As I wrote:
In "eqNames" you have to change these two lines from "red" to "blue".
And do NOT copy them from this forum thread. They will include HTML garbage.

I displayed the entire code to show that it is 11 times the same procedure.
Have look at the manual to understand, what the functions ("rand", "strlen", "substr", "str_replace") do to better understand the logic..

Addition:
And if you switch back to "question by question" you have to move
"selNames" before "eqNames"; and of course hide these question after you tested.

Joffm

I got the names working perfectly earlier. I had 11 names appearing at random as desired, as well as randomiziation of task type. I recreated the same kind of code for the behavior task (note, this is split into three groups, with labels N, P, or B plus the corresponding number). I checked if the system was working early into inputting the N behaviors, and it was working perfectly, doing exactly as I wanted (the same as the names, selecting a predefined number n at random from a larger number of possible choices N). Once I finished inputting all the N behaviors, I went to check again how the survey itself was working, when I discovered that the behaviors went from being selected as desired, to having selBehaviors_999 being selected, to having nothing at all selected. Further, the names are also now messed up; before, the list at the beginning indicated which names had been randomly selected, and these mapped nicely onto the labels I had applied to each name (numbers 1-9 or letters a-x). Now, the 'selected names' list is just random strings of letters/numbers/symbols that don't map on, and the number of names that appears within the survey itself varies from none to 5. 
I wonder if you could take a look and see where I went wrong on the coding? I'm trying to use ChatGPT for suggestions but its hard to understand what it's telling me to do 
 

Oh, something quite different.

At first: syntax errors.
And you copied the equation, didn't you.
Look what you got
 

And the second?
How many elements do you want to show of your bejhaviour questions?
5? There are 15 subquestions (5 x 3)
6? You join 6 numbers
7? You calculate up to 7 numbers
 

Now: Logic.
Remember, the idea of the logic was
B1, B2, B3 and Graph are shuffled around the Names  groups.

Fortunately there are 4 groups.
Because all are in the same randomization group and only one of the behaviour groups is displayed, there can be
2 groups before the names groups
2 groups after the names groups
1 before and 1 after the names groups

Here one outcome of the shuffle.
May be like this
 
If the random number is equal 1, the order is "Graph - Names - B1"
If the random number is equal 2, the order is "Graph - B2 - Names"
...

But in this your new survey design you again have three questions in each Behaviour group.
As in your very first approach this doesn't work.
Again you try to do a "group in group" approach

You can't display the "Names" groups and the "Behaviour" groups in random order.
What you can do is:
1. Use only one question per behaviour; use an question of type array.
Here a real nice solution, though with javascript
 
You insert some subheaders for your questions.
And to extend the first questions to seven points shouldn't be an issue.
I already complained about the silly 5-point-choice

As you can't use javascript, you may style the left side of this array with some css,
roughly like (without better styling)
 

2. You use micro-tayloring.
This means, you do not shuffle questions but only the question text according to your randomization.
This will shorten your survey immensely, but - especially in your case - is a huge amount of work to restructure your data before you are able to analyse.

Only a quick answer.

And I do not know, why you don't create a small prototype the test your ideas, especially when you implement something different.

Joffm

BTW:
What's the reason to randomize "Behaviour" and "Names".
Is it in your hypotesis that there is an influence if the respondents see the names first vs last?
But then you should control this. At the moment you do not know in which order the questions and groups were displayed.

I'm trying to use ChatGPT for suggestions but its hard to understand what it's telling me to do 

Oh, yes, do it.
This is usually a source of endless laughter from our side when ChatGPT talks about LimeSurvey.
It always sounds pretty logical, but it's usually just outrageous nonsense.




 

You see, all your "Behaviour" groups and "Graph" are in the same randomization group (RG2)
So "Graph" may appear somewhere in the middle of your "Behaviours".
Is this what you try to achieve?

And here a slightly better styled array


Joffm

Hi Joffm,

I'll try to answer everything from your last two messages.

Indeed, I copied the syntax, but I since have managed to fix the issue and all is working as I want it to.
I hear your concerns about the randomization and I understand that if the individual questions are to be randomized, the overall tasks cannot be. I think it's more important to randomize the questions themselves (though this needs to be passed by my supervisor so could change). Doing it on a question-by-question basis also seems to have the benefit of keeping each trio of questions together (which is perhaps the most important thing), as well as making it more straightforward to present the instructions before each relevant task.
Unfortunately, the question design is fairly rigid, e.g., when there's a five point scale it needs to be 5 points, not 7. This is to do with how people interpret the question differently. I agree with you in terms of presentation but its important to keep the scales as they are. 
And the randomization issue I think I've explained above, but essentially all tasks have a set up where there is a group of questions that must be presented together but only a selection (e.g., 11/33) of the questions will be shown to participants. And so this leads me to my question :

I followed the example you provided for the Names to apply the same logic to the Behaviors task with great success, and now I am trying to do the same kind of thing to the Graphs task (this task follows a similar set up except that there are 6x graph types, each with a positive and negative version. I want participants to be shown one version of each 6 graphs. I have it set up currently so that there are different question groups for each graph version (12x total), and am trying to apply the randomization so that for example, graph1positive gets the randomization group label RG5, and then recieves the strpos(selG1pos_999,'X')>0 that was used in the other tasks. I also altered the code in the equation question accordingly and set up a selGraphs question block too. The issue is that while one graph is being selected, as I want, it is not being displayed in the actual survey itself. This is inspite of the fact that I can see which graph has been randomly selected on that very first screen labeled Only for your information that you put in the last time.

