Skip to main content
Open Access Publications from the University of California

UC Berkeley

UC Berkeley Electronic Theses and Dissertations bannerUC Berkeley

Race as a Dependent Variable: Three Papers on the Social Predictors of Racial Attribution and Identification in Brazil


Racial classification is relational, and it is constructed and contested as both an identity and as an ascribed category. In this three-part dissertation, I examine racial classification in São Paulo, Brazil, and argue that we can leverage survey data on racial classification to reach a more nuanced understanding of racial boundaries.

Paper 1: Because racial classification is relational, we must examine both sides of the relationship, and yet most studies on ascription do not consider the ways in which the ascriber’s traits matter for classificatory outcomes. Using a quasi-experimental design, I examine how statuses of both the classifier and of the person being classified matter for racial attribution. I find that high-status biographies paired with a racially ambiguous face have a higher likelihood of white (versus pardo) racial attribution. I also find that educational achievement is significantly predictive of seeing whiteness in others—irrespective of the phenotype, self-identification, or the social status of the person being classified.

Paper 2: Drawing on a large-scale survey linked to census-tract data, I examine how classification can operate asymmetrically across different racial boundaries. I compare the effects of various individual-level and neighborhood-level traits on the chances of having one’s racial identity contested by an observer. I show that social status does not have a uniform effect, mattering greatly for the boundary between white and pardo, and much less for the boundary between pardo and black. I also find that individuals living in richer or more educated neighborhoods—even when controlling for individual status—are more likely to be whitened, and less likely to have their white identities challenged.

Paper 3: In this paper, I demonstrate how we can leverage racial mismatch data to study the everyday impact of institutions (in this case, the police) on individual racial identity. With a combination of survey and census data, I show how self-reported contact with the police in the twelve months prior to the survey is associated with a lower chance of identifying as white, even when controlling for individual status and observed race. I also show how neighborhood characteristics can further shape racial identity: Proximity to a favela, the neighborhood rate of economically motivated crime, and the racial makeup of the neighborhood all have separate and independent effects on racial identity.

Main Content
For improved accessibility of PDF content, download the file to your device.
Current View