The day after his 2004 re-election, in which George Bush narrowly defeated John Kerry 51%–48% and in which his party won both houses of Congress, the President held a news conference where he explained: “I earned capital in this campaign, political capital. And now I intend to spend it.”1 Many Democratic pundits balked at the idea of a mandate based on such a slim margin. Washington Post columnist E.J. Dionne wrote: “Blue Americans… are amazed that a majority was not concerned about heaping a huge debt burden on our children… A 51%–48% victory is not a mandate.”2 Eight years later, things have come full circle. Following an election in which Barack Obama defeated Mitt Romney 51%–47% while Republicans held the House of Representatives, the President and many Democrats were suggesting a mandate.3 Now it was Republicans who objected.
Political interpretations of the meaning of elections tend to be, well, “political.” This article examines the 2012 election by the numbers and in historical context in an effort to gain some distance from partisan claims and interpretations. In studying American presidential elections, political scientists interested in testing hypotheses face a limited number of observable cases. There have been just 57 presidential elections in US history, and even this number is greatly inflated for purposes of comparison. A presidential election today is in many respects hardly the same kind of event as it was a century or two centuries ago, and elections within the same era differ according to type (whether an incumbent is running or not) and to sequence in a cycle (how many terms the incumbent party has been in office). In addition, each presidential election is an idiosyncratic event with its own historical character and cast of candidates. The analysis that follows attempts to be sensitive to these differences and to recognize the limits of drawing inferences based on a small number of comparable cases. At the same time, there are some trends and patterns in presidential elections that can help place the current election in a larger perspective and evaluate competing political claims.
The Magnitude of the 2012 Victory
Elections serve the constitutional function of selecting the persons who will hold office. A mandate is an extra-constitutional concept that victors use to try to supply added energy and legitimacy for their agenda. A mandate, if it exists, is based on an interpretation of what was said during the campaign and on the size of the victory. In an effort to provide a factual basis for these interpretations, this section explores the magnitude of Obama’s 2012 victory by comparing it to the size of victory of other modern presidents, to the size of victory of other incumbents who sought a second term, and finally, within that same group of incumbents, to their record in holding party control in Congress.
Like George Bush’s re-election victory in 2004, the margin of Barack Obama’s victory in 2012 (3.85%) is modest by historical standards, ranking 24th out of the 30 elections since 1896 (see Table 1).4 Presidential elections in this period may be categorized as landslides (victories by more than 11 points), substantial wins (5–11 points), competitive wins (2–5 points), and dead heats (<2 points). Obama’s victory in 2012 ranks as competitive, in contrast to his substantial win (7.3%) in 2008. Within the competitive group, and among other postwar elections, Obama’s 2012 victory is situated between Bush’s 2004 win over Kerry (2.5%) and Truman’s 1948 win over Dewey (4.5%). In terms of Electoral College vote share, the 2012 victory (61.7%) is only slightly higher-ranked at 22nd out of 30.
For those who followed the campaign closely, the Obama victory may have “felt” larger than it was, by outperforming many expectations. In the last months of the campaign, pollsters saw the 2012 election as heading toward a “dead heat,” at least in the popular vote. The RealClearPolitics average of polls in the final week before the election had Obama up by just 0.7%. Obama’s victory margin of 3 points more than projected had its effect. Similarly, in the months leading up to the election, the two candidates’ campaigns, and the media coverage of the election, focused on a handful of swing states: Florida, Ohio, Colorado, Virginia, New Hampshire, Nevada, Iowa, and, nearer to the end of the campaign, Pennsylvania, Wisconsin, Michigan, and Minnesota. In the greatest surprise of election night, Obama won them all. Watching these states go for Obama one by one was like witnessing a juggernaut.
A glance at Table 1 also suggests a significant development across this 116-year period. Earlier in this period, margins of victory tended to be greater, and the swings between parties in consecutive elections much larger, than is the case today. The average margin of victory in the first 23 elections, through 1984, was 12.65%. In the last seven elections, the average margin of victory has fallen to 4.95%, and the margins in all of these elections are below the median. Obama’s 2008 victory, ranked at 19th, was the largest of this group.
