The Victory Lab (31 page)

Grebner went on to build Michigan’s first statewide voter file, an arduous task in a state famous for its decentralized registration policies and poor record keeping, and then did the same in Wisconsin. He happily sold the files to Democrats and Republicans. It was a decent business, run out of a second-floor office suite above a pinball parlor, in the back of which Grebner maintained a small apartment he insisted was not his primary residence. (“I don’t live here exactly, but I spend the night here,” he explains, by way of not explaining all too much.) Having his own voter file also gave
Grebner a new playpen for his extraprofessional curiosities. In the language of statistics, he had six million data points (Michigan voters) and fifty replications (elections available in individual vote histories), and he enjoyed wading through them in different combinations, looking for patterns to who votes and when. “It’s like a pig farmer who just naturally has pig shit on his boots,” says Grebner. “I just naturally have data clinging to me.”

Grebner never stopped thinking about the 10 percent of the electorate that regularly claimed, falsely, to have voted. He would look for their traces in the voter rolls, like a detective always waiting for the one fresh clue that could take a vexing cold case and heat it up again. Grebner certainly understood why some people voted, and he could comprehend some of the reasons why others didn’t. But the 10 percent who existed in this liminal space between voting and not voting seemed to offer Grebner a skeleton key to unlock the internal arithmetic of voter behavior that confounded political scientists. In some ways, the liars seemed to have the most finely tuned valuation of the costs and benefits of voting. “Why do people lie? They’ve discovered it’s a secret whether or not you vote! You discover you can get out of voting by lying, and once you discover that why waste half an hour?” asks Grebner. “Fifty percent of Americans say they go to church on Sunday. Go count cars in parking lots. They must all walk to church!” Grebner, whose instinct for blasphemy is so refined that when he wants to describe something unfavorably he reflexively analogizes it to organized religion, pauses as he winds toward a constructive solution to this problem. “If they believed they couldn’t get away with it, how would that affect them?” he asks. “How about if we threaten to expose them?”

In the spring of 2004, Grebner picked up the new book
Get Out the Vote!
, a compendium of campaign field experiments by Don Green and Alan Gerber, who had included in their text an open call for new collaborators to contact them. Late on Friday night at the beginning of the July Fourth weekend, Grebner wrote a long e-mail to Green and Gerber to describe an experiment he planned to run in the next month’s primary: mailing entire Michigan blocks what he called voting “report cards.” Before
the primary, everyone would receive a list of the block’s residents with blank boxes next to their names, and then after the election he would send another round in which those who vote would get a checkmark. “The question I’m looking at is different from the approaches you have tested: stressing the importance of voting,” Grebner wrote. “Instead of appealing to their civic responsibility, I plan to try to work their sense of shame.” He told the Yale professors he would be back in touch at year’s end, when he would send over the results of his test.

But the academics didn’t want to wait that long to hear more about his ballsy experiment, and within days the Michigander was predictably doing most of the talking in a two-hour phone call as Gerber scribbled notes trying to keep up. Grebner explained that before he had even read
Get Out the Vote!
he had tried to run his own field experiment shaming nonvoters but had not collected the results in a form useful for analysis. (“It was a personal project, designed to learn rather than to convince anybody else,” Grebner explains later.) Nonetheless he sent the materials he had used to New Haven. “His original mailings were quite a quirky bundle,” Green observed. He and Gerber thought Grebner had muffled their potential effectiveness, especially by prominently labeling the letters with the word
experiment
. To ensure that the findings from Grebner’s next mailing passed scholarly standards, Gerber and Green proposed that they design the experiment from Yale and contract with Grebner’s firm to prepare the list and handle all the mail. To refine the approach, they postponed Grebner’s planned August launch and looked ahead to Michigan’s 2006 primary, and began familiarizing themselves with the psychological underpinnings of shame. Gerber opened the
Handbook of Social Psychology
to identify what he called the “active ingredients” of social pressure, and drew up different letters that would trigger them separately so that the experiment could compare their effects.

Gerber wrote four letters, each with a reminder of the upcoming August 8 election and the sign-off “DO YOUR CIVIC DUTY—VOTE!” The first message, the least provocative,
was designed to gently scold.
“Why do so many people fail to vote? We’ve been talking about this problem for years, but it only seems to get worse,” it said. “Your voice starts with your vote.” The next was inspired by a legendary industrial study at an Illinois factory called the Hawthorne Works, which found that worker productivity changed when the subjects knew they were under observation. “YOU ARE BEING STUDIED!” the so-called Hawthorne treatment letter warned, noting that the analysis would rely solely on public records to track why people do or do not vote and offering his reassurance: “Anything we learn about your voting or not voting will remain confidential and will not be disclosed to anyone else.” (As with all university-funded research, Gerber and Green had to get this experiment approved by the Yale Human Subjects Committee.)

The other two letters were written to jolt nonvoters out of their complacency. One was based on
the idea of “norm compliance,” a theory associated with Robert Cialdini after he used a littering experiment to demonstrate that people adjust their behaviors to match what they think their peers do. “WHO VOTES IS PUBLIC INFORMATION!” announced the letter addressed to a whole family at a single address, above a chart naming each voter in the household and whether they voted in elections held in August and November 2004. “The chart shows your name from the list of registered voters, showing past votes, as well as an empty box which we will fill in to show whether you vote in the August 8 primary election. We intend to mail you an updated chart when we have that information.”

The most incendiary letter included a similar chart, but instead of revealing a recipient’s vote history to other members of the household, Grebner found all the registered voters on the block and included theirs. “We’re sending this mailing to you and your neighbors to publicize who does and who does not vote,” the letter explained. “The chart shows the names of some of your neighbors, showing which have voted in the past. After the August 8 election, we intend to mail an updated chart. You and your neighbors will all know who voted and who did not.”

