L2 National Voter File: Data Collection and Processing
The L2 National Voter File dataset is a census of all U.S. registered voters gathered from the following sources:
- State- and county-level registered voter files: These contain self-reported data like voter history, party affiliation and address
- Current U.S. Census data
- Election return data: L2 obtains results from every county in the U.S.
- Five national telephone source files
- L2-owned lifestyle and issue data: Much of this information comes from third-party syndicated datasets like Experian; the commercial data is updated at least every six months on a rolling basis and much of the L2-owned commercial data is updated quarterly. Political issues data like support for or opposition to gay marriage, taxes or guns is self-reported data as of the last general election cycle.
For more information on how the data is collected, see the L2 Voter File page on the L2 website.
To create the dataset, L2 uses a proprietary compilation and development process that involves both human and automated control checks. In addition, it ensures the accuracy of its dataset by scrubbing its records against the National Change of Address (NOCA) database and the deceased voters list.
Rhiza then scrubs the voter registration files of all sensitive contact information, effectively making the data anonymous.
For more detail on how L2 processes the data it collects, see the L2 Voter File Enhancements page on the L2 website.
A note on Data Modeling
L2 models the data for a few attributes in order to tell a complete national story. In particular, the Parties_Description attribute data is modeled for the following states: Hawaii, Washington, Montana, North Dakota, Minnesota, Wisconsin, Michigan, Vermont, South Carolina, Missouri, Alabama, and Texas. None of these states registers voters by party and the official state and county voter files contain no assigned party registration. (Two of them -- Texas and South Carolina -- do report which party ballot was picked up and used in a primary; that information is used in the algorithms for the party modeling.)
© 2016, Rhiza, Inc. All rights reserved. Last updated March 22, 2016 07:31:27 AM.