By Nate Silver – – – – –
We’re going to spend a lot of time over the next 87 days contemplating the possibility of a Donald Trump presidency. Trump is a significant underdog — he has a 13 percent chance of winning the election according to our polls-only model and a 23 percent chance according to polls-plus. But those probabilities aren’t that small. For comparison, you have a 17 percent chance of losing a “game” of Russian roulette.
But there’s another possibility staring us right in the face: A potential Hillary Clinton landslide. Our polls-only model projects Clinton to win the election by 7.7 percentage points, about the same margin by which Barack Obama beat John McCain in 2008. And it assigns a 35 percent chance to Clinton winning by double digits.
Our other model, polls-plus, is much more conservative about Clinton’s prospects. If this were an ordinary election, the smart money would be on the race tightening down the stretch run, and coming more into line with economic “fundamentals” that suggest the election ought to be close. Since this is how the polls-plus model “thinks,” it projects Clinton to win by around 4 points, about the margin by which Obama beat Mitt Romney in 2012 — a solid victory but a long way from a landslide.
But the theory behind “fundamentals” models is that economic conditions prevail because most other factors are fought to a draw. In a normal presidential election, both candidates raise essentially unlimited money and staff their campaigns with hundreds of experienced professionals. In a normal presidential election, both candidates are good representatives of their party’s traditional values and therefore unite almost all their party’s voters behind them. In a normal presidential election, both candidates have years of experience running for office and deftly pivot away from controversies to exploit their opponents’ weaknesses. In a normal presidential election, both candidates target a broad enough range of demographic groups to have a viable chance of reaching 51 percent of the vote. This may not be a normal presidential election because while most of those things are true for Clinton, it’s not clear that any of them apply to Trump.
A related theory is that contemporary presidential elections are bound to be relatively close because both parties have high floors on their support. Indeed, we’ve gone seven straight elections without a double-digit popular vote victory (the last one was Ronald Reagan’s in 1984), the longest such streak since 1876-1900.
As with other theories of this kind, however, there’s the risk of mistaking what’s happened in the recent past for some sort of iron law of politics. Historically, the U.S. has ebbed and flowed between periods of close presidential elections — such in the late 19th century or early 21st century — and eras in which there were plenty of lopsided ones (every election in the 1920s and 1930s was a blowout).
These patterns seem to have some relationship with partisanship, with highly partisan epochs tending to produce close elections by guaranteeing each party its fair share of support. Trump’s nomination, however, reflectsprofound disarray within the Republican Party. Furthermore, about 30 percent of Republican or Republican-leaning voters have an unfavorable view of Trump. How many of them will vote for Clinton is hard to say, but parties facing this much internal strife, such as Republicans in 1964 or Democrats in 1972 or 1980, have often suffered landslide losses.
Perhaps the strongest evidence for a potential landslide against Trump is in the state-by-state polling, which has shown him underperforming in any number of traditionally Republican states. It’s not just Georgia and Arizona, where polls have shown a fairly close race all year. At various points, polls have shown Clinton drawing within a few percentage points of Trump — and occasionally even leading him — in states such as Utah, South Carolina, Texas, Alaska, Kansas and even Mississippi.
Just how bad could it get? Let’s start by giving Clinton the 332 electoral votes that Obama won in 2012. That’s obviously not a safe assumption: The race could shift back toward Trump, and even if it doesn’t, Clinton could lose states such as Iowa or Nevada, where her polling has been middling even after her convention bounce. But as I said, we’re going to focus on Clinton’s upside case today.
So I’m going to list the states Romney won in order of how easy it is for Clinton to flip them, according to our polls-only model.1 The number in parentheses by each state represents the point at which the model estimates it would flip to Clinton, based on her lead in the national popular vote. For instance, South Carolina (+9.5) means that Clinton would be favored in South Carolina if she leads by at least 9.5 percentage points nationally, but not by less than that. These projections are based on where the model has each state projected currently, along with each state’s elasticity score, a measure of how responsive it is to changes in the national environment. Here goes:
North Carolina (+3.2): It wouldn’t be any surprise if Clinton carried North Carolina, which Obama narrowly won in 2008. But Obama lost North Carolina in 2012 despite winning by about 4 percentage points nationally. This year, it looks like Clinton would win North Carolina with a 3 percentage point national victory. In other words, North Carolina has drifted slightly bluer relative to the rest of the country and is closer to being a true tipping-point state this year.
