Get our latest essays, archival selections, reading lists, and exclusive content delivered straight to your inbox.
Each American presidential election eventually turns into a story. In 1960 Nixon stumbled in the debates and lost to a more vigorous Kennedy; in 1988 Dukakis was self-defeatingly passive in response to an aggressive Republican campaign; in 1992 Bush lost core supporters when reneging on his “read my lips” pledge came back to haunt him. These stories about the meaning of the election begin to coalesce during the campaign. Once the votes are counted, the stories solidify into conventional wisdom and supply convenient ways to judge what the election was about, why it came out the way it did, and what the result suggests about the future. Because these stories become part of the public understanding, they have real political importance. And because they are so important, there is strong pressure to provide explanations as soon as the election is over; people debate the future by arguing about what just happened.
Political scientists often complain about these quick and simple accounts. While piles of data are available—tracking polls, exit polls, election results themselves—commentators often fixate on a single piece of data and exaggerate its significance in order to produce a crisp and usable story. In 2004 a poorly worded exit-poll question (mixed with some editorial-page hyperventilation) made “values voters” the key to Bush’s reelection. Subsequent research failed to confirm this (see Stephen Ansolabehere and Charles Stewart III, “Truth in Numbers,” Boston Review, February/March 2005). Similarly, commentators often seize upon a “key moment” or a turning point in a campaign: Lyndon Johnson’s “Daisy” ad is frequently cited in accounts of the 1964 election, even though Johnson’s lead over Goldwater did not change at all after the advertisement aired.
The 2008 campaign was, in this regard, little different from its predecessors. Like them, it produced a variety of narratives, some of which rapidly morphed into the stories that now circulate about the outcome. But a broad survey of available evidence tells a different tale, one that is also more equivocal than popular discourse currently allows.
Nearly one year after the election, we can significantly revise some of the now-conventional narrative threads. The Obama victory was historic, but it was not surprising; Obama shifted, but did not redraw, the electoral map; race and class mattered, but not in the way people assumed they would matter; and partisan loyalty was powerful even as partisan defections like Colin Powell’s garnered headlines. Furthermore, it is simply too soon to tell how and how much campaign tactics mattered, whether this election’s outcome constitutes a realignment in voting behavior, and whether Obama has emerged with a mandate.
• • •
Everyone knew that 2008 was a bad year to be a Republican. Less well-known—certainly less well-understood—was how bad. Because expectations of the Republicans were so low, early in the campaign some commentators were perplexed that Obama was not doing better. In August 2008 New York Times columnist David Brooks suggested that the election, by rights, “should be a Democratic wipe-out,” but, because the race was fairly close, he speculated that voters might not be ready to trust Obama.
Brooks’s premise was wrong: there was no sound basis for this expectation and thus none for speculations about lack of trust.
Political scientists and economists have built models that forecast presidential election outcomes based on “fundamentals”—such as the state of the economy, approval of the incumbent president, and casualties during war—that derive from theories of voting behavior. These models show that the incumbent party does better when times are good and especially when the economy is strong.
Unsurprisingly, the forecasting models predicted an Obama victory—not a landslide, but a victory nonetheless. There are many such models, and, on average, they predicted that Obama would win 53 percent of the vote: precisely what he won. Obama did as well as would be expected for a challenger in a weak economy. In the standard models, electoral performance depends on economic performance early in the year of the election. The first two quarters of 2008 were bad, but they were not disastrous, so a modest Obama victory was predicted.
At least in this respect, the election’s outcome was not surprising: Obama performed no better or worse than the fundamental conditions in the country suggested he would. This does not, however, detract from Obama’s historical significance. Electing an African American as president was a big step in our long and tentative progress toward racial equality. But we should not lose sight of how much Obama and the Democratic Party simply benefited from the nation’s ills.
• • •
The day after the election, Obama’s top pollster, Joel Benenson, declared, “what this election definitely says is that Democrats set out to compete in places where Republicans said it was impossible, and we won and redrew the map.” Beneson seems to suggest that the Obama campaign crafted a strategy that would be especially effective in “red” states such as Virginia, and that, as the term “redrew” implies, this strategy may have created a permanent shift. Whether the shift is permanent depends, of course, on what happens in future elections. But what about the idea that the Obama strategy was especially successful in states that had previously voted Republican?
