Eight years ago, Donald Trump stunned election experts and pundits everywhere by overcoming Hillary Clinton to become the 45th President of the United States. Trump won by acting as a lightning rod for discontent among the working class, especially in the Rust Belt, where he picked up the states of Pennsylvania, Michigan and Wisconsin, which had been in Democratic Party hands for decades.
Trump ran for a second term in 2020, where he was defeated by Joe Biden in what will go down as the strangest election in American history, and certainly among the most contested. To this day, debate rages about the circumstances under which Biden won back those same states that Trump picked up four years prior, in an election that saw a mammoth turnout thanks to temporary voting measures to combat Covid restrictions.
Now, in 2024, Trump is running for a third time, trying to achieve what only Grover Cleveland has done before him: spend two non-consecutive terms in the White House. His opponent is a different candidate again, this time in the form of current Vice-President Kamala Harris.
Even for American politics, it has been a strange campaign period. The Republican primary campaign was largely ignored by the man who ended up winning it all, who didn't appear at any of the scheduled debates for it, and who had the narrative around him shift from "he can't possibly win it again" to "he's obviously going to win it again" in record time.
The Democratic primary campaign was comfortably won by the sitting President of the United States, who then got kicked out of the nomination by his own party in favour of his Vice-President, who has never won a presidential primary, and her nomination did not even take place through the traditional method of an open convention floor, with the party leaders deciding to install her on their own.
Since becoming the nominee, Trump has been shot once and nearly shot at least once more, and has appeared on a multitude of podcasts on the advice of his eighteen-year-old son, Barron. Harris has been in much safer territory, so much so that she went months as a nominee without holding a single press conference or interview, and only recently deigned to be interviewed in a setting that wasn't incredibly friendly to her.
Oh, yes, and Trump has also been dealing with a number of criminal and civil cases, brought against him by some especially vindictive Attorneys-General. You may have heard a little bit about those. If he wins, he will be breaking all kinds of new ground as a convicted felon.
As I wrote recently, this election feels like a full establishment vs rebels factional pitched battle, with the mass media, big tech, celebrities, universities - basically anyone in a position of societal power - largely arrayed behind Harris, and opponents of 'the system' (in myriad stripes) joining forces with Trump.
But if that is what this election is, it hasn't come out of nowhere. If anything, it is the apex of a three-part story that began in the lead up to the 2016 president election, the third act of a quite incredible tale.
Understanding the Third Act
Historians are always trying to tell stories. Stories are the way in which we understand our world, and the past is full of stories waiting to be told. My election predictions are really that of an overly enthusiastic historian attempting to figure out the ending to the story before it happens.
Donald Trump is, unquestionably, the lead character in this current story of the United States, which is part of a many-chaptered story of that country and the rest of the world with it. Your view on what kind of lead Trump is will determine what kind of story is being told. Is he a tragic comic figure? A villainous protagonist? An all-conquering hero?
Perhaps we won't really know until the third act is complete, but the first two acts give us some important context as we examine this election.
The 2016 contest was against the grain for most political watchers - do excuse me, I feel a sudden need to clear my throat here - but that's because many of them weren't quite paying attention to the right things, and were instead assuming that the winner of a presidential election must be a certain kind of person, with certain kinds of policies and a certain way of speaking.
Once they missed that Trump was appealing heavily to a large group of people (the white working class) in a region that would be able to shift the election without winning the overall popular vote (the mid-west), they were going to miss that he had a real chance of winning. This victory was even pre-empted earlier that year, when Britain voted to leave the European Union. Both elections were the result of the same groups of voters voting against the establishment grain, and came as a shock to that very same establishment.
This result was a joyous occasion for his supporters, and a terrible shock for his opponents, and was the highpoint of the first act. But if his supporters thought this would be the start of something amazing, of a Trump Train without any brakes, they were in for a surprise of their own when, over the next four years, his foreign policy successes were outshone by his being dragged down by domestic disputes, with an unhelpful Congress from both sides of the chamber not doing him any favours.
