Late Polling Shift? What the YouGov Survey Shows
With the 2016 campaign in its final days, the set of questions being asked by political junkies, party professionals, and ordinary voters are tending to dovetail. Mainly, all of us want to know two things.
First, are the pre-election public opinion surveys that show Hillary Clinton with a diminished but lingering lead correct or is there some sort of “shy Trump voter effect”—or any other phenomenon—that’s been misleading us all?
Second, has the drip, drip, drip of continued WikiLeaks revelations, along with FBI Director James Comey’s announcement that the bureau has reopened the investigation into Clinton secret email server, altered the landscape in a way that suggests a Trump upset?
The first question hinges on a dubious theory for which there is only sketchy and anecdotal evidence. Still, the answer is unknowable and will be revealed in the fullness of time—Tuesday, Nov. 8, to be exact. Data does exist on the second question—partial data, anyway—but first a bit of background.
Barack Obama won re-election largely because 5 to 6 percent more Democrats went to the polls in 2012 than Republicans, and because the president persuaded well over 90 percent of them to vote for him. Thus, even though Mitt Romney did very well among Republicans, he had to make up that 6 percent gap by drawing more Independents. He won a majority, but not enough to capture the presidency. So what does this mean for Donald Trump?
In September 2015, we reported from the YouGov recontact poll (the same 5,000 respondents interviewed monthly since May of 2015) that going into the election cycle, the Democratic Party had retained its nine-percentage-point advantage over Republicans. The implication here is that if Clinton could do as well with Democrats as Obama had, Trump would be in trouble. Specifically, he’d have to (a) attract as many (or more) Republicans to his side as Romney did in 2012; and (b) win about 60 percent of Independents.
A late August-early September 2016 poll showed that Trump had work to do with Republicans. But so did Clinton with her party. Neither nominee had yet sealed the deal with their own partisans, with Clinton at 73 percent of Democrats and Trump at 77 percent of Republicans.
The most recent YouGov surveys are out, including the recontact survey finished on Nov. 1, and Democrats still hold a sizeable advantage among registered voters over Republicans (eight to nine points), with Clinton now attracting 91 percent of the Democrats to her side. Also following the normal pattern, Republicans “came home” to Trump: He went from 77 percent Republican support to 86 percent.
While the movement toward Clinton is favorable, she is still a little short of Obama’s level of party support, and Trump has also not quite met the standard set by Romney. Clinton is doing well among women, African-Americans, and Hispanics but not as well as Obama did in 2012. Trump is winning men but not by as much as Romney did in 2012.
Why is this? The answer is that education levels are a more significant factor this year. Obama won a majority of those with a high school diploma (or less) in 2012, while Romney won college-educated voters. This year the numbers are reversed. Among white voters with only a high school education, Trump leads by over 25 points. Among whites with a college degree, Clinton leads by about 10 percent.
This is the first time since serious polling began in 1952 that this has happened. The traditional pattern of Democratic support among blue-collar workers this year follows the high-school-or-less pattern with white, blue-collar workers preferring Trump to Clinton. These unique combinations of less support for traditional party loyalties across education levels accounts for the narrow Clinton lead of four points in our latest poll. Another example, which makes the point, is that women with less education are voting for Trump while college-educated women are very strong for Clinton, which cuts into the gender gap and makes Clinton’s lead among women slightly less than Obama’s was in the previous election.
Neither the various WikiLeaks revelations and other hacked Democratic Party emails, nor Jim Comey’s announcement about reopening the FBI’s investigation of Clinton’s email practices, have induced voters who identify as Democrats to change positions. Ninety-eight percent of Democrats who preferred Clinton prior to the first debate are still with her. (The same is true among Republicans supporting Trump.) Moreover, 93 percent of Democrats who reported voting for Obama in 2012 are voting for Clinton and 85 percent of Republicans who voted for Romney are voting for Trump. In short, among partisans, events do not trigger position changes very often, and that phenomenon has continued even in this remarkably contentious election season.
This leaves Independents, who almost by definition are more susceptible to alterations in the political environment. Trump has made some progress since the last poll, going from nine points ahead of Clinton to 11 points. Partly, this is because the number of voters saying they were undecided or voting for third party candidates has been cut in half. By Election Day, their ranks will be thinned even further.
So we pressed the holdouts: “If you had to vote for either Hillary Clinton or Donald Trump, who would you vote for?” The answer is that Clinton continues to receive more of this vote than Trump, which somewhat offsets Trump’s lead among Independents and—coupled with Democrats’ advantage in party identification and Clinton’s stronger support within her own party—indicates that Trump’s lead among Independents is probably not large enough to give him a victory on Nov. 8.