The Victory Lab (6 page)

Standards were changing rapidly, and it was no longer acceptable for a professor to publish a credible paper on public opinion that used data gathered by his own students, as Gosnell had in 1923. In the late 1950s, Gosnell, then working as a State Department analyst, approached pollster Clyde Hart to propose a reprise of his Chicago voting studies and suggested he might be able to raise ten thousand dollars to fund it. “He looked at me as if to say, ‘Where have you been, Rip Van Winkle?’ ” Gosnell recounted to a gathering of pollsters to which he had been invited by an old friend, Elmo Roper, who had been a pioneer of national surveys during Franklin Delano Roosevelt’s bid for a second term.

“Don’t you remember Elmo’s famous comment after the 1948 election?” Gosnell went on. “This was a priceless comment about the price of
polling. What were the lessons learned from that slight discrepancy between the polls and the election results in that Dewey-Truman contest? Was it that last minute change in calculating the turnout? No, none of these things. Elmo put it in a nutshell. The 1948 polls showed that polling is a complicated business. It is going to cost customers more.”

The costs did not dissuade Campbell, who was looking to develop the first systematic effort to explain how presidential elections were decided.
His 1952 survey came with a $100,000 price tag, covering interviews with 1,900 subjects and asking 224 different questions. The project, which later became the American National Election Studies, grew into the definitive data resource in political science: a massive biannual polling project that was expensive to collect but created a permanent repository of data on who voted and what they said about why, all with a consistency that made it easy to track changes through a campaign and from one year to the next.
Campbell’s questionnaire took an expansive view of its subject, with questions about not only the election under way but practical matters of political behavior (“did your coworkers’ opinions influence you?”) and philosophical approaches to citizenship (“should one vote if his party can’t win?”).

The responses guided Campbell toward nothing less than an all-encompassing theory of how elections are decided. Along with colleagues Philip Converse, Warren Miller, and Donald Stokes, Campbell concluded that a person’s partisan identification was the strongest predictor of how they would vote in national elections, even better than asking them where they stood on any particular issue. Parties were glorified social clubs, pulling people in because of class, regional, or religious ties and keeping them for the long term—with a sort of thoughtless choice resembling the inertia that led people to take the same jobs as their relatives. Individuals rarely switched parties over the course of a lifetime. Campbell and his colleagues described individual voting decisions with the image of a funnel: citizens’ social and psychological loyalties narrow them into a party, which usually guides them toward a candidate.

For all that, though, there were short-term disruptions that pushed
voters toward a candidate of the other party. After all, the same voters who had decisively elected Franklin D. Roosevelt four times swung broadly behind a Republican, Dwight Eisenhower, less than a decade later. Voters were attached to parties, but those bonds were generally breakable, the Michigan scholars argued, and sometimes a candidate comes along who is so appealing that his personal attributes overwhelm partisan loyalties. Converse liked to compare it to a big wind sweeping through a field of wheat, which leaves every stalk leaning in the same direction, although bending some more sharply than others.

In 1960, Campbell and his colleagues introduced this metaphor in the book
The American Voter
, the first universal, data-intensive study of electoral behavior, but the argument would prove poorly suited for its era. American politics convulsed in the late 1960s and 1970s, following the passage of the Civil Rights Act, and a partisan-driven model seemed tragically anachronistic. Within two decades, the South had become the base of Republican presidential coalitions even as most of its residents remained Democrats, and political scientists began to thrash about for a new way to explain the American voter. It became popular to say that people deserved more credit for the political choices they made. “The perverse and unorthodox argument of this little book is that voters are not fools,” V. O. Key Jr. wrote to start his 1966 treatise,
The Responsible Electorate
, which argued that many voters were “switchers,” rationally alternating between parties each election to find the candidate closest to them on the issues.

