Five days before the failed vote to recall Wisconsin Gov. Scott Walker, Rob “Biko” Baker sat behind the wheel of his Ford Fusion in a working-class neighborhood of Milwaukee’s West Side and watched two of his college-age volunteers knock on doors. They were doing what canvassers for the League of Young Voters, the group Baker serves as executive director, often do before an election: reminding urban, working-class black voters of an upcoming vote and exhorting them to participate.
“Official records show that you are registered to vote. We’re visiting voters like you who care about our community to remind you about the special recall election on June 5!” the canvassers began, after asking to speak with a specific member of the household. “It looks like a lot of people will be voting this year in the special recall election, and we hope our community turns out, too!”
Only this time there was something new in the scripts that Baker’s canvassers had affixed to their clipboards. “Check this out,” they read. “Have you heard that they’re trying to take away our right to vote?” The canvasser was then supposed to go on and explain what had happened that year in Wisconsin: A law had been passed to require voters to present ID, but a court had intervened to halt its implementation. “Our power to change laws is directly tied to voting. Imagine if you didn’t have the right to vote,” the script read, before delivering some stage directions: GIVE PERSONAL STORY.
Baker looked out on the scene. “After 2010, we’ve been on the defensive so much about voter ID,” said the 34-year old Milwaukee native, a former hip-hop promoter who did social media for Snoop Dogg before starting work on a doctorate in history. Founded in 2003 as part of a Bush-era renaissance in liberal organizing, the League of Young Voters had shifted its emphasis to mobilizing specifically urban minorities in six states, relying on the efforts of mostly young, black volunteers. It was a task Baker thought had taken on an unanticipated urgency with the spate of new state laws requiring that citizens present ID cards to vote. “When I left school I didn’t think my job would be convincing people to get drivers’ licenses,” he said. “That’s not why I left graduate school.”
Baker was nonetheless approaching the voter ID question in a rather scholarly way. When it looked as though the Wisconsin law would be enforced in 2012, he wondered if it might present a new opportunity for his group: Could raising hackles over the law’s perceived injustice help to mobilize infrequent voters who might otherwise feel they don’t have a stake in election outcomes? Baker sought out partners to help him design a field experiment that could be administered in the June recall to yield lessons for November. The League of Young Voters ended up partnering with two leaders of Washington’s lefty empirical-electioneering establishment: the New Organizing Institute and the Analyst Institute, which specializes in randomized-control trials to measure the impact of campaign activity. “I’m like a backyard mechanic who’s got a little bit of schooling,” says Baker, “but I’m not a political scientist.”
The New Organizing Institute helped create walk lists that included only targets who had been identified by voter data as likely to be African-American, a group the league believed would be the most outraged by new voting laws. The Analyst Institute oversaw the process of randomly assigning Milwaukee’s 318 wards to three separate groups: 104 would get a standard get-out-the-vote canvass, 104 would get scripts with the additional language about voting rights and ID laws, and 104 would be in a control group that was ignored altogether by the league’s canvassers.
The streets of Wisconsin were a muddle that weekend, with canvassers from established institutions (like unions and Democratic Party organizations) mixing with those from pop-up groups founded solely to battle Walker; all were boosted by the arrival of out-of-state volunteers. The experimental design couldn’t control for the activities of these other groups, so the best Baker could do was to train his two dozen paid staffers and their teams of often teenage volunteers to stick closely to their instructions when they headed out in to knock on doors.
Over the summer, it became clear that presenting IDs wouldn’t be a concern to Wisconsin that fall, but that it would be in other states that mattered a lot to Democratic fortunes, like Pennsylvania. As Baker waited for the results of his field experiment to be tabulated, a growing constellation of liberal operatives, strategists, and donors had already reached the conclusion that the new issue would permit them to go on offense, mobilizing the most disaffected corners of their coalition.
“It’s going to be an amazing get-out-the-vote effort for us because there will be stories about people being denied the vote,” John Anzalone, one of Obama’s pollsters, said of voter ID laws at a panel discussion hosted by National Journal during the Democratic Convention in Charlotte. “This has the ability to really piss a lot of people off.”
