The groundwork for the Democrats’ get-out-the-vote operation began in earnest months ago in the offices of Civis Analytics, in a seventh-floor loft of an old office building in Chicago’s Greektown and in a townhouse near Washington’s Dupont Circle. Under contract to the DSCC, Civis Analytics started to model the electorates in states with the most competitive Senate races.
Civis is a descendant of Obama’s 2012 campaign. One of its founders is Dan Wagner, who began as a volunteer in the 2008 campaign and by 2012 was overseeing the sprawling data and analytics operation that helped guide many of the decisions in the campaign’s effort to find and turn out as many voters as possible.
“Campaigns do five things,” Wagner said. “They register people who like you; they turn out people who like you at higher rates; and they persuade people who don’t like you to get them to like you. Those are the three big things. Then to support those three things, they build a fundraising organization and then build a voter contact organization that makes all those things possible.”
Modeling produces indexes on a 1-100 scale for each voter. One index estimates how likely someone is to support a particular candidate; another measures how likely someone is to vote; the third projects how open to persuasion someone is. A campaign would target someone with a high support score and low turnout score but would not go after someone with a high turnout score but low support score.
At its core, what Democrats are trying to do is replicate the electorate that helped Obama win by maximizing turnout among young people, African Americans, Hispanics and unmarried women.
“Figuring out how to win is actually not the difficult part,” said Paul Dunn, the DSCC’s national field director. “It’s easy to do a model and put all the model into the calculators and it all spits it out. . . . The strategy isn’t the challenge. It’s enacting the strategy.”