Choice and Chance in the 2012 Presidential Election
By John Sides, Lynn Vavreck
PRINCETON UNIVERSITY PRESS Copyright © 2013 Princeton University Press
All rights reserved.
Here is a number: 68. That is how many moments were described as "game-changers" in the 2012 presidential election, according to an exhaustive search by Tim Murphy, a reporter at Mother Jones magazine. A few of these allegedly game-changing moments were cited in jest. The writer for the Celebritology blog at the Washington Post joked that the troubled actress Lindsay Lohan's endorsement of Mitt Romney could be a game-changer. A few of these moments were plausibly important, like the raid that killed Osama bin Laden. But the rest of the list is comprised largely of blips that failed to transform the race: the attempted recall of Wisconsin governor Scott Walker, Ann Romney's speech at the Republican National Convention, the Des Moines Register 's endorsement of Mitt Romney, a video of a Barack Obama speech from 1998, and so on and on. All told, Murphy found that the term "game-changer" had been mentioned almost twenty thousand times in the ten months before the election. It was, according to one reporter, the single worst cliché of the campaign.
The search for "game-changers" may make for grabby headlines, but it does not really help us understand presidential elections in general and the 2012 presidential election in particular. American presidential elections are rarely decided by a single moment that changes everything. Instead, presidential elections depend on national conditions, especially the state of the economy, and the candidates' efforts to mobilize and persuade voters. Both are important. Our goal in this book is to understand what role these factors, including some of the 68 alleged game-changers, played in deciding the 2012 election and what that might teach us about the dynamics of presidential campaigns. In other words, how and why did Obama win again? Our answer turns more on the advantages incumbent presidents have even in slowly growing economies and less on television advertising or field organizations, although we will investigate the effectiveness of both of these, too. Could Romney have won? To do so, he would have needed large advantages in campaign effort that were hard to achieve in a presidential election that pitted two qualified, well-financed candidates against one another. That neither Romney nor Obama had a large and sustained advantage helps explain how the $2.3 billion that they and their allies spent did not move the polls very much. Presidential campaigns do affect people, but because Obama and Romney matched each other almost dollar for dollar, the effects of this campaign were often in equal and opposite directions.
Choice and Chance
We think of an election outcome as having two ingredients: choice and chance. By choice, we mean the choices that the media make—in terms of who and what to cover, and how to cover them—and the choices that candidates make, such as their message, when and where to run advertisements, and how much to invest in a field organization. By focusing on choice, we zero in on many of the things that were called game-changers—if not quite all 68. By chance, we mean the circumstances in which candidates find themselves and over which they have less control, such as how well the economy is doing, who their opponent is, who endorses them, and the balance of Democrats and Republicans in the electorate. We call these circumstances the "fundamentals" of elections. They are the national conditions that set the stage for the campaign. By studying choice and chance simultaneously, we can compare their impact. We can evaluate how much work the fundamentals do to influence elections alongside the work that candidates do to win votes. This helps us answer such questions as: roughly how much campaigning might be necessary to overcome a disadvantage in the fundamentals?
Our story has three central lessons. First, as political science research has long shown, the fundamentals structure outcomes in advance of the campaign. They provide predictability and stability to how people vote and thus to who wins elections. The fundamentals typically do not make for headlines about "game-changers." But they make a difference. In the Republican primary election, support within the network of party leaders strengthened Romney's campaign and helped him survive a series of challenges from the other candidates. In the general election, people's longstanding identification with a political party was a powerful influence: most people who actually voted knew how they would vote a year before the general election. Moreover, the advantages based on economic fundamentals were hard to dislodge. In 2012, a slowly growing economy was enough to make Obama the favorite, and that advantage was borne out on Election Day.
Second, media coverage and campaign strategies can shift votes, thereby creating volatility in the race. But this happens only when the coverage and campaigning favor one side—such that one candidate dominates the airwaves, for example. In the Republican presidential primary, news coverage and intense campaigning produced instability because lopsided moments occurred often, creating the surges of many different candidates. But in the general election, news coverage and intensive campaigning were accompanied by stability among voters precisely because Obama's and Romney's efforts were so balanced. We liken the general election period to a "dynamic equilibrium," whereby a vigorous and dynamic campaign often produced a stable equilibrium in the polls.