By then, the numbers had come in from Milwaukee, and the Washington institutions that had taken on responsibility for the analysis had begun circulating draft memos and PowerPoint presentations. At first glance, the voter ID script appeared to have had an impact: 29.9 percent of Milwaukeeans assigned to hear it turned out to vote in the recall, compared to the control group’s rate of 28.8 percent. But the residents who received a standard get-out-the-vote message from league canvassers voted at a 30.7 percent rate. In other words, the voter ID script actually did worse than the traditional mobilization method, although not in a way the Analyst Institute assessed as statistically significant. “We cannot conclude that Voter ID messaging dampened turnout, but we also can’t conclude that it increased turnout,” an institute memo read.
But Baker and the New Organizing Institute, which was created to nurture high-tech improvements to field operations and has close ties to Obama’s campaign world, downplayed that conclusion and focused on the one demographic subgroup where the voter ID messaging did seem to have power. Voters over 55 who had heard it were more likely to vote than those who got the standard get-out-the-vote nudge, an impressive 2.5 percentage points above the control. "Anecdotally I believe the reason why 55-and-up respond to the voter ID message is the legacy of the civil-rights movement,” Baker speculated. “That’s been hammered home more than with younger voters.”
This attempt to rescue good news from the experiment troubled Analyst Institute executive director Jennifer Green, a political scientist who had run voting experiments among rural villagers in India before taking her post in Washington. She knew the Milwaukee experiment would generate more than just academic interest. There was a community of liberal donors—some unenthused about Obama and only halfheartedly invested in his re-election—who had latched onto the voter ID laws as a place where they could help the broader Democratic cause. "Everyone is really concerned about the impact of voter ID laws and everybody wants to do something,” she said. “They want to know ‘How can we get people really angry to get motivated?’"
Such eagerness to simply do something could actually prove counterproductive to liberal efforts to mobilize voters, Green feared. She set out to caution progressive allies not to be drawn in by the finding about older voters or misinterpret the 1.1 percentage-point boost it delivered to the broader population as a sign that such messaging was worth deploying nationally in November. "The results support the hypothesis that the message is ineffective and could be harmful if used in place of something we know works,” she said.
Green is steeped in the body of experimental get-out-the-vote research, which—if it amounts to any sort of overarching theory of what works—makes the case that appeals to civic duty or the power of democracy have little place in motivating non-voters to go to the polls.
In 1998, postcards were sent to New Haven, Conn., voters as part of the first randomized trial ever run by field-experimental pioneers Alan Gerber and Don Green, who had been Jennifer Green’s mentors at Yale. (The Greens are not related.) One card sent to New Haven citizens warned “when people from our neighborhood don’t vote we give politicians the right to ignore us,” while another featured an image of Iwo Jima with the slogan “They fought … so we could have something to vote for.” The content may have differed, but the impact on turnout rates didn’t; both proved relatively ineffective at mobilization. By 2004, analyzing the results from dozens of similar trials for their book Get Out the Vote, Gerber and Green concluded that when trying to trigger turnout “the message does not seem to matter much.”
The adjustments to language that did matter were not political messages but ones that toyed intimately with individual voter psychology. In 2005, behavioral psychologist Todd Rogers ran his first experiment challenging a common bit of conventional get-out-the-vote wisdom. For years, campaigns would recite statistics about how many potential voters had stayed home in the last election. Don’t be part of the problem, they would beg. Rogers knew of examples in other fields where people had been coerced to good behavior by the opposite tack: hotels had increased reuse of towels by letting guests know how many other guests did, rather than cautioning them about the ecological peril of low recycling rates. When he and Gerber pitted two equally accurate get-out-the-vote reminders against one another, the one that cast turnout patterns in a way that made voting sound popular got better results. Rogers, who went on to be the Analyst Institute’s first executive director, later credited this to “the basic need for belonging.”
A legacy of that message endured in the GOTV script used in the Milwaukee experiment. “It looks like a lot of people will be voting this year in the special recall election, and we hope our community turns out, too!” canvassers were instructed to say. It was that intervention, and other subtle behavioral nudges that have become standard language in the left’s canvassing scripts, that Green credited with the only unambiguous impact that the League of Young Voters had had in mobilizing voters before the turnout: the 1.9-percentage-point effect of the standard GOTV script over the control group. That was by any measure an impressive result, particularly useful since few prior experiments had been focused uniquely on urban African-American neighborhoods. “People get approached by so many people in these national elections, but it’s different when it’s a kid from your neighborhood,” Baker said.
Many of those in Democratic politics had, reflexively, assumed that it was best to make that kid deliver the most politically engaging message at hand. But instead of telling voters there was an effort under way to disenfranchise them and other members of their community, it may have been better just to encourage everyone to go along with the crowd.