Published online by Cambridge University Press: 06 March 2017
In this research note, we explore compare and contrast three methods for measuring race. We utilize as our baseline, or “true”, measure expert coded racial categories, and to this compare two alternatives. The first is a hybrid Bayesian analysis of racial/ethnic surname lists and population distributions, which allow us to develop a race probability score for each candidate. The second is a novel and innovative crowdsourcing method that allows many contributors to classify the racial identity of candidates. We analyze and discuss the potential benefits, pitfalls, and tradeoffs of each method. We conclude with the implications of these new measures for future election research as well as race and politics scholarship more broadly.