After Richard Nixon’s reelection in 1972, Michigan’s Warren Miller desperately rewrote the
American Voter
theory to keep up with changing times. The survey data for that year had shown, for the first time in the two decades of the national election studies, that party influence over how voters chose among presidential candidates had diminished markedly. The “issueless” fifties, as they put it, had been followed by a decade in which the country was politically riven on fractious matters of war and peace, identity and liberty, that crossed the old party lines. Voters ditched their
social clubs for the candidate who stood closest to them on policy. The Michiganders explained this by pointing to a way the electorate had fundamentally changed: Americans were better educated than before, and went to their polling places with a more enlightened interest in affairs of state. “Voters with a college education are better informed politically,” they wrote, and “therefore, more likely to make a vote decision on the basis of policy preferences than are less well-educated individuals.”

Such academic theories were barely acknowledged by those who worked in politics, and when they were it was often with skepticism. In fact, those on both sides of the Nixon reelection battle scoffed at the
American Voter
team’s reading of the landslide. The president’s pollsters, Bob Teeter and Fred Steeper, disputed the idea that “the 1972 patterns portend great ideological battles for future presidential elections,” as they wrote, “and that the political parties must change their issueless ways in order to cope with an increasingly polarized electorate.” Relying on their polls for Nixon, Teeter and Steeper delivered a new theory for what had prompted so many Democrats to unmoor from their party and dock with Nixon. They suggested that swing voters were no longer voting on issues, such as Vietnam or the economy. They may not have even had strong ideas of the right and wrong positions on the issues. Now they were giving their votes to the candidate who seemed best able to “handle” those challenges. Nixon, like Eisenhower, had established himself as a more credible leader on the issues of the day. This “candidate-induced issue voting,” as Teeter and Steeper called it, had as much to do with the candidates as the issues.

They found an unlikely ally in Samuel Popkin, a University of California, San Diego, political scientist who served as a campaign adviser to the man Nixon had defeated, George McGovern.
In 1972, Popkin had been a Harvard statistics professor when three of his undergraduate students, including Pat Caddell, sold the South Dakota senator his first poll for five hundred dollars. Soon Caddell was the chief strategist for the Democratic nominee’s campaign and enlisted Popkin, who was only thirty then
but still nearly a decade older than his excitable protégé, to join the recently minted Cambridge Survey Research trio as an in-house wise man and extra hand with the numbers.

After the campaign, Popkin aggregated all the polling data and tried to answer the same question the Michigan scholars had tackled: why had McGovern lost to Nixon by more than twenty points? McGovern’s early polls suggested the race should be competitive. But as it went on, Nixon’s lead widened, and the issues alone couldn’t explain such a gap. Even those who agreed with the dovish, liberal McGovern on his top foreign and domestic priorities were drifting away. In September, McGovern led Nixon among voters who considered Vietnam the most crucial issue, believed that the United States should withdraw immediately, and supported a guaranteed family income, by a margin of 52 to 38. By the end of the campaign, McGovern had lost them all. His internal polls showed him trailing even among those who thought the military budget should be drastically reduced.
McGovern hadn’t lost voters because he was out of sync with them on issues, Popkin argued, but because they thought he wouldn’t be able to do anything about those policies. They watched McGovern during the campaign and concluded he was incompetent.

Popkin thought voters were much savvier than the Michigan studies had initially cast them, but that even those with college diplomas could never gather all the information necessary to weigh the entire set of costs and benefits attached to each issue or candidate. They weren’t making a buying decision, because they wouldn’t get the product they eventually chose. Instead, thought Popkin, it made more sense to think of them as investors, who knew whatever information they gathered to inform their decision making would require time and effort. So when it came time to choose a candidate, they relied on shortcuts. They interpreted symbols and looked for cues where they could find them, and then extrapolated. In one of Popkin’s favorite examples, when voters saw Gerald Ford fail to shuck a tamale before biting into it, they interpreted it as a sign that he did
not understand issues facing Latinos. (Popkin had worked as a campaign adviser to Jimmy Carter in 1976.) Popkin called this “gut reasoning.”