Third, when imbalances during the campaign do occur, their effects are often short-lived. In the primary, the surges of the candidates were sustained by favorable coverage that came after the media's "discovery" of the candidate's potential. But within a few weeks, increased media scrutiny of the surging candidate helped bring about the candidate's decline. In the general election, where the media coverage did not systematically favor either candidate, it was the political advertisements whose effect did not last long. Any advertising advantage that Obama or Romney opened up in the battleground states on a Monday would shift votes in his favor that day. But most of that shift would be gone by Tuesday unless the campaign sustained its advantage. And because both Obama and Romney, combined with their allies, spent roughly a billion dollars each, it was tough for either candidate to dominate the airwaves for long. The Obama and Romney campaigns largely neutralized each other's efforts, leaving Obama, the candidate who benefited most from the underlying fundamentals, the eventual winner.
Our Moneyball Approach to Understanding Political Campaigns
"We're picking up steam," a tired but enthusiastic Romney campaign staffer told Vavreck in September. "The rallies seem really energized. People love him. We're going to win this thing!" According to a top Romney strategist, staff who traveled to battleground-state rallies from the campaign's Boston headquarters came back and said the same thing as Election Day approached. One staff member who attended a rally in Philadelphia said, "That is not what a losing campaign looks like." Yes, it just "felt" like Romney had momentum. But believing in the size of the rallies showed how feelings and gut instinct could lead you astray.
These stories reminded us of Moneyball: The Art of Winning an Unfair Game. In this book, Michael Lewis describes how Billy Beane, the now famous general manager of the Oakland Athletics, turned baseball into "moneyball" by eschewing expensive superstar players—which the team could not afford given its small payroll—and building a team of bargain players who had undervalued skills. Beane found these players by using data to identify the specific skills that were most important for winning games and then finding lesser-known players who had those skills. His old-fashioned scouts were skeptics of this approach, rejecting the data for what they "knew" in their gut to be true about players.
Today, virtually every baseball team uses advanced statistical analyses to evaluate players, and moneyball has bled into other areas, including elections. More and more, campaigns are not relying on intuition but on data about what does and does not work to persuade and mobilize voters. The Obama campaign pushed further in this direction than any previous campaign. For example, they sent out different fund-raising emails to random groups of supporters to see which subject lines and messages would generate the most contributions. Before these tests, Obama's staff members sometimes bet on which email would be the most effective. As one later reported—in a telling indictment of relying only on instinct—" We basically found our guts were worthless."
Moneyball may have infiltrated campaigns but unfortunately not a great deal of political commentary. The same bits of folklore are trotted out as if they are fact in election after election. For example, "undecided voters break for the challenger" emerged again in 2012, especially among Romney supporters who believed it meant they were destined to win. One of our major motivations in writing this book was to inject a bit of "moneyball" into our understanding of presidential campaigns—to do for explanations of why the winner won what forecasters like the New York Times 's Nate Silver or the Huffington Post's (and fellow political scientist) Simon Jackman did to predict who would win. Our approach is deeply informed by data and seeks to test the instincts and pet hypotheses of commentators, practitioners, reporters, and academics alike in a rigorous and scientific manner. The best way to know whether Obama won because of his ground game or Romney lost because he was perceived as an out-of-touch plutocrat is to delve into hard numbers on field organization or voters' perceptions of Romney and estimate their effect, while simultaneously accounting for competing explanations. Other books may spell out why the campaigns did what they did, but The Gamble demonstrates whether what they did made any difference.
Our approach goes beyond firsthand accounts from campaign insiders. While insiders can provide valuable insight into why decisions were made, their assessments of whether those decisions made any difference may be biased. To be fair, a strategist's job is only to win an election. In the heat of a campaign, they do not have the time to conduct a science experiment to figure out what is working. Nevertheless, after the campaign, insiders often act as if they know exactly what worked and why. If their candidate won, the argument too often goes, then clearly what they did worked. They have an incentive to exaggerate the campaign's impact and the genius of their own strategizing. After all, if people perceive them to be brilliant, it helps ensure they get hired when the next election comes around. We are not suggesting that all campaign consultants do this. But the temptation is always there, and it makes relying on only paid strategists to understand how much the campaign affected voters a bit like relying on doctors who are paid by a drug company to tell us whether its new drug works. The incentives for campaign consultants are all wrong.
