Inside the Black Box:

Survey methodologists borrowed the think aloud method from cognitive psychology in order to make respondents’ thinking explicit in a variety of tasks. The method has become a central feature of cognitive interviewing to pretest questionnaires. In adapting the method for pretesting, survey researchers changed it in important ways that may have implications for data quality, either for better or worse. The procedures described by Ericsson and Simon (e.g. 1993) for eliciting verbal reports limit the interviewer/researcher’s role to providing initial instructions to the subject and, during the laboratory session, reminding the subject to keep thinking aloud. Under this approach, the content of the verbal reports should be largely unaffected by the interviewer/researcher’s actions. In cognitive interviewing, the interviewer’s role has generally been more active and more interactive. The interviewers query respondents’ understanding of survey questions, request they perform additional tasks beyond thinking aloud, e.g. paraphrasing the question, and interact with respondents, for example, conversing about what has already been said (e.g. U.S. Bureau of the Census, 1998; Willis, DeMaio & Harris-Kojetin, 1999).