VRscores Occupations Parquet 2012-2024

Occupation-year panel dataset derived from the VRscores 2024 release. Each record corresponds to an O*NET SOC occupation and calendar year from 2012–2024 with at least five matched workers across any year. Variables include the occupation code and title, worker counts, raw and imputed L2 voter registration counts by party, average match quality, raw and imputed partisan shares, two-party and overall margins, partisan diversity indices, effective number of parties, and a processing timestamp. The dataset contains 4,979 occupation-year rows covering 383 occupations per year. It is built from a matching of employment records from Revelio Labs (April 2025) with voter registration data from the L2 voter file (November 2024) using ensemble linkage techniques. Small occupations (<5 matched workers) are excluded, one-to-one matches are enforced, and the VRscores methodology and working paper document the data processing and definitions. If you use this dataset, please cite Kagan, Max; Frake, Justin; Hurst, Reuben (2025). "VRscores: A New Measure and Dataset of Workforce Politics Using Voter Registrations" (SSRN 5104795) and Frake, Justin; Hurst, Reuben; Kagan, Max (2025). "Political Segregation in the US Workplace" (SSRN 4639165 / DOI 10.2139/ssrn.4639165).