Values are saved on moving backwards too - so if a user gives a value at all it must be within required parameters.

Alright, that makes sense.
But then there shouldn't be a redirect to the previous page after the Javascript alert. That is just something that irritates a bit imho (or I just do not get it completely ;-)).

Mallo, you have to give a valid (e.g. sufficient number of) answer(s) before you are able to proceed (meaning clicking next or previous.

This behavior is a little weird, I would also expect to be able to move backwards when not having entered any data.

c_schmitz, lemeur do you think we can/should change this?

Mazi, did you check it?

We have to decide here if we prefer:

• to ensure that the temporary data recorded into DB are always consistent with min/max constraints
  OR
• we prefer to ease the participant's navigation

In any case, I agree that showing the warning message and yet going backward is a weird behaviour.

I personally vote for the solution proposed by Mallo: max/min scripts could be bypassed when going backward. This of course would authorize Non-Completed responses to be inconsistent with min/max constraints, but anyway I always consider these incomplete responses as possibly inconsistent...

Carsten, Mazi your thoughts ?

I vote for 'to ensure that the temporary data recorded into DB are always consistent with min/max constraints' - I rather not have any invalid data in the tables.

From my point of view it's netter to not irritate the user by showing alert messages. If you didn't fill out anything and just want to move back you don't expect to have to fill out all questions at first. Having a survey in group-by-group mode with many questions really causes a problem then.

I think there will be only some very few cases where users enter some answers, then want to go back, maybe change a question so that following questions which were answered previously are hidden by cvonditions.
Only in this case invalid data will remain in the DB I guess?!

So I vote for improving the usability.

From my point of view it's netter to not irritate the user by showing alert messages. If you didn't fill out anything and just want to move back you don't expect to have to fill out all questions at first.

I totally agree on that. A question responses being still empty (although mandatory) should be fine. But invalid values should should not be tolerated.
We could quietly discard these (not preferred) or force the user to give a valid value.

lemeur, do you think you can have another shot at this for CI 1.92?

This is fixed in 1.92. There are no longer any pop-up messages for min/max answers - instead the validation criteria are color-coded green/red.

1.92RC3 released