California Center for Population Research

I investigate the extent of ambiguity in racial classification using a national representative survey of Brazilian urban areas. Ambiguity is operationalized as the lack of consistency between racial classification by interviewers (categorization) and respondents (identification) using the categories, white, brown, and black. Racial classifications are consistent in 79 percent of the study sample. However, persons at the light end of the color continuum tend to be consistently classified while ambiguity is especially great for those at the darker end. Using statistical estimation techniques, the findings also reveal that consistency varies from 20 to 100 percent depending on one’s education, age, and sex and the racial composition of local urban areas. For example, only 20 percent of high educated females that self-classified as black were classified as black by interviewers while classification as white was nearly always consistent in predominately white urban areas. Also, the direction of the inconsistencies to lighter or darker categories depends on these variables and whether the reference is intervie wer or respondent classification. For example, interviewers "whitened" the classification of higher educated persons who identified themselves as brown, especially when such persons resided in mostly nonwhite cities. Finally, I discuss the role of the Brazilian state in constructing race, and understandings of race and racial groups and comparative studies of race relations.