The Many Dimensions of Race: A Latent Variable Approach for Quantitative Research
Aliya Saperstein, University of California, Berkeley
Typically, race is measured through self-identification and included in quantitative research as a set of “dummy” variables that serve as controls in regression-type analyses. This implies several strong assumptions, including that race is a static characteristic of individuals and that it can be described by mutually exclusive and exhaustive categories. I propose a new approach to analyzing race in quantitative research in the hopes of bringing research practice closer to current social science theories of race. Using a statistical technique called latent class analysis, I combine measures of observed race, self-reported race and self-reported ethnic origins into a single variable that better captures the complexity of an individual’s experience of race and ethnicity in the United States. The latent variable consists of multiple racial classes, or categories, and can be used in subsequent studies as either a dependent or independent variable.
Presented in Session 42: Measurement Issues in Race/Ethnicity