Election Update: Is Trump Getting A Convention Bump?
Nate Silver
Polls taken during and after the Republican National Convention, which concluded on Thursday in Cleveland, generally show Donald Trump continuing to gain ground on Hillary Clinton, making for a close national race. But it’s customary for candidates to receive a “bounce” in the polls after their convention. There’s not yet enough evidence to come to firm conclusions about the size of Trump’s convention bounce, but the initial data suggests that a small-to-medium bounce is more likely than a large one.
Before we run through the polls, a note of caution: The convention bounce is going to be harder than usual to study this year. That’s because in contrast to 2012, when the polls were extremely steady for weeks before the conventions, they were on the move heading into the RNC this year. In particular, they were on the move toward Trump — or away from Clinton — with Trump whittling down what had been a 6- or 7-percentage point lead for Clinton in late June into something more like a 3-point lead by mid-July.
So when you see a new poll suggesting that Trump has received a bounce, or failed to receive one, you’ll want to be mindful of when the previous edition of the poll was conducted. If the pollster had last surveyed the race in June, odds are that Trump has made some fairly big gains. Some of those were probably realized before the convention and not because of it, however. But if the previous edition of the poll was in July,1 his gains are likely to be smaller.
Keeping that in mind, here’s the data we have so far. First, there are three post-convention national polls, meaning that all of their interviews were conducted after Trump’s acceptance speech on Thursday night.
◾A RABA Research poll, conducted on Friday, shows Clinton 5 percentage points ahead of Trump, 39-34, with a large undecided and third-party vote. That sounds bad for Trump, but the trend line in the poll is favorable for him: The previous edition of the poll, conducted two weeks ago, had him down 12 points.
◾A Gravis Marketing poll, conducted Thursday and Friday, has Trump 2 points ahead of Clinton. Their previous national poll, in late June, had Clinton up 2 points instead.2 Note, however, that Gravis has generally shown better results for Trump than most other pollsters.
◾Finally, an Echelon Insights poll, conducted on Thursday and Friday, shows Clinton up by 1 percentage point, although Clinton’s lead grows to 4 points if Libertarian Gary Johnson and the Green Party’s Jill Stein aren’t included. Echelon Insights had not previously polled the election.
You see what I mean? Measuring the convention bounce isn’t so straightforward. At first glance, it appears that the RABA Research poll is good for Clinton and the Gravis Marketing poll is good for Trump. But the RABA Research poll shows Trump making big gains and, furthermore, doing so in comparison to a July poll. By contrast, a 2-point lead in a Gravis Marketing poll is a pretty “meh” result for Trump, given that it has generally shown Trump-friendly results and that Trump didn’t improve all that much from its previous poll in June.
Meanwhile, Echelon Insights hadn’t conducted a pre-convention poll, so putting its numbers into context is hard. It’s interesting, however, that its poll shows a relatively large difference based on whether or not Johnson and Stein were included. It’s plausible that some of Trump’s post-convention gains will come from wayward conservatives who were thinking about voting for Johnson, so polls showing third-party candidates may show a larger bounce for him than those that don’t. Likewise, liberal voters who were contemplating a Johnson or Stein vote may move to Clinton after next week’s Democratic convention.
There are also a couple of polls that contain a mix of post-convention, pre-convention and during-convention data:
◾The USC Dornsife/LA Times tracking poll, conducted from last Saturday through Friday, has Trump up by 2 percentage points. That suggests very little bounce, given that Trump had been up by 1 percentage point in the poll before the convention.
◾However, the Reuters/Ipsos tracking poll, conducted from Monday through Friday, shows Trump making major gains, trailing Clinton by less than 1 percentage point. Clinton had generally been up by around 10 percentage points in that poll before the convention began. Clinton’s lead is slightly larger, 3 percentage points, in the version of the poll without Johnson and Stein.
◾Finally, a Rasmussen Reports poll, conducted during the convention on Monday and Tuesday, shows Trump ahead by 1 percentage point. That’s actually down from a 7-point lead for Trump in Rasmussen’s poll last week, although that poll had been a big outlier.
This data is also pretty confusing. The massive gains Trump made in the Reuters/Ipsos are unabashedly good news for him. But their polling earlier in July had been something of a pro-Clinton outlier, so some gains for Trump were probably inevitable. Likewise, even though Rasmussen Reports has a long history of polls that show a statistical bias toward Republicans, a 7-point lead for Trump was a bit rich, even by Rasmussen standards, and the poll was likely to regress to the mean somewhat regardless of how effective Trump’s convention was. The trend lines in the USC poll aren’t great for Trump. But it only started publishing data a week or so ago, and the poll doesn’t include third-party candidates.
The FiveThirtyEight forecast models are helpful at times like this, but even they’re going to have trouble sorting everything out. Earlier this week, I wrote that “it would be a bad sign for Trump if he can’t at least tie Clinton in polls conducted in between the RNC and the DNC.” Our now-cast, which is very aggressive and addresses the question of what would happen in a hypothetical election held today, shows Clinton up by about 1 percentage point, so Trump has almost brought the race to a tie, but not quite. If Trump still trails in the now-cast after we get the next couple of polls in, the convention might qualify as disappointing for him, but it’s too soon to come to that conclusion. Meanwhile, Trump has continued to gain in our polls-only model, which is less aggressive than the now-cast but takes the polls at face value, whereas the trend in our polls-plus model, which builds in a convention bounce adjustment (it assumes that an average convention bounce is 3 to 4 percentage points) has been flat over the past few days.
I wish I had more definitive answers for you. But the data we’ve gotten so far is inconsistent and comes from a weird group of pollsters, several of which had shown outlier-ish results in one direction or the other before the convention began. I think we can probably rule out Trump getting a huge, 8-point bounce or something like that, and I think we can probably rule out his getting no bounce at all, but beyond that, we’re just going to have to be patient.
In another sense, however, the story isn’t so complicated. Whereas June’s polls suggested a potential blowout for Clinton, July’s polls have shown a highly competitive race. We’ll see what August’s polls bring, after the Democrats have held their convention and the bounces have died down.