The Victory Lab (32 page)

But the Bush administration radicalized Thaler and many of his peers, who were liberal in their personal orientation but sympathetic to markets and saw a Republican Party seemingly resistant to the type of empiricism that defined their worldview. In 2004, Fox and his mentor Daniel Kahneman, whose Nobel Prize two years earlier had helped to validate the study of flawed decision making in economics, were lamenting that John Kerry had allowed himself to be defined as a flip-flopper. Fox and Kahneman thought it was typical of what they thought of as ruthless, disciplined Republican messaging, like the famous example of pollster Frank Luntz rebranding the estate tax as a “death tax,” and they thought that as psychologists they had insights that could help Democrats respond. Fox started attending Los Angeles fund-raisers in the hopes of meeting prominent Democrats. At one, he encountered Congressman Patrick Kennedy and handed him a brief memo outlining what he thought behavioral science could bring to politics, which in turn found its way to Steve Elmendorf, a deputy campaign manager for Kerry. Fox and Cornell psychologist Tom Gilovich and several peers met with Elmendorf in Washington, where they presented two white papers on the psychology of messaging. “Our style was to be intellectually honest but try to offer advice. If you qualify everything, and put too much jargon in, people aren’t going to use it,” says Fox. But Kerry’s campaign never appeared to adopt any of the advice. “In hindsight it was spectacularly naïve to think we’d have any effect at all six weeks before the election,” Fox says. “They had lots of issues they were dealing with at the time.”

After the election, Fox kept the group together as the Consortium of Behavioral Scientists, and as the 2006 midterms approached, the group’s
e-mail list became a lively forum for trading notes on election-year strategy. For many of the participants, who had never before been involved in politics, the group’s activity augured new relevance for their once-marginalized work. “Why is it psychology and other behavioral sciences have less of an impact than economics on policymaking?” laments Fox. Because consortium members knew their involvement made them easy targets for critics—who happily ridiculed the approach as highbrow nannystatism or even ivory-tower-approved government mind control—many adopted a
Fight Club
mentality toward publicity, even refusing to speak the group’s name in front of strangers. But when Fox met Senator Hillary Clinton as part of a 2006 fund-raising effort, he told her about his consortium and the science of messaging. Several weeks later, she called back to invite the group to brief the Democratic Senate leadership.

Fox extended the consortium’s circle, in essence preparing to lead a high-level delegation from academia for a strategy session with national opposition leaders, and planned a daylong meeting in New York to prepare. As he looked over his ranks, Fox realized he had one Nobelist and multiple members of the National Academy of the Sciences but hardly anyone who had ever seen the inside of a campaign headquarters. “Most of us were amateurs at politics,” Thaler recalls. “We knew our stuff, but it was good to have somebody who knew the other side as well.” Bazerman proposed that Rogers, the only graduate student on an e-mail list that didn’t even include any assistant professors, join the group in Washington. The twenty-nine-year-old had already impressed him with not only a high quality of scholarship (which wasn’t unusual among Bazerman’s students) but natural political instincts (which were). “There are a lot of people who assume he’ll be a congressman or a senator someday—by which path no one knows,” says Bazerman.

Less than two weeks later, a small group of consortium leaders took their seats across from Clinton, Senate Minority Leader Harry Reid, and Senators Debbie Stabenow and Byron Dorgan, both members of the party’s leadership, in Reid’s office suite. Over the course of a campaign year,
politicians hear many supporters give them unsolicited advice about lines they should use in their speeches and ads they should run, but these presentations may have had an unusually prestigious patrimony. New York University psychologist Susan Andersen told the senators their messaging could be more emotionally resonant if they stayed away from aspirational themes. People are more sensitive to losses than equivalent gains, Andersen explained, so Democrats should focus their language on restoring what’s been lost. Promise to “reverse erosion of our security,” she proposed, or a “return to a balanced federal budget.” When it was Thaler’s turn, he looked at Clinton and employed a pedagogic tool he has honed over three decades of teaching a “Managerial Decision Making” course to cocky students at the University of Chicago’s Booth School who arrive in his class believing that they are too smart to make the mistakes that Thaler believes infect markets. He begins the inaugural session of his semester-long course with a rapid-fire series of stunts and quizzes he privately calls “the magic show,” which is designed to humble his students into understanding that their brains are as feeble as everyone else’s. Now Thaler read the senator a brief excerpt from a speech in which entreaties to vote were paired with statistics warning against low turnout, before citing Rogers’s research that argued for the opposite approach. Clinton, sheepishly, recognized the words from one of her own speeches. “I wanted to show her how easy it is to get this wrong,” he says. “This is from long experience teaching people about decision making. You first have to show people that they make this mistake.”

