Our proposal centers around a new item to measure respondents’ racial/ethnic self-identification. This item enables a test of a lingering gap between theory and practice. Theoretically, social science accepts a constructivist view of group identities like race and ethnicity – that they are socially defined and vary across contexts of structural constraint, social interaction, and individual choice. Empirically, while most social surveys now allow respondents to identify with more than one race/ethnicity, the underlying concept in extant measurement remains fixed and categorical. If a more contingent, continuous reality were measured categorically, this gap between theory and practice would be potentially consequential – both to our study of racial/ethnic politics and to any aspects of electoral politics more generally where a person’s race/ethnicity is a contributing factor.