In this article we show how, whether the goal is reflecting or creating reality, research interviewers must pay closer attention to the particular trajectories of the interactional events in which they collect their data. We focus on two guidelines that research interviewers often use - the injunction to maximize similarities of social identity between interviewer and interviewee, and the injunction to share personal stories as a means of building rapport - and we show how following the same guideline in the same way can yield dramatically different results from one interview to the next. Data is drawn from research interviews conducted by young African American men with young African American men who have become fathers as teenagers. The empirical analysis shows that bids for similarity of identity within the research interviews are sometimes accepted and sometimes parried, depending on the particulars of the interactional event, thus illustrating the complexity of "similarity" of identity in research interviews.