Even so, he was on track for re-election until the onset of Covid. Going along with most of the advice he was given from Dr Anthony Fauci and others within the federal health department - against his own instincts - tanked the economy, which had been his greatest vote-winner in the polls to that point. It also protected his opponent, Joe Biden, from needing to appear in public too often, and enabled the creation of certain temporary electoral measures that ended up having a profound impact on the way votes were cast.
This confluence of circumstances delivered an eventual victory being given to Biden, much to the frustration of those who suspected something was especially wrong with what had taken place.
In any case, this means that everything is set in place for the third act. The circumstances of the first act had now been flipped around, and the protagonist was facing the biggest crisis of his story. Everyone knew he would try again one more time in 2024, but should he have been allowed to? Would his opponents be able to stop him, employing any and all means available to them? What would happen if he lost? What would happen if he won?
If there's anyone in the modern world who can make his own life seem like something out of Shakespeare, it'd be Donald Trump. He is larger than life, and for the time being the rest of us seem to just be waiting to see how it all ends.
One way traffic
I'll be honest: election prediction can usually feel like a fool's game. While data helps a great deal, ultimately you still rely on your gut feeling about which way the wind is blowing. Sometimes the data is so sparse that you end up coming up with two scenarios as likely as each other, because you're not quite sure where one particular group of voters will end up.
For example, in 2017 I said that the ruling Liberal-National government in Western Australia would lose power - which was correct - but that the governing parties would lose ten fewer seats than they ended up losing. That election ended up being one of the biggest electoral defeats for a sitting government in Australian history.
But in my article, I pointed out that there were a bunch of seats that could easily fall to Labor if their swing in established suburbs was just a touch larger. If one went, they would all go - and that is exactly what happened. That is often how it is in the world of predictions: you figure out which scenarios are plausible and implausible, and then do your best to pick the most likely of the plausible ones, but sometimes you'll zig instead of zagging and the final result will be different to what you said it would be.
This is why I would argue that my 2016 prediction was not as impressive as it may seem, and my 2020 one is not as wrong as it appears. Both were within a range of possibility, and it was simply about which one I would calculate was the most likely to happen within that range.
In many ways, the official result of the 2020 election feels more like a vindication than anything, because the unusual turnout was the only part of the equation that was a moving and largely invisible target. I would suggest that a standard turnout election in 2020 would have turned out largely as I suggested, except for the three states I argued as being too difficult to call due to the noise and lack of data - Nevada, Minnesota and New Hampshire, all three of which I suspect I had the wrong way around. Nevada was one of the few states to move towards the Republicans in the final figures, while Minnesota and New Hampshire moved the other direction.
But it wasn't a standard turnout election, so it wasn't as I called it. Such is life.
Nevertheless, in order to even know which scenarios are the plausible ones in any election, you first need to get a 'feel' for the momentum of the election campaign. This can be tricky, as relying on media coverage alone leaves you vulnerable to their biases. Relying on polling data alone will leave you confused, as headline figures will be often quite different depending on the firm, because each of them also has their own biases. And relying just on you own eyes and ears leave you vulnerable to your own biases, which are the strongest of all!
A big part of the reason I call this the third act of a story is that it has many of the same qualities as far as media coverage, polling, and my own sense of the race. In 2016, the media coverage of Trump was largely negative and doubtful even as he kept winning, and polling was terribly off-base, missing his final vote share by significant margins, especially at state level. My sense of the race was that low-propensity voters that Trump was appealing to were being missed by pollsters, taken for granted by Clinton, and ignored by the media who largely live in a very different world to them.
2020 was a repeat of this, but even more magnified. After four years of deriding him at every opportunity, media coverage of Trump was treating the election as over before it began, and polling was posting absurd, double-digit margins for Biden until the last week, when they began their (cowardly) herd towards somewhat neutral territory - and still managed to be off-base by even more than they were in 2016. My sense was that turnout for Trump would be even greater than it had been before, and this would be enough to overcome whatever the media and polling felt they were seeing.
Putting shenanigans to one side, this was largely true, but there was one group that I underestimated: the younger voter in 'growth' corridors, living in apartments or suburban homes with a young family, especially if they were first time voters.
