By Nate Silver – – – – – –
There weren’t a lot of polls this past weekend, but the ones we saw didn’t have a lot of good news for Donald Trump. Instead, Hillary Clinton maintains a national lead of about 8 percentage points, and Trump’s chances of winning the election are down to 11 percent in our polls-only model — his low point of the year — and 21 percent according to polls-plus.
Instead of focusing on the details, let’s zoom out and ask a few big-picture questions about where the election stands. When we asked these questions a month ago, before the party conventions, Clinton held a lead of 3 to 4 percentage points over Trump, but Trump seemed to have momentum. Now, the landscape is very different:
Who’s ahead in the polls right now?
Clinton, by a lot. National polls that include third-party candidates have Clinton with an average of 44 to 45 percent of the vote, Trump at 36 to 37 percent, and Libertarian Gary Johnson at roughly 9 percent. State polls tell a broadly similar story. Trump’s low percentage of the vote is noteworthy: Every major-party nominee since 1928 has received at least 36.5 percent of the vote. (Democrat John W. Davis got just 28.8 percent in 1924.)
What’s the degree of uncertainty?
Moderate-to-high, although decreasing. The polls are often highly volatile around the party conventions, but they come out of the conventionsconsiderably more accurate than they were beforehand. The question is whether we’re far enough away from the conventions — which concluded on July 28 — that we can no longer attribute Clinton’s lead to some sort of “convention bounce.” There isn’t a hard-and-fast answer to this question, but for what it’s worth, the convention-bounce adjustment applied by our polls-plus model will begin wearing off this week. More importantly, we’re approaching the point where a Trump comeback would be relatively lacking in precedent.
But still, the number of undecided and third-party voters in the polls remains high, which has historically been an indicator of higher volatility. And the conventions concluded relatively early this year, so the comparison to post-convention polling from past years is somewhat imprecise.
What’s the short-term trend in the polls?
There’s room to ask whether Clinton’s lead over Trump has peaked and begun to decline slightly, or whether it’s continuing to grow. The evidence is mixed: You can find polls like this one and this one, where Clinton is at or near her all-time high, but you can also find others like this one and this one, where she’s still doing well but is a bit down from her peak. Overall, our models see Clinton’s position as fairly steady, although it’s possible that Trump is continuing to lose ground to third-party candidates and to the undecided column.
What’s the medium-term trend in the polls?
Clinton has unambiguously gained ground since the conventions, holding a lead of about 8 percentage points now as compared with 3 to 4 points just before the conventions began. There is, however, still some question aboutwhere the equilibrium in the race stands. Clinton held a mid-to-high single-digit lead over Trump for most of March, April and June — and now she has one again in August. But the election was apparently closer for much of May and July.
Which states shape up as most important?
The swing states have sorted themselves into order. According to our polls-only model, Clinton has a lead of at least 9 percentage points in states collectively worth 273 electoral votes, including New Hampshire,Pennsylvania, Virginia, Colorado, Michigan and Wisconsin. Ohio, Florida,Iowa, Nevada and North Carolina are closer, by contrast.
This leads to a situation where no one state is overwhelmingly more important than any other. But still, Florida, Pennsylvania and Ohio hold the top spots according to our tipping-point index. A loss in Florida or Ohio would make Trump’s situation almost impossible. Clinton has a few winning maps without Pennsylvania, Florida and Ohio, but they’d require her to win at least one state that Barack Obama lost in 2012, most likely North Carolina.
Does one candidate appear to have an overall edge in the Electoral College, relative to his or her position in the popular vote?
This is an academic question until and unless the race tightens substantially. And for the time being, an Electoral College landslide against Trump is at least as likely a prospect as the election coming down to the wire. But if the race does close, Trump is more likely to win the Electoral College while losing the popular vote than the other way around, according to our models. Still, this is a relatively narrow advantage and it has been diminishing as Clinton’s polling strengthens in states such as Pennsylvania.
How do the “fundamentals” look?
In the absence of polls, we might look toward factors such as economic conditions to predict the election. They imply that the race ought to be competitive. A stronger economy generally helps the incumbent party’s candidate — in this case, Clinton. But the economic data has been mixed, with relatively strong job growth, a healthy stock market and low inflation on the one hand, but tepid income and GDP growth on the other hand. Still, you should be wary of economic determinism. “Fundamentals”-based models that don’t look at polls have a fairly bad track record, even in years that aren’t as crazy as this one.
How do FiveThirtyEight’s forecasts compare against prediction markets?
Betting markets put Trump’s chances at around 20 percent, which is similar to our polls-plus forecast but more optimistic for Trump than polls-only. I’ll take a pass for now on the question of whether these markets are too optimistic or too pessimistic about Trump’s chances.
What would keep me up late at night if I were Clinton?
At some point, complacency could become an issue, although it’s probably too early to worry about that. In the nearer term, I’d be worried that the race has been so volatile. Sure, things look good now. But conditions in May, and then again in July, produced a close race. Is there anything inherently preventing those conditions from arising again? I suppose I’d wonder aboutwhat Wikileaks has up its sleeve and what sort of geopolitical events could work in Trump’s favor.
What would keep me up late at night if I were Trump?
I might not be sleeping at all. The tactics that helped me to win the primary don’t seem to be working in the general election. My position in the polls is deteriorating from middling to dire. Most acutely, I’d worry about getting cut off by the Republican National Committee or thrown under the bus by down-ballot Republican candidates who are rightly concerned about their own survival. If those things happen, I might not have the resources to win the election even if the debates and the news cycle turn in my favor in September and October.
Nate Silver is the founder and editor in chief of FiveThirtyEight.