I'm working in the validation of a 15 question long, Likert scale based questionnaire. The general steps I have followed are more or less described here: https://www.methodspace.com/validating-a-questionnaire/.
A subset of the questions (4 of them) can only be answered if the respondent either smokes or drinks alcoholic beverages. After noticing this overlook in the first day of pilot testing, we modified the instrument accordingly to consider that branch, but now we don't know how to handle the skipped questions, since this isn't random missing data. The first possibility that I envision is running the statistics and analysis for the general questions, and validate those 4 questions separately. After that, merge the results. Even if I validate each subset of questions individually, I don't know if it's methodologically correct to just merge them in a single questionnaire later.
Good URL for you :-
How to Choose the Most Effective Survey Question Types.
https://surveyanyplace.com/survey-questions-types/
Dear Alvarez
In my opinion you validate your questionnaire in 11+4 format. Prepare separate norms for each set. Club both and specifically mention, answer these items if you either smoke, drink or both (it depends upon sample size, larger the size better the outcomes in this format). This is one option. Another option is to strike off those 4 items but with a catch that it is likely to compromise your objectives.
I recommend that you analyze the ratings of the respondents who answered "yes" the four conditional items independently from the respondents (presumably larger) who answered "no," and then you run the same analysis, be it factor analysis or otherwise, for both larger and small samples.
- Badges
- Science topic
- Similar topics
- Engineering
- Materials Engineering
- Powders