Most Voters Haven’t Changed Their Minds All Year

For many Americans, and not a few journalists, this election can’t end soon enough. Fifty-nine percent of Americans reported that they were “exhausted” by election coverage — and that was in early summer. Since then, those following the campaign have been drinking from a fire hose of election news: the conventions, the Khan family, Hillary Clinton’s pneumonia and stumble, Donald Trump’s taxes and the Access Hollywood tape, and James Comey’s latest announcement, to name just a few.

But how much have these story lines actually changed the race? That’s a hot topic these days, with some polls showing dramatic swings toward Trump while others have hardly budged. Writing at YouGov, Ben Lauderdale and Doug Rivers make the case that a lot of what we are seeing is driven by changes in who responds to surveys, not in who voters intend to support. If so, this presidential race might be more stable than you would think by looking at polls on a given day.

Imagine for a moment that we lived in a democracy with snap elections, and that Trump and Clinton faced off in January of this year. Would the distribution of public support have differed much from what it is today? With a team of researchers including my University of Pennsylvania colleague Diana Mutz and the University of Massachusetts’ Seth Goldman, I’ve been conducting the Institute for the Study of Citizens and Politics panel, a survey which has asked the same respondents questions about politics over almost nine years. So we know what our respondents thought, both in January/February 2016, just before and after the Iowa caucuses, and in mid-October, when our last wave of the survey was fielded. (One caveat: Since the panel began more than eight years ago, we are missing any Americans who are now under age 26, making this poll likely to skew slightly against Clinton.)

The table below shows the responses for the 1,227 respondents who completed our October poll, with the rows showing their candidate support in January and the columns indicating candidate support this October.¹ When asked about a matchup among Clinton, Trump and other candidates², a significant majority of the respondents gave the same answer both in both months: 37.2 percent backed Clinton both times, 30.9 percent backed Trump both times, and 12.7 percent opted for another candidate or said they wouldn’t vote both times. In all, that means that around 80.7 percent of respondents gave the same answer in both surveys. (Of course, that doesn’t rule out the possibility their support shifted in between — just that they settled back where they were.)

Notice another thing: There is very little movement between the two major-party candidates. 0.5 percent of respondents were with Clinton in January and are now with Trump, while only 1.5 percent of our respondents moved from Trump to Clinton. There is far more movement from not supporting either candidate to backing one than from supporting one major-party candidate to supporting the other. That, in turn, stabilizes the horse race to some degree: Voters who can be persuaded to switch sides are twice as influential, since they affect both candidates’ tallies simultaneously.

One general expectation is that over a campaign, the number of people who are undecided or planning not to back a major-party candidate should decline as people learn more about the candidates. One unusual aspect of this election is that the number of people who refuse to back Clinton or Trump has remained high as the election approaches. And our panel data reinforces that conclusion: Almost equal numbers of people have moved away from the major-party candidates as have moved toward them.

Let’s say we restrict our sample to the 933 survey respondents who we know voted in either 2008 or 2012, based on post-election voter file updates. The number of people who refuse to back either candidate drops somewhat, to 10.1 percent, meaning that the people who refuse to back a major-party candidate are lower-propensity voters, on average. But four key features of the data — a Clinton lead among this population, a small number of people moving between the candidates, a large number of people who won’t commit to a major-party candidate, and little change in the total number of those unwilling to commit — remain.

To appreciate these results, it is worth comparing them to those of 2008, when our survey asked respondents about whom they supported in the late spring and summer as well as after Labor Day. Below we see the same table looking at support for President Obama and for John McCain in those two waves of the panel for our same 1,227 respondents. Notice that overall stability is comparable, with 38.1 percent of people always backing Obama, 33.5 percent of people always backing McCain, and 8.7 percent of people always refusing to support either. But here, we see a tad more movement between the candidates, with 2.3 percent of people going from McCain to Obama and 1.7 percent going the opposite way. What’s more, there is the expected asymmetry in who refuses to back either candidate, with more people moving toward the candidates than away from them. What makes 2016 distinctive is that voters are not increasingly gravitating toward the candidates as the election nears. Pundits commonly call people who won’t commit to a major-party candidate “undecided,” but this data suggests that many of these folks have decided. They’ve simply decided to not support either Trump or Clinton.

Of course, this evidence does not imply that people’s preferences have remained fixed since January. It could well be that all of the negative news cycles surrounding accusations of sexual assault and poor debate performances by Trump meant that our mid-October survey caught him at his low-water mark, one that happened to match opinions at the outset of primary voting. It’s also key to acknowledge that the people who remain in a panel for years may differ from others, and that, too, could lend their views a stability not found in the broader electorate. But from these respondents alone, we see an electorate that appears surprisingly stable.

With Nov. 1 happening to fall on a Tuesday this year, Election Day is slated for Nov. 8, the latest date on which elections can take place under the “first Tuesday after the first Monday” rule. And with daylight saving time providing an extra hour this weekend, it feels a bit like the fates have conspired to make this an asymptotic election, one that gets closer but never ends. Still, from this data, it looks as if a snap election in January might just have produced similar results.

Nate Silver: Trump’s chances of victory have doubled since last debate