Changing the response to some question may impact the choices and previous answers to some other questions (eg. having a driving license will show later the car driving mode choice, or some questions are asked on the first page if hh size > 1, to reduce the number of questions per person in next page)
Currently, the conditional is the place where we would handle the side-effect and potentially change the values of previous answers. Validations and choices function can make a question invalid or change the visible choices. But for those functions to run, the widgets have to be visible on the current page.
But the questions with side effect may not be on the page. For example, changing if a person has a driver license should invalidate previous trips by this person with the carDriver mode, but as long as the person does not go back to the trips section (if the section was previously completed and valid, there's no reason to go back to it), the now invalid questions will not be changed.
The survey framework should plan for a way to detect those side effects and automatically handle them. Also, if side effects are "important" (need to define important), there should be a warning. The case of changing driving license ownership if the person has carDriver segments is one such case.
Changing the response to some question may impact the choices and previous answers to some other questions (eg. having a driving license will show later the car driving mode choice, or some questions are asked on the first page if hh size > 1, to reduce the number of questions per person in next page)
Currently, the conditional is the place where we would handle the side-effect and potentially change the values of previous answers. Validations and choices function can make a question invalid or change the visible choices. But for those functions to run, the widgets have to be visible on the current page.
But the questions with side effect may not be on the page. For example, changing if a person has a driver license should invalidate previous trips by this person with the carDriver mode, but as long as the person does not go back to the trips section (if the section was previously completed and valid, there's no reason to go back to it), the now invalid questions will not be changed.
The survey framework should plan for a way to detect those side effects and automatically handle them. Also, if side effects are "important" (need to define important), there should be a warning. The case of changing driving license ownership if the person has carDriver segments is one such case.