Previous work has shown that stance—the way speakers position themselves with respect to what they are talking about and who they are talking to—provides powerful insights into why speakers choose certain linguistic variants, beyond correlations with macro-social categories such as gender, ethnicity, and social class. However, as stancetaking moves are highly context-dependent, they have rarely been explored quantitatively, making the observed variable patterns difficult to generalize. This article seeks to contribute to this methodological gap by proposing a formal guide to coding stance and demonstrating how it can be operationalized quantitatively. Drawing on a corpus of eight individuals, self-recorded in three situations with varying levels of social distance, we apply this method to variation between English complementizers that and zero (i.e. no overt complementizer), providing a replicable and theoretically grounded protocol that incorporates both quantitative and qualitative analyses in a variationist sociolinguistic study. (Stance, complementizers, that, English)*