"You're supposed to say White": Epistemic Ambiguity and Ethnoracial Classification.

Quantitative analyses of inequality typically take ethnoracial classification for granted without considering how such data comes to be. We argue that criteria for categorization is often unclear, creating ambiguity about who has the authority to determine an individual's race or ethnicity in the U.S. context. Using more than 300 survey interview transcripts from the UC Berkeley Social Networks Study (UCNets), we explore how respondents and survey interviewers negotiate both epistemic ambiguity and "White" identity. We identify multiple strategies of discursive distancing from a straightforward "White" response and trace how respondents and interviewers pass responsibility for classification, deferring to each other and appealing to external authorities, before answers are recorded. We also demonstrate this combination of distancing and epistemic ambiguity can result in respondents with qualitatively similar responses receiving diverging race/ethnicity codes. We conclude with recommendations about how to navigate epistemic ambiguity for users and producers of ethnoracial data.