I get that an interview is a piece of discourse in and of itself that's worth being analyzed as such (that is, not only for mining data from the interviewee's responses). What the interviewer says is bound to help produce what the interviewee says and vice versa. Looking into this is a good way of finding something out about whatever construct you're studying.
Questionnaires are a kind of interaction, too, though, no? The writer's audience is always a fiction, and a questionnaire is a piece of writing* so the writer of the questions is already making certain responses possible/impossible.
(* Maybe I am wrong about this, but I tend to think that anything that has, like, words in it, orthographically, is writing. I was pretty surprised to run into a linguist last year who told me that blogging wasn't writing. What else is it?)