PREFACE TO THE PAPERBACK EDITION

The scientific study of presidential elections arguably dates back to a 1948 book called The People’s Choice. Its authors, Paul Lazarsfeld, Bernard Berelson, and Hazel Gaudet, set out to understand the 1940 presidential race between President Franklin Roosevelt and Wendell Willkie in an unprecedented fashion. Setting up shop in Erie County, Ohio, they interviewed a sample of residents every month from May through November of the election year, thereby tracking the opinions of the same people interviewed at multiple points in time.

What they found was surprising to many. At a time when the rise of fascism and the development of mass media—especially radio—had raised fears about people’s susceptibility to propaganda, Lazarsfeld and colleagues found that people’s views of the candidates were mostly stable. Very few people switched their support from Roosevelt to Willkie, or vice versa. Campaign propaganda did not seem as powerful as many believed it would be.

But the campaign still mattered, just in more subtle ways. For one, it helped to reinforce the opinions people already had. If you were a Roosevelt supporter, the campaign solidified this choice. The campaign also pushed undecided voters to the candidate that they were already predisposed to support. Undecided voters whose demographic profile made them look like Democrats mostly ended up supporting Roosevelt, and those who looked like Republicans mostly ended up supporting Willkie. Lazarsfeld and colleagues showed that a campaign could be consequential, even in an electorate in which many voters were not up for grabs.

In the years after this book, other scholars began to use social science methods and data to study elections. Lazarsfeld would help author a second book, this one about the 1948 election. Not long afterward, a team of social scientists at the University of Michigan wrote other seminal books about voting behavior and elections—particularly The American Voter—as well as articles about individual presidential elections that were published not long after the elections themselves. More recently, there have been notable book-length studies of the tumultuous 2000 presidential election and the historic 2008 election, when the first African-American, Barack Obama, was elected president.

The Gamble fits squarely in this tradition. It is the story of how Republicans nominated Mitt Romney to challenge Barack Obama in 2012, and how Obama ultimately won reelection. We tell this story using similar kinds of tools as these earlier books, including quantitative data and statistical methods. Indeed, one of our central sources of data is a survey not unlike the one in Erie County, Ohio: multiple interviews with the same set of voters over the year before the election. The main difference is whereas that 1940 survey focused on 600 voters in one county, ours includes 45,000 voters across the United States. These tools help us to identify why voters decided as they did and how the campaign affected them along the way. This, in turn, helps explain why Obama won and what implications his reelection has for party competition in future elections.

There is a parallel tradition of campaign narratives written by journalists. In fact, perhaps the canonical book in this tradition—Theodore White’s The Making of the President—was published not long after the first social scientific accounts appeared, when White documented the 1960 election between Richard Nixon and John F. Kennedy. White would write other installments after the 1964, 1968, and 1972 presidential elections.

Journalists working in this tradition rely on different sources of information than do social scientists. Rather than crunch data, they spend many months on the campaign trail, following the candidates. They conduct interviews with the candidates, their campaign strategists, and sometimes a small number of voters. These books seek to answer a similar question about the election—why did the winner win?—but mainly by focusing on the decisions the candidates made. These books thus paint vivid pictures of the characters in the campaign—their personalities, their strengths and weaknesses, their foibles and eccentricities. These books dwell on dramatic moments. They tell lots of good stories. After the 2012 election, several such books were written, including Dan Balz’s Collision 2012 and Mark Halperin and John Heilemann’s Double Down: Game Change 2012.

The Gamble differs in key respects from these books. We lack the access to the candidates that journalists can negotiate; and thus we cannot ask the candidates why they did what they did. What we can do, however, is figure out whether their actions made an impact on voters. That is where those 45,000 voters, a few graphs, and the statistics waiting in the appendix become useful.

To take one example, consider a storyline being pushed by Romney after his loss. He blamed his loss on the fact that Obama had given “gifts” to key constituencies. For instance, Obama’s signature achievement in his first term, health care reform, mandated insurance coverage of contraceptives. Romney argued that “free contraceptives” helped Obama woo female voters. In Double Down, Halperin and Heilemann report Romney’s claim, but lacking any means of testing it and any countervailing data from the Obama campaign, they let the claim stand. By contrast, we use our survey data to examine how much the debate over reproductive rights affected voters. Contrary to Romney’s claim, we found that the issue of reproductive rights did more to keep male voters from leaving Obama than it did to attract women to him. Most women for whom this issue was important had already chosen their candidate before the debate about contraception and the controversies that followed throughout 2012.

