Ancestral ambivalence and racial self-classification change

In survey research, race is often treated as an exogenous control variable, which assumes that response instability in racial self-classification represents random response instability only (error). With census figures suggesting that only 3% of Americans change their racial self-classification in the short-term, potential violations of this assumption may seem harmless. Advances in genetic research together with other factors, however, may render racial self-classification increasingly flexible in the future, which in turn may increase the risk of observing systematic response instability.This study subjects the research hypothesis of systematic response instability to conservative empirical testing using a representative telephone survey with repeated race questions (n = 1200). The results suggest that knowledge about ancestors of other backgrounds (‘ancestral ambivalence’) may predict response instability in racial self-classification. Results of a college student experiment (n = 555) further suggest that survey content can induce ‘ancestral ambivalence’ and short-term racial self-classification change. The implications of these findings are discussed.