Once the letters had been prepared, Green set up a betting pool for himself, Gerber, Grebner, and Christopher Larimer, a doctoral student
they had enlisted to work on the project, to wager on which they thought would have the biggest impact on turnout. They all agreed that the “neighbors” mailing, the most heavy-handed of the bunch, would lead, but they differed slightly on how large the effects would be. The four estimated the neighbors letter would increase turnout by 3 percent, which would put it ahead of the straightforward get-out-the-vote mailers Gerber and Green had tested elsewhere. As a result of that estimate, they set up their experiment with massive samples, totaling eighty thousand across treatment groups, so that if one piece of mail was four times more effective than the next, the numbers would be large enough to calibrate meaningful distinctions. Grebner prepared the letters on laser printers in his office, and as he often did, hired local high school students to sit around an ovaloid table and run the baby-blue sheets through a folding machine.

A few days after he had wheeled the last of the letters to the post office, Grebner learned that he and his collaborators had all dramatically underestimated their ability to alienate voters. Grebner had suspected voters might not take well to the bullying tone of the neighbors letter, and the academics had reason to concur. The behavioral science literature was split between case studies in social pressure, where outside attention performed a suasive role, and of “reactance,” in which people respond to overbearing advice not to drink or smoke by doing ever more of it. But tobacco warnings rarely include a phone number for the surgeon general; Grebner had listed his own after the words “For more information,” along with his e-mail address and a return PO box, on each of the letters he sent out.

Grebner was never able to calculate how many people took the trouble to complain by phone, because his office answering machine filled so quickly that new callers were unable to leave messages. (When he tried, unscientifically, to match complainants to particular mailings, it looked as though those who got the neighbors letter were angriest.) Then there were the people who found a live line for Practical Political Consulting. One man, who Grebner’s office staff said sounded drunk, announced that he was in Bay City and headed to East Lansing to get Grebner. Then he called later
to report he was in Saginaw en route, and then in Flint. Grebner knew the geography of his state well enough to become nervous. “Then he called us from Durand and said he had sobered up and he wasn’t going to come down, and please tell the police he was sorry that he threatened us,” Grebner recalls.

Grebner found more satisfaction when he learned, the day before the election, that Nolan Finley of Livonia, an upscale Detroit suburb, had received one of the twenty thousand neighbors letters. Finley was the editorial-page editor of the conservative
Detroit News
, a favorite antagonist of Michigan Liberal, a group blog that Grebner had helped to launch the previous year. In a signed editorial in the
News
, Finley communicated his outrage at having his personal vote history—which showed his dutiful turnout for general elections every two years but poor participation in primaries—made public to those around him. “It turns out for all of Nolan having written probably a thousand editorials telling people to vote against every millage and vote against these candidates for judge, he actually hadn’t voted in a primary himself in twelve years, because he’s too fucking lazy,” says Grebner, in a voice that does not suggest the dispassionate social science experimenter. “His wife votes in elections, and his kids vote in elections, but Nolan is such a dishonest piece of shit that he hadn’t actually voted. Which is as far as I’m concerned a good thing! Nolan gets this thing dropped on him—I don’t know if his wife saw it, but it basically says: ‘Your husband’s a lying asshole! It just says that right there: Your husband is a lying asshole!’ ”

A few months later, Grebner smiled when he saw that Finley had cast his first primary ballot since 1992. (Finley told the publication
Stateline
that he didn’t typically vote in primaries because identifying with one party would compromise him as a journalist.) The state had updated its voter file, with records of which Michiganders turned out in the primary, and Grebner was busy matching them against names of residents of the 180,000 households in the study, which represented nearly 350,000 people. When he calculated turnout percentages for each group, the results were staggering. The control group had voted at 29.7 percent, those who received the
“civic duty” message at 31.5 percent, the Hawthorne treatment 32.2 percent, and the “self” mailer that showed other members of their household at 34.5 percent. Among those who received the neighbors letter, 37.8 percent voted—which meant it was roughly three times more effective at increasing turnout than any other piece of mail ever tested. It was several times better to deliver a threatening letter to a nonvoter than to have a neighbor sweetly remind her of the importance of voting in the upcoming election. At one point, Gerber calculated that the neighbors mailing had increased turnout at the enviable price of two dollars per marginal vote, and began to estimate that they could have reached the entire electorate for half a million dollars—and, for instance, swung the outcome of a Republican senate primary. “Alan,” Grebner said, “if we had spent five hundred thousand dollars and covered the whole state, you and I would be living with Salman Rushdie.”

They agreed that they had isolated the single strongest tool ever recorded for turning nonvoters into voters, but they already knew that it would be unlikely that anyone—a candidate or an independent group—would ever put its name on such a mailer. “How,” asked Green, “can you take this static electricity and turn it into useful current?”

IN THE SUMMER OF 2006
, Bazerman invited Rogers to join the Consortium of Behavioral Scientists, a secret society formed by University of California, Los Angeles, psychologist Craig Fox to help Democrats apply cutting-edge academic research to liberal politics. Many of the signature discoveries of behavioral science had clear relevance for policymaking, from automatically enrolling workers in retirement plans (and putting on workers the onus to opt out as opposed to opting in) to showing consumers their neighbors’ energy-use levels as a form of social pressure. Thaler and a University of Chicago colleague, Cass Sunstein, had begun writing of these “nudges”—the government gently pushing people to make better
decisions—comprising the core of an ideology they liked to call “libertarian paternalism.” But even so they avoided partisan activity. “I kind of stayed explicitly away from public-policy debates,” says Thaler. “I did not want behavioral economics to be viewed as a politically motivated field of study. It was hard just getting taken seriously, and there were some people here who always suspected that behavioral economics was just a sheen for government interference.”