Data & Data Practices

With years of experience in electoral and demographic data, explore how the Lab answers questions about the best ways to use the data we have.

Reports

• Disaggregation of Unsorted Votes

  Some modes of voting, including absentee and early voting, are only reported at the county level. We call these votes unsorted as they have not been sorted into respective precincts. In this report we compare several different methods for county-to-precinct vote disaggregation and discuss various methodological considerations for practitioners.

• Conventions for Constructing Demographic Categories

  In this note we explain the construction of VAP and CVAP from Census variables, using both the ACS and the Decennial (PL 94-171) as sources.

• MGGG States

  One of the main challenges in studying elections is processing election results into geospatial files. Precinct boundaries, the smallest unit at which elections are reported, are determined by localities, and thus the quality of the boundary data varies widely. Early on in the Lab, we spent an inordinate amount of time and effort cleaning these precinct data products, which were stored in a GitHub repository called MGGG States.