When looking at counties that shifted from 2016 to 2020, this is the only group that routinely exists in counties that changed from the Republicans to the Democrats. While Biden made up a little bit of ground over all with working class voters in the Rust Belt, especially in Pennsylvania, the sheer amount of turnout for Trump from among them could have overcome that. But this other group of urban and suburban, often new voter was experiencing a serious crisis as an adult, perhaps for the first time, many having still been in school at the onset of the Global Financial Crisis, and were among those most eager to vote by mail.
While the method of voting and the direction of those votes seem almost inextricably linked, those very same voters also provide some insight into the state of the race in 2024. They were evidently over-represented in turnout in the previous election, but with Covid now no longer a pressing issue, there are other concerns on their minds.
Polling has shown that the economy is the single biggest issue for voters, and when split into multiple categories it dominates the top issues collectively as well. Jobs, inflation, cost of living, all these categories are top of mind for voters, and Trump is trusted more on them by some distance. Immigration is also a pressing concern, with even greater trust levels in favour of Trump.
Threats to democracy is more varied in polls - some have it higher, some less so - but it is one that is, interestingly, largely equally split and heavily depending on party identification. Abortion is also more varied, generally appearing higher in polls that have democracy lower, and vice versa.
In media resistance
Not that you'd necessarily be aware of any of this if you only watch traditional broadcast media or read their news websites. As in the previous two elections, they have been almost entirely uniform in their approach to the two candidates, and not in Trump's favour. Conversely, if you only use alternative media - podcasts, YouTube, X, smaller news websites - the opposite may be the case.
This is highly relevant to the election, as that new voting bloc mentioned above is also full of voters that don't engage with traditional media. In fact, much of the electorate doesn't engage with traditional media. It's well established that levels of societal trust have dipped quite low, and the media - once trusted to be the ones holding the establishment to account - are now seen by many as part of that establishment.
This is a tricky accusation for them to escape. Firstly, if we just focus on the United States, American mass media companies do tend to report the same things from the same perspectives, often because their journalists and producers and executives all believe those things. Secondly, it is in their interests - whether the accusation is true or not - to attack alternative media sources as untrue and spreading falsehoods, because they are a threat to the power and income of the mass media.
In a sense, this election is make-or-break for the media's role in deciding and influencing elections. It used to be that if the mass media said you were done, you were done. If they didn't pay attention to you, you couldn't break through. Oil billionaire Ross Perot was perhaps the first in the modern era to truly break through barrier in 1992, but undid himself by suspending his campaign while leading the polls, then un-suspending it.
Ron Paul was the next to gain real traction in 2008 and 2012. Treated largely as a joke by the media and the establishment Republicans, Paul gained a significant following on the internet and won huge grassroots donor support. But social media was only just starting to break through, and was therefore unable to shift the narrative. Paul was also not a natural at campaigning, nor was he able to do much in the way of self-funding.
Come 2016, the tide was shifting just in time for Donald Trump and Bernie Sanders to take centre-stage. But Sanders, while winning the support of the majority of Democratic voters in the primaries, was seriously disadvantaged by a) the Democratic Party's system of superdelegates, which ensured the establishment leaders could essentially veto anyone they didn't like, and b) Donald Trump sucking up so much of the oxygen in both mainstream and alternative media spaces that Sanders could not make the primaries all about his own grassroots campaign, and which essentially required the Democrats to be the establishment party in order to win (or so they thought).
But even then, Trump was making the most of his mainstream media connections. He appeared on The View, and on The Tonight Show. He would do interviews with the main network television news programmes. He would call in to shows like Fox and Friends constantly.
But 2020 saw a fundamental shift in the way people absorbed media in general, including news media. Broadcast television was no longer that important. Everyone was stuck at home, and they were streaming shows and watching YouTube videos. They were watching Twitch, they were scrolling TikTok. Neither party quite knew how to engage with this because every American state had a different approach to doing life that year, often divided by party lines.
But one thing Trump did know is that the mainstream media hated him, and he hated them back. After election night that year and the days that followed, his fury at them (including, and perhaps especially, Fox News for calling Arizona early, an incredibly irresponsible decision on their part that caused voters waiting in line to leave) meant he would be trying alternative angles in the future.