I know you said ChatGPT is useless, and it can be quite unhelpful sometimes, but it helps when your knowledge of coding is negligible. So I had asked, and gotten some useful tips with the behaviors set up, and decided to try asking it about the graphs also. Unfortunately again its' proving quite useless, though it did say something about it being unnecessary to provide the same kind of coding as for the other tasks, since they selected a larger number of question groups, and with the Graphs task im selecting only 1 out of 2 for presentation. Do you have any insights on that? Or is that just random gibberish? 

I think you can safely ignore anything that ChatGPT tells you in terms of Limesurvey. If something was correct, it was out of sheer luck on ChatGPTs side.

Don't get me wrong. I use ChatGPT too. But as it has the tendency to invent things when it has no information about (instead of just saying: "Sorry, I have no idea"), it proves to be more dangerous than helpful in many scenarios. Especially when you have no clue about coding, it can tell you anything and it sounds so convincing and confident, that you will just believe it.

When you have decent knowledge about something, then ChatGPT can be helpful to accelerate processes, but if you have no idea of the topic, it is just too dangerous.

I don't think that ChatGPT was ever trained on anything that has to do with Limesurvey. So it probably just takes its knowledge from other areas and tries to apply it to Limesurvey, without sounding like a human that would let you usually know that he/she doesn't exactly know if it is correct. ChatGPT just assumes and states these assumptions as facts.

So, please, be very, very careful with anything ChatGPT and Limesurvey, at least for now.

So, what design do you achieve now?

As far as I understand, you

• select randomly and also display randomly 11 out of your 33 "names" groups (each containing three questions)
• select randomly and also display randomly x out of your y "behaviour" groups (each containing three questions)
• select randomly something of your "graph" groups (6 positive vs. 6 negative  or 6 graphs randomly positive or negative?)

Well, this way you can't randomize the groups "names", behaviour", "graph".
If you really think it to be necessary, you may create 7 surveys
6 surveys that display these three "groups of groups" in one of the 6 possible orders.
1 survey that only creates a random number  from 1 to 6 and in the end-url links to the matching survey.

As usual:
If there are still questions, show what you did by lss export.

Joffm

LimeSurvey Cloud version: 6.4.4
Survey theme/template: fruity twentyThree
============
Hey, Joffm,
I've read your discussion and tried to understand the basics of randomisation.
You gave me an advice in another topic (about the mismatch). I followed the suggestion of tpartner and removed group randomization. And once I did that, the survey ran correctly.
However, the randomization and the conditioning of question. You proposed not to use randomization function built-in Lime Survey and write a code.
So, I am not familiar with Java Script. I ran the survey you've wrote for AideenW and found that I have the same type of question he had: Q1 (Tamms) -> Q2 (TammsReminder) + Q3 (TammsJa). 
I linked the necessary question to the target ones but they appear only occasionally. Despite the logic is ok as well as syntax, the survey doesn't work correctly. Is it the problem of condition? Do I have to write a condition manually to make cascade questions emerge?

I saw you stated the "relevance" to the questions. Do I have to apply this option to every linked question within the group?

Thank you for advance!
 

Hi,
where did I propose not to use randomization function built-in?

Well, 
if you want to keep these three questions together, but display these "groups" in tandom order,
you have to put them into several gropups and add a "randomization group name" on group level.

Best you create a small example where you can test and study what "randomization group name" means.

Joffm

Yes, you're right you haven't wrote that. I misunderstood.

I reread the topic. Aideen had similar problem and the creation of additional groups seems to be the only solution.
But it will be too much groups because of the number of stimuli.

Is there no any other (more optimal) solution?

Because wthout randomization the survey works well and I supposed that conditioning (if "yes" - show the question) ties the one question to another. In some question it even works...

Thank you for your time and pieces of advices in advance.

Danił.

Hi Guys,

It's been a while- but I've been busy! My survey now works perfectly (you are right in your description of my design, @Joffm).

I hope this will be the last question that I have (As I am nearly ready to publish it). I have a consent form at the very beginning, and it's very important that people who do not consent (i.e., select 'no') not be shown any more of the study, barring the final debriefing message that thanks participants for their time. How would I set this up?

Kindly,

Aideen

P.S., I'm a girl

Also- is there a way to test how the survey will appear on smaller screen (i.e. phones, tablets) before I publish it??