A survey can be easy for respondents to complete while producing awkward data to analyze. Branching and unfinished responses are useful things to inspect early.
What would you test in a small trial? Would you include deliberately incomplete responses so you know how they are represented?
.
