By Ronald Brownstein
Analysis of exit-poll results from the last eight presidential elections shows how demographic groups are—and aren’t—shifting in their voting preferences.
After Barack Obama won his historic victory in 2008, the opportunity to reshape the landscape of American politics glimmered before him and his advisers.
CHARTS: Presidential Voting By Demographic Group, 1980-2008
In his ardent, pathbreaking campaign, Obama captured 52.8 percent of the popular vote against John McCain, becoming the only Democratic nominee since World War II other than Lyndon Johnson in 1964 to exceed 50.1 percent. Obama ran best among young people, minorities, and college-educated white women, all groups that were growing in numbers. That “coalition of the ascendant” offered the new president the prospect of a steadily expanding base—even as his dramatic victory provided him the opportunity to court voters dislodged from the Republican coalition by disillusionment with outgoing President Bush. For all of these reasons, the 2008 results allowed Obama’s team to dream, much like the French generals at Verdun, of breaking through a grinding stalemate—in the current case, the rigid and narrow divide that had characterized American politics for the 12 years since Bill Clinton’s reelection in 1996.
Four years later, those dreams of realignment have, for now at least, turned to dust. After four grueling years of economic turbulence and partisan conflict, no one in either party expects Obama to consolidate a commanding new majority coalition this fall. Instead, his team is struggling against fierce economic headwinds to marshal a bare majority that relies less on converting ambivalent swing voters than on maximizing turnout and the president’s advantage among his core supporters.
Lifted by economic discontent, Republican contender Mitt Romney is running better in polls this summer than McCain did among the groups that were always the most skeptical of Obama, particularly older and blue-collar white voters, as well as well-educated white men. But Romney, too, has directed his agenda almost entirely at his party’s core supporters. And that has made it easier for Obama to hold his most reliable groups despite the economic anxiety. The net result is an election that today appears on track to more resemble the 50-50 division of Bush’s slim victories in 2000 and 2004 than Obama’s 365-vote Electoral College blitz in 2008. Absent a dramatic late development, small shifts in the preferences or turnout of virtually any group in the electorate could decide this election.
No Democrat since Lyndon Johnson has carried even a plurality of whites.
To understand the boundaries and dynamics of this struggle, National Journal has updated a project we conducted in 2008 that examined, in unprecedented depth, the fault lines and cross pressures among American voters. In that initial report (“The Hidden History of the American Electorate,” NJ, 10/18/08, p. 14), we examined the results from the general-election presidential exit polls conducted across the country by a consortium of news organizations from 1988 through 2004. In this update, National Journal adds the results from the 1980, 1984, and 2008 elections to produce an expanded look at the hidden history of the electorate over the past eight presidential elections. (The organizations performing the exit polls have changed over the years; the 2008 survey was conducted by the Edison Media Research/Mitofsky International National Election Pool.)
The calculations were performed on exit-poll data files by Emory University political scientist Alan Abramowitz for 1980 and 1984; Ruy Teixeira, a public-opinion analyst and author at the Brookings Institution and the Center for American Progress, for 1988 to 2004; and Joe Lenski, executive vice president of Edison Media Research, for 2008. National Journal, however, analyzed the results for each year.
As we noted in 2008, the exit polls allow for finely grained analysis of specific subgroups because they survey such a large sample of voters (17,836 in the 2008 version). Most readers, for instance, are familiar with the gender gap— the tendency of men to vote more reliably Republican than women and women to vote more reliably Democratic than men—or Hispanics’ strong preference for Democrats. But in this report, NJ highlights much more precise trends, such as the differences between college-educated and noncollege-educated white women, and those between Protestant and Catholic Hispanics. In its unique detail, this analysis spotlights both the changes in, and the durability of, the preferences among the key groups in the electorate.
It also shows how shifts in the electorate’s composition have altered the political balance at least as much as have changes in voters’ attitudes. Because of the steady growth of the minority population and Romney’s failure so far to crack those voters, Obama could prevail in November with an 80/40 solution: winning about 80 percent of the vote among minorities and about 40 percent among whites. Yet, hard times could put even that modest showing with whites beyond Obama’s reach. Heading into its final months, the 2012 campaign still looks like a titanic collision between the economy and demography.
