This paper introduces VRscores, a workplace-level measure of employee partisanship constructed by linking U.S. voter registrations to electronically available worker profiles covering 2012 to 2024. The resulting organizational-level data set captures the partisanship of 24.5 million workers across more than 534,000 employers with at least five matched employees. We release this employer-level data set along with parallel data sets reporting VRscores at the firm, occupation, industry, and metropolitan statistical area levels. We show that VRscores cover substantially more employees and organizations than donation-based approaches to measuring political ideology. We also show that VRscores are more representative of the U.S. workforce in terms of partisanship, seniority, occupation, and industry. Finally, we demonstrate that VRscores and donation-based measures are only moderately correlated (Formula: see text) and that they classify one in five organizations differently with respect to whether they lean Democratic or Republican. Funding: We acknowledge generous financial support from the University of Maryland Smith School of Business, the Michigan Ross School of Business, and J. Frake acknowledges funding from the University of Michigan’s Year of Democracy Research Grant. M. Kagan acknowledges funding from the National Science Foundation Graduate Research Fellowship Program (GRFP) Grant 2146752 and the National Science Foundation Learning the Earth with Artificial Intelligence and Physics (LEAP) Science and Technology Center (STC) Grant 2019625 and the Institute for Humane Studies Grant 019144. Supplemental Material: The online appendix is available at https://doi.org/10.1287/orsc.2025.20402 .
Kagan et al. (2026) studied this question.