The evidence does not support this claim. If we look at the 2008 electoral results, we find that Obama shifted the overall map in his favor: he did not redraw the boundaries by performing especially strongly in previously red states. More precisely, the economy shifted the whole terrain in the Democrats’ favor and Obama took advantage of this.
If we define the white working class broadly, then there is little evidence that it has shifted to the Republican Party.
Consider how Obama did relative to John Kerry in 2004. Simply put, he did better in nearly every state. Even though we typically hear about the state-by-state battles that make up the election—Florida in 2000; Ohio in 2004—recent elections, including 2008, have each been dominated by a nationwide swing. Once we account for this national swing, the outcomes in most states were very close to the 2004 pattern. Despite the focus on battleground states, there were only small shifts attributable to state-level characteristics, as opposed to this national swing (for example, Obama did no better in states with poorer economies, whether measured by unemployment or home foreclosures). This fits the recent pattern: state-by-state election swings have generally been declining over the past few decades. The traditional red-blue map is much more stable from election to election than it used to be.
• • •
Few aspects of the 2008 campaign received more attention than the roles of race and class. Once it became clear that Obama would do very well among black voters, this narrative focused almost entirely on the behavior of white voters, particularly working-class white voters. Indeed, the white working class regularly figures in campaign commentary and reporting, based on the belief that working-class whites are “swing voters.” As the story goes, they present problems for Democratic candidates, having moved to the Republican camp in recent elections on the basis of racial and moral issues even though the Democratic Party better represents their economic interests. Some writers, notably Thomas Frank in his widely discussed What’s the Matter With Kansas?, have made this claim. Many white voters, especially among the working class, were also assumed to harbor enough racial prejudice to make them reluctant to support Obama. The relentless attention paid these voters implied that they were numerous enough to hurt his ability to win. The Democratic primary only intensified this focus, as Hillary Clinton frequently bested Obama among white, working-class voters, leading many to suspect he would struggle to win these voters in the general election.
This familiar characterization of white, working-class voters fundamentally misrepresents their partisan preferences. If we define the white working class broadly—as those below some income threshold, or those without a college degree—then there is little evidence that the white working class has shifted to the Republican Party. In fact low-income white voters are more likely to vote Democratic than are high-income white voters. If we define the white working class more precisely—as those in particular occupations, for example—then there is some evidence that white, working-class men have moved away from the Democratic Party.
But by the time the white working class is defined based not only on education attainment and income, but also on occupational sector, it is numerically smaller and less electorally consequential. The working class shrunk considerably over the past few decades and is likely to shrink further in the years ahead. This is not to suggest that the working class will become irrelevant to either political party, but certainly there is and will be less reason to privilege the white working class as a component of either party’s coalition.
It is no surprise, then, that in 2008 the behavior of white voters and white, working-class voters belied the assumptions embedded in the standard narrative. First, Obama actually did better than Kerry among these voters. Relative to John Kerry, Obama gained three percentage points among whites en route to capturing the votes of 43 percent of whites. Relative to Kerry, Obama also gained roughly the same amount among white voters both with and without a college degree. Second, differences across “classes” were quite small and did not indicate any consistent disadvantage for Obama among the white working class. Obama did only slightly worse among whites without college degrees (40 percent) than among white college graduates (47 percent). He did only slightly better among white voters earning under $50,000 (47 percent) than among those earning more (43 percent). Ultimately, the class cleavages among white voters were just not that large. Obama did not win a majority among any of these groups, but he did not need to.
The media’s single-minded focus on the white working class obscured more than it revealed in 2008. Most importantly, a central fact sometimes seemed to elude commentators, especially when Joe the Plumber was in the headlines: on average, Democrats, not Republicans, do better among lower-income voters.
This was true in 2008 as well. The only difference is that Obama did better than Kerry or Gore with every income group. Obama did particularly well among the richest voters, who provided him a ten-point bump over Kerry in 2004. (In absolute terms, voters earning more than one hundred thousand dollars per year were evenly divided in their support for Obama and McCain.) We do not yet know whether the Democrats’ improved performance among rich voters will prove durable. It could reflect a temporary reaction to the economic crisis, which is affecting many higher-income voters with significant assets in the stock market. Or it could reflect a reaction to the candidates themselves.
