This paper explores why respondents to a telephone public-opinion survey often give reasons for answering as they do, even though reason-giving is neither required nor encouraged and it is difficult to see the reasons as attempts to deal with disagreement. We find that respondents give reasons for the policy claims they make in their answers three times as frequently as they give reasons for value or factual claims, that their reasons tend to involve appeals to personal experience, and that they (...) often talk about their thought processes, especially when the evidentiary stakes are high. We then explore several ways of explaining these findings. We suggest that one useful approach is to see the reason-giving in the survey interviews as deliberative, reflexive argumentation of the sort described as `critical thinking. We further suggest that the reason such argumentation is often conducted out loud in the interviews, rather than internally, is that it functions in the service of rhetorical ethos, in particular the need to display the fact that one is human, with human autonomy and agency. Doing this may be particularly important in contexts such as anonymous survey interviews in which people are at risk of being treated like machines. (shrink)