If the electorate and Obama’s share of the non-white vote are resemble 2008, a Romney victory would require an extraordinary performance by Romney among whites. Is the former likely and can Romney do better than any Republican since 1984 among white voters?
It's worth noting that there are charts and graphs that didn't make it onto this diary, so you might want to read it in original form here:
http://electionate.com/...
http://electionate.com/...
electionate.com/2012/04/19/does-obama-have-a-demographic-firewall-part-iii-whites/
-----------
There were times when the race seemed genuinely close during the 2008 campaign. It’s easy to forget that McCain led in early September, before Lehman clarified the scale of the financial crisis and Couric clarified Palin’s preparedness. But when I opened up a spreadsheet and started entering exit poll data in the early hours of November 5, I began to wonder whether the outcome of the race was ever really in question.
The composition of the 2008 electorate was unlike any in American history. Whites represented just 74% of the electorate and the burgeoning non-white vote offered 81% of its support to Obama. As I fiddled around with the spreadsheet, it became clear that a McCain victory would have required a historic performance among white voters. Indeed, even if McCain had repeated Bush’s performance among whites, McCain still would have lost by nearly 5%. For McCain to prevail, he would have needed to reduce Obama’s share of the white vote to just 37%.
It would have been extremely difficult for McCain to hold Obama to 37% of the white vote. Michael Dukakis was far from a vaunted candidate, but even he won 40% of the white vote in 1988. Even when Obama appeared to trail in early September, he probably would have won more than 37% of the white vote, and probably the Presidency. A Pew Research Poll, taken September 9-14, showed Obama up 2% with 46% nationally, but with 38% of the white vote and 10% undecided. Similarly, a Washington Post poll, taken 9/7-9/9, showed McCain leading by 2%, but with Obama at 38% of the white vote. After allocating undecided voters, Obama would have easily won the necessary number of white votes, assuming that the non-white vote played out as it eventually did.
In the modern political era, it has taken extraordinary circumstances to hold Democrats to 37% or less of the white vote. The last Democratic candidate to fall so low was Walter Mondale, who only won 35% of the white vote in 1984. In 2010, House Democrats only won 37% of the white vote, demonstrating that the GOP can still hold Democrats to 37% of the white vote. On the other hand, it required exceptionally favorable conditions and, by most accounts, low Democratic turnout. Given Democratic enthusiasm, the prevailing political climate in 2008, and Obama’s then untarnished appeal, it seems almost inconceivable that McCain could have produced a similar performance among white voters.
The prevailing narrative of the 2008 election – that the election was close until the financial crisis – is probably accurate. But in this instance, close might not have meant especially competitive. Obama probably would have held a narrow, but nearly unshakable lead based on historically high-levels of support and turnout from non-white voters, even without the flood of swing voters that accompanied the financial crisis or successful debate performances.
If one assumes that Obama was assured more than 37% of the white vote in 2008, then Obama was also assured victory, assuming eventual non-white turnout and non-white support for Obama. In my view, these assumptions were probably accurate, and combined to constitute a demographic firewall. Obama’s demographic firewall was not impenetrable, but it was high enough that a McCain victory required an extraordinary performance among whites, more like landslide blowouts in 1972, 1984, or 2010 than 2004 or even 1988. Given the circumstances, such an extraordinary performance was nearly inconceivable.
In 2012, it is not clear whether the Obama campaign can recreate the demographic firewall. Certainly, Obama’s support and turnout among non-whites was at least somewhat reliant on historic circumstances. Even if the demographic firewall does reemerge, it’s less clear whether it would still be high enough. The GOP limited Democrats to 37% of the white vote in 2010 and one would be unwise to assume that it is impossible for Romney to repeat the GOP’s success.
Regardless, if non-white voters turnout and vote as they did in 2008, the GOP will need to win historic numbers of whites. The next two posts will assess each question in turn: 1) whether non-white turnout and support for Obama will match or exceed 2008 levels and 2) whether the GOP can win the requisite share of the white vote necessary to prevail. Of course, those are two big “ifs” – there is no guarantee that non-white voters will turnout or offer their support to the President at the same rate that they did in 2008.
