VRscores Methodology

Understanding how we measure employee partisanship in the American workplace

Overview

VRscores represent a novel approach to measuring workplace politics by linking voter registration records to online employment data. This methodology provides a more comprehensive and representative view of political affiliations in the US workplace compared to traditional donation-based measures.

Data Sources & Linking

We use public voter registration records and online employment data to create the VRscores dataset. Our data linking process employs Splink, based on the Fellegi-Sunter model, to match records.

Key Matching Fields

  • Metropolitan Statistical Area (blocking variable)
  • Last Name (blocking variable)
  • First Name
  • Middle Name or Initial
  • Gender
  • Birth Year

Data Standardization

  • Names converted to lowercase
  • Punctuation removed
  • Geographic indicators normalized

Coverage & Limitations

The public dataset only includes measures for organizations where we observe at least 10 Republican or Democratic employees. This ensures that we have a sufficient number of observations to make reasonable estimates.

The VRscores dataset covers all major industry sectors and includes both large and small employers. However, coverage varies by state, and not all employees are registered voters. Further, not all states record the party registration of their residents. In these cases, we impute party registration.

For the visualizations on this website, we only include employees who are registered with one of the major parties (Democrats or Republicans), but our full public dataset will also include a measure that incorporates independent and third-party voters.

Users should also be aware that party registration may not perfectly align with political views.