SEPTEMBER 21, 2012
Wednesday morning, in a small conference room in its Washington, D.C. office, NARAL Pro-Choice America rolled out a plan to clinch the election for President Obama. That’s an ambitious goal for an organization pushing a niche issue in a contest dominated by the economy, but they’re hoping to win big by thinking small. Their plan is to target what they’ve termed “Obama Defectors”—pro-choice women who supported Obama in 2008, but are now poised to vote for Mitt Romney—and win them back. Nationally, they've identified 5.1 million across the country, 1.2 million of whom live in swing states.
But their real focus is actually on a much smaller number: a mere 338,020 women who live in swing counties in Colorado, Florida, Iowa, North Carolina, New Hampshire, Nevada, Ohio, Virginia, and Wisconsin. By their math, targeting this highly prized—and potentially persuadable—demographic in the right places could be enough to push Obama over the edge.
This micro-targeting strategy isn’t anything new, but it’s the first time it’s being employed by a group with such narrow interests and with such a small target group. “I don’t know if anything like this has been done—at least not this election,” boasted Drew Lieberman, the pollster behind the model. In fact, it’s a descendant of an approach that was initially utilized by Karl Rove to devastating effect in 2004 against John Kerry. Its architect was veteran Republican pollster Alex Gage (who actually worked for Romney in ‘04); he wanted a way to connect with a particular type of voter, which would require information beyond the standard voter file. So he proposed borrowing from the consumer data-mining practices of the private sector. (A more detailed account of this move towards data-driven campaign thinking can be found in Sasha Issenberg’s new book, Victory Lab, recently reviewed by my colleague Nate Cohn.)
In 2004, this meant creating messaging so hyper-specialized that the Bush campaign was sometimes printing mailings in batches of only a few hundred fliers. By 2008, the Democrats had joined the game, creating an independent voter database to gather and house the exploding amount of data on voters (independent so that it could work with campaigns and independent progressive groups—if it was owned by the DNC, it couldn’t give the data to PACs). Team Obama brought their data analysis to a new level, and, using a database called Catalist, applied behavioral science to identify and communicate with voters at a level of specificity that would have seemed like science fiction a generation ago. The NARAL model, which also builds off of Catalist data, is taking things a step further, trying to drive a hard wedge into a must-win swath of voters that are otherwise poised to vote for Romney. The difference this year: it can now deliver its messages directly to the voters they were meant for, without also broadcasting to voters who disagree, or simply don’t care.
NARAL is basing its strategy on a 20,000-interview survey conducted by former Clinton pollster Stan Greenberg’s firm (Greenberg, Quinlan, Rosner), which used roughly 500 data points—like whether they’re cable subscribers, what publications they read, even what products they like—to identify their target voters. While Obama leads among women, they found that he's polling eight points lower than he finished in 2008, creating a seemingly up-for-grabs voting bloc that both sides are pursuing aggressively. From there, NARAL narrowed the group down even more by dividing their so-called “Obama Defectors” into persuasion voters, who were likely to show up on election day, and Obama supporters, who were less likely to head to the polls, and focusing on the former. This segment, it turns out, tends to be white, independent female voters younger than 40.
As NARAL takes advantage of the micro-targeting trend, it’s also taking advantage of another trend of this election cycle: campaigning on abortion rights. While the economy is unquestionably the dominant issue, the GOP “war on women” has allowed NARAL and others to make abortion a vote-deciding issue this year, when it arguably would’ve been pushed to the background under other circumstances. (In fact, the GQR survey found that 40 percent of its target group would vote for a candidate that was pro-choice, even if they disagreed with the candidate on every other issue. They also found that Obama was able to pick up six points when his position on abortion rights was explained and compared to the Republican platform.)
To be sure, the subset of the electorate NARAL has identified is small, as low as a few thousand in some counties; and in an election of unprecedented spending, especially in swing states, it’s possible that NARAL’s highly targeted campaign won’t have the hoped-for effect.
Nevertheless, NARAL is optimistic, claiming their approach will be “leaner, smarter, more efficient than the campaigns have been in years past.” While messaging is still being field tested, the program is meant to go live around Columbus Day, unleashing a torrent of phone calls, direct mailings, and ads on programming that caters to this demographic—TLC’s “Say Yes to the Dress,” for instance—before NARAL turns to a more traditional GOTV operation in the weeks directly before the election.
But more than simply wielding a new weapon in the increasingly sophisticated arena of political campaigns, the NARAL model is an experiment in the effectiveness of this kind of targeting—and an opportunity for further refinement. It has siphoned 60,000 of its target voters into a control group, to test the impact of its methods (surveying their votes after Election Day). That means by 2016, this model will be probably be history.