Researchers in several disciplines have argued that a social science interview should be seen as a product of situated interaction, rather than as the elicitation of the interviewee's pre-existing cognitive state. We propose an approach for analysis of this interaction based on studies of stance-taking and of responses to questions. Studies of stance-taking can help us understand the process of elicitation; in these studies, expressing one's evaluation of an object is seen as an inherently dialogical act that involves positioning oneself, defining a shared object, and aligning or not with previous stance-taking. Studies of responses to questions can help us apply these general analyses of stance-taking in dialogue to the specific genre of the research interview. In this article, we analyse existing transcripts from the Qualidata Archive and focus on the devices interviewees have for showing that they are taking up the question, and for aligning or disaligning with the stance projected in it, and the devices used by interviewers in follow-up questions to recalibrate the object of stance-taking. Since we focus on the lexical and syntactic form of elicitations, responses, and follow-ups, the approach can be applied, with some important caveats, to typical social science research transcripts, not just to specially retranscribed interviews.