Coincidentally, one crucial path to opening up those alternative angles to such a large audience came through Twitter. After Elon Musk's purchase and renaming of the company, the X algorithm was vastly different to Twitter's, and a host of alternative media voices that were either banned or subdued on Twitter were now welcome and thriving on X. This accelerated the shift towards alternative media, and accelerated the need for mainstream media to stop their rivals, and stop Trump.
One particularly unusual feature of 2020 and 2024, though, has been the paucity of one-on-one time that media outlets have been able to spend with the Democratic candidates. Both Joe Biden and Kamala Harris have been rather camera shy, especially when not carefully stage-managed. But the traditional media has largely been able to fill in the spaces for them, often running that day's talking point from the Democratic campaign without critique.
Thus, if you only take in information from the mainstream, you will have a largely negative image of Donald Trump, and a largely positive one of Harris. Given that their favourability ratings are essentially equal, this means that people are not listening to the traditional media sources in large numbers. They are actively distrusting the institutions that were previously trusted to give them much of the necessary information for making an informed voting decision.
Given the strength of Trump's popularity in much of alternative media (though, it should be stressed, certainly not all - there are plenty of emerging liberal and socialist sources that don't necessarily like him a great deal), one can quite easily assume that this is where people are getting their information and forming their opinions from.
If Trump wins not just the election but the popular vote as well - and, to be honest, even if he doesn't - these traditional media sources will have to seriously think through their approach. Either they will need to reform themselves to reach a larger number of people, or they will continue to shrink and wither away. They certainly wouldn't be the first media organisations to do so. Many of the newspapers and pamphlets in circulation two centuries ago were gone a century later. Perhaps these companies, too, are reaching the end of their life cycle.
Pools of polls
Public polling comes in three broad groups: companies hired by media organisations; universities; and independent pollsters who are usually trying to advertise their services for private polling.
It hasn't been a great decade for polling companies; arguably, it's been longer than a decade of misses, as many struggled in 2012 as well. But the Trump-era has been particularly hard on many of them. 2020 was such a disaster that most of them received cuts in funding, although Morning Consult, one of the most inaccurate of the bunch, got a hefty raise. Go figure.
There are extenuating circumstances for them, though. Those employed by media organisations tend to be on a deadline, and many of those companies may want the polls slanting a particular way due to their own biases. If they have to ship off dodgy data that they'd rather re-do if given the choice, well, too bad - they get the money either way. Those coming out of universities, meanwhile, often have slimmer budgets to work with, which means they simply can't afford to re-sample or over-sample to get a good poll out. They, too, can be affected by their biases, of course.
As, for that matter, can independent polling firms. But unlike the other two, these tend to be freelancers who depend on the quality of their work to survive. You could argue the other types of pollsters do too, but it's just not the same. If the media gets a poll they like the look of, the accuracy is secondary. Universities will always fund polling if they think it's worthwhile. Independent firms live or die on what they put out.
The problem for many of them has actually been the same thread since at least the Obama years: overstating the vote of college-educated whites, and understating the working class.
This happens consistently due to the latter being harder to reach than the former. College-educated voters will often be quite happy to tell you who they're voting for and why, and are especially likely to respond to polls when they feel their preferred candidate is doing really well. Working class voters (or those who are otherwise more suspicious of polls and politics) are harder to catch - some don't want to talk politics, and some just want to eat dinner.
And sure, you can account for over-sampling one demographic to an extent with weighting the polls (not every firm does this - some just try to get more samples instead, but that costs money and time), but the higher the weight ratio, the greater the margin for error. This becomes even more problematic when you also have to try and estimate the likely turnout of the electorate, and sometimes the weight of response bias (when some people are really, really excited to report that they are voting for a certain candidate) just overwhelms the ability of polling companies to match the true, likely electorate.
And sometimes, polling companies are being deliberately misleading because they want a particular candidate to win or lose. If one is to be cynical, it's rather curious that polling outliers in 2016, and 2020, and even now in 2024 have almost entirely been in favour of the Democratic candidate, and when polling firms break from a consensus that - in the light of election results - leans too far Democratic, they are criticised as being "Republican" pollsters, rather than congratulated for having the gumption to go with what their data tells them.