Even when there were close elections in the earlier period, they often represented halfway points between large electoral swings between the two parties. When Kennedy beat Nixon in a dead heat in 1960, the result still represented a 16-point swing in favor of the Democrats from the previous election, and was followed by another large swing, a 23-point win, for the Democrats in 1964. Similarly, when Nixon beat Humphrey in a dead heat in 1968, the result represented a 23-point swing for the Republicans, and was followed by a 23-point re-election win in 1972. Whereas double-digit swings between the parties were once the norm, the contemporary electorate has not moved by 10 points or more since 1992. This development, as we will later suggest, may be tied to growing partisan polarization.
Change in Support from First to Second Election
The presidency, as Woodrow Wilson observed, is a personal office.5 What this means – and what electoral analysts have always recognized – is that presidential elections with an incumbent constitute a distinct type, in which the central question posed, as Hamilton explained in the Federalist Papers, is whether or not “to continue him in his station.”6 Slightly more than half (32) of all presidential elections have featured an incumbent. In five of these cases the incumbent was a vice president who had ascended to the presidency, and in two cases – FDR in 1940 and 1944 – the incumbent was running for a third and fourth term. This leaves 25 elections where an elected incumbent president was running, as with Obama in 2012, for a second term. Table 2 ranks the electoral contests in this last group, since the rise of mass voting, according to the popular vote share gained or lost between the president’s first and second election.7
President Obama made history in 2012 by becoming the first president to win a second election while losing vote share. All other victorious incumbents until now increased their electoral strength. Those who lost votes had always also lost the presidency. In one sense, this makes Obama the weakest of the re-elected presidents. It also, however, shows his strength – and the strength, perhaps, of current partisan attachments. Despite losing favor with many voters, Obama was able to maintain a sufficient base of support to win. The President, of course, had the luxury of being able to give votes away from his substantial victory in 2008. His electoral campaign was predicated on the classic military strategy of yielding ground but saving the city. Obama lost vote-share among almost every demographic group, Hispanics being the most notable exception. Yet the strategy succeeded.
The overall strength of a president’s victory is also measured by how his party fares in Congress. Table 3 lists the 11 victorious presidents from Table 2 showing their party’s performance in Congress between their first and second elections. Nine of these two-term presidents initially came into office with their party in control of both houses of Congress. Only Nixon and Reagan, who were elected in the era of Democratic congressional dominance, entered office under divided government – Nixon with neither chamber and Reagan with just the Senate. The overall record of the nine presidents in maintaining unified government is otherwise not that impressive: five did so and four did not. Of those who gave up unified government, Eisenhower and Clinton lost control of both chambers, while Wilson and Obama lost control of the House.8 Judged by the strength of his performance in holding Congress, Obama as an incumbent ranks closer to the bottom than the top of this group. Republicans during his term also made huge gains at the state level, both in the number of gubernatorial seats that they control and the number of state legislatures under Republican control.
|Re-election year||President||Change in unified (U) or divided (D) government||Number of congressional houses gained or lost||Change in senate from first election||Change in house from first election|
Incumbency and American Presidential Elections
Many Republicans were optimistic about their chances to oust President Obama in 2012, and some pundits have blamed Governor Romney in the aftermath of the election for losing a seemingly easy victory. But the historical record shows the difficulty of unseating an incumbent, especially in the modern era. No doubt this race was winnable for Romney, but it is an error to think that he had a simple path to victory if he had only run a better campaign.
