Overview
Since our last update in mid-August, Trump has cut into Hillary Clinton’s lead in national and state polls, but continues to trail. At the present time, Hillary Clinton still maintains small to moderate leads over Donald Trump nationwide and in nearly every battleground state.
On the whole, the map is little changed from last month. One state is moving from Clinton’s column to Trump’s, the first such blue-to-red shift I’ve made since January, and the first time any state changes hands since June. Five other states are moving in Trump’s direction this time, while one key state is growing a bit more secure for Hillary Clinton.
The relatively minor changes to my projections may surprise some people, considering the significant shift in polling, but that’s how these projections are supposed to work. Last time several commenters said my predictions were too cautious considering Clinton’s lead at the time, but I was expecting that her numbers could fall somewhat, and felt that there was a fair chance that Trump could gain support. Similarly, at the moment I feel that polling is a little bit tighter than the “true” state of the race: that as people get to see these candidates in the debates and over the home stretch of campaigning, both of them should gain some support, but probably more of the undecided/third-party vote will gravitate towards Clinton than Trump.
Additionally, Clinton’s stronger ground game and better organization give her a built-in advantage: the people who actually show up to vote will probably be slightly more favorable towards her than election-eve polling indicates. Not because the polls are “skewed,” but because some Trump fans who would vote for him if the Republican GOTV effort was stronger might not vote at all, while some Clinton supporters (or undecided Democratic leaners) who otherwise might stay home will turn out for Hillary thanks to her team’s efforts. I should say, though, that the size of this advantage is not known and may end up being negligible, and, it’s possible that polling averages are skewed in favor of Clinton (as Trump’s team has been known to allege), although it’s equally possible that they’re biased in Trump’s direction.
Definitions
Safe: the favored party's candidate is extremely likely to win this state, even if the opponent wins in landslide nationally (99-100% chance the favored party will win this state in 2016)
Solid: the favored party's candidate is extremely likely to win this state, except in a wave election the other way (90-98% chance)
Likely: the favored party's candidate is likely to win this state (75-89% chance)
Lean: the favored party's candidate has the advantage, but this state could go either way (60-74% chance)
Edge: this is a tossup state, but I believe one party/candidate has a slight advantage (50-59% chance)
Current Projection
Hillary Clinton: 341 Electoral Votes
Donald Trump: 197 Electoral Votes
(270 Electoral Votes needed to win)
Popular Vote:
Hillary Clinton 49% (+/-3%)
Donald Trump 44% (+/-4%)
State-by-State Forecast Changes and Updates
Before we get to the forecasts, let’s do a little thought experiment. We’ve all heard from multiple sources that it’s best not to pay too much attention to polls with shocking results (as outliers are less likely to be accurate), and instead to look at the polling averages. It’s often hard to resist the temptation, though. So...for our experiment, what if we were to assume that the most competitive poll or polls, rather than the polling averages, actually showed the true state of the race?
If that were true (it’s not, but let’s just pretend), then the map would look much different. We’ll begin by defining a “safe state” as a state in which one candidate has led by double-digit margins in every single poll. Sounds reasonable enough, right? Well, I looked at the data*...if that’s the standard we use, there would only be eight safe states out of the fifty—three red, and five blue. Trump could rely on Idaho, North Dakota, and Wyoming, for a total of 10 electoral votes. Clinton would hold California, Hawaii, Maryland, Massachusetts, Vermont, and the District of Columbia, for 86 EVs.
Things get even better for Clinton when we move to the next two categories: let’s define Lean states as those in which one candidate has led in every poll, but sometimes by less than 5%; and Likely states as those in which one candidate has led by more than 5% in every poll, but sometimes less than 10%. Clinton would add five Likely states (Connecticut, Illinois, Minnesota, New York, and Washington) for 78 more EVs, and three Lean states (Delaware, New Jersey, and Rhode Island) for 21 EVs, giving her a total of 185 electoral votes: about two-thirds of the way to the 270 EVs she needs.
Trump, meanwhile, adds five Likely states (Alabama, Arkansas, Louisiana, Oklahoma, and Tennessee) for 41 EVs, and two Lean states (Mississippi and West Virginia) for 11 EVs, giving him a total of 62—less than a quarter of the way towards 270.
The remaining 27 states, collectively worth 291 electoral votes, are toss-ups: they are states where at least one poll has shown a tied race, and states where both candidates have led in at least one poll. Clinton would need 85 EVs out of these toss-up states, while Trump would need a whopping 208. You can view this map here.
