So, unstructured interviewing is supposed to be for cases of exploring new arenas, right? Well, semi-structured interviews work to “combine the flexibility of the unstructured, open-ended interview with the directionality and agenda of the survey instrument to produce focused, qualitative, textual data at the factor level” (SSL 1999, 149). Factors, of course, being upper-mezzo levels that work to describe social phenomena, where we can clarify key patterns and concepts, operationalize our variables, develop and respond to hypotheses, and develop qualitative bases to inform ethnographic surveys.
To make “good” semi-structured interview questions, SSL recommend to:
– Make questions and concepts accessible to all respondents, with special attention put on the language, cultural background, and other relevant characteristics of potential interviewees
– Avoid biased questions or positive/negative examples or associations in questioning/elaborating on questions.
– Avoid double-barrelled questions (items that ask two questions in one) or double-negatives, as they’re confusing and may not offer accurate responses.
– Avoid asking informants to rank, perform several tasks, or respond yes or no.
– Construct interview schedules (list of questions to be asked) from earlier events to more recent events, and likewise, from simpler topics to more complex or controversial ones.
– Order questions from the more concrete to the more abstract.
*** However, some researchers argue that you should ask more intellectually difficult and controversial items first, as to not have interviewees skip over elaborations as they tire throughout the interview.