Two-thirds of all incumbents – far from an overwhelming number – have won their second term bids (see Table 4, which includes successor vice presidents). The incumbency advantage, however, has ebbed and flowed across time. It was strong at the outset from 1789 to 1832, when all seven incumbents ran for a second term and when all those without the name of Adams were re-elected. With the development of mass parties and two-party competition in the 1830s, the incumbency advantage weakened. For the remainder of the 19th century, incumbent presidents, now subject to the will of national party conventions, were regularly denied a chance at a second term by their own party, and they often lost when they did run. During this period, eight presidents either were denied, or chose not to seek, re-nomination by their party. Only five incumbents were re-nominated, and just two of them won re-election: Lincoln and Grant. Since 1896, with the rise of the modern personal president who stands above his party, the incumbency advantage has returned. All incumbent presidents – including successor vice-presidents – have been re-nominated for another term, and they have won three quarters (15 out of 20) of these elections.9
|President||Elections||Re-nominated||Won second election|
|J.Q. Adams||1824, 1828||X|
|Van Buren||1836, 1840||X|
|Cleveland||1884, 1888, 1892||X|
|B. Harrison||1888, 1892||X|
|F.D. Roosevelt||1932, 1936, 40, 44||X||X|
|G.H.W. Bush||1988, 1992||X|
|G.W. Bush||2000, 2004||X||X|
The five incumbents who were defeated in the modern era are Taft, Hoover, Ford, Carter, and Bush Sr. Three factors help account for these losses. First, in all of the cases except Hoover, whose 1932 defeat during the Depression really explains itself, there were important intra-party primary challenges. Taft faced Teddy Roosevelt in 1912, Ford faced Reagan in 1976, Carter faced Ted Kennedy in 1980, and Bush Sr. faced Buchanan in 1992. Each of these challenges showed weaknesses in, and created problems for, the incumbents, foreshadowing defeat in the general election.
Second, in two of the cases there was an important third-party candidate who drew support more heavily from the incumbent than the challenger. In 1912, after his failed Republican primary challenge, Teddy Roosevelt led the Progressive Party in a campaign that captured most of Taft’s followers (he finished third), and in 1992 the independent candidacy of Ross Perot pulled votes away from Bush. These challenges garnered the greatest vote shares of all third-party candidacies since the Civil War.
Third, all of these incumbents except Carter were seeking another term at the end of long periods of White House control by their party, when general sentiment for change most likely increases. Taft was going for a fifth consecutive Republican presidential term, Hoover and Bush Sr. for a fourth, and Ford for a third.
Obama enjoyed the incumbency advantage in 2012 without encountering any of these three weakening factors. He faced no primary challenge; indeed, he was extremely popular in his own party. No major third party was operative, and the Democrats had only been in office for one term. If the historical record suggests a default position, it was in favor of a re-election of Barack Obama.
There is, of course, a fourth possible factor to consider: the performance of the economy. It was largely on the basis of a weak economy that many counted on a Romney victory, and there is no doubt that making the election into a referendum on the President’s handling of the economy was the central element in Governor Romney’s campaign strategy. A poor economy is a frequently-used predictor for why party control of the White House switches hands. Moreover, in the case of two of the incumbents who were defeated (Hoover and Carter), a judgment of a poor economic performance was the most influential factor in determining the outcome.
The problem in analysis of this kind is less the importance of economic performance for elections than the question of exactly how it is to be measured and judged, and especially how the voters parse the responsibility. Purely mechanical formulations and measures have proven unhelpful even when they are not downright silly. With the unemployment rate hovering around 8% by the end of Obama’s first term, some were foolish enough to find significance in the claim that no president since WWII had won with unemployment above 7.4%, which just happened to be the rate in October 1984 before Ronald Reagan was re-elected in a landslide. Usually omitted from this observation is the fact that only one President had tried: Jimmy Carter in 1980, with an unemployment rate of 7.5%.
A more reasonable test of the importance of unemployment would have set the bar at 7%, a threshold that would have brought in seven incumbents during the “modern” period: Hoover (18.8%), FDR in 1936 (16.6%), FDR in 1940 (14.6%), Ford (7.7%), Carter (7.5%), Reagan (7.4%), Bush Sr. (7.3%), and Obama (7.9%). The result of an incumbent running for re-election with unemployment above 7% is an almost even split of three wins and four losses. Examining these on an individual basis shows that the economy was not Ford’s only problem in 1976 nor Carter’s in 1980. Conversely, the unemployment rate did not doom re-election bids for FDR or Reagan. Both were able to argue that the poor economy was the fault of the previous administration and that it had improved on their watch. Obama followed the same formula, though with less success. Exit polls indicate that a weak economic recovery did work in Mitt Romney’s favor, just not nearly to the degree that some expected.