The polling averages over the last six months are, frankly, strange viewing. After a slew of polls came out following the Biden/Trump debate that had Trump up by double digits, Biden was dumped, and given how friendly many of these polls had been to him in the years prior (including in 2020), it's easy to wonder whether they were used as a pretext for dumping him, rather than being an accurate representation of the electorate. Of course, they could have just been overwhelmed by response bias in favour of Trump, which would be the first time perhaps ever in national polling that that's happened.
That these same polls then often showed a massive jump in support for Harris post-Democratic National Convention, only to now come back to anywhere between a tie and a Trump lead is especially questionable. Public support just doesn't rise and fall so much in such a short space of time. People are not that fickle, especially when the election is really akin to a referendum on the previous two presidential terms. Voters know what they're getting, and have mostly already decided who to vote for. It's just a matter of whether they turn out to vote, and that is not going to result in an eight-point turn in the national vote percentage in the space of a week.
After the miss in 2020, polling firms have done almost nothing to fix their mistakes. The only change for many has been the addition of 'voter recall', where likely voters tell firms who they voted for in 2020, so that they can try and map the electorate to how it voted last time. This is potentially useful in figuring out crossover voters, but it assumes that people aren't going to lie about their votes last time. If they regret who they voted for, some of them absolutely will lie, and if they're going to lie about who they voted for in the previous election when there's an unpopular sitting president, it's usually because they vote for the president.
Just to make things even tougher for pollsters, people often also don't tell the truth about their party registration, though this is less due to shame and more due to how they perceive themselves. We'll talk a little more about self-perception and ID later, but what can often happen is that people will give their party registration according to the person they're voting for, rather than what the data on the electoral roll actually says. For example, a registered Democrat may tell the pollsters that they are a Republican because they're going to vote for Trump, but they may also be voting for Democrats downballot. This can make polling the presidential race tricky, and downballot races even tougher.
On the other hand, polling at national level is easier than at state level. While national polls at the past two elections have missed to some extent - often saved in the final tallies by late herding, but showing up worse in rolling month-long averages - state polling has routinely been terrible. This is because in national polls, you can fill in certain demographics by polling states that don't get much attention. Need a rural Republican? Hit up someone in country Arkansas. Need a young tech worker? Try the suburbs of Seattle.
But in state polling, you're calling on a much smaller pool of people, and when there's a smaller pool, you have to cast the net more narrowly as well. Problems with sampling particularly demographics are bound to pop up, and by the end of the campaign it's often the case that the groups that are hardest to poll in the first place are just not going to talk any more. They're often sick of being contacted by polling firms! And guess what? These groups tend to be men, tend to be Republican, and tend to be working-class.
Needless to say, given all these issues and the problems that polling companies had in 2016 and 2020, there's good reason to think that if there's going to be a polling miss this time, it will again be because they are missing the real size of the Trump vote, and the question will be whether it's a 1-3 point miss, as in 2016, or 3-6 points off, as in 2020. The former was defended by some for being within the margin of error, but again, that was only due to late herding. In both elections, aggregators that attempted to weight polls according to whether they were 'high quality' ended up with a worse record than the Real Clear Politics average, which does not weight and was therefore brought closer to reality by the group of polling firms that strayed from the norm.
Amusingly, RCP is currently not included in the Wikipedia aggregate of aggregates, because they include these same polling firms (such as Rasmussen and Trafalgar) that other aggregators claim are "low quality", despite having better records in the past decade than "high quality" firms.
This time around, many polls appear scared of their own shadow. Having done the loop-de-loop from Trump winning to Biden dropping out to Harris becoming the nominee and then the second debate, the averages are ending up unusually close between the candidates and to each other. Many are tied. This seems anomalous in its own way - where are the outliers?
State polling, meanwhile, is almost schizophrenic. New York Times/Siena had a seven-point difference between results in Arizona and Nevada, Harris ahead in North Carolina but Trump ahead in Georgia, and the two candidates tied in Michigan and Pennsylvania but Harris ahead in Wisconsin. None of these combinations make sense. CNN/SSRS is similarly confusing, with Harris up by more than the margin of error in Michigan and Wisconsin, but tied in Pennsylvania.