But the overall effect of income on presidential voting does not tell the whole story. Rich and poor voters behave differently in rich and poor states. While higher-income voters are more likely to vote Republican within any given state, they are especially likely to do so in poor states. Thus, in Mississippi, a relatively poor state, rich voters are far more likely to support Republican candidates than are poor voters. But in Connecticut, much richer than Mississippi, the differences in voting behavior among rich and poor voters are smaller. This pattern has been noted since the 1980s, and 2008 was no exception.
In addition to concerns over whether white, working-class voters would support a black candidate, another racially inflected vein of commentary focused on the “Bradley effect” or “Wilder effect.” The idea is that black candidates do worse in elections than they do in pre-election polls, presumably because some white voters tell pollsters that they will support the black candidate but ultimately do not. Even before 2008, however, evidence of the Bradley effect was thin. In recent elections black candidates running against white opponents fared, on average, no better or worse on election day than pre-election surveys predicted. This proved true in 2008 as well, both during the primaries and in the general election. As in earlier years, polls taken a few days before the general election were very close to the national vote and to the votes in the states. Obama’s share of the vote was not consistently higher or lower than his poll numbers.
Even more important is the actual impact of racial prejudice on voting behavior. Most social-scientific research shows significant reservoirs of racial prejudice among white Americans, and early results from the campaign were hyped to suggest a potentially devastating outcome as a result. One article, summarizing the results of a late-August/early-September 2008 AP/Yahoo poll, stated: “Deep-seated racial misgivings could cost Barack Obama the White House if the election is close” and “such numbers are a harsh dose of reality in a campaign for the history books.”
These conclusions face two objections. First, racial identities are not the only relevant identities in elections. In fact, people’s partisan identities, as Democrats or Republicans, are especially important. Approximately 90 percent of the public identifies with or leans toward one of the two major parties, and the vast majority of partisans supports their party’s presidential candidate. The loyalty of partisans actually has been increasing over time. The 2008 election continued the trend. Eighty-nine percent of Democrats supported Obama, while 90 percent of Republicans supported McCain. For most voters, partisan identities trumped racial prejudice. One study showed that 62 percent of the most prejudiced Democrats said they voted for Obama. And, among Republicans, even the least prejudiced were still loyal to McCain, voting for him in 89 percent of cases.
Second, while these results suggest that, because of his race, some voters did not support Obama, they do not tell us anything about the net effect of race on the outcome of the election, a factor that the AP poll ignored. Racially prejudiced voters are not distributed evenly across the country: they are more likely to be found in rural areas and in the South. And, as it turns out, Obama outperformed Kerry in virtually every county in the country, except for Republican-leaning poor counties in the South. We can also approximate Obama’s vote share among non-African Americans, by county. Red counties—where whites were especially unlikely to vote for Obama—are mostly in red states, where the race was never very competitive to begin with. Racial prejudice, to the extent that it was operating, may not have altered electoral college math very much.
The researchers found that 11 percent saw Obama’s race as a reason to vote against him, but roughly three times as many saw his race as a reason to vote for him.
It turns out that Obama’s race actually may have helped him more than it hurt him. Although his share among white voters was only three percentage points higher than Kerry’s, Obama significantly outperformed Kerry with every other racial group. Compared to Kerry, Obama gained seven percentage points among blacks, twelve among Asians, and thirteen percentage points among Latinos. As Ansolabehere and Stewart wrote in the January/February 2009 issue of Boston Review: “Obama gained not only by bringing new minority voters into the electorate, but also by converting minority voters who had previously been in the GOP stable.” Even if Obama’s race does not explain why he gained votes among these groups, it certainly was not a hindrance.
