Hi,
as it is frequently recommended "In a good questionnaire there is one question per page", you obviously may merge the two groups.
I really don't know why the developers of LimeSurvey decided to set "group by group" as the default setting.

And as long as there is no flexible "group concept", like a question type "group" that may contain several other "questions", as you see here in a commercial survey tool, it is difficult.

You see there are normal questions like "NumSlider", "SimpleMulti", but then there is a group that contains two questions, which you may display horizontally or vertically.
But you also see, that this commercial tool shows question by question by default, what is the natural way to answer a questionnaire.

I personally don't like to scroll down, I don't like to scroll up again, because I missed one answer option in an array at the top of the page, then down again.
And of course, you may get a bias, if people see which question will follow and - if there are filters - see the results of their choices. I mean, if I see that I later have to answer questions to each object I selected in a multiple question, the number of answers will decrease.

Just to consider.

Joffm

holch wrote: What I am saying is, that IF WE WOULD HAVE a page concept, besides the group concept, it would be easy to do what OP wants to do. Not more, not less.

But here : it's not simple like you say/write.

OP want something really specific

Position of a group at a randome position followed by another group.

Have a group + page concept give some facility, but even with page + group + randomization group : we don't have an easy solution ....

Hey Joffm,

Thank you for your contribution. But I have my reasons for the study setup.

Actually, the way you order your questions depends on the research you do. In my case, the mentioned bias is desired. I investigate cognitive biases in energy-related consumer behaviour, e.g. the study design hides effects in follow-up tasks for control and treatment groups.

OP want something really specific

Position of a group at a randome position followed by another group.

Have a group + page concept give some facility, but even with page + group + randomization group : we don't have an easy solution ....

I don't see a problem here. I would put A and B into the same logical group, but separate them onto two pages. When I randomize the logical groups, they are always together, because they are in ONE logical group. Problem solved. Or am I missing something?

From what I understand, having a page concept AND a group concept in parallel would be perfect for these kind of cases.

I don't see a problem here. I would put A and B into the same logical group, but separate them onto two pages. When I randomize the logical groups, they are always together, because they are in ONE logical group. Problem solved. Or am I missing something?

Can you specify what you mean by a "logical group"?

Sorry rakittam, I am discussing with Denis a theoretical concept that I had suggested to be implemented into LS a while ago. IT IS NOT IMPLEMENTED, so it is no solution for you. This is a "theoretical no problem" if we would have the page concept in addition to the question group concept implemented, not a "real life" "no problem".

Currently we have "question groups" as a concept. They can either be used as a visual group (== page) or as a logical group (questions that belong together logically, but not necessarily visual - on the same page).

My proposal was something like this:

Thank you for this addition.

Actually, you have my vote; this would be a useful extension of the existing 'question group' concept.

Or probably better something like this:



But as said, this was a Feature request that I created quite some time ago (additional "Page concept"), which was not implemented (yet). And probably never will.

In LS 4, the question groups have been partially renamed to "pages", but we still have to decide whether we want to use the current concept for grouping questions on pages or if we want to use it for logical grouping of questions.

The approach highlighted by Joffm is also interesting. Generally people want to display question by question, I think. But sometimes you would like to have two questions on one page, for a specific reason (e.g. a Text display and then an evaluation question, etc.). Currently, because you might have one case where you want to do this, you'll have to wrap each question into its own group, just to be able to have 2 questions in one of the groups/pages. If you have a very long survey, this might add a lot of extra steps, just to be able two show two questions on one page in one occasion throughout the questionnaire.

E.g. my company has a standard questionnaire that has about 26 questions. The general mode is to show everything question by question, just the gender and the age are shown on one page, generally.

In some tools I only create 26 questions and than join 2 of them onto 1 page.

In Limesurvey I need to create 26 questions and 25 question groups if I want to achieve the same result. If someone creates a lot of surveys, this can have a significant impact on how long it takes to program such a questionnaire. If you only create 1-2 new questionnaires per month, not a big deal. If you are programming 2-3 questionnaires per day, this has a strong impact on your workload.

Yes it would be,
but holch and I are talking about it for years.

But you can merge the two groups.

Simple by inserting a simple question between the two parts (you should insert the same question at the end of your other groups as well)

Then something like:
Group1 Q1 <- is_empty(QX)
Group1 Q2 <- is_empty(QX)
..
Intermediate question QX
Group2 Q1 <- !is_empty(QX)
Group1 Q2 <- !is_empty(QX)
..





And it should be possible to hide this intermediate question after clicking it by javascript.

Now it's up to you.
Either you wait until somebody will develop something,
or you review your design
or you use a different tool,
or...


Joffm

Actually, you have my vote; this would be a useful extension of the existing 'question group' concept.

Feel free to vote here:

bugs.limesurvey.org/view.php?id=9416

The feature request is from end of 2014 (almost 5,5 years ago) and it has since been "acknowleged", whatever this means.

But while I think it would be a perfect extension for Limesurvey and allow a lot of flexibility, I am also realistic enough that this would probably touch many different parts of Limesurvey, thus would be probably quite some work to implement. And of course this reduces the chances of implementation. The higher the complexity of something, the lower the chances that it will make it into LS, that is just natural. Development resources are limited.

Of course this won't prevent me from highlighting the advantages of such a concept.

I understand, they should really consider your concept! I'm keeping my fingers crossed for you guys.

Thank you all for your time. I will use the information you have all provided and create a makeshift solution.

Again and again : it's more easy to construct something complex with a lot of little part than something doing whole in one big part ...

Worpress without plugin are ... near unusuable, but it's one of a more used CMS used. Drupal without plugin or extra developement are ... near unusuable etc ....