Replication Data for: Measuring Descriptive Representation at Scale: Methods for Predicting the Race and Ethnicity of Public Officials

Ethnicity and race are crucial variables for understanding a wide range of political outcomes. Although data on individual-level ethnic and racial identification are not readily available, recent statistical advances have allowed researchers to impute racial and ethnic classifications based on publicly available information. However, these predictions vary in their accuracy and their susceptibility to biases in downstream analyses. We provide an overview of the estimation methods commonly used in the social sciences, including Bayesian approaches that utilize names and geographic data, as well as machine learning techniques that use names or images as input. We propose and test a hybrid approach that improves upon existing methods by combining surname-based Bayesian estimation with the use of publicly available images in a convolutional neural network. We find that the proposed approach not only reduces bias in downstream analyses, but also improves predictive accuracy in a sample of over 16,000 local elected officials. We conclude with a discussion of ethical and practical considerations and describe domains and settings where the hybrid approach is especially suitable.