Two political scientists performed a particularly telling experiment aimed at discovering whether Obama’s race had a positive or negative effect on his candidacy. Stanford’s Simon Jackman and UCLA’s Lynn Vavreck provided respondents a list of considerations and asked how many of them were reasons to vote for Obama and how many were reasons to vote against him. One half of respondents saw this list: economic plan, party, Iraq policy, health care plan, and speaking ability. The second half of respondents saw that list plus the phrase, “He’s black.” Because respondents specified the number of items seen as positive or negative, rather than the particular items, they did not have to reveal any sensitivity to Obama’s race. Comparing the average number of items considered positive or negative by each group of respondents tells us the fraction that considered Obama’s race a positive or negative factor. The researchers found that 11 percent of the sample saw Obama’s race as a reason to vote against him, but roughly three times as many saw Obama’s race as a reason to vote for him. This experiment does not provide definitive evidence, but it suggests that the effects of Obama’s race were two-sided. He may have won more votes because of his race than he lost.
• • •
After an election ends, explanations of the outcome frequently center on the campaign itself, crediting (or blaming) decisions that the candidates made, a “moment” or event, or some piece of campaign propaganda. It is hardly a surprise that the winning candidate, according to these post-election narratives, almost always ran a better campaign than the losing candidate and benefited from—or at least was unharmed by—events that may have been out of both candidates’ control. The day after the 2008 election, the front page of The New York Times had this headline adjacent to a photo of a victorious Obama: “Near-Flawless Run from Start to Finish Is Credited in Victory.”
Understanding how campaigns and unforeseen events matter is more difficult than it might seem. Most casual commentary relies on loose standards of evidence or on conjectures that cannot be proven: “If not for Willie Horton . . .” In the 2008 election, the September financial crisis was crucial in the minds of some observers. University at Buffalo political scientist James E. Campbell called the Wall Street meltdown a “game changer.” A New York Times article described McCain’s comments that accompanied the Lehman Brothers failure (“The fundamentals of our economy are strong”) as a “potential turning point.” This is all plausible. But it is hard to prove. The general success of pre-campaign forecasts suggests that campaign effects may be predictable, with events such as the choice of Sarah Palin as a running mate being less “game changers” than excuses for voters to return to their predictable preferences.
There is little evidence that the financial crisis and the candidates’ reactions to it marked a major shift. Obama, who, predictably, slipped in the polls after the Republican convention, had already retaken the lead before the collapse of Lehman Brothers. Moreover, one would expect a sharp break in polling trends had Lehman’s failure or some other event directly moved voters. Instead, Obama’s margin increased relatively steadily until mid-October. Does this mean that the financial crisis—and the candidates’ reactions to it—were irrelevant? Not necessarily. If nothing else, they kept voters’ minds focused on the economy, which was already a weak point for the Republicans. Nevertheless, the crisis (as opposed to the underlying poor economic performance) may have been less consequential than commentators assert.
While individual campaign moments are probably overrated, voter contact and mobilization efforts may shed some light on Obama’s victory. The political parties are newly interested in old-fashioned, shoe-leather forms of campaigning—the so-called “ground game,” as compared to the “air war” of television spots. Much recent political-science research shows that voter contact can increase turnout when done correctly. Did contact matter in 2008? The evidence here is circumstantial, but suggestive. In Colorado, Florida, Indiana, and North Carolina, Obama outperformed Kerry by a greater margin in counties with Obama field offices than in counties without them. Obama also did particularly well, relative to Kerry, in counties with extensive voter contacts by his campaign and other progressive organizations. Obama outspent Kerry by approximately $400 million, so it makes sense that outcomes in battleground states would shift in his favor. But we need to look more closely before we can be confident that his election was determined by campaign strategy.
• • •
As soon as the election was over, the debate about whether or not Obama has a “mandate” began. What does it mean to have a “mandate”? Does Obama have one?
According to one understanding of the term, mandates are signals from voters. A winner has a mandate when we can interpret the outcome of the election as an endorsement of the winner’s policy proposals. This definition is implied by New York Times colunist Paul Krugman’s post-election op-ed:
This year . . . Mr. Obama ran on a platform of guaranteed health care and tax breaks for the middle class, paid for with higher taxes on the affluent. John McCain denounced his opponent as a socialist and a ‘redistributor,’ but America voted for him anyway. That’s a real mandate.
Similarly, liberal blogger Matthew Yglesias wrote:
People want Obama to implement his agenda, and his agenda is a progressive one—cutting carbon emissions, expanding access to health insurance and early childhood education, making the tax code more progressive, and spreading the wealth around building broad-based prosperity (strikeout in original).
