In mid-January, Wisconsin Democrats unloaded 128 cardboard boxes from the back of a moving truck outside the Government Accountability Board, which is just around the corner from the state capitol that served as a stage for last summer’s epic legislative fight over union bargaining rights. Democrats were optimistic about the boxes’ contents, reams of petitions to set in motion the process that will subject the state’s governor, Scott Walker, to a recall vote. When, two months later, the board formally accepted the petitions, they confirmed that the truck’s arrival had consummated a staggering feat of political organizing: 900,039 valid signatures collected by 35,000 unpaid volunteers over two months. That was well over one and a half times the number required to trigger a statewide recall ballot, and nearly matched the number of votes the Democratic nominee for governor received in 2010. Walker won that election with only 1.2 million votes cast in his favor.
That math may herald the premature end of Walker’s term when he has to defend his office next month. (Democrats will select their nominee in a primary today.) But there’s a consolation for Republicans: They’re thrilled to get their hands on the petition documents, and the 900,039 names they believe will help them make sense of the state’s political geography on behalf of Mitt Romney this fall. Indeed, the boom in large-scale signature-gathering efforts—most visible in 16 different recall campaigns filed in Wisconsin over the past year, and similar efforts elsewhere in the Midwest—has fixed attention on a largely unmined source of political data that can help clarify fault lines in a difficult to gauge electorate. “They’ve just handed us the names of 900,000 people who are known, or are likely to be, anti-conservative voters. It’s a huge favor they’ve done.” says Rick Wiley, the political director of the Republican National Committee, which has 22 offices open in Wisconsin coordinating anti-recall efforts. “Without it you were stuck with somehow IDing these voters.”
Usually such information comes at great cost, and especially so in Wisconsin. Voters do not register with a party and can participate in either primary, although unlike in many states with similar laws, no record is kept of which ballot they choose. The only way to confidently sort individual voters by partisan attitudes is to contact them individually and ask. Wisconsin Republicans worked out early deals with conservative allies who would identify voters’ positions on key issues, like abortion, guns, and taxes, and add the information to voter lists. But many of the groups were restricted by the tax code from asking about partisan campaigns, and a sequence of uncompetitive statewide elections in the middle of the last decade meant few Republican organizations had the resources to do the work on their own.
Campaign tacticians faced similar problems in Michigan, which also doesn’t permit registrants to associate with a party. In the early 1990s, Mark Grebner, an East Lansing consultant who had created the state’s first voter file during the previous decade, decided to see if he could find clues in the reams of petitions submitted by candidates looking to qualify for the ballot and activists seeking to place issue questions there. The documents tended to stack up at the offices of local election authorities, largely unexamined; when people looked closely at the names and addresses, it was usually to disqualify an opponent from running.
Grebner flagged the records of voters who had signed different petitions and set out to see what he could learn about the signers’ political attitudes. Signing a petition, after all, was not the same as calling yourself a Democrat on a registration form or picking a Republican ballot in an open primary. In fact, it represented a political behavior entirely detached from voting: just because someone will succumb to a petition gatherer when accosted at a supermarket doesn’t mean he will make a special trip to an elementary school on a Tuesday evening to cast a ballot. Signing a petition is no commitment to policy, only an endorsement for it to come up for later consideration, and even when a signature plainly backs a partisan cause, like Walker’s recall, it’s not quite a vote: There are no limits to how many petitions a voter can endorse, making a signature nonexclusive. A voter can theoretically sign simultaneous petitions to legalize gay marriage and ban gay marriage, and do so in good faith.
Grebner found that some petitions were useless at predicting partisanship, while others proved valuable. Petitions with strong ideological content, like those to get a partisan candidate on the ballot or change abortion laws, significantly increased Grebner’s ability to predict how a signee would vote. A voter who Grebner estimates as having a 50 percent probability of picking the Democratic candidate can jump to 62 percent based on the addition of a single new data point. “By itself, it's not terribly powerful, but if you collect a few million of them,” Grebner says of the signatures, “you start to think of them as real data.”
Grebner began making petition information a staple of his voter file, digitizing images of petitions and sending them to Bangalore firms to reduce the cost of processing costs to around 5 cents per signature. When Grebner expanded his franchise to Wisconsin in 1999, he replicated the practice there—and the last year’s convulsions in the state’s politics have created a bonanza of new signatures for him to analyze. “The recall eruption is a whole new thing in Wisconsin, providing a large amount of accurate data in a state where partisanship is very hard to discern from public records,” Grebner says.