Since the advent of reliable unemployment data at the turn of the 20th century, there have been 12 elections where an incumbent has run for re-election. Table 5 organizes these 12 electoral contests according to the unemployment rate. Certainly, the change in unemployment rate over the course of a president’s first term can tell us more about the public’s perception of an incumbent performance than the raw unemployment rate alone. Yet even ranking electoral contests according to this change in unemployment over the preceding 4 years is inconclusive: the three presidents who saw unemployment decrease won re-election, but five of the eight presidents who saw unemployment increase (including Obama) also won re-election.
In Obama’s case, moreover, the increase was known to be fictitious. It had come down from a high of 10% that was considered to be still chargeable to the economic crisis of 2008. More important than the overall change between the first and second election is for the unemployment rate to be moving in the right direction in the months leading up to the election. This was the case for Obama. The unemployment rate dropped below 8% in October 2012 for the first time since Obama took office in January 2009, a move that surely helped Obama’s campaign.
In comparing the 2012 election with the comparable cases in the past, the best guess from all of the data analyzed is that an Obama victory was more probable than not. First, it is not surprising that an incumbent president strong enough to avoid a primary challenge was also strong enough to defeat his challenger in the general campaign. Second, it is not surprising that even with a weak recovery, a president with an economy that had slightly improved, and whose weakness in any case could still be plausibly attributed in part to his predecessor, was re-elected. Republicans can certainly rue their defeat and lament an opportunity that could have brought victory. But they are too hard on themselves if they conclude that they lost an election that they should have won.
Changes in State Electoral Behavior
A widely noted feature of the modern electorate – and a factor that may help explain the smaller victories and diminishing swings of recent elections – is partisan polarization. One manifestation of this phenomenon is the greater fidelity of geographic units to one political party or the other. Over the last two decades, states have sorted themselves between the two parties in consistent ways. Consider this fact: from the 1950s to the 1980s, virtually every state in the nation was a “swing state.”10 This was true in the North, East, South, and West. A few examples will illustrate.
Today, Louisiana is no longer considered a swing state, but from 1952 to 1980 Louisiana never stayed with the same party two elections in a row. Vermont, also no longer a swing state, once swung wildly between the two parties with large margins of victory for whichever party it happened to favor that year: Republican Eisenhower won the state by 44 points in 1956, Democrat Johnson by 33 points in 1964, and Republican Nixon by 26 points in 1972. Swaying back and forth with each year’s particular winner, Arkansas bounced between Johnson, Nixon, Carter, Reagan, and Clinton. 2008 and 2012 marked the first elections that Arkansas sided with the loser since the 1950s. Similarly, California went for Eisenhower in 1952 by a 15 point margin, Johnson in 1964 by 18 points, Reagan in 1984 by 17 points, and finally settled down with the Democrats and Clinton in 1992 by 13 points.
Swings of this magnitude have been absent over the past seven elections. Only a handful of states are now considered swing states, and they typically give only modest margins of victory to the winner. In 2012, Obama largely won the same states that he won in 2008, that Kerry won in 2004, that Gore won in 2000, and that Clinton won in 1996. Similarly, Romney largely won the same states that McCain won in 2008, that Bush won in 2004 and 2000, and that Dole won in 1996. Only two states changed parties from 2008 to 2012: North Carolina and Indiana moved from Obama to Romney. This shift of just 4% of all the states (2 out of 50) is the smallest amount of change since George Washington won all states back-to-back in the first two elections.
