Hello Polina,

While a number of IRT calibration software packages can handle the "item not reached" missing values, most factor analysis software packages have either no or very clumsy options for handling missing data. The big issue I see is, do people for whom a given section is relevant respond differently to common items than people for whom a given section is not relevant? That could be an example of DIF (differential item functioning), which poses a problem.

For more common types of analysis, such as estimated score reliability, and dimensionality of specific condition item sets (e.g., for those who do have stress), the only option you have is to estimate technical indicators using just those who do elect to respond to the items.

Good luck with your work.

Very interesting question. I am interested in knowing about it

With skip logic, you can change what question or page a respondent sees next based on how they answer the current question. No matter what the use case, adding conditional branching makes the survey process quicker and smoother for your respondents—just one more way to easily extract insights.

you can validate an instrument, an the remain instrument you can use to compare people who are clinical or not clinical sample but I think you are thinking two diferente phases of your research (but maybe someone here knows better than me).