By contrast, journalists and commentators—who also opine on why an election turned out as it did—are more likely to provide something campaign operatives often cannot or will not: actual scrutiny of what campaigns did and said, which sometimes sheds light on the true consequences of campaigning, including ones that the operatives would prefer to downplay. But journalists also face conflicting incentives. After all, it will not make the front page to write that nothing important is happening on the campaign trail or that the candidate gave the same speech today that he gave yesterday and the day before that. So at times journalists seem to "root for the story"—suggesting that campaign moments may be, could be, might be critical. Maybe even game-changers. As one writer for a major political publication told us during the campaign, "I generally try to spot potential trends before they become conventional wisdom." Of course, trying to "spot potential trends" can mean pushing a dramatic thesis well before the evidence actually backs it up. Journalistic accounts of campaigns also tend to put journalists themselves in the background. These accounts implicitly suggest that journalists are merely observers of the story of an election rather than characters in that story—even though, as we will show, the tenor of news coverage can profoundly affect the public's views of the candidates, especially during primary elections.
In explaining the 2012 election, we take conventional wisdom and political reporting seriously, investigating many campaign moments to see how they may have moved voters. But we also draw on a rich research tradition in political science and other social sciences that evaluates the effects of presidential campaigns. This research identifies patterns that occur and reoccur over many elections, giving us a sense of what to look for and where to look for it. Indeed, we are but one in a long series of political science studies of individual elections, stretching back to the 1940 presidential race.
We draw on multitudes of data from the 2012 election and previous elections. Historical data on the economy, presidential approval, and presidential election outcomes. Data from hundreds of public polls. Data on the individual opinions of approximately 45,000 voters who were interviewed three times in YouGov polls—once in December 2011, once again between January and Election Day, and once again after the election. Data on campaign news coverage in roughly 11,000 different outlets, including how often they reported on each candidate and how favorable or unfavorable that coverage was. Data on the millions of political advertisements aired, including when they aired, where they aired, and how many were on the air. Data on the location of candidate field offices.
These data enable us to document quantitatively the state of the economy and other fundamentals, the volume and tenor of news coverage, the ad blitzes and ground game in the battleground states, and the shifts (or not) in the polls. These data allow us to arrive at firmer conclusions about why things happened and especially why Obama won. They allow us to adjudicate among alternative explanations. In these data we will find many correlations, but we aim to go beyond these relationships to figure out which correlations are more likely to imply causation. Adjudicating among explanations also means going beyond platitudes like "campaigns matter at the margins" or "in close elections, campaigns can make the difference." We will present much more specific estimates: if Obama aired 1,000 ads in a media market and Romney aired none, how many points did that earn Obama in the polls if nothing else changed? The result is a picture that—whether or not it confirms earlier political science research, folklore, conventional wisdom, or casual punditry—is empirically richer and more robust.
None of this is to suggest that quantitative data can answer every question about the effects of a campaign. Campaigns are not experiments. Often we cannot easily sort out correlation from causation or we simply do not have the exact data we need. Although the Obama campaign was willing to conduct small-scale experiments with fund-raising emails, neither they nor the Romney campaign was conducting large-scale experiments with campaign messages or campaign advertising. Campaigns do not place ads randomly and evaluate whether voters who saw ads voted differently than voters who did not see them. They place ads where they need votes or, more precisely, where they have some chance of gaining votes. Because campaigns are strategic our job is harder and our analyses will come with some uncertainty.
Furthermore, although we can measure a lot of what happened and whether or how it appeared to affect the election's outcome, we cannot know what would have happened if various events had never occurred. Absent a time machine and the ability to rerun an election under different conditions, we cannot observe the counterfactual path that never materialized. But we can make educated guesses about how alternative scenarios might have played out. We can do things like estimate how many more ads Romney would have had to run to beat Obama, how many fewer field offices Obama could have had and still beat Romney, or whether Obama's barrage of advertising about Romney's time at Bain Capital really defined him. In addressing these questions, we can never know for sure what might have been, but data help us evaluate the plausibility of different scenarios.
In primary elections, the candidates are often relatively unfamiliar and voters cannot use their own identification with a political party to choose among the candidates. Because of this, they rely on other information, which is supplied by the news media and the candidates' electioneering. Unlike in general elections, the news media and electioneering are often imbalanced. At different points in time, one candidate may receive much more, and more favorable, news coverage than the others. Some candidates have enough money to run well-funded, professional campaigns, while others are running campaigns on a shoestring budget. As a result, when voters are inundated with information favoring one of the candidates, they often gravitate toward that candidate. In 2012, that happened repeatedly, creating so many ups and downs that the polls seemed almost random. But there was an underlying logic at work. (Continues...)
Excerpted from The Gamble by John Sides, Lynn Vavreck. Copyright © 2013 Princeton University Press. Excerpted by permission of PRINCETON UNIVERSITY PRESS.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.