Determining whether non-white turnout and support for Obama will match 2008 levels is extremely challenging. Analysis is clouded by a host of potential variables, since depending on whether non-white voting behavior was the result of an especially favorable environment, demographic trends, youth enthusiasm, Obama’s historic candidacy, or Obama’s formidable ground operation, one might draw very different conclusions about the likelihood of a repeat in 2012. Further complicating matters is the diversity of the non-white vote. The factors underlying African American turnout and support were probably different, at least in relative weight, than those that motivated other non-white voters. Even if there was a clear answer, the course of the campaign, including Romney’s Vice Presidential selection, could undermine today’s conclusions before November. Nonetheless, the best evidence suggests that non-white support for the President should approach 2008 levels. While judging turnout is more difficult, non-white turnout should approach or exceed 2008 levels.
The best evidence for unwavering non-white support (note: not turnout) for Obama is in the 2008 polling data and 2010 election results. Even when Obama was at his nadir in early September 2008, non-white voters continued to offer overwhelming margins to Obama. Two polls following the RNC from the Washington Post and Pew Research showed Obama receiving 66% and 61% of the Hispanic vote. Those numbers are marginally short of Obama’s eventual 67%, but considering the large number of undecided voters, I think it’s probably fair to suggest that Obama would have received something near 67% of the Hispanic vote, even at the low point of his campaign.
Similarly, Democrats still received overwhelming margins among non-white voters during the 2010 midterm elections. Harry Enten has already elaborated on this point, but the short story is that Latino support for Democrats was comparable in 2006 and 2010, and only somewhat lower than in 2008, depending on whether you use Latino Decisions or network exit poll data. If non-white voters held firm for Democrats in 2008, they are likely to again offer overwhelming support to Obama. Compounding the issue is Romney’s position on immigration-related issues, which might risk an even worse performance than McCain.
Additionally, it is difficult to imagine that Obama will receive substantially less than the 95% of the African American vote he won in 2008. Even Al Gore received 91% of the African American vote, and Obama will do better. Although it’s possible that some of Obama’s marginal supporters were only committed to electing the first black President, many of those same voters can be expected to defend him in an extremely controversial, negative, and polarizing election.
While Obama can expect a similar share of the non-white vote in 2012, turnout is more questionable. Non-white voters have lower turnout rates and there are reasons to believe that 2008 was unique. In particular, exceptional levels of African African turnout were likely driven by historic circumstances – the chance to elect the first African American President. However, Obama’s challenge with African American turnout should not be overstated. To be sure, some 2008 African American voters may only have been inspired to elect the first black President, and as a consequence, I think all would concede that black turnout rates are likelier to fall than not. The question, however, is by what amount, and my suspicion is that it won’t fall by precipitously.
Some recent polls show that black voters are among the most enthusiastic, and some polls even show that they remain the most enthusiastic, although it’s worth noting that there are conflicting data points. There’s reason to expect interest to increase further, since a heated, negative, and controversial campaign will probably reignite 2008 passions and arouse old supporters to turnout in Obama’s defense. Given that African Americans were 11% of the 2004 electorate, compared to 13% of the 2008 electorate, at least 12% seems quite reasonable, including the very real chance of repeating 13%. At the same time, one should not understate the importance of any decrease in black turnout. For every one point drop in African American turnout, Obama’s national popular vote percentage dips by just over one half of one percent.
In addition to historic African American turnout, high minority participation was also dependent on youth turnout, since young voters are disproportionately diverse, and most, although not all, of the growing minority voting age population comes from young voters. Obama certainly held tremendous sway among young voters in 2008, but at least some of their enthusiasm for his candidacy has faded. A tarnished Obama might be less likely to drive out young voters, who are among the most difficult to turnout.
However, there’s reason to question whether non-white youth turnout will drop. Perhaps contrary to expectations, 18-30 year olds were 17% of the electorate in 2008, just like they were in 2004. Given overall increases in turnout, youth turnout was high in absolute terms, but was not so high as to increase their share of the electorate. As mentioned before, a bitter and controversial campaign will generate high turnout, period.
Even if youth, non-white turnout decline, it may not be sufficient to outweigh the growing non-white share of the voting-age eligible population. As mentioned in the last post, the white share of the electorate declined by 3% in every Presidential election since 1996 – from 83 to 80 to 77, and, in 2008, 74%. This long term trend suggests that higher minority turnout might not have been as much the result of unique circumstances as many believe, but instead a product of broader demographic trends. Similarly, between 2006 and 2010, the white share of the electorate in off-year congressional elections fell from 79% to 77%. The decline in the white vote was not as much as the Presidential elections, but enough to demonstrate that the 2008 electorate was also a product of demographic changes, not just one-time circumstances.