Many political scientists and journalists have speculated that Obama’s 2008 and 2012 victories mark a party realignment in American history. Yet from a geographic perspective, Clinton’s wins in 1992 and 1996 were much more realigning. Obama has won only one state in both 2008 and 2012 that did not go for Clinton: Virginia. Clinton, on the other hand, flipped New Jersey, Connecticut, Delaware, Maine, California, Illinois, and Vermont, all of which have become solidly Democratic. The geographic sorting of the 1950s through the 1980s has hardened into place in recent decades.
State voting behavior has changed dramatically over the past half-century: states are safer for one party or another. Table 6 offers evidence of this shift by providing electoral data on four presidential elections over the past half-century in which – to supply “controls” – there were similar margins of victory and no significant third-party activity: 1960, 1976, 1988, and 2012. There has been a sharp increase in the number of states won by large margins. Numerous state landslides, for both parties, was one of the salient facts of the 2012 election. For example, in an election decided by only 3.85%, Obama won Vermont by 36 points, New York by 28, Maryland by 26, and Massachusetts and California by 23. Romney won Wyoming by 41, Oklahoma by 34, West Virginia by 27, and Nebraska by 22. Going back to the issue of sorting, the table shows a marked decline over this period in the average state swing between the parties from the previous election.
|Election||Total margin of victory||Number of states with margin <4%||Number of states with margin 4%–8%||Number of states with margin >8%||Average state swing from previous election|
A presidential election in the US is a contest of winning states. Republicans, who will be struggling to come back in 2016, will find that the current shape of state voting behavior offers grounds for both hope and despair. Every Republican victory this year, except in North Carolina, was by 8 points or more. The core of GOP support appears safe. The problem for Republicans is that Democrats have a similar base that looks equally secure. There is little reason to think that the contours of the next election will look that much different from this one. It will be a battle for the battleground states, with Republicans having a lot of territory to make up.
“Transcript of President Bush’s News Conference,” New York Times, 4 Nov 2004.↩
E.J. Dionne, Jr., “…He Didn’t Get It,” The Washington Post, 5 Nov 2004.↩
Senator Bernie Sanders wrote: “President Obama and the Democrats won a decisive victory on Election Day. The people have spoken and the Democratic Leadership must make it very clear that they intend to…hold the line.” “The People Have Spoken,” Daily Kos, 29 Nov 2012.↩
Historians of American politics often point to 1896 as the beginning of modern presidential elections.↩
Woodrow Wilson, Constitutional Government in the United States. New York: Columbia University Press, 1908, p. 54.↩
Alexander Hamilton, Federalist 72, 1788.↩
Prior to the advent of mass popular voting in the 1830s, popular vote data is sketchy and not as comparable with later elections. Still, we know that all five re-elected Presidents in this era – Washington, Jefferson, Madison, Monroe, and Jackson – increased their number of electoral votes. It is likely that Madison and Jackson both lost popular vote-share even though both probably increased their number of popular votes. In Jackson’s case, a third-party candidate run by the Anti-Masonic Party split the vote in 1832, which means that Jackson probably increased his margin of victory despite losing vote-share.↩
Wilson lost the House in his 1916 re-election campaign, while all the others lost chambers in the prior mid-term election.↩
Interestingly, no vice president, after finishing his president’s term, has run for more than one election. Theodore Roosevelt, Coolidge, Truman, and LBJ all won an election in their own right as incumbents, but none went on to stand for a second election. LBJ was the closest to pursuing this option in 1968, but withdrew in the face of stiff resistance from his primary challengers.↩
One exception is the District of Columbia, which has voted in favor of the Democratic Party’s candidate in every election.↩
Verlan Lewis is a PhD candidate in the Department of Politics at the University of Virginia.
James W. Ceaser
James W. Ceaser is Professor of Politics at the University of Virginia and a Senior Fellow at the Hoover Institution, Stanford University. He is the author of, among others, Nature and History in American Political Development (Harvard University Press, 2006), and After Hope and Change: The 2012 Election and American Politics, with Andrew Busch and Jack Pitney (forthcoming from Rowman and Littlefield).