Here at FiveThirtyEight, our favorite election-related chart is what we officially call the “winding path to 270 electoral votes” and unofficially call the snake. Designed by my colleague Aaron Bycoffe, it lines the states up from most favorable for Hillary Clinton (Hawaii, Maryland) to best for Donald Trump (Wyoming, Alabama) based on the projected margin of victory in each one. The snake is bisected by a line indicating 269 electoral votes: cross this line — meaning you get 270 electoral votes — and you win the election.
Right now, Clinton is over the line by exactly one state. As of this writing, that state — what we also call the tipping-point state — is New Hampshire. But a group of states are closely lumped together, and Pennsylvania, Colorado and Wisconsin have all taken their turn as the tipping-point state in recent weeks.
If she wins all those states and everything toward the blue end of the snake, Clinton would finish with 272 electoral votes, even assuming she loses the 2nd Congressional District of Maine. (Maine and Nebraska split their electoral votes by congressional district.) That’s two more than she needs to win the election.
But in different ways, that both understates and overstates how precarious Clinton’s position is. It understates it because Clinton has no margin to spare. Clinton’s polling has been somewhere between middling and awful in most of the other swing states lately, and they all at least lean toward Trump at the moment, narrowly in some cases (such as Florida) and more clearly in others (such as Iowa). If Clinton loses any of the states on the blue side of the snake without picking anything up on the red side, she’ll be stuck on 269 electoral votes or fewer.1
On the other hand, Clinton’s leads in the states she needs to win appear to be pretty solid. As of late Thursday afternoon, she’s ahead in our forecast by 3.1 percentage points in New Hampshire, and by slightly more than that in Colorado (3.3 points), Pennsylvania (3.4 points) and Michigan (also 3.4 points).
Put another way, you might as well saw the snake in half. There’s a big gap between the New Hampshire/Colorado/Pennsylvania group of states and the next ones, where Clinton is losing.
So does that mean Clinton could win the Electoral College despite losing the popular vote? Yes, it’s possible. If every state swung by exactly 2.5 percentage points toward Trump, for instance, that’s exactly what would happen.
But that exactitude is a big assumption, especially given that we still have 47 days — and three debates — to go until the election. Even if it were Election Day, in fact, it would be unrealistic to expect such high precision. State polling averages have been pretty good for the past few presidential elections, but “pretty good” still provides for plenty of times when they miss by 2 to 4 percentage points. If one of those misses is in Trump’s favor in Pennsylvania or New Hampshire or Colorado, especially if the race shifts a bit further to Trump overall, then Clinton will go from being in a pretty good Electoral College position to having a total mess on her hands.
Furthermore, the way in which the states are aligned right now doesn’t make all that much sense based on their demographics. In the table above, you can find regression-based estimates for the outcome in each state, which are derived based mostly on how they voted in 2008 and 2012 and to a lesser degree based on their region and demographic makeup. (This regression is combined with the polling average in each state to make our forecasts, although it receives a low weight in well-polled states like these.)
There’s an apparent discrepancy between Pennsylvania and Ohio, for example. In our polling average, they’re separated by 5.2 percentage points: Clinton leads in Pennsylvania by 3.5 points but trails in Ohio by 1.7 points. In 2012, by contrast, the states were separated by only 2.4 points, and so our regression model is confused as to why the gap is so much bigger this year. Perhaps that means the polling gap between them will close — good news for Clinton if Ohio moves toward Pennsylvania, but a problem if it’s the other way around.
Likewise, how safe can Clinton feel in Colorado given her poor polling in Nevada? Can she be entirely comfortable in New Hampshire given that Maine is surprisingly close? Does the sharp tilt toward Trump in Iowa tell us that Wisconsin or Minnesota have the potential to turn red?
Between the unusual nature of Trump’s electoral appeal, the disagreements between pollsters, the large number of undecided and third-party voters, and the wider range of swing states than in 2008 or 2012, this is not a year to expect tremendous precision. Instead, the actual map is likely to be a little messier than polling averages indicate, with the probability of modest errors in either direction. For the time being, Clinton is more likely to be hurt by those errors than to be helped by them. She has one really good Electoral College path, but it’s only one path, instead of the robust electoral map that President Obama had in 2008 and 2012. That’s why our models estimate that Trump is more likely to benefit from an Electoral College-popular vote split than Clinton is, although either outcome is possible.