Will these demographic trends outweigh a decline in non-white turnout rates? Probably. Assume, for the moment, that if non-white turnout rates remained constant, that demographic trends would reduce the white share of the electorate by another three percent, to 71% in 2012. A three percentage point increase in the non-white electorate is an 12% percent increase – and it seems highly unlikely that non-white turnout could decrease by more than 12% in a competitive and controversial election cycle. Now, whether demographic changes should result in an 11% increase in non-white turnout is a fair question, but no matter how one examines the question, it takes a pretty meaningful drop off in non-white turnout to match demographic gains, let alone rollback the non-white share of the electorate.
This chart might help visual thinkers. The two lines represent (DISCLAMER: the data is not real, and for illustrative purposes only, although I do think it roughly approximates reality) two age lines for non-white eligible voters. On the left, the line for 2008 – as you can see, old people are disproportionately white and young people are relatively non-white. Over the last four years, people have aged and the line consequently moves four years to the right. There is a new wave of disproportionately non-white voters who have since entered the electorate. The decrease in non-white turnout rates must be large enough to overcome the influx of non-white voters who have become eligible over the last four years. By some estimates, this will increase the non-white share of the electorate by 1 or 2% in most swing states.
Finally, Latino turnout is unlikely to decline. Much like the youth vote, Latino turnout held steady at 9% of the electorate, suggesting that while Obama’s candidacy generated uniquely high black turnout, Latino turnout was less dependent on Obama’s historic candidacy. Given that Latino registration rates remain far lower than the rest of the population and forthcoming efforts to register new voters, combined with Latino population growth, it’s tough to imagine lower Latino turnout, and quite easy to envision further increases.
While there are many conflicting trends, in the aggregate, non-white participation and support for Obama should remain similar to 2008. Substantial declines in turnout seem relatively unlikely, and long-term demographic trends are probably sufficient to mitigate, if not overwhelm, any decrease in participation rates. Although it is impossible to precisely estimate the composition of the electorate one week before an election, let alone seven months, it is highly unlikely that non-white turnout will revert to anything reminiscent of 2004, and there is a good chance that non-white turnout will match or exceed 2008 levels.
Regardless of the exact turnout, the electorate and Obama’s share of the non-white vote are likely to resemble 2008. If so, then Obama’s 2008 demographic firewall is likely to reemerge. It is too far from election day to judge more exactly, but Obama’s strength among non-white voters will require an extraordinary performance by Romney among whites. However, it would be unwise to assume that such a performance is impossible, and the next post will consider the difficulty of Romney’s challenge, and, if I may so, that’s a really interesting question.
Unfortunately, recent Presidential elections do not provide a clear reference point for weighing Romney’s chances of holding Obama to less than 38% of the white vote. Walter Mondale was the last Democratic Candidate to do so poorly, and no Democratic candidate has since fallen beneath 40%, except Clinton who was held to 39% of the white vote by Perot’s third party candidacy. Obama’s strength among non-whites has forced Republicans into uncharted territory and there is considerable uncertainty; we don’t know what this looks like or what it takes.
Despite the uncertainty, a number of data points help clarify Romney’s challenge. The best starting point is the 2010 Midterms, when Republicans held Democrats to 37% of the white vote – precisely where Romney must hold Obama. Unfortunately, midterm results are not entirely analogous to Presidential results. First, parties usually nominate candidates well-suited to their districts, and House Democratic candidates exceeded Obama’s expected share of the vote in Appalachia and the white south. On the other hand, off-year turnout is not representative of Presidential election year turnout. One easily identifiable example is the youth vote. In 2008, 18-29 year olds constituted 17% of the electorate, compared to just 12% in 2010. There are changes in other Democratic-leaning categories as well, including a decline in Democratic partisan-ID and a decrease in Labor turnout. On the other hand, other areas show a more favorable electorate for Democrats. For instance, there were a disproportionate number of college educated whites, and Obama does better among them than their high school educated counterparts. Conflicting indicators aside, most agree that turnout was tilted toward the Republicans. Regardless of the details, it would be unwise to look at the 2010 midterm results and conclude that it represents how Obama would have performed among in similar circumstances.
