I've been struggling with a really strange behavior.

I've boiled it down to this (see attached survey):

1. Question group: User inputs household size. I compute a variable from this via an equation.
2. Question group: User inputs income (plus a few other things). I compute a standardized income from this ("ComputeIncome"). It is definitely a number that I can further computer with other things (like "Position")

This all works fine.

3. Question group: Here I need to do further things with ComputeIncome. But it seems Limesurvey changes its variable type. While it's value seems to stay the same, I cannot anymore e.g. divide it by 2 (result is NAN).

Here is the thing: This behavior in the 3. question group only occurs when question groups are loaded on different webpages. When I run the whole survey on one page, everything works as expected.

Does anyone have any idea how I can solve this problem?

In your example there is no variable called "IncPosEstimate" which is used in equation "Bias".
Please send the revised survey.

Furthermore you should revise your equations.
You use obsolete conditions
{if(RandEU.NAOK == "true", if(ComputeIncome.NAOK/ComputeHHfactor.NAOK <= 8406, 0, if(ComputeIncome.NAOK/ComputeHHfactor.NAOK > 8406 and ComputeIncome.NAOK/ComputeHHfactor.NAOK < 10961, 10, if(ComputeIncome.NAOK/ComputeHHfactor.NAOK >= 10961 and ...

The red part is not necessary because it is the FALSE part of the first comparison. This first failed therefore IS greater than 8406.

And with numerical codes instead of "A1", "A2", ... the first equation could be done with one simple formula:
{(SDHHSize+1)* 0.5 - SDChildren*0.2}

Some first findings:
equation RandEU should be changed; as RandEUStep has a fixed value you my just say {if(RandEUStep>3,1,0)}.
equation ComputePosition: there is no value if RandEU!=true -> undefined.

Joffm

Thank you Joffm, this is really helpful.I've changed the things you suggested.The basic behavior has not changed. In the final question group, Limesurvey correctly displays the value of ComputeIncome and HHsize, but apparently is unable to perform computations on them (like divide them by 2). Again, it works as expected once I choose to show the survey on one page instead of separate pages.

Because it's NOT really an integer …

Solution 1 (checked): use dedicated equation part : you'r sure you din't have HTML extra part : www.limesurvey.org/manual/Question_type_...ation_.28equation.29
Solution 2 (not tested) : use Number only : www.limesurvey.org/manual/Question_type_...y_.28numbers_only.29

Thanks Denis, this was a very good hint. After deleting the HTML comments, everything works.

I am still irritated because the HTML comments per se did not contain any mistakes, as far as I can see, and also the bug did not occur whenever I would run the survey on one page. So it seems it to me it's a Limesurvey bug.

For me : there are an issue too :
Seem data saved in DB are OK (are only number), but unsure …

Can you try without the comment only ? Just leave p tage ?

Then : you can report the issue about comment difference