But do we really know what people want just because they support one candidate over another? We cannot assume that people voted for Obama based on detailed knowledge of his policy positions. As Princeton political scientist Larry M. Bartels argues:
We expect voters to listen carefully to what the candidates say and weigh the candidate’s positions in comparison with their own convictions and make a choice of candidates on the basis of their issue position. And then we expect the election to enforce responsiveness by having put the candidate in office who’s closest to the voter’s issue positions, who then implements those policies, and so people get policy outcomes that are close to what they wanted in the way of policies with respect to all the issues that they care about. That’s mostly not what happens.
Instead people often choose a preferred candidate and then rationalize issue positions to fit this preference. So we cannot interpret an election outcome as a wholesale endorsement of the winner’s policy proposals (or as a wholesale rejection of the loser’s). This is as true in 2008 as in 2004, when President Bush declared, “I earned capital in the campaign, political capital, and now I intend to spend it”—and followed up this pledge with a failed attempt to alter the Social Security system.
In Mandate Politics, Lawrence Grossback, David Peterson, and James Stimson offer a second definition of mandates. They say that what voters do or do not want is less important than what policymakers believe voters want. After elections perceived to convey a “mandate,” members of Congress appear to change how they would normally vote on policy, thus shifting the overall direction of policy. However, in the authors’ account, only the 1964, 1980, and 1994 elections were truly “mandate” elections. Moreover, the effects on Congressional voting were short-lived.
How should we evaluate the “mandate character” of the 2008 election in light of past results? Obama’s 365 electoral votes fall short of Johnson’s (486) and Ronald Reagan’s in 1980 (489). Obama’s popular vote margin is comparable to Reagan’s victory over Carter. More importantly, Obama’s victory, like Reagan’s in 1980, coincides with his party’s two successive gains in the Congressional vote. This fact seemed to elude some observers in 2008. Charlie Cook of the influential Cook Political Report wrote: “Given the strength of the top of the ticket nationally, one might have thought that the victory would have been more vertically integrated . . . what happened down-ballot was not proportional to what happened at the top.”
Cook’s theory—that support for Obama did not translate into support for Congressional Democrats—is not supported by the data. Obama won 53 percent of the two-party vote; Congressional Democrats averaged 56 percent. The swing of 5.7 percent from Democratic Congressional candidates in 2004 to Democrats in 2008 was actually greater than the popular vote swing of 4.5 percent from Kerry to Obama. Typically, Congressional voting has been much less volatile than presidential voting over the years. This makes the Democrats’ 5.7-point gain over two elections even more impressive. Democratic gains in 2006-2008 were virtually equal to Republican gains in 1994.
In sum, the 2008 election, in combination with 2006, represents a partisan shift in the Democrats’ favor. But does it constitute a mandate? The question is how Republican leaders respond. “It is consensus on electoral message that carries force in Washington and moves politicians to rethink strategies,” Grossback and colleagues argue (emphasis in original). “Consensus happens when both sides agree. Thus, the losing side need only deny electoral messages to keep consensus from forming.” It appears Republican leaders in Congress—who have vigorously opposed the Obama administration’s early initiatives, especially the stimulus package—will be part of no such consensus.
• • •
The notion of “realignment” is a staple of post-election analysis. The day after the 2008 election, The New Republic’s John B. Judis wrote, “[Obama’s] election is the culmination of a Democratic realignment that began in the 1990s, was delayed by September 11, and resumed with the 2006 election.” A New York Times article declared, “In a striking measure of potential political realignment, the gap between self-described Democrats and Republicans has grown.” This is quite a turn-around, seeing as 2004 was also described by some as a realignment. For example, in 2005, the New America Foundation’s Michael Lind wrote, “Karl Rove is an evil political genius, but he is a political genius. As he hoped, 2004 was a realigning election like 1896.”
The promiscuous use of “realignment” reflects the amorphous quality of the term. At times journalists and commentators use it to mean “something big has happened.” After the election a Boston Globe article referred to “a profound political and social realignment in America.” At other times, it seems to mean “the winning candidate’s supporters will form a long-term majority coalition.”
Political scientists have traditionally defined a realignment as a dramatic shift in the party coalitions, which ushers in an extended period of party control, which in turn brings with it a notable movement in policy: the Republicans dominated national politics after 1896, the Democrats after 1932. However, the concept of realignment is not in such good standing anymore. Nailing down evidence of large and stable shifts is quite hard. And, in the post-War era, party control of government has oscillated back and forth.