Although there are limits to extrapolation from off-year election results, they are still representative enough to glean limited insights. After all, the exit poll found Obama’s approval rating at 44% among 2010 voters, just under the Democrats’ 45% of the overall vote and Obama’s 45% in the RCP approval average, although that similarity fades if one relies on net-approval. In the aggregate, demographic imbalances in the 2010 midterms might roughly balance out; Obama would have underperformed House Democrats in the South, but would have benefited from a Presidential-year electorate with a greater number of young white and low-interest white Democratic voters. The 2010 midterm elections thus provide a route to hold Obama beneath 38% of the white vote, even though no Republican Presidential candidate has done so in recent memory. The existence of the possibility, however, does not meaningfully inform probability, and that, unfortunately, is a subjective and interpretative question.
Part of the issue is whether the 2010 midterm elections should be considered an anomalous, best case scenario for Republicans. Certainly, the 2010 midterm elections were conducted under extraordinary circumstances. The RCP right/wrong track average showed that 63% of Americans believed that the country was on the wrong track. Similarly, the President’s approval rating was beneath 45% and fervent opposition to the President’s health care and spending policies drove historic turnout among GOP voters. To the extent that congressional approval ratings are relevant, suffice it to say, they were really bad.
As bad as these numbers were, none are worst-case. In October 2011, after the debt ceiling crisis and Obama’s failed jobs bill, the number of voters who said the US was on the wrong track surged to 75% and Obama’s approval rating fell to approximately 42% in the RCP averages. Today, the fundamentals aren’t significantly better. The right/wrong track numbers are approximately where they were on Election Day 2010, and Obama’s approval rating is marginally higher at 47%. If the 2010 elections were extraordinary, anemic economic growth, employment figures, approval ratings, and right/wrong track numbers suggest 2012 could be extraordinary as well.
But while the environment appears conducive to an extraordinary GOP performance in 2012, such a performance will still be exceptionally challenging. After all, there is a reason why fatally flawed candidates like Dukakis managed to exceed 37% of the white vote, or why Mondale still received 35%; such weak performances require Democrats to lose voters who lean toward the Democratic Party. Unfortunately, the 2010 exit polls didn’t breakdown party identification by whites so, instead, consider the 2008 exit poll as an example. If white partisan ID held constant from 2008 (31D, 39R, 31I), a GOP victory would require Romney to win a substantial number of self-identified Democrats, nearly every true independent voter, as well as independents who probably lean toward Democrats in national elections. Here’s one scenario for a bare victory: the GOP holds Obama to 37.2% of the white vote by limiting him to 81% of white Democrats, 5% of white Republicans, and 33% of white independents. Now, presumably the partisan leanings of the white vote have shifted since 2008, and while that should reduce the number of white Democrats and increase the number of white Republicans, it ought to produce a more solid Democratic base than in 2008, which would lessen the impact of changing partisan ID. Put differently, whether 08-Democrats call themselves independent or not, the GOP needs to win over many voters who considered themselves Democrats four years ago.
Superficially, it sounds challenging for the GOP to win such a broad coalition, but is it out of reach in the current political environment? Pew Research’s comprehensive demographic breakdowns help illustrate. In particular, Pew provides a breakdown of Obama’s approval and head to head numbers by white partisan leanings, rather than among whites or a given party as a whole. These polls portend ominously for Romney – his chances are in the hands of Democratic-leaning whites, and even at Obama’s worst moments, these Democratic-leaning whites appear willing to reluctantly endorse Obama. Throughout early 2012 and late 2011, Obama meaningfully outperformed his approval rating among whites in head to head match-ups with Romney for a relatively straight forward reason: plenty of Democrats were unwilling to say they approved of Obama, but willing to vote the party line. Unfortunately, not all Pew Polls include comprehensive data on both approval and the horserace, but these four polls include both.
