The truth about Obama’s victory wasn’t in the papers
Andrew Gelman and John Sides
.........
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.
.....
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.
.......
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.
......
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.
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.
....
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.
.....
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.
bostonreview.net |
