Final 2024 Predictions
A low confidence Election Day forecast.
Since starting this blog in 2001, I’ve posted my Electoral College for every presidential election but one.
In 2004, I predicted Bush 286, Kerry 252, going 51-for-51. I should have stopped there.
In 2008, I predicted Obama 325, McCain 213. That actually undershot Obama’s performance by 40 Electors because Virginia (13), North Carolina (15), Indiana (11)—all historically red states—went for him. He also picked up one Nebraska district.
In 2012, I predicted Obama 290, Romney 248. That was an even bigger miss, as Obama won by an additional 42 Electors. I mistakenly thought Florida (29) and Virginia (13) would go back into the Republican column. Steven Taylor made the exact same prediction. The late Doug Mataconis correctly gave Virginia (or mutual home state at the time) to Obama. None of the other predictions we rounded up did any better.
In 2016, for whatever reason, while I reluctantly endorsed Hillary Clinton at the 11th hour after having been anti-Trump since his announcement, I never got around to making a prediction. It was just as well, as I would certainly have had Clinton winning which—spoiler alert!—she did not. Both Doug and Steven went 323-215 for Clinton. I suspect I’d have been wrong in equal measure.
In 2020, I was wildly off, predicting a 389-149 Biden wave, overstating Biden’s total by 83! I foolishly believed Texas (38) and Florida (29) would turn blue and that North Carolina (15) would revert to Obama-era form.
I’ve never had less of a sense of how the election is going than this cycle. With next to zero confidence, though, I’m going this way:
That’s closer than it should be by any reasonable measure but, essentially, I’m giving Trump North Carolina, Georgia, and Arizona while giving Harris Nevada, Wisconsin, Michigan, and Pennsylvania. (That we don’t mention the other 43 states, where almost the entire population lives, in these assessments, is less-than-ideal artifact of our system.)
Trump has been melting down on the campaign trail—most recently promising to put the wife-beating ignoramus Herschel Walker in charge of our missile defenses!—and is clearly campaigning as someone who knows he’s losing. But, as much as I’d like to predict a wave election, I just don’t see it.
I’d say this is her best-case scenario:
This has her also picking up Georgia, which Biden won in 2020, and North Carolina, which Biden lost but without an albatross of a GOP gubernatorial nominee dragging down the ticket.
Alas, I can also see this scenario:
This gives Trump all of the states where he’s currently leading—even by the narrowest of margins—in Nate Silver’s aggregator.
Silver’s analysis is paywalled but
At exactly midnight on Tuesday, we ran our simulation model for the final time in this election cycle. Out of 80,000 simulations, Kamala Harris won in 40,012 (50.015%) cases. She did not win in 39,988 simulations (49.985%). Of those, 39,718 were outright wins for Donald Trump and the remainder (270 simulations) were exact 269-269 Electoral College ties: these ties are likely to eventually result in Trump wins in the U.S. House of Representatives.
Talk about your nail-biters.
OTB Blogger Emeritus Alex Knapp (via email) goes slightly higher than my best case for his prediction:
This is, basically, the 2020 map minus Nevada, which I have going for Trump this cycle based on polling, early voting and registration numbers.
Some quick thoughts on my process here, even though most of the polling aggregators have the race a virtual tie.
1. Trump has never taken more than 47% of the popular vote. This automatically makes him an underdog. In the past 2 elections he has taken 46.1% and 46.8% of the popular vote, despite the fact that quite a few late-breaking polls in 2020 also had quite a packet that had him in the 47-49% range. Now, it’s true that he DID overperform the polls in battleground states (we’ll get to that) but almost always within the MOE. Despite this, reporting by Nate Cohn at the NYT and other places have quoted pollsters as really weighing the polls out of concern that they undercount Trump again. They don’t want to be wrong.
2. The Republican Party is more divided than it was in 2020. Despite the many problems of the Trump Administration’s first term, a lot of Republicans still, for the most part, stood by him. In this cycle, Trump faced a heavily contested primary that he may have had a challenging time winning if not for the rally round the flag effect of his felony indictments or if Ron DeSantis had a decent campaign staff. Even after Nikki Haley dropped out in the first week of March, something like a quarter to a third of Republican voters still cast their vote for her in the primary. An October survey of Haley voters had about 36% of them saying they would cast a vote for Harris and about 13% were undecided. This isn’t great for Trump because….
3. …. Independents have been breaking late for Harris. Many surveys are showing Harris winning a majority of independent votes and a significant majority of women independent voters. A good rule of thumb is whoever wins independents wins the election. Now combine that with Republican defectors (even those who ultimately decide to leave the top of the ballot blank) and that’s a big advantage for Harris.
4. Polling in battleground states significantly undercounted Democratic votes in 2022. If you believe the polls, Republicans should control the Senate right now. Polls had Oz winning in Pennsylvania, Walker winning in Georgia, Laxall winning in Nevada and an extremely tight race between Masters and Kelly. In the end, the Republicans lost all of those races and in Arizona it was by a comfortable margin. I have looked for reporting on any pollsters who decided to weight their models differently to account for this and found none – but found plenty of reporting they were weighing their models in an effort to not undercount Trump’s vote this time. I think there’s a good chance we’ll see a similar undercounting of votes for Harris this cycle as a result.