More importantly, there is little evidence that the 2008 election constitutes a realignment. Why? First, even if it did, we would not know for years. Second, voting behavior did not change that much. Obama did win states that Democrats had not won in a while, and demographic trends suggest that Democrats have a chance to win those states in the future. But there were only small shifts in state-level vote margins. Two thousand-eight looked a lot like 2004 and did not signal any wholesale change in partisan loyalties or party coalitions. Moreover, it does not appear that Obama “realigned” specific groups of voters. The widespread fixation on carving the electorate into its constituent groups misses the crucial fact that Obama did better than Kerry in nearly every possible demographic: rich, poor, white, black, Protestant, Catholic, men, women. The differences are not always large, but they are consistent. There was a “national swing” among groups of voters, as there was among the states. Voters of all stripes were displeased with the economy and President Bush and so voted for the opposing party’s nominee.
Because political preferences formed during early adulthood tend to be stable, most young people will likely continue to identify with the Democratic Party.
There was, however, a particularly noteworthy shift among young voters that could signal something more permanent and profound. The enthusiasm of young people for Obama was obvious in the early going, but after the election, some observers downplayed it. For example, political consultant Mark Penn wrote on the New York Times Web site: “Sure, young people voted heavily for Mr. Obama, but they voted heavily for John Kerry.” Was Penn right?
Contrary to Penn’s remark, in 2008 the gap between young and old increased—a lot. Obama beat McCain two-to-one among voters under 30, a margin among any age group unprecedented in recent years. The turnout of young people was not as significant as their strong tendency to support Obama. According to the Current Population Survey, the turnout of those under 30 years old increased only by about two points between 2004 and 2008.
Because political preferences formed and developed during early adulthood tend to be stable over the lifespan, most young people will likely continue to identify with the Democratic Party. The strong youth vote by no means assures an era of Democratic dominance, in part because subsequent generations of voters may trend toward the Republicans. Nevertheless, the Democratic Party has likely won a majority of this cohort of voters for the foreseeable future.
• • •
Our story of the 2008 campaign confirms some parts of the journalistic narrative and refutes others. Yes, the economy was important; yes, young voters swung to Obama and the Congressional Democrats; yes, Obama did particularly well among minorities (Latinos and Asians as well as African Americans), even beyond the Democrats’ usual strength among these groups; yes, the Democrats made new inroads among the most affluent voters. But no, working-class whites did not run away from Obama; and no, Obama did not redraw the electoral map. Since 2004 the Democratic Party gained about five percentage points of the vote both in presidential and Congressional elections: not a landslide but a large swing by historical standards. The chief lesson for Obama’s first term is that the fundamentals will rule. Future elections will likely turn on the economy’s performance under the new administration.
Andrew Gelman is Professor of Statistics and Political Science at Columbia University. His books include Bayesian Data Analysis, Teaching Statistics: A Bag of Tricks, and Red State, Blue State, Rich State, Poor State: Why Americans Vote the Way They Do. He also blogs at Statistical Modeling, Causal Inference, and Social Science.
John Sides is Assistant Professor of Political Science at George Washington University and coauthor of The Gamble.
The media can and must do better.
In 2008, there were 2 million more black voters.
The failure of the economy to move the campaign is a puzzle in its own right.
We must not dismiss the idea that Obama created a new electoral map.
In elections, what is certain is almost never new.
Vital reading on politics, literature, and more in your inbox. Sign up for our Weekly Newsletter, Monthly Roundup, and event notifications.
But I do miss the hymns, / the small, hard apples with their dimpled skin. I do miss / things.
The vast hinterlands of the Global South’s cities are generating new solidarities and ideas of what counts as a life worth living.
Protests in China are shining a light not only on the country’s draconian population management but restrictions on workers everywhere.
Austerity is not the only way to save our overextended planet. A simpler life might be both more pleasurable and more equal.
We must reject the legal liberalism that attempts to cordon off constitutional questions from democratic politics.
The United States ranked first on health security; then came COVID-19. In place of technocratic hubris, we need robust new forms of democratic humility.