Hi,

I'm entirely new with LimeSurvey. I've put up a questionnaire for my doctorate research in psychology. It has 12 vignettes, each one being a group that has 3 sub-questions. I found out how to present those 12 groups randomly, hurray! Now, I want to offer randomly to participants 4 different versions of the first sub-question of each of the 12 vignettes. The participants would randomly be presented with only one of the 4 versions. Is this possible and if so, how?

I hope I made this clear...I thought of doing other randgroups, but than all the 4 versions would be presented to all the participants, but randomly.

Does someone have an idea for me? Thanks a lot for your advice.

Hi,
first we should clarify:
You have 12 groups (you call them vignettes) which you display in random order. (by entering the same name in "group randomization name)
Each of these groups contains 3 questions (not subquestions; subquestions are e.g. the items of a question type "array")
But there are 4 versions of each first question. Only one is to be displayed.

To achieve this there are two options:
a. micro-tayloring - if only the question text is varied.
b. 4 different versions of this question. (G01Q01a, G01Q01b, G01Q01c, G01Q01d, and G02Q01a, G02Q01b,...)

In both cases you have to create a random number using the function "rand(x,y)".
Now you have to decide:
Do you want to display the same 1st question in each of your groups (always question "a"), or o you want to display always a randomly selected 1st question?

In the first case it is sufficient to create ONE random number at the start of the survey
and with the surrounding construct: {if(is_empty(randnumber),rand(1,4),randnumber)}
Otherwise create 12 random numbers:
{if(is_empty(randnumber1),rand(1,4),randnumber1)}
{if(is_empty(randnumber2),rand(1,4),randnumber2)}
...


If you use microtayloring enter the question text like
{if(randnumber1==1,"This is my text of version 1",if(randnumber1==2,"This is my text of version 2",if(randnumber1=,"This is my text of version 3","This is my text of version 4")))}

Of course it is is possible for answer texts as well.

If you want to use four different questions,
set the relevance equation of your 1st questions
G01Q01a: randnumber1==1
G01Q01b: randnumber1==2
G01Q01c: randnumber1=
G01Q01d: randnumber1==4

G02Q01a: randnumber2==1
G02Q01b: randnumber2==2
...


If it is always the same 1st question:
G01Q01a: randnumber==1
G01Q01b: randnumber==2
G01Q01c: randnumber=
G01Q01d: randnumber==4

G02Q01a: randnumber==1
G02Q01b: randnumber==2
...

Joffm

It is always recommended that you attach a prototype of your survey as lss export (only the relevant questions)

And - please - do not double post.

Another alternative way to show 1 question in 4

1. Create a group with relevance set to 0
2. Create you 4 questions
3. Set random group to you 4 questions
4. Move 3 of this question in the group with relevance to 0.

It work

Oops... here is the file.

Thank you both Joffm and Denis for your answers! Sorry for the double post, I've misunderstood the forums' functioning.

Right, I talked about questions, not subquestions AND I want to display always a randomly selected 1st question in each of the 12 groups.

I joined a .lsg file of one of the vignette groups.

I'll explain the goal and the structure I made, maybe there's another way to do it.

Goal: Explore knowledge and perceptions of Canadian mental health professionals toward workplace bullying in 12 clinical vignettes.

I had 2 variables to include in the vignettes:
- The most important one: the nature of the interaction (Implicit = I; Explicit = E; NonBullying = N);
and
- The characters' gender: consulting client and colleague (Man = M and Women = W)

So I made a total of 48 vignettes : 4 different versions of Implicit, Explicit and Non bullying interactions, each version in four gendered versions (Man with Man = MM, MW, WM and WW).

First, I wanted the participant to view in random order all of the 12 versions of interactions (4 I; 4 E and 4 N), but only 1 gendered version for each of them, randomly chosen by LimeSurvey. That would have given me 4 answers for a type of interactions (I, E and N), in randomly gendered versions.

QUESTION 1: Is there a way to do that ??

I could not find how to do that, so instead I chose only one of the 4 versions of interactions (I, E and N) in 4 gendered versions (MM, MW, WM and WW). That gave me 12 vignettes (IB-MM, IB-MW, IB-WM and IB-WW; the same with EB and NB).

Here is the questionnaire's structure:

PART 1 : 1 group of 9 sociodemographic questions (code SD)

PART 2: 12 groups of clinical vignettes, each including 3 questions
- first question includes the vignette and one question;
- 2 other questions (always the same for the 12 groups) explore secondary variables.

PART 3 : validated questionnaire of 9 items
PART 4: 3 final questions

I will try to apply your solutions to the vignettes' groups.

QUESTION 2: The 12 groups already have a relevance equation pertaining to Part 1 ((SD03.NAOK == "R1") and (SD04.NAOK == "R2")) to eliminate participants who do not meet the research criterias. So how do I proceed to add something else to the relevance equation?

Hope I gave you enough informations. Thanks a lot for your time!

Hi,
please only attach lss files.
Reason: lsq and lsg files are language sensitive, meaning we have to guess which base language your survey had, then create a survey with that language to import your group resp. your question.
But lss files contain all of your settings.

The 12 groups already have a relevance equation pertaining to Part 1 ((SD03.NAOK == "R1") and (SD04.NAOK == "R2")) to eliminate participants who do not meet the research criterias.

To screen out persons you'd better use a quota.

If I understand your explanation correctly, it is sufficient to have your 12 groups displayed randomly by giving them the same randomization group name.
And there is only a text variation depending on consulting client and colleague, you may use micro-tayloring in the first question as I showed in the blue part of my first post.
So the question text may be something like this:

{if(randnumber==1,"In order to complete his assignment, your client needs data from a colleague. Every time he tries to approach his colleague, the latter says he’s not interested in talking to him or that he has more urgent matters to attend to. Therefore, your client cannot progress in his assignment.",if(randnumber==2,"In order to complete her assignment, your client needs data from a colleague. Every time she tries to approach her colleague, the latter says she’s not interested in talking to her or that she has more urgent matters to attend to. Therefore, your client cannot progress in her assignment.",... )))}

How serious do you rate this situation or what the person experiences in this context?
(This is the general part of the question and therefore not part of the if-construct)


Best, you attach a lss export containing at least one of each interaction type groups and some text examples of the gendered variations.

On Monday I'll send a small example.

Joffm

Hi,
here the promised example.


To save some time there are only 5 vignettes included.
In the first one I used the text of your export. The rest is text to show what happens.
And please read the text of first text display question.

Joffm

By the way:
In your settings you set an ordinal scale for SPSS, what is absolutely correct.
But your answer codes are strings ("A1","A2",...). So SPSS will import as string variables -> categorical data.
Therefore you will not be able to calculate statistical measures and tests like mean, t-test, ANOVA, ...
The mean of "A1" and "A4" is not "A2.5".
So please change to numerical answer codes now or you have to recode everything in SPSS.