Arizona (+7.1): Arizona and Georgia have been flickering between light blue and light red in our polls-only projection recently. That’s because the model figures each state would be a tossup with Clinton ahead by about 7 points nationally, and that’s where the forecast has been for the past few days. Arizona is the fourth-most-Hispanic state after New Mexico, Texas and California, although historically its Hispanic population has voted at relatively low rates. A strong Hispanic turnout, perhaps coupled with gains for Clinton among Mormon voters (about 6 percent of Arizona’s electorate), might swing the state to her.
Nebraska’s 2nd Congressional District (+7.1): Nebraska and Maine award one electoral vote to the winner of each congressional district. That came in handy for Obama in 2008, when he won Nebraska’s 2nd Congressional district, which consists of Omaha and most of its suburbs. District boundaries were redrawn after the 2010 Census to make them slightly tougher for Democrats, but Omaha’s highly-educated demographics — we estimate that 47 percent of voters in the district have a college degree, comparable to Virginia or Connecticut — could wind up being favorable to Clinton. There’s been no polling in the district yet, so its position on this list is based on the model’s guesses based on its demographics and voting history.
Georgia (+7.2): In some ways, Georgia might be more promising than Arizona for Democrats’ long-term future. It has more electoral votes — 16 to Arizona’s 11 — and could serve as part of a bloc of states (along with Virginia and North Carolina) that could eventually offset losses for Democrats in the Rust Belt. It’s easy enough to see how Georgia’s demographics are favorable for Clinton: It has a substantial black population, but also an increasingly well-educated white population, with lots of migration from the Midwest and the Northeast.
Let’s pause here to see what the map would look like if Clinton wins by 8 percentage points nationally — close to where her lead in the polls has been over the past week or so. This map you see below is worth 375 electoral votes, close to the 365 electoral votes Obama won in 2008 when he beat McCain by 7.3 percentage points. In fact, the map is identical to 2008 but for three changes: Georgia and Arizona turn blue, while Indiana (which surprisingly went for Obama in 2008) remains red:
But let’s say Clinton continues to build her lead, instead of Trump rebounding. Which dominoes might fall next?
South Carolina (+9.5): Public Policy Polling caused a big stir on Thursday when it published a poll showing Clinton down just 2 percentage points in South Carolina — but the result shouldn’t have been all that shocking. South Carolina was only a couple of points redder than Georgia in 2012 and 2008, so if Georgia has moved to being a tie, you’d expect South Carolina to follow just a half-step behind it. True, South Carolina doesn’t have a metropolis like Atlanta, but a relatively high percentage of white voters there have college degrees.
Missouri (+10.3): It’s surprising to see Missouri, once considered a bellwether state, so far down this list. Bill Clinton won it twice, and Obama came within 4,000 votes of winning it in 2008. But now we estimate that Hillary Clinton would need to win by about 10 points nationally to claim the state. Note, however, that the recent polling in Missouri has been mixed, with polls showing everything from a 10-point lead for Trump to a slight edge for Clinton.
There’s something of a gap after South Carolina and Missouri before the next set of states. Thus, Trump might be able to hold Clinton below 400 electoral votes even if she won by 12 points nationally:
But after that, the floodgates would really open, with lots of traditionally red states in all parts of the country potentially turning toward Clinton:
Mississippi (+12.3): I’m skeptical about this one, since Mississippi presents something of a modelling challenge. You can see why it’s an attractive target for Democrats, in theory: It has the highest share of black voters in the country (after the District of Columbia). But in 2008, only 11 percent of Mississippi’s white population voted for Obama. Clinton trailed Trump by just 3 percentage points in the only poll of Mississippi, taken in March. In that poll, Clinton got 20 percent of the white vote. If she can replicate that on Election Day, the outcome could be close.
Indiana (+13.2): Obama’s win in Indiana in 2008 — one of just two timesDemocrats have won the state since 1940 — might be hard to duplicate. He benefited that year from investing in the ground game in a state that is usually ignored, and from Indiana’s connections with Chicago. Plus, Indiana Gov. Mike Pence is Trump’s running mate. Still, if Clinton stretches her national lead into the teens, Indiana could be competitive.
