Is there anyway I can ask a question in different ways depending on the answer of a previous question but store in the same idcoder? For example:
[idcoder fishdry] 1) Did you eat fish last dry season? ( ) Yes ( ) No
If Yes [idcoder fishflood] 1) And in the last flood season, did you also eat fish? ( ) Yes ( ) No
If No [idcoder freqdry] 2) And in the last flood season, did you eat fish?
So, although I have three question (two options for the second question), my final table would only have two columns: fishdry and fishflood
Is that possible?
As you only want two fields returned, could you have a calculate that returns the extra text and insert that in the label?
[Idcoder calcflood] If ${fishdry}='yes' 'also ' ''
[Idcoder fishflood] And in the last flood season, did you ${calcflood} eat fish?
Uh, thanks for the tip. That could work. But I actually gave an example and some sentences I wanted basically change the whole wording but not the options or the idcoder. Example, in one I want to first ask if they have bathroom in the house and in the second how is the bathroom (if they have on the house) or how they do to go to the bathroom (if they do not have bathroom). Thus, idk if calculate could help with that
If your question field name and the responses are fixed and only the label changes based on a prior response then I think that could still work, you could calculate the entire label based on the previous response, not part of it, then the label is only "${calculatedlabel}", not "What is the ${calculatedlabelportion} in this case?"
Oooh nice, i will try that then! Thank you so much
Depending on the complexity you desire, you could either change the second question label dynamically - as @ahblake correctly suggests. Another alternative is to combine & consolidate response from multiple questions (where all but one hidden, via appropriate relevance) with a calculation row; eg using an if() statement, and use the result of this calculation as your final (second) response.
Is coalesce a better option for this consolidation case where only one of two will be populated? Or are there pitfalls there (nested ifs / progressive coalesces for >2 questions) I'm not considering?
Yup, a bunch of nested coalesce()'s should do the trick just as well as a long nested if(), given you are basically performing null/empty checks for everything [and your relevance's should ensure only 1 isnt, so the order they are specified shouldnt matter...]. And may come out looking a bit cleaner.