Rogers had happened upon the nexus of two scholarly minded subcultures lurking in the shadows of the Democratic Party’s exile from power: the behavioralists who thought they could direct government to help citizens make better decisions, and the political scientists teaching campaign consultants how to manage field experiments. By the time Democrats won back Congress in November 2006, as many as sixty people were coming to Mike Podhorzer’s geek lunches, and participants had begun calling themselves the Analyst Group. The following summer, Rogers, who had not attended any of the sessions, was invited to speak. He gave a
presentation called “The Behavioral Science Toolbox,” in which he argued for a “psychologically enriched understanding of why people vote.”

Any Analyst Group lunch attendees with a preconception of what being lectured by a psychology graduate student would be like were probably surprised once Rogers walked to the front of the room. He maintained an athletic build typically hidden behind billowing dress shirts and pleated khakis, and possessed an intense ebullience that came out as soon as he began talking about his research. Even though he was fully committed to the Gerber-Green experimental method, Rogers was trying to subtly distance himself from one of the conclusions that had emerged from their body of work. In 2004, when they had reviewed the findings from their experiments and those of acolytes, Gerber and Green had concluded that when it comes to GOTV, “
the message does not seem to matter much.” Instead, they then attributed the tremendous disparity in effects—an
average 8-percentage-point boost from in-person canvasses, compared with less than 1 point from paid phone calls or direct mail—to means of delivery. “
Face-to-face interaction makes politics come to life and helps voters to establish a personal connection with the electoral process,” Gerber and Green had written. “
To mobilize voters, you must make them feel wanted at the polls. Mobilizing voters is like inviting them to a social occasion.”

Rogers didn’t quite disagree with the Rockwellian glow that brightened these sentiments, but he thought that there were other, perhaps darker social dynamics that could mobilize someone to vote. Political scientists who had concluded that message didn’t matter had just been testing the wrong ones. Rogers argued that campaigns needed to treat being a voter not as a temporary condition that switched on or off each election day, but as a form of identity crucial to individual self-consciousness. “Changing how people see themselves can change behavior,” Rogers said, citing an experiment in which people who had been informed by a survey taker that they were an “above-average citizen” turned out at a higher rate than those told they had an “average likelihood of voting and participating in politics.” He summarized the experiment that Grebner had run in
Michigan, which preyed on a similar vulnerability in voters: their desire to conform to what they thought were their peers’ expectations for good citizenship. Rogers emphasized that the Grebner test was also important because it considered an individual’s mind-set before, during, and after election day. Campaign operatives could no longer think of voter contact as a series of discrete communications, Rogers argued, but needed to consider voting as a “social behavior extended over time.”

Rogers described an experiment that he and Gerber had conducted just a few months earlier, in the run-up to Kentucky’s gubernatorial primary. The week before election day, 660,000 households in the state received an automated phone call that announced itself with a recording: “This is Voter Roll Call with a two-minute opinion survey about Tuesday’s election. If you are registered to vote in Kentucky, press 1.” The 68,490 people who did so and stayed on the line were asked how likely they thought they would be to vote. If they signaled an interest, an exchange ensued that, amid the frenzy of political calls that weekend, was notable only for what it did not ask. Unlike pollsters’ calls, there was no question about which candidate voters preferred in the Democratic and Republican primaries, or why. And unlike canvassers for either candidate, or the unions backing them, callers did not offer a ride to the polls or last-minute help turning out to vote. The recorded voice could have been asking about any election anywhere.