Texas (+13.8): Democrats have long talked about turning Texas blue — or at least purple — but the truth is they haven’t come anywhere close. Obama lost Texas by 12 points in 2008 despite his near-landslide margin nationally, for instance. But Clinton has a number of factors that could work in her favor. We estimate that about somewhere between 37 and 40 percent of Texas’s electorate will be Hispanic, black, Asian-American or Native American, depending on turnout. A high proportion of its white population has college degrees. And Trump has run afoul of locally popular politicians, such as Ted Cruz and George W. Bush. Previous polls of Texas had shown Trump with only a mid-single digit lead there, although a more recent survey had him up by 11.
Montana (+14.1): Obama also nearly won Montana in 2008, losing by just 2 percentage points. But Montana is historically an anti-establishment state, and Trump led Clinton in the only poll we can find — which, granted, was way back in November 2015 — by 21 percentage points. A winning scenario for Clinton would probably involve Libertarian Gary Johnson getting a substantial portion of the vote: Montana was Johnson’s second-best state, after New Mexico, in 2012.
Utah (+14.2): People are fascinated by Clinton’s prospects of winning in Utah, which went for Romney by 48 points in 2012. But it’s hard to say just how realistic those are. The polls-only model has Clinton just a couple of percentage points behind in the polling average in Utah, but its demographic model projects her to lose it by 16 points — a lot better than 2012, but not particularly close. As with Mississippi, therefore, the odds you assign to Clinton in Utah are highly sensitive to your choice of assumptions. She’s taking her chances seriously enough to make some efforts to campaign there, but is it a wild goose chase — like when Dick Cheney visited Hawaii in 2004 — or part of long-term plan to swing Mormons into the Democratic Party?
South Dakota (+14.9): Less excitingly, Clinton could win South Dakota in the event of a national rout, as the state seems to have become the slightly bluer of the two Dakotas after North Dakota’s oil boom. Perhaps South Dakota has a soft spot for Clinton, having voted for her in the Democratic primary in both 2008 and 2016, when Obama and Bernie Sanders won almost all the surrounding states.
Kansas (+15.6): Polls have had Kansas surprisingly close — with one surveyin June even having Clinton ahead. One can squint and make an argument for it: Kansas is relatively well-educated, and Republican Gov. Sam Brownback is extremely unpopular. But note that Kansas polls badly overstated Republicans’ problems in 2014, when both Brownback and Sen. Pat Roberts won re-election.
Alaska (+15.7): I doubt that Alaskans have much affection for Clinton, but the state is idiosyncratic enough that I don’t really know what they think of Trump, who lost to Cruz in the state’s Republican caucuses. As in Montana, a Clinton win would probably depend on Johnson sucking up a lot of Trump’s vote. Clinton trailed by just 5 percentage points in the only poll of Alaska in January, which didn’t include Johnson as an option.
Nebraska’s 1st Congressional District (+15.8): As goes Omaha, so goes Lincoln? Here’s what the map might look like if Clinton won by 16 percentage points nationally, along with all the states we’ve mentioned so far:
That would work out to 471 electoral votes, to 67 for Trump, which would be fairly typical for a win of that magnitude. Dwight D. Eisenhower won 457 electoral votes when beating Adlai Stevenson by 15 points in 1956, for example. And Franklin D. Roosevelt won 472 electoral votes in 1932, in an 18-point win against Herbert Hoover. Clinton would be a ways short of Ronald Reagan’s 525 electoral votes in 1984, however.
All right, let’s stop there. I’m trying to encourage you to keep an open mind. The way the polls-only model thinks about things, Clinton is ahead by 7 or 8 percentage points now, and the error in the forecast is symmetrical, meaning that she’s as likely to win by 14 or 16 points as she is to lose the popular vote to Trump. There have even been a couple of national polls that showed Clinton with a lead in the mid-teens. But my powers of imagination are limited. Other than losing North Dakota to go along with South Dakota, or perhaps the statewide electoral votes in Nebraska to go along with the congressional district ones, it’s hard for me to envision Trump doing any worse than this — unless he really does shoot someone on 5th Avenue.
- Calculations are based on the polls-only model as of noon on Friday.
Nate Silver is the founder and editor in chief of FiveThirtyEight.