WHITE MEN AND WHITE WOMEN
In 2008, Obama became the first presidential nominee ever to lose white voters by double digits and still win the White House: McCain beat him among white voters, 55 percent to 43 percent. But Obama’s vote share among whites wasn’t unusually weak for a Democrat; in fact, his 43 percent tied Clinton’s 1996 showing as the party’s best performance among whites since 1980. But Clinton’s losing margins among whites were much smaller than Obama’s (just 1 percentage point in 1992 and 3 points in 1996) because independent candidate Ross Perot siphoned many of them away from the GOP. Other polling measures that predate the national exit polls show that no Democratic nominee since Johnson in 1964 has carried even a plurality of whites. Polling this year suggests that even if Obama wins, he could well run below his 2008 showing among whites.
Gender, class, and cultural affinities (such as marital status and religious practice) all consistently shape whites’ voting preferences. Democrats do better among women, single voters, those who attend church less frequently, and those with more education. Republicans run best among men, married voters, regular churchgoers, and people without advanced education. All of these influences intersect and reinforce each other: In the past two elections, Republicans have won about two-thirds of married, noncollege white men, while Democrats have won almost two-thirds of single, college-educated white women.
Over the past three decades, gender has been the most extensively discussed divide in the white electorate. In the past eight elections, the Democratic nominee has averaged just 36.1 percent of the vote among white men; every Democratic nominee over that period has lost white men by double digits except for Clinton in 1992 (when the Perot effect was strongest). The modest 41 percent that Obama won among white men was actually the Democrats’ highest share since 1980.
For most of this period, Democrats have struggled among white men of all education levels. But the past two elections have seen a divergence. White men without a college degree are now solidly Republican. Since 1980, Democratic nominees have averaged just 35.9 percent support among these voters; Obama’s meager 39 percent share of this group was the party’s best showing over that period. For Democrats, the low point came in 1984 when Walter Mondale, even while offering the classic New Deal message of economic fairness, attracted just 31 percent of noncollege white men. Polling this year shows that Obama could approach that nadir—or even sink beneath it.
In the 1980s, college-educated white men were equally, if not more, difficult for Democrats to win: In the three elections of that decade, the party’s nominees averaged just 31.3 percent support among them. Even through Clinton’s two elections, Democrats ran slightly better among white men without a college degree than those with one. But those lines roughly converged in 2000 and then crossed: John Kerry in 2004 and Obama in 2008 ran 4 percentage points better among white men with a degree than those without one. Obama reached 43 percent among college-educated white men, the best showing for Democrats over this period. The effect of this “class inversion” since the 1980s has been even more powerful with white women. But in polling this year, the class inversion is widening among men, too: Polls frequently show Obama running between 8 and 10 percentage points better among college-educated men than those with no college. That’s not because Obama is improving his backing among the better-educated men—he is just losing less ground with those men than with the blue-collar white men stampeding toward Romney in Reagan-size proportions.
Marital status also matters among men. Since 1984 (the first year that the exit poll tracked marital status), no Democratic nominee has won more than Obama’s 38 percent among married white men. By contrast, in each election since 1988, the Democratic nominee has won at least 40 percent of single white men, topping out at 46 percent for Kerry in 2004 and 48 percent for Obama in 2008. Class and marital status reinforce each other in predictable ways. Kerry and Obama faced an average deficit of nearly 40 percentage points among married, blue-collar white men, while running almost exactly even among their single, college-educated counterparts.
Over this period, white women have been more receptive to Democrats and also more unsettled in their preferences. Only Clinton in 1996 carried a plurality of white women, although he ran about even with them in 1992, as did Gore in 2000. (Obama captured 46 percent of white women; only Clinton in 1996 and Gore in 2000 won more over this period.) The Republican share among white women has oscillated from Reagan’s 62 percent in 1984 to a low of 41 percent in the Perot-influenced 1992 race.
The same factors that shape the preferences of white men are even more powerfully evident among white women. In the three elections in the 1980s, Reagan and George H.W. Bush carried college-educated white women by an average margin of 9 percentage points. But Democrats have captured a plurality or majority of those well-educated women in each election since then, except in 2004 when they split between Kerry and Bush. Over these past five elections, Democrats have averaged 49.2 percent among college-educated white women, topping out at 52 percent of the vote behind both Gore and Obama.