Based on other Pew data, we can infer that the gap among independents probably comes from Democratic-leaning independents. In a summer 2011 report, Pew found that 39% of white voters were either Democrats, or leaned toward Democrats. Considering the similarity between that figure and Obama’s tallies among whites, it seems reasonable to assume that much of the deviation between Obama’s approval rating among independents and success against Romney comes from Democratic-leaning independents. It is worth noting that Pew found 44% of whites leaning toward or identifying with Democrats in 2008, and that same number ultimately voted for Obama in the Presidential election.
Unfortunately, other polls do not provide similarly detailed demographic data. For instance, in the Gallup tracking poll, Obama has averaged 79% approval among Democrats since the debt ceiling debacle, a figure which would seem consistent with the Pew data. However, the breakdown by race also shows Obama’s approval at 85% among blacks and just 52% among Latinos over that same time period, suggesting that much of Obama’s low approval rating among Democrats could be attributable to non-white voters. It is also worth noting that other polls have not consistently shown a large gap between Obama’s approval rating and performance against Romney, such as PPP and Quinnipiac, although without intra-white demographic breakdowns, top-line numbers are alone insufficient to contest Pew’s findings.
Another way to consider Romney’s challenge is in terms of the number of white voters who appear unwaveringly committed to Obama. Take October/November 2011, for instance, when the political climate was at its worst for the President. If there was ever a time when support for Obama should have cracked, it was then. However, even at the lowest moment of his Presidency, Obama was near the requisite number of white votes to secure the Presidency. Pew Research actually found Obama at 38% of the white vote, even though only 33% approved. Gallup’s tracking poll tends to confirm that Obama’s weakness was mainly due to disaffected Democrats, as they found Obama’s approval rating among whites at 32%, with just 72% of Democratic voters approving. Similarly, PPP found Obama holding 35% of the white vote against Romney, but with 11% of whites undecided, far more than the 4% found by Pew. If PPP was accurate, Romney would have needed more than 73% of the undecided white vote to hold Obama beneath 38%. This is not impossible, but reasoned minds can seriously debate whether it is actually realistic.
Recent polls, which show Romney competitive or leading nationally, actually show Obama near or above the requisite number of white voters to win the popular vote. On average, the five polls in the RCP average with appropriate demographic breakdowns (that I can find!) found Obama with 38.8% of the white vote. If one holds 2008 non-white turnout and support for Obama constant, that gives Obama 50% of the popular vote, somewhat higher than the 48% of the vote in these five polls, presumably due to undecided non-white voters.
In these polls, Obama already has the requisite number of white voters, even before securing any undecided voters, many of which voted for him in 2008. If accurate, then Romney has an uphill battle. Not only would Romney need to sweep the undecided voters, he would need to rollback Obama’s support among whites by approximately 1%. Alternatively, Romney would need to reduce Obama’s share of the non-white vote or construct a whiter electorate. All of these options are difficult, but not impossible. Serious analysts can differ over whether any are especially realistic.
As these numbers show, it is not impossible for Republicans to hold Obama beneath 38% of the white vote, even though no Democrat has performed so poorly since 1984. Obama continues to face an extraordinarily difficult political climate, and the 2010 midterm election results demonstrate that such a performance is conceivable in the modern political era. It is also worth emphasizing that the “requirement” for Romney to hold Obama to beneath 38% of the white vote is inherently imprecise. Even though non-white voters seem likely to vote as they did in 2008, no single voter group, let alone an artificial category encompassing all “non-white” races and ethnicities, will behave precisely the same way in consecutive elections. Subtle changes in non-white voting patterns proportionately alter the requisite number of white voters Romney must win to secure the Presidency, and those changes could potentially be to Romney’s benefit, although they could also work the other way.
Even so, Obama’s demographic firewall places Romney at a serious disadvantage. To compensate for high non-white turnout and support for Obama, a bare Romney victory requires an overwhelming share of the white vote. Given the exceptionally favorable conditions and recent polls, Romney should probably do better among whites than any Republican candidate in recent history. However, if Romney pushes past George H.W. Bush’s share of the white vote and into a close race nationally, his path to the Presidency becomes increasingly difficult. After maximizing his share of white Republicans and genuine swing voters, Romney will need to trudge through Democratic leaning whites. Recent polls provide cause for optimism on both sides. While many of these voters do not approve of the President, they may not necessarily disapprove and currently appear willing to revert to their partisan leanings when forced to choose between Obama and a Republican, even when Obama’s Presidency was at its lowest point.