Can We Trust Race Prediction?

TL;DR

This paper develops a BiLSTM-based model and comprehensive datasets to improve race prediction accuracy from voter registration data, providing benchmarks for future research.

Contribution

Introduces a novel BiLSTM ensemble model, extensive surname and first name databases, and a benchmark dataset for race prediction from US voter data.

Findings

BiLSTM ensemble outperforms existing models by 36.8% in F1 score

Created the most comprehensive US surname and first name databases

Provided a high-quality benchmark dataset for future research

Abstract

In the absence of sensitive race and ethnicity data, researchers, regulators, and firms alike turn to proxies. In this paper, I train a Bidirectional Long Short-Term Memory (BiLSTM) model on a novel dataset of voter registration data from all 50 US states and create an ensemble that achieves up to 36.8% higher out of sample (OOS) F1 scores than the best performing machine learning models in the literature. Additionally, I construct the most comprehensive database of first and surname distributions in the US in order to improve the coverage and accuracy of Bayesian Improved Surname Geocoding (BISG) and Bayesian Improved Firstname Surname Geocoding (BIFSG). Finally, I provide the first high-quality benchmark dataset in order to fairly compare existing models and aid future model developers.