Single white women have changed their allegiances even more dramatically. Reagan carried three-fifths of these women in 1984, but they gave George H.W. Bush only a small majority in 1988. By 1992, they preferred Clinton by double digits; and in the past four elections, Democratic nominees have averaged 56.5 percent among single white women, with Obama setting the standard at 59 percent.
The story reverses with the opposite groups of women. The Republican nominee has carried married white women in each of the past seven elections and averaged 57 percent of their votes in the past three.
Noncollege white women, the so-called waitress moms, have also leaned Republican, but they have been much more volatile. Recent polls, for instance, have Obama regaining some lost ground with them.
These women are typically economically strained, and although they are often more culturally conservative than their white-collar counterparts, they are less likely to vote based on social issues because they face so many pocketbook concerns. The Republican edge among noncollege white women averaged nearly 20 percentage points in the 1980, 1984, and 1988 elections. Clinton ran virtually even among them in 1992 and then, in 1996 with his “tools for parents” agenda, carried them by a solid 48 percent over Bob Dole’s 41 percent. (Both in terms of the 7-point margin and share of the vote, that showing remains the Democratic high point with noncollege-educated white women since 1980.) After Clinton’s victories, though, the waitress moms moved back sharply toward the GOP, preferring George W. Bush each time and then backing McCain over Obama by a resounding 58 percent to 41 percent. That wasn’t much different than McCain’s advantage among blue-collar white men. Indeed, in each election since 1980, the gender gap has been wider among whites with a college education than those without one.
As with men, these tendencies powerfully interact. Since 2000, Republicans have averaged 61 percent among white noncollege married women; Democrats have drawn a comparable 65 percent among single college-educated women. Cross-pressured by culture and class, married women with college degrees have divided almost exactly in half in four of the past five elections. Single blue-collar women consistently lean Democratic.
Regular churchgoers vote overwhelmingly for Republicans.
Below, we note dynamics among whites relating to religious practice and partisanship. But as we found in our initial analysis of these trends four years ago, these results underscore the extent of the class inversion among whites that has reshaped the electoral landscape. As we wrote then: “In the middle decades of the 20th century, when economic class served as the principal glue for the two parties’ coalitions, Democratic presidential nominees Adlai Stevenson, John F. Kennedy, Lyndon Johnson, and Jimmy Carter all ran at least 13 percentage points better among white voters without a college education than among whites with college or postgraduate degrees, according to the University of Michigan’s American National Election Studies, an exhaustive post-election poll.”
But as cultural and foreign-policy disputes between the parties have assumed greater prominence since the 1960s, Republicans have gained among blue-collar whites, and Democrats have made mirror-image inroads among white-collar whites, producing the class inversion. In 1988, 1992, and 1996, Dukakis and Clinton each ran almost equally well among college and noncollege whites. In 2000, Gore ran 4 percentage points better among college whites; Kerry widened the margin to 6 points; and in 2008 Obama ran 7 points better among college-educated than noncollege whites.
That gap will almost certainly widen in 2012, especially if Obama wins. Polls now consistently show the president running below—often well below—his 40 percent showing among noncollege-educated whites in 2008. His fate could depend on how close he comes to matching the 47 percent he won among college-educated whites four years ago. Most surveys this summer have shown Obama losing some support among college-educated white men but running very close to his 52 percent from 2008 among college-educated white women; those women tend to be both socially liberal and more open than other whites to an activist role for government.
If Obama survives, the shifting composition of the electorate will also be a—and perhaps the—critical factor. In 1984, whites without a college degree represented 61 percent of all voters, and college-educated whites just 27 percent. But by 2008, noncollege whites had plummeted to just 39 percent of all voters, and college-educated whites increased to 35 percent. Noncollege white men, consistently the GOP’s best group, fell from 28 percent of all voters in 1984 to just 18 percent in 2008; the waitress moms also slipped from 33 percent to 21 percent over that period. White men with a college degree, who traditionally lean Republican but did so by smaller margins in the past two elections, essentially maintained their share of the electorate over those years, increasing from 16 percent to 18 percent.
