Final Predictions: 2024 Presidential Election

If you’ve followed the Partisan Gravity Presidential Election Forecast, then you know well that we’ve had this race categorized as a toss-up as the election has drawn closer. In fact, the race has been so close during the model’s duration that it has almost only depicted the race as a toss-up, only diverging from this race status momentarily for a few days where the race tilted ever so slightly toward the Democratic nominee, Kamala Harris.

Our final model run is now available, but if you don’t feel like reading the extra details about the model’s methodology, we’ll spare you by including the model’s topline just below:

With 10,000 simulations:

HARRIS WINS 5282 (52.82%)

TRUMP WINS 4578 (45.78%)

NO WINNER 140 (1.40%)

Ultimately, this race will conclude with a definitive winner, and although the model highlights a close contest, we are committed to offering a set of informed predictions. It is highly improbable, however, that each of these projections will prove accurate; some will almost certainly miss the mark. If that’s not the case—well, then welcome to the most precise forecast available. That said, let’s be transparent: these predictions are tinged with a degree of bias and influenced by anecdotal evidence and a touch of wishful thinking. So, consider these insights with a healthy pinch (or perhaps a full tablespoon) of salt.

Prediction

There’s certainly a plausible path for former President Donald Trump to win, but I think Harris has several more viable routes to victory. Some indicators should concern the former president, and many of these are reflected in our daily Presidential Election Model updates.

Trump has a five-alarm fire brewing in the Midwest. In fact, some of the highest quality polling is demonstrating vast slippage for the topline Republican in Kansas, Iowa, and Ohio. On top of that, Harris’ polling in Wisconsin has been surprisingly strong and consistent. This is something that I had imagined was something of an artifact – perhaps some consistent polling error undercounting Republicans – but the best pollster, Ann Selzer, calmed these nerves with her bombshell poll showing Harris a full 3 points ahead in Iowa. Let me repeat: Harris ahead in Iowa, of all places.

The Republican response was to release internal polls showing Trump ahead by five to seven points, but even these numbers indicate erosion in Iowa, a state he won by eight points in 2020. Combined with troubling signs in Kansas, Ohio, and Wisconsin, there’s reason to believe Trump faces real challenges here.

The Map

If I had to guess, the election will likely follow the contours of the map in the image above, with Harris expected to win a robust 308 Electoral College votes.

Let’s start with the basics. There are seven traditional swing states this cycle. But, if you’re curious, I include some others in my model summary that are especially interesting. The model gives the following chances of winning each state by candidate in the table below.

The race is a toss-up.

Why the changes between model and prediction?

Now, you’ll perhaps note that I differed a little bit from the model. That’s because I hold a set of priors. We incorporate some of these in the “lightly” Bayesian model.

Below are some core assumptions that help translate from model outcome to election prediction. This is an honest assessment of the data that we have, which, as we know, is far from perfect. At the end of the day, there are several more indicators in favor of Harris and her campaign. Whether it’s her favorability advantage, strong GDP and jobs growth under Biden’s tenure, the aforementioned recent polls, her “Blue Wall” polling advantage that remains solid, Pennsylvania’s Puerto Rican community, celebrity endorsements, and polling momentum with the highest quality pollsters, there are a lot of reasons for confidence.

But, there are also plenty of causes for despair for Harris and Democrats. Trump’s campaign is likely feeling jubilant with a flurry of low-quality polling suggesting a Trump lead in several of the swing states, and this is bolstered by even high quality pollsters showing a narrowing national polling lead for Harris.

Moreover, democracies around the world currently hate their incumbent parties that served during the most brutal COVID-19 years. In the United States, we played Biden’s polling issues off as being related to his age (hint: they are to some degree), but that was far from the full picture. Canada looks poised to toss Justin Trudeau and the Liberals out the door in their upcoming election. Britain threw out the Tories. One was a landslide, the other looks like it will be. France saw its ruling coalition nearly shatter. Germany’s Social Democrats look like they’re on life support. Poland’s Donald Tusk and his coalition of parties are now polling behind the conservative coalition. Speaking colloquially – you see my point?

How is that for some American Exceptionalism? America’s two party system is powerful, and it’s downright remarkable that the Democratic party, when faced with near-certain defeat, chose to bench the sitting President and leader of the party. Desperate times, desperate measures.

The Assumptions

Wisconsin

Wisconsin will be closer than the polls show – it was one of the very closest states and tends to be decided by a percentage point or less. However, the model might be more on the money than I am given the trends in the similar states like Iowa and Kansas. In 2022, the Democratic Senate Candidate Mandela Barnes lost by 1% to Republican Ron Johnson. In 2020, Joe Biden won by 0.7%.

Expected margin: Harris + 0.5-1%

Georgia

Momentum doesn’t usually just appear, and given the continued demographic momentum in Georgia, I think that Harris is more likely than not to eke out a win. After electing Biden and two Democratic Senators, we have a strong recent data point to work with alongside the demographic momentum. But, I think it’s going to be close.

Expected margin: Harris + 0.5-1%

North Carolina

North Carolina may just tilt ever so slightly toward Harris, despite the polling that has her merely a breath behind Trump. This state has been perpetually frustrating to Democrats. It’s always close – very close, but usually not attainable. But, I’m counting on something lacking in empiricism: reverse coattails from gubernatorial candidate Mark Robinson’s stark unfavourability. The Republican is 15-20 points behind the Democrat in the race for Governor, and although we don’t have much evidence for this extreme case, this may – just may – be enough to lead to a voting enthusiasm gap that would possibly favor Harris. And, remember, these North Carolina races only need a very marginal shift to flip.

