Overview: A growing literature suggests that darker-skinned, less European-looking Blacks and Latinos face greater discrimination and experience worse socioeconomic outcomes than their lighter, less racially or ethnically “prototypical” counterparts (Weaver 2012; Maddox and Gray 2002; Espino and Franz 2002). Attesting to the importance of this line of research, the U.S. Bureau of Labor Statistics and the directors of the General Social Survey and the American National Election Study have begun collecting skin-color data, and the Equal Opportunity Employment Commission has issued guidance about color- and phenotype-discrimination cases under federal law. However, both the science and the law of phenotype discrimination have been hindered by serious measurement problems. The standard measure of skin color, called the Massey-Martin Scale, is not reliable. There is no generally accepted measure of racial/ethnic prototypicality as such. Social psychologists have measured racial prototypicality using Likert scales and, as coders, convenience samples of college students, but this measurement strategy assumes that everyone sees race in the same way, and that “apparent race” is mainly a function of objective, physical features of the person being observed. Yet social theorists have long argued that race and ethnicity are “socially constructed” categories (Lopez, 1994; Omi and Winant, 1986), and that racial perceptions are likely to be affected by the social status, dress, expressions, and behavior. To establish solid measurement foundations for research on phenotype discrimination, our project will use Bayesian pairwise comparison models, surveys, and experiments to develop new measures of skin color, apparent racial/ethnic stereotypicality and ancestry. The ancestry measure is designed to capture stable features of the subjects physical appearance, and the stereotypicality measure to capture the broader set of cultural associations with race/ethnicity. Intellectual Merit: Our measurement strategy addresses all of the known or suspected problems with existing measures, including: (1) the possibility of systematic differences across demographic groups in coder perceptions; (2) ambiguity about whether the measures track innate physical features of the person being observed, or facial expressions, dress, and indicators of social status; (3) possible bias from extraneous information, such as the order in which subjects are observed, or shocks to the mood of the observer; and (4) a lack of information about the precision of estimates of skin color or prototypicality. Our studies will also make contributions to the sociology and psychology literatures on the social construction of race, and the political science literature on race discrimination by voters. The PIs, who have different but complementary backgrounds, are well qualified for this work. Abrajano, a leading authority on racial and ethnic politics, has done important work on the measurement of minority political preferences. Quinn, a methodologist who bridges the worlds of political science, statistics, and law, has done foundational work on latent-attribute scaling methods similar to those we will use. Elmendorf, a law professor who studies voting rights, has written a number of papers using empirical methods and experiments to answer legal questions. Broader Impacts: Our project has potentially transformative implications for social scientific research, the practices of government agencies, and law and public policy. It will enable social scientists to quantify the strength of “treatments” in experimental work on race/ethnicity discrimination, and to better estimate the relationship between skin color, racial/ethnic prototypicality, and socioeconomic outcomes. As for government agencies, our research will provide the first comparative evaluation of the Massey-Martin Scale (now used on several important surveys) and certain promising alternatives. Finally, our pr