am looking for.

I appreciate your comments again.

As @holch said,
with built in features you will not be able to control exactly this
4 people answer the rotation 1
4 people answer the rotation 2
...


Either you do it the "normal" possible way, creating a random number (1-21)
{if (is_empty (randnumber.NAOK), rand (1, 21), randnumber.NAOK)}
and displaying the different questions by relevance equation
Rot1: randnumber==1
Rot2: randnumber==2
...

Okay, you will get a little overquote (5 answers in one rotation, or so. But what? Remove randomly these respondents.
Or you check the progress and the filling of the rotations.
If Rot1 is full (4 respondents) you may switch the randnumber==1 to another rotation that is a bit low
Like
Rot1: randnumber==0 (never selected anymore)
Rot13: randnumber13 OR randnumber==1
This way you are able to control, though it is time consuming, and difficult if a lot of people answer the survey at the same time.

The next option is more sophisticated:
You may use an ajax call to a remote php script.
This script queries the database and selects the least filled rotation, whch is returned to the survey.
Here you need access to the database and some knowledge in php / MySQL.

But even with this approach you will NEVER be able to get this quote exactly.
Why?
Imagine the following situation:
All rotations are filled, except rotation 12 with 3 respondents.
Now a respondent starts the survey and is leaded to rotation 12.
Another one starts the survey and also is leaded to rotation 12.
This because the program can only use data of completed surveys. Who tells you that neither of these two will terminate?


Joffm

I repeat @holch's question - and I agree, this is a typical design of a paper-pencil questionnaire.
Why is it absolutely necessary to stay to these 21 rotations? Why is here in this online survey a - built-in - randomization not possible?