Please help us help you and fill where relevant:
Your LimeSurvey version: LimeSurvey Community Edition   Version 6.3.4+231108 
Own server or LimeSurvey hosting: Own server
Survey theme/template: TW_Main
==================
Hi! I'm trying to set up a survey where participants will see the same set of questions twice, but will only be able to select answers the second time through meet certain criteria. 

Someone on here helped my coworker set up our TrueWiring DISC survey a few years ago and this survey uses a similar model. I'm using an array by column format: type H. For each question, there are 4 answers that correspond with 4 overall categories. Participants will get 4 answer options and they need to select the one that they most identify with and the one they least identify with. After the first pass through all of the questions, I use equations to figure out how much participants responded to each of the 4 categories answer options fall into and category with the most "most" responses will be their number one category and the one with the most "least" responses will be their fourth or lowest category. On the second pass, I want to be able to have participants only see the 2 answer options that come from the categories that were not selected as their overall most and least. I will then use these answers to calculate their 2nd and 3rd place categories. 

I hope that makes sense. 

Any help would be much appreciated! Thanks! 

 

Hi,
can you please explain this

On the second pass, I want to be able to have participants only see the 2 answer options that come from the categories that were not selected as their overall most and least.

So let's say:
In "Pass 1 - Coordinator" the respondent selects
most: You want to designate someone to lead  (A)
least: You have identified someone to lead (E)

Now the overall most is R and the overall least is X
Does it mean in "Pass 2 - Coordinator" again you display these two items (A & E)?

Joffm

I actually want the opposite. My hope is to do the first pass through all 14 questions and find the overall scores for A, E, R, X. If A scores the highest "Most A" and then E scores the highest "Least E", then on the second pass, I only want answer responses for R and X available to determine those scores.

It's possible that someone will choose R and X responses in the first round but then will see those same options to choose between in the second round.

Am I making sense?

That's what I thought and you confirmed now.
Would be rather easy with a different question type.
Array per column neither supports relevance in any of the axes nor array filter.

As long as you do not use the "MaxDiff" script you have to validate that "most!=least".
Use an array and validate "two subquestions with different answer options".

I'll create a small prototype later.

Joffm 

Here's a small sample
 

Don't forget to hide the questions of group GCalc

Joffm
 

If you use a "normal" array as pass 2 (this is sufficient as it is a 2x2 array) you may use the array by column in pass 1.

Sample 
 
I added the script "MaxDiff" in pass 1 which blocks the same selection of the same item in "most" and "least"


Joffm

Thanks! This has been super helpful.

I set up the survey, but it's telling me I am not responding in a valid way once I get to the group 2 questions. It doesn't matter how I respond. If I pick one most, one least it's invalid, if I pick both as most or both as least it's also saying invalid.

Additionally, would I need I second group calc after the second round of questions to determine the highest and lowest score of the group 2 questions?

Thanks again!

Sorry,
add  ".NAOK" to the validations of the second pass -> unique(self.NAOK)
And validate all of them!

would I need I second group calc after the second round of questions to determine the highest and lowest score of the group 2 questions?

I do not know your goal.
Do you want to use this in the survey?
Is it sufficient to calculate during analysis?

On the other side it is rather easy.
Nearly the same as the first calculations.
Now you know how to use the "countif" function
Like
Most2A : {countif("SM",Q1b_A.NAOK,Q2b_A.NAOK,Q3b_A.NAOK,Q4b_A.NAOK)}
...
Least2X : {countif("SL",Q1b_X.NAOK,Q2b_X.NAOK,Q3b_X.NAOK,Q4b_X.NAOK)}
And the rest.

But did you consider how to handle ties?
You may get a distribution like this
MostA: 5
MostE: 3
MostR: 1
MostX: 5
What to do?

Joffm

By the way:
 {if( 0 > round(sum(((15 - LeastA)/15)*100),0), 0, round(sum(((15 - LeastA)/15)*100),0) )}

What do you sum here?
sum(((15 - LeastA)/15)*100)

How can this result be less than 0?
if( 0 > round(sum(((15 - LeastA)/15)*100),0),
You only have 14 items (in your sample survey), so LeastA can be 14, but not more, so the least (rounded) value is 7.
 

Thanks for clarifying on the second pass on how to fix the validation.

We would use the survey to generate a report so I imagine the second calculation would be helpful as well. I will ultimately need to know how to rank all 4 categories based on the two rounds of responses.

As far as a tie, you bring up an interesting point. A, E, R, X are 4 categories but someone who scores high in A should not score high in X. In the example you gave, I would choose A because E is higher than R showing that they are more likely on the lower end of the 4 categories. Ideally, people would score in group X. I'm unsure how to resolve a pattern of 5, 2, 5, 2, or 5, 2, 2, 5. I need to think more about it.