What does this mean? Probably not too much—it’s giving outliers way too much influence. But if that’s what you want to do, be evenhanded about it: if Utah and Texas are toss-ups, then so are Oregon and Michigan. If Mississippi and West Virginia are competitive, then so are Rhode Island and New Jersey.
*Polling data taken from fivethirtyeight.com
So...back to a more realistic approach...here’s how much predictions have shifted in the last few weeks:
Iowa flips to Edge Republican (was Edge Democrat). In the most significant move this week, I’m shifting Iowa into Trump’s column. Trump has led in most of the recent polls here—often by very slim margins—and the aggregators don’t agree on who is ahead (HuffPo estimates that Clinton leads by 1.1%, while Real Clear Politics has Trump up by 0.8%). FiveThirtyEight has swung Iowa back and forth, but at the present time Trump is favored in all three of their projections. There is also some reason to believe that the polls are overstating Clinton’s support in Iowa: in the Iowa caucuses she underperformed expectations in 2008 and again this year, and in the 2014 Senate race, Republican candidate Joni Ernst won a much wider victory than polling had indicated, in part because non-college white voters have moved significantly towards the GOP. While Iowa looks to be one of the closest contests in the nation—and it certainly could go either way—at the moment I’m not very confident about Clinton’s chances.
North Carolina moves to Lean Democrat (was Edge Democrat). While polls have narrowed here—and indeed Trump has led some recent ones—Clinton’s numbers haven’t fallen in North Carolina as they have in other states. Looking at the averages, Hillary has led rather consistently, and with time starting to run short, it’s looking less likely to me that Trump can pull this one out. At this point, I’m slightly more concerned about Ohio (which nevertheless remains a Lean Democratic state for the time being).
Maine shifts to Likely Democrat (from Solid Democrat), and Maine’s 2nd District moves to Edge Democrat (from Likely Democrat). Clinton still looks like a good bet statewide, but Trump has a real chance to win the one electoral vote allotted to the Second District. He’s led in the three public polls in the Second District. I’m not ready to flip the district to his column yet, though...I do think that third-party voters will return to the Democratic tent if it appears that Trump is in danger of winning...but right now this district looks rather tough for Hillary.
Indiana shifts to Solid Republican (from Likely Republican). The perfect storm that led to a Democratic victory in 2008 is unlikely to be replicated this year, especially with Indiana’s own Mike Pence on the ticket.
Arkansas, Mississippi, and North Dakota move to Safe Republican (from Solid Republican). We have a lot more polling data in these states now, thanks to 50-state polls from Google, Ipsos, and SurveyMonkey. Trump appears to be headed for a very strong victory in North Dakota. His lead in Arkansas and Mississippi has been narrow in some polls, but even an optimistic view would have Clinton falling a bit short.
Looking Ahead
In the next two weeks, we should find out some hard data on how Hillary’s pneumonia and her “basket of deplorables” have had on the race...I expect a significant but brief dip in her numbers. Trump has some rough waters of his own to weather, as questions mount about his business ties, his charitable donations (or lack thereof), the Trump Foundation, and his own health. Probably by the time of the first debate, things will stand roughly where they are now.
The debates, of course, could have a significant impact on the race—I don’t expect either candidate to lose much support even if they perform badly, but whoever is perceived to have won could receive a big boost from voters who are currently undecided or backing a third-party candidate.
There are a few potential endorsements that could affect the race, too. While prominent Democrats are almost uniformly supporting Hillary Clinton (she even won the support of Joe Lieberman, who previously had said he was considering Trump), many Republicans have yet to endorse. While Romney, Kasich, Cruz, and the Bushes may not choose to involve themselves any more than they already have, if any of them do come out in support of Clinton or Trump—or even if they were to endorse Gary Johnson or Evan McMullin, or something else (like a boycott), they could have a substantial effect on the outcome. Other Republican or nonpartisan figures (Colin Powell, Condoleeza Rice, Arnold Schwartzenegger, etc.). I’ll also be looking for any endorsement changes (if a a candidate loses the support of an important endorser, that’s a strong sign that he or she is in serious trouble), as well as endorsements by newspapers and other organizations.
This election isn’t a toss-up, but it’s not a done deal, either. If Trump were to capture the states/districts that are Lean and Edge Democratic, he’s at 266 Electoral Votes, and would need to add just one of the Likely Democratic states (other than Maine at-large).