5. Harris is winning reliable voters–and has a much better ground game. Here’s a fun fact: the last time the 65+ age cohort voted for a Democrat was the year 2000. And even that was a fluke of the Clinton era – he won the senior vote in both his elections but in the 20 years prior Republicans dominated this age group. But Biden before he dropped out and Harris both consistently polled as winning the 65+ vote. This is THE most relaible bloc of voters. Harris is also winning women, another reliable bloc of voters, and college-educated voters – again, very reliable. The only bloc of reliable voters Trump is winning is evangelical Christians. The rest of the groups voting for him – young men in particular – are the kind of folks you need a dedicated GOTV effort for. But Trump’s is a mess mostly outsourced to Elon Musk’s America PAC, which started late and is staffed by a bunch of ex-Desantis guys…. remember how well he did?
6. Objectively speaking, the economy rules right now. And Harris is a quasi-incumbent. In most election cycles, that’s enough. And since the Trump campaign has basically abandoned this as an issue, a huge mistake given unhappiness with inflation.
7. Abortion is an underrated issue for both pollsters and pundits. Whenever abortion is on the ballot, support for abortion rights massively overperforms the polls. There is perhaps no better example of this than my own state of Kansas, In 2022, there was an initiative to amend the state Constitution in such a way that it would allow the legislature to place restrictions on abortion. Polls ahead of the vote had it winning by 5 points. It lost by 18. Abortion is on the ballot in Arizona and Florida. Strict abortion restrictions are sparking backlashes in Texas and Iowa. This may well impact the presidential race.
8. Polls show a weird disconnect between Senate and Presidential races. In several battleground states, Democratic Senate candidates are out-performing Harris. For example in Michigan we see polls showing Slotkin +5 and Harris +1…. but it would be very historically unusual to see voters split like this. Who are the Trump/Slotkin voters? Polls in Pennsylvania similarly show a sizable lead for Casey. Split tickets are much rarer on the federal level than they used to be and this may also serve as a good sign for Harris.
He also has a Harris 270, Trump 268 scenario in which Arizona, Georgia, and North Carolina go red and a Trump 283, Harris 255 scenario in which Trump adds Michigan.
From your mouth to God’s ears, James.
I pray we continue to have a democracy at day’s end.
I’m going to go out on a limb and predict that she’ll sweep all seven battleground states for a total of 319 EVs.
The momentum in the final days has been entirely in Harris’s favor. That was very much not the case with Hillary in 2016. If there’s a “Comey Letter” equivalent this time, it’s the Puerto Rico comment at Trump’s Nuremberg rally in New York.
One of my biggest sources of concern over the past week and a half was Jon Ralston’s somewhat dire warnings about possible Democratic underperformance in the early vote in Nevada. But now he thinks Harris will carry the state. His record of predicting the state based on early vote is impeccable so far.
I also think Ann Selzer’s poll of Iowa is a sign of what many of us have suspected, that the regular polls are underestimating Harris, in part due to the continuing impact of Dobbs. Unlike much of the commentary I’ve been reading, I choose to take seriously the possibility that the poll is literally correct and that, despite not campaigning there, Harris could in fact carry Iowa. If that’s the case, we could be looking at an electoral blowout, Harris picking up states outside the Big 7 that few saw coming, such as Florida or Texas or Ohio. But even if Selzer is underestimating Trump despite her long record of dead-on accuracy and picking up things other pollsters missed, it’s unlikely Trump is going to do as well in Iowa as he did in 2020 (that would require an 11-point miss, an unusually large error for any pollster, and which would be totally unprecedented for Selzer). I think that has ramifications for other states, not just in the Midwest, but also in states like Arizona where abortion is on the ballot, or North Carolina where it might as well be, due to Mark Robinson.
Everything seems to be falling into place when it comes to Harris’s campaign. She has better ground game, is much more popular, and has a lot more enthusiasm behind her than Trump. The landscape couldn’t be more different than in 2016. Trump isn’t new or exciting in the world of politics, he’s a fading Vegas act. There’s no complacency on the part of Dems, no assumption that Trump can’t possibly win and therefore that people don’t have to bother getting off their butts and voting. There’s also no sign of substantial third-party voting. Voters are highly engaged, and the signs are that turnout is through the roof. I believe she has this, and that we’ll finally at long last see the American public come to a strong and decisive verdict to end the reign of Agent Orange.
@Kylopod: All of that would have made great sense prior to 2016; it just doesn’t seem like normal rules apply to Trump. But it would be great to have a return to normalcy in that way.
I enjoyed hearing Trump threaten everyone with actual death unless they voted for him.
“If you don’t vote for me, you’re all going to die.”
And he was so low-energy when he said it.. you’d almost think he knew he was lying and projecting.
I don’t have a guess, I have scotch. Scotchy, scotch, scotch. Here it goes down, down into my belly.