The VPC introduced measurable list-based approaches to voter registration, vote by mail and voter turnout based upon modern commercial direct marketing techniques. At the outset, the VPC learned that ensuring the quality and enhancing the value of lists were essential to lift voter participation and increase cost efficiency and programmatic effectiveness. Since then, the VPC has made enormous progress in converting low quality commercial data into high quality lists of unregistered Americans. That progress has been hard won, detail by detail, over an eight-year period.
One of the fundamental advances the VPC has made is to take a list–based, universal approach to registration that focuses on demographics and much less on particularly dense geographic locations. To do this, the VPC had to improve the reliability and accuracy of lists to understand who was truly registered. That was achieved by testing conventional wisdom and using new tools:
- Multi-Sourcing: The VPC has consistently tested conventional wisdom about list generation and use. Through this testing, it became clear that any name and address that appears on only one commercial list and does not have one or more confirming matches on another commercial list is far more likely to be an incorrect name and address and far more likely to fail to respond to any mailing. Essentially, the VPC discovered that records of individuals which are compiled from multiple sources are more likely to be records of real people and more likely to respond to voter registration mail. Later, the VPC confirmed this same principle with registered voters on the voter file, with single source registered voters 25 percent more costly than multi-sourced registered voters at producing a net additional vote. This finding, in conjunction with past findings that helped the VPC evaluate address quality have been a great step forward both in improving the accuracy of VPC lists, modeling response and in lifting response rates. Multi-sourcing is now a protocol in every VPC program from voter registration to turnout.
- Historical databases: The VPC also used new tools — databases of historical information — to determine if an individual was already registered, either at the same or at a different address. By using these databases, the VPC discovered that a quarter or more of the individuals that some lists identify as unregistered, are actually already registered. These databases containing years of address history proved to be much more accurate in matching voter registration lists against commercial data. In 2010, the VPC was able to further refine these techniques.
Though no process is 100% foolproof, the VPC is as thorough as possible in filtering out inaccurate information. Individual records are analyzed with a statistical modeling algorithm to identify records that are unlikely to represent a real person. These records are then removed from the mailing list. Individuals are also removed from the mailing list if an address is determined undeliverable or if they have been reported as deceased in the Social Security Administration’s Death Master File.
The investments in list enhancement allow the VPC to achieve dramatically higher ROI by excluding people who should not be mailed and experiencing better response rates from those who are mailed. These enhanced lists have also been used as a source of important research about target audiences; for example, better data provides more insight into the impact of mobility on the drop-off rates and voting practices of the RAE. The VPC discovered that someone who moves during an election cycle is 12 times more likely to be a drop-off voter than the average consistent voter.