Expected margin: Harris + 1%

Arizona

Arizona still beats me a little bit. The polling may suggest a 3-5 point win for Trump. This would be somewhat surprising given that Biden won the state, not to mention the growth of suburban Phoenix which would theoretically be great for Democrats. I’m willing to think that the polls are underestimating Democrats a little bit here, but I’m not sure I can go that far – and nor should I, as we never can safely estimate the direction of a future polling error. So, I think Harris loses Arizona by under a point.

Expected margin: Trump + 1.5%

Nevada

Nevada gives Democrats the benefit of the Reid Machine – an excellent party infrastructure that the late Senator Harry Reid fostered that does a simply excellent job curing signatures on ballots and churning out the vote within Clark County. Look, polls here are pretty decent for Harris. It’s close, and some folks were worrying over early vote returns even though these are historically poor indicators since it would require guesses about the electorate that we simply can’t support. I think Harris wins the state, but I think it will be close. How close? The incredible journalist Jon Ralston has been calling Nevada elections for decades, and he thinks within 0.3%.

Expected margin: Harris + 0-0.5%

Pennsylvania

Pennsylvania has spent most of its time in the model as a “Lean Harris” state, with the exception of a few days during the last week as pollsters published essentially statistically impossible improbable results showing tied races with no statistical dispersion. It’s clear why, in this prediction, it lands in the Democratic column. There are some concerning trends for Democrats in the state. Chiefly among those is party registration becoming a bit more Republican. Much of that may well be offset by younger voters beginning their careers as voters as unaffiliated and still voting for candidates who support liberal causes, or, in this case, perhaps abortion.

I think that the best case that you can make for Trump is that Pennsylvania is tied. The worst polls for Harris among the high-quality pollsters show a tied race. But, more often than not, the high-quality and local pollsters show her leading. It’s true: some polls show Trump ahead. These typically are worse pollsters. Let’s also keep in mind that Democrats have done pretty well in the state in recent elections. Democratic Senate Candidate John Fetterman won the race by around 5 points in 2022, and, so far as we hear, his team is heavily involved in Harris’ ground game. Moreover, Joe Biden won the state in 2020.

Expected margin: Harris + 1%

Michigan

Michigan has also spend the last month and a half comfortably within the “Lean Harris” column. By comfortable, I mean sitting above a 55% chance of winning for Harris. Often, she has been around 60%. That is far from a guarantee, but Trump’s odds in the state hinge on a polling error in his direction. I think it’s more likely that he loses by 2-3 points in the state.

Expected margin: Harris + 2.5-3%

Texas

Harris is expected to lose Texas around 85% of the time. Could she win? Sure. It’s unlikely, though. The Senate Candidate, Colin Allred, has a better shot…and he still is more likely to lose than not.

Expected margin: Trump + 4%

Florida

Harris is slightly more likely to win Florida than she is to win Texas. That’s rather remarkable given the demographic changes in Texas that had Democrats thinking winning was in their “destiny” at some point in the near future. That seems unlikely for 2024, and as Trump is poised to make some small inroads with Hispanic voters, may also be somewhat unrealistic in 2028.

Florida, however, seemed as though it would move away from Democrats. Perhaps it is. In voter registration, that is readily apparent. That said, Florida is very urban and diverse and is extremely pro-abortion. With abortion on the ballot in the state this November, this is an extremely high salience election that may see higher turnout and a benefit for Harris as voters make their voices heard on policies they care about. Be that as it may, the partisan gravity working against Democrats in Florida is very strong and Trump is much more likely to win.

Expected margin: Trump + 5%

Iowa

What a fun state to write about. There is no world in which this state would have earned a mention until earlier this week when Selzer & Co. released their final Iowa poll showing Harris 3% ahead of Trump. Our model clearly still expects Trump to win the state. My priors conform with the model’s result. That said, people who bet against Selzer usually end up looking like fools.

More likely than not, Iowa is going to be much closer than Republicans (and Democrats) ever expected during this Presidential cycle. Maybe Harris could have won the state if her campaign knew to invest there. In fact, maybe she still will win the state. But, there are many more reasons than not to assume that Trump still wins the state that is very white and rural.

Expected margin: Trump + 3%

Final Thoughts

This election is a nail-biter. That it is this close is both surprising and, to many, infuriating. I have worked hard to demonstrate the uncertainty inherent to this election over the last couple of months through consistent model updates, and, much to my dismay, a high level of uncertainty remains even as the race comes to a close.

Don’t take the model output or prediction as fact. It’s a data-driven guess with biased priors working upon data from pollsters who themselves may often smudge the ink on their surveys to herd toward an average or please a partisan sponsor. Even so, the model hopefully has served as a shield against notions of a dramatically changed race that have often filled the media sphere. The discourse online and on TV has shifted constantly, but if you look at this model and others, you’ll see that very little has changed.

Maybe that means that all of the partisan punditry is even more disconnected than we thought. Maybe voters simply have not really been changing their minds. Maybe scandals don’t mean much. Maybe endorsements don’t either (though researchers would disagree).

In fact, the biggest change came from Democrats replacing their nominee. Harris’ favorability skyrocketed nearly overnight and her polling in all of the swing-states increased substantially in an instant. Her chances in this race are far, far ahead of where Bidens were and would have been.

As we enter the darkness tonight and hopefully emerge from it tomorrow, try to keep all of this uncertainty in mind. Try to remain humble. For all the bluster of the campaigns and partisans in the media, no one knows where this race is headed.

You just have my best guess.