But the biggest gains have been recorded among two groups that favor Democrats. With women now earning about three-fifths of college degrees, college-educated white women have increased their share of the vote from 11 percent in 1984 to 17 percent in 2008. (These upscale, Democratic-leaning women could conceivably cast more votes than blue-collar men this year.) Minorities have grown even faster, rising from 12 percent of the electorate in 1984 to 23 percent in 2004 and 26 percent in 2008. Recent analyses of census data show that minorities make up 29 percent of 2012’s eligible voters (although it’s unclear whether that population growth will produce comparable gains in the electorate.) And with that burgeoning population, Republicans now see very few openings.
This year, the key question about minorities is not how they will vote but how many will vote. Either a depressed minority turnout or a huge outpouring among conservative whites could stall the steady growth in the minority vote share since the 1980s. But their loyalties don’t seem much in doubt. In 2008, Obama captured a combined four-fifths of all nonwhite voters, winning not only 95 percent of African-Americans but also 67 percent of Hispanics, 62 percent of Asian-Americans, and 66 percent of all others. Although exit polls showed the Democratic share among all minorities slipping to 73 percent in the 2010 House elections, almost all surveys this year show Obama nearing, or even exceeding, his overwhelming advantage among those voters from 2008.
African-Americans show no cracks in their Democratic preference. From 1980 through 2000, GOP nominees performed somewhat better among black men than women, but that gap narrowed in 2004 and essentially vanished by 2008. Likewise, through 1996, Republicans usually ran somewhat better among African-Americans with college degrees than those without them, but that gap has also disappeared. Among blacks, the modest marriage gap evident in earlier years has also closed.
Hispanics have shown more cracks. Overall, Democrats have carried these voters in every race since 1980, capturing at least three-fifths of their vote in six of the eight contests. George W. Bush, who made a major play for Hispanic support, reached 43 percent in 2004 (although some Latino analysts continue to dispute that exit-poll finding), but after the GOP rejected comprehensive immigration reform in Bush’s second term, McCain fell back to just 31 percent support.
Republicans have run more strongly among Hispanic men than women in each of these eight elections except 1992, although the difference usually has been modest. The class pattern in Hispanic preferences isn’t as consistent. In 1988, 1996, and 2000, Republicans ran better among college-educated than noncollege Latinos (re-creating the pattern evident among whites from the 1940s through the 1980s). But that gap narrowed in 2004 and reversed in 2008, with Obama running more strongly among college-educated than noncollege-educated Hispanics.
Religion eclipses these other differences. The Democratic nominee ran at least 9 percentage points better among Hispanic Catholics than Hispanic Protestants in each election from 1980 through 2004 except 1984. The Democratic advantage among Hispanic Catholics (who represent about three-fifths of all Hispanics, according to the Pew Hispanic Center) is overwhelming: The Democratic nominee has averaged about 70 percent of their votes since 1996, and even George W. Bush never attracted more than about one-third of Latino Catholics.
Bush made his greatest gains among Hispanic Protestants, winning just under half of their votes in 2000 and a 54 percent majority in 2004. Protestants make up about one-fifth of all Hispanics, and evangelicals, preponderantly socially conservative, represent about two-thirds of that group. The exit poll did not include enough members of this group to provide precise figures on their behavior in 2008, but they remained more Republican than Hispanic Catholics.
Amid widespread economic discontent, exit polls showed Republicans increasing their share of the Latino vote to 38 percent in the 2010 House races, even though Democrats maintained larger advantages in several key Senate races. But Romney wielded unwaveringly conservative positions on immigration to cudgel his rivals during the GOP primaries, and surveys since then give Obama a plausible chance of exceeding his 67 percent among Hispanics from 2008. (The latest Univision polling shows Obama attracting three-fourths of Hispanic Catholics and almost three-fifths of Hispanic Protestants.) That’s a remarkable prospect, given Latinos’ economic distress. Republicans appear paralyzed between their intellectual recognition that their party must improve its appeal among the rapidly growing Hispanic population and the political constraints imposed by their reliance on blue-collar and older-whites, groups that polls show are the most uneasy about the demographic change remaking the country. (See “Separate but Equal,” NJ, 6/4/11, p. 17.)
The religious alignment of American politics once set Democratic-leaning Catholics (many of them immigrants) and Southern evangelicals against the Northern Protestants who founded and long dominated the Republican Party. The past few decades have scrambled that alignment, as Southern evangelicals and religiously devout white Catholics have largely realigned with the GOP, while Democrats have offset the losses with the growing population of both the nonwhite (many of them religiously devout) and the nonreligious.