To be sure, both kinds of books about an election are valuable. It is important to understand what was going on inside campaign war rooms and what was going on inside the voting booth. Each tradition—social science and journalism—can complement the other. When they diverge, however, we think that science should be privileged, to the extent that it rigorously tests competing claims using carefully collected data. Interestingly, this belief has become prevalent not only among scientists, but also among some journalists and a fair bit of the population. Hard data and scientific approaches increasingly animate the conversation within journalism, perhaps most visibly in outlets like Nate Silver’s 538 website (which lived at the New York Times during 2012) and the wonkish blogging of Ezra Klein (in 2012 at the Washington Post). Campaigns are increasingly scientific themselves—gathering data, running experiments, and constructing statistical models to predict voter turnout and preferences for the candidates. Modern campaign strategy increasingly resembles what Sasha Issenberg provocatively called “The Victory Lab” in his book of the same title. There is every reason to think, then, that the understanding of campaigns and their effects will draw more and more on approaches similar to ours.

Part of The Gamble’s contribution draws directly on its social science lineage. In essence, previous studies told us where to look to identify important patterns and the likely impact of the campaign. For example, 70 years after the Erie County study, we find that many of the same patterns still exist. Preferences for Obama or Romney tended to be stable for the vast majority of Americans, and relatively few switched their support from one to the other. At the same time, the campaign helped to reinforce the views of those who appeared to make up their mind early on. And for those who remained undecided, the campaign led them to a predictable choice: undecided Democrats mostly gravitated to Obama and undecided Republicans to Romney.

We also push in directions less well-explored in the previous literature. One direction concerns presidential primaries, which remain vastly under-studied relative to presidential general elections. Indeed, the last major study of campaign dynamics in presidential primaries—Larry Bartels’s Presidential Primaries—was published in 1988. The 2012 presidential primary was a propitious one to study, as it featured a highly fluid Republican race with multiple frontrunners. We are able to show that the ups and downs in the polls followed a predictable pattern, driven largely by news coverage. We also show that Mitt Romney—who never experienced any real ups or downs and was often underestimated by commentators—remained the frontrunner by several key measures. Part of our contribution stems from new technologies that automatically gather and analyze news coverage from thousands of media outlets. This allows us to ascertain how much and what kind of attention the media paid to each candidate, and to do so in nearly real time as the campaign unfolded.

Another less-explored direction concerns the general election campaign. Only recently have scholars been able to measure the volume of advertising and field activity with anything approaching precision. In 2012, we marry data on both ads and field mobilization to our survey data and to data on the election results. This allows us to evaluate the impact of various forms of campaigning simultaneously and in comparison to other factors that affect election outcomes, such as the state of the economy.

Three key findings emerge. First, we show that when either Obama or Romney was able to out-advertise the other, the polls could move—but the effects of the ads typically wore off quickly. This contrasts with a piece of conventional wisdom from 2012: that the Obama campaign’s decision to air an early advertising blitz in the summer helped defeat Romney. In all likelihood, the effects of the summer ads on vote choice had long worn off by November.

Second, the Obama campaign’s field organization appears to have earned Obama votes, and more than Romney’s field organization earned him. But the number of votes won was not likely large enough to determine the outcome in crucial battleground states. This again contrasts with the conventional wisdom, which attributed his victory, implicitly or explicitly, to the strategic mastery of the Obama campaign on the ground.

Finally, the kinds of campaign effects that we identify—which certainly appear real—are small compared to the fundamental factors that affect elections. Romney would have needed vast quantities of additional campaign resources—over and above the $1.2 billion he and his allies already had—to offset the advantages Obama had as an incumbent president presiding over a slowly growing economy.

Our account, however, remains necessarily limited. The scientific understanding of an election takes time, and we will be learning more about 2012 for years. For one, 2012 certainly showed that money raised by outside groups is becoming a larger part of the money spent in an election campaign. The question is what impact it has, and whether its impact differs much from money raised by the candidates themselves. For example, during the Republican presidential primary, a few wealthy supporters of Newt Gingrich and Rick Santorum contributed millions of dollars to outside groups supporting them. Did this encourage Gingrich and Santorum to stay in the race longer, even though winning the nomination was always a long shot? Groups supporting Romney also raised a great deal of money during the primary and general election campaigns. How much did this help him? Would it have been preferable to raise more of his own money and rely less on these groups, given that Romney’s campaign team could have directly controlled how that money was spent?

The impact of new campaign tools also remains a contested subject. The tools of “The Victory Lab” have led to many innovations, especially in Democratic campaigns. In 2012, the Obama campaign used experiments with its fundraising emails to determine, for example, what kinds of subject lines would generate the most donations. Other innovations of the Obama campaign—such as having its supporters contact targeted voters via Facebook—also received a great deal of press. However, we know little about the impact of many of these innovations. There is no question that campaigns will continue to innovate—in how they deliver advertising to voters, in social media, and in other kinds of tactics and strategies. Measuring effectiveness, however, remains a significant challenge.

On the first page of The People’s Choice, the authors write: “Every four years, the country stages a large-scale experiment in political propaganda and public opinion.” This statement reminds us that, however consequential Barack Obama’s victory in 2012 proves to be, this particular election is just one of many that have occurred and will occur. Future elections will likely refine and improve on what we have learned to date.