Rogers did not care what voters’ answers were to the three questions, only whether they had any. He was testing a psychological concept he referred to as the “plan-making effect,” which suggests people are more
likely to perform an action if they have already visualized themselves doing it. There was a rich history of testing these “implementation intentions,” as Rogers described them, in areas other than politics. One experiment conducted on a college campus involved a set of reading materials that students were told were optional for a course. One set of students was asked to develop a plan for when they would drop by their teaching assistant’s office to pick them up: when they would go, the way they would get there, how long it would take. Another set was not asked to formulate a plan for picking up the materials, and eventually did so at a lower rate. Rogers thought he might be able to nudge people to vote by tricking them to rehearse parts of their election-day regimen in advance, inadvertently forcing them to develop a plan.

The presentation before the Analyst Group was something of an audition for Rogers. The lunches’ reach had become so broad that their participants needed a structure to coordinate new research projects and share findings, and Podhorzer was scheming to transform the informal Analyst Group into the Analyst Institute. The new entity would operate with the sensibility of a think tank but the closed books of a private consulting firm. As Podhorzer, the institute’s chairman, searched for an executive director—reviewing résumés (from the political professionals he talked to for the job) and curricula vitae (from the academics)—he appreciated how much the pick would indirectly determine the nascent group’s priorities.

Many of the recent lunchtime presentations had featured microtargeting projects, which had been the source of so much innovation and investment on the left, and had become much easier for many after Catalist developed a common data resource. In his job interviews, Rogers made clear his utter disinterest in building new microtargeting models, which he acknowledged had predictive power but couldn’t actually conclude what changes human behavior, and his sole interest in running randomized field experiments. He eventually won over the Analyst Institute’s new board by arguing that randomized tests were the area where the new entity could quickly deliver the most value. Rogers liked to describe their purpose as
intellectual arbitrage, bridging the gap between the prolific output of academic experimenters like Gerber and Green and their lack of penetration into the world of political practitioners. “The modeling stuff seemed to be pretty quickly fetishized as the answer to all campaigning needs,” says Judith Freeman, the executive director of the New Organizing Institute, who interviewed Rogers. “What Todd brought was not just the ability to run experiments or build models but to think critically about what was going to help us. He brought us back to the idea that we need to understand what motivates individual people.”

Given his limited background in politics, Rogers initially appeared an unlikely candidate to run the Analyst Institute. But his itching desire to humiliate experts proved him to be a perfect fit. Podhorzer approached Rogers with the executive director post as he was being pursued by Harvard’s Kennedy School of Government and Carnegie Mellon University. Rogers summoned Bazerman, who had never before had a student become anything other than a professor, for coffee in Harvard Square to discuss his options. Rogers was drawn to the advantages of doing academic-style work outside the academy: he could commission experiments without nonprofit funding restrictions or the bureaucracy involved in a human-subjects review board. He also realized that if he were to assume a teaching post and see one of his own students take the Analyst Institute job he would feel jealous. “The professor in me was slightly disappointed,” says Bazerman, “and the citizen part of me was thrilled.”

Rogers’s hiring heralded a major shift in the left’s research attentions. He ordered boxes of Cialdini’s
Influence
and Sunstein and Thaler’s
Nudge
, and began handing them out to his new colleagues as a sort of intellectual calling card. He scheduled coffees with representatives of as many liberal groups and party committees as would meet him, to convince them of the value of randomized trials and discuss possible ways to build experiments into their existing programs. Rogers had arrived in Washington with a long list of psychology concepts he hoped to test in the political arena, many of which were altogether foreign to many of those in the Analyst Group’s
world. Rogers felt that as long as he could demonstrate their effectiveness in the campaign context, they would be an easier sell to consultants and operatives who were pragmatic about vote-winning than they would be to political scientists whose theories of voter behavior were hardened with the stubbornness of philosophy. “We can bring political science theories to these topics, but I’m not sure how valuable those theories are and after eight years we had tried a lot of them,” says David Nickerson, a Gerber-Green protégé who went on to teach at Notre Dame and serve as a senior adviser to the Analyst Institute. “Now here comes Todd with a literature in social psychology that we hadn’t ever really seen.”