Election scholars had ignored large swaths of modern psychology, which was increasingly identifying ways in which people were neither socially preprogrammed toward certain attitudes nor walking calculators able to make perfectly rational choices. In other academic disciplines those theories of human behavior had long fallen from vogue, replaced by a less elegant one. In the 1970s, two Israeli psychologists, Amos Tversky and Daniel Kahneman, began to document the ways that people were incapable of deciding rationally, and in fact kept making the same mistakes over and over again. Around 1980, a young economist named Richard Thaler began translating these insights to the way people handled money, and it became readily apparent that people weren’t as rational as economists imagined them to be. When forced to make decisions, people lacked a steady set of preferences. What they had instead were unconscious biases that made them bad at assessing situations and accurately judging costs and benefits.

But even a decade later, this basic insight—people are flawed, if well-intentioned, beings—had barely penetrated the political science department. “The science half of political science is to some extent a bit of a misnomer,” says Thaler, who in 1995 began teaching at the University of Chicago, just blocks from where Gosnell and Merriam had designed their field experiment to study voter behavior seven decades earlier. “At least no one has been quite ready to agree on what the science part of it is.”

In his 1991 book
The Reasoning Voter
, Popkin introduced a theory of voter activity equally informed by behavioral psychology and his own experiences within presidential campaigns. “
These contests are commonly criticized as tawdry and pointless affairs, full of dirty politics, dirty tricks, and mudslinging, which ought to be cleaned up, if not eliminated from the system. In their use of sanitary metaphors, however, many of these critiques confuse judgments of American culture with aesthetic criticisms
of American politicians,” Popkin wrote. “They do not look closely at how voters respond to what they learn from campaigns, and they do not look closely at the people they wish to sanitize. If campaigns are vulgar, it is because Americans are vulgar.”

This was a theory of the electorate that could make political professionals, increasingly under attack as overpaid Svengalis of spin, feel good about what they did for a living. In December 1991, Popkin wrote an oped for the
Washington Post
whose headline blared “We Need Loud, Mean Campaigns.” Paul Begala, a Democratic consultant, clipped Popkin’s article from the paper and handed it to his partner, James Carville. Carville and Begala, who had recently joined Bill Clinton’s campaign as lead strategists, were both loud, occasionally mean, and always unrepentant about the clangorous tone of the campaigns they ran. Popkin’s essay offered affirmation. Carville called Popkin to request a copy of
The Reasoning Voter
. Not long after, Popkin joined the campaign as an adviser. “He’s one of us,” says Begala. “He gets it.”

Popkin spent much of 1992 collecting polls and past election results to build simulations of electoral-college scenarios that could get Clinton to the necessary 270 votes. The results informed key strategic decisions: which states would get offices and staff, visits from the candidate and his family, and a precious share of the campaign’s budget for paid campaign communication with voters. “It’s an important decision in any war,” says Popkin. “Who’s going to pick the theater of operation?”

But as those strategic choices atomized into a series of tactical options, Popkin was amazed at how little he actually felt he knew. For two centuries Americans had been electing presidents, and for half of one century specialized scholars had been trying to rigorously study that process. Yet they had accumulated little information useful in deciding how to spend campaign dollars. As one of the few political scientists with access to a presidential campaign’s war room, Popkin had his feet in the worlds of people who practice politics and the people who study it, and neither field impressed him with its ability to judge what actually won votes. Popkin thought campaigns
had learned to be smart about how they picked their theaters of conflict, but once assigned to one, a general had only his instinct to rely upon in deciding whether to battle in the air (buying TV and radio ads) or on the ground (the hand-to-hand mobilization known as field).

“It’s the all-time question of every defense department in the world: army versus air force. What is the ultimate value in any war of a soldier versus a bomber?” asks Popkin. “You can target a state, and everyone could say ‘the swing voters are in Peoria’ and ‘
Oprah
costs this many dollars.’ ” But that information alone was of little use. “No one has any idea of the value of the ad versus a phone call from a friend,” he goes on. “If you have a dollar to spend, do you spend it on an ad or do you spend it on a phone call? And if you only have money to spend on ads, do you give one person fifty ads or two people twenty-five? Nobody knows.”