White Protestants, split between evangelicals and mainline denominations, have provided the GOP substantial margins in each election since 1980. The Democratic vote among white Protestants has ranged between 31 and 36 percent (except for 1984 when it sunk to 28 percent). Evangelicals, who represent just over half of white Protestants nationwide, are especially reliable conservatives. To underscore the power of evangelical belief in shaping behavior, consider how powerfully it bends even class divisions. In 2008, McCain beat Obama, 71 percent to 26 percent, among the nearly half of noncollege whites who also identify as evangelicals. But among noncollege whites who don’t identify as evangelicals, Obama drew 50 percent to McCain’s 48 percent. College-educated whites display similar patterns.
White Catholics have recently behaved less predictably than white Protestants. The Republicans hit their high point with white Catholics in 1984 and 1988, when they carried these voters by identical 14-point margins. Clinton won them in 1992 and 1996, but they tilted back narrowly to Bush in 2000, and then decisively in 2004. In 2008, white Catholics gave McCain a slim 52 percent to 47 percent edge over Obama.
In both groups, religious practice, rather than religious belief, increasingly looms as the dividing line. The University of Michigan’s American National Election Studies show little difference during the 1950s and 1960s between the voting preferences of Americans who attended religious services regularly and those who did not; like the class inversion, a religious-attendance gap emerged as cultural issues loomed larger, beginning in the 1970s. Today, the more often voters attend religious services, the more likely they are to vote Republican.
In each of the past three elections, exit polls show, the GOP nominee has defeated the Democrat among white Protestants who regularly attend church (once a week or more) by almost exactly 3-to-1. By contrast, Democrats have attracted two-fifths of white Protestants who don’t regularly attend. Republicans since 2000 have likewise outpolled Democrats 3-to-2 with white Catholics who regularly attend church. But white Catholics who don’t regularly attend church narrowly backed Gore in 2000 and Obama in 2008, and only slightly preferred Bush in between.
Overall, then, Republicans have carried whites who regularly attend church by about 2-to-1 in each of the past three elections. But over that same period, Democrats have averaged 57 percent support among whites who don’t regularly attend church. Similarly, Democrats since 1988 have averaged 61 percent of the vote among whites who declare no religious preference, reaching their contemporary high of 71 percent behind Obama in 2008. In each of the past two elections, Democrats also have amassed at least 80 percent of the vote among minorities with no religious preference. In 2008, voters with no religious preference represented 12 percent of the electorate, slightly more than double their share in 1980 and 1984. For Democrats, the secular are an answered prayer.
Political strategists disagree on how many voters who call themselves independents are really reliable partisans who just don’t want to publicly wear their team’s jersey. But self-identified independents still represent a big enough share of the vote to tip a competitive election. Since 1980, Bush in 2004 was the only candidate to lose independents and win the White House (Gore also won the popular vote while losing them).
Through much of this period, Republicans performed better with independent men than independent women. (Obama, however, narrowly carried both groups in 2008.) Every Democratic nominee since 1980 has run better with college-educated than with noncollege independents. Race and ideology loom as critical factors, too. Democrats have dominated among nonwhite independent voters throughout the period—even during the Reagan landslides. White independents gave the GOP crushing margins during the 1980s, moved slightly toward Clinton in the 1990s (with many detouring to Perot), and have leaned slightly to the GOP since (they preferred McCain over Obama by just 2 percentage points).
Liberal independents have preferred Democrats by at least 3-to-1 in each election since 1988. Conservative independents have provided the GOP nominee comparably emphatic margins in all eight elections.
Independents who describe their ideology as moderate—arguably, the bull’s-eye at the dead center of the electorate—may be a group especially worth watching. Reagan carried them by roughly 20-point margins in his two races. They broke narrowly for George H.W. Bush in 1988 and more substantially for Clinton in his two victories. George W. Bush lost moderate independents by just 3 and 4 percentage points, respectively, in his two races, but they tilted sharply to Obama, providing him an emphatic 16-point edge over McCain.This will likely be one of many groups among which Romney must meet or exceed the second Bush’s performance to overcome Obama—and the ongoing demographic changes that have made the road to the White House even slightly steeper for Republicans than it was in 2008.
This analysis was first published on the National Journal in August
Leave a Reply