Always a good idea to delete the redirect to your panel provider...

For me, the information seems to be saved just fine:

 

Tested with the lastest version of LS 3.x
Version 3.28.1+220229

Thank you for the tips, Holch!

Yes, but only in this copy we created to send here. We imported and exported the original survey, and then deleted the other questions. Still, same age and agegroup question and coding and the agegroup answers come as the code we assign...

We´ve had other surveys with the same questions and length and this is the first time we are experiencing this issue.

In both of your files I don't see anything e3xcept that you obviously changed the equation a bit.

@holch already showed our reesults.

Here a sample of my idea.
 

Thank you for the sample. 
It works with the new answers. However, I imported the vvfile with previous answers and only a few responded to it, as in the attachment bellow: 

 

Well, then something was wrong in the previous question. But as we do not have access to this, we can't tell.

Hi guys!

Thank you so much for your help!
We were able to fix the age grouping issue for new respondents.
However, the same people who came with the adjusted age group are not answering which city they live in.
We have a question about the state and then we have another question about the city.

Can you please help us again?
I'm sending the file here in Iss. 

Remove all relevance equations you entered by "condition builder".
Use "Expression Managaer". 
It's much easier.
Only enter e.g. in "rs" state==43

By the way:
Here you see the advantage of bumerical codes:
You can shoren your equation in "region" a lot. (lt = "less than")
{if(state.NAOK lt 20, "N",  if(state.NAOK lt 30, "NE", if(state.NAOK lt 40, "SE", if(state.NAOK lt 50, "S",  "CO"))))}