Morning Consult, who I ragged on earlier, made a sudden, consistent multi-point shift towards Trump late in the piece, putting him ahead in almost every swing state. For that I give them kudos, as they must have done so through gritted teeth. Ann Selzer, the supposed doyen of pollsters who releases monthly in Iowa, has staked her legacy on a Harris +3 in Iowa, which goes against all indicators, including the enormous Republican lead in registrations and national polling, and appears to have polled a completely different electorate than any other polling company - the economy ranks below democracy and abortion in this poll, and seniors vote for Harris by over 20 points more.
National polling is useful for understanding what happens in states from election to election. They give a baseline to work with. For example, in 2016, Trump won Pennsylvania by less than 1%, and Biden won the state in 2020 by just over 1%. The difference between the two elections nationally was 2.5%, and Pennsylvania shifted by about 2%. While swing is not uniform from state to state, you can see how closely aligned they are in this case.
Trump won Iowa by 10% in 2016 and 9% in 2020, which is also a shift with the national result. For Selzer's Iowa poll to be right, the state would have to have completely changed its voting habits in the past four years, and Harris would need to be leading nationally by some distance.
Here's the kicker: polling in Iowa both elections had a close race, and Selzer had it tied in September 2020 before moving to Trump +7 in October. As I've said, people don't change their minds that quickly. State polling has been more consistent in understating Trump on aggregate in the past two election cycles, but also with how wildly different results are from company to company compared to national polling.
What we can say, though, is that given where the national polling is at compared to the past two cycles - even if they are spot on this time around - that would put Trump ahead of where he was in either of the past two elections in the final tally.
The race that stops the nations
Listening just to the media, then, will give you a jaundiced view of the race and the candidates involved, depending on which form of media you absorb. Listening to the polls has proven to be a risk, because they have consistently understated Trump's support. This is where our own judgement has to come in to play.
We can glean nuggets of information from the past, like the fact that the Rust Belt states of Michigan, Pennsylvania and Wisconsin have voted together since 1988, that the direction of voter registration trends tend to point to the candidate that will win, or that the Electoral College equation favours non-college educated voters. We can observe rallies, media events and the online debate, to see where the tone of the campaign is at. Where are candidates going? Which stories are staying in the news cycle, and are they actually reaching people? Which campaign seems happier with their position?
Trying to 'feel out' where the campaign at can seem incredibly subjective. If I say that the Harris campaign appears tired, worried and angry, based on how her surrogates are acting and news that dribbles out to reporters, is that an objective fact, or just how I'm interpreting things because of my own biases? Otherwise, the candidates seem mostly like themselves. Trump is a little older than in 2016, without quite the same one-liners that got him in the headlines during his first election campaign. But the Harris campaign's push to make him seem like he's 'losing it' really hasn't gained much traction, especially thanks to his hours-long podcast appearances, where it's been clear - as it was in his debate against the current president - that there's quite a difference between him and Biden, who is definitely suffering from senility.
Harris has never been particularly well-liked by the public in any office she's held, but she's been able to hold her own when dealing with people on a personal level. However, on any serious matter, and when she's received pushback or come across a situation she did not prepare for, she has regularly looked all at sea. It has been hard to escape the sense that, after that initial rush of media-generated momentum, the Harris campaign has been returning to a kind of equilibrium that isn't altogether different to where the Biden re-election campaign was at a year ago.
It's certainly the case that the Trump campaign seems mostly on the offensive, pushing rallies in New Mexico and Virginia within the last two weeks of the campaign, well beyond what are considered the core seven swing states. It's also the case that the Harris campaign has been shrinking their focus down as the campaign has worn on, with rallies largely centred on the three northern swing states.
Aside from polling which candidate voters are likely to vote for, there are other data points we can use to measure enthusiasm and direction of 'mood'. We've mentioned already that polls ask which issues matter to voters and which candidate voters trust more to deal with those issues, and Trump leaders solidly in the most important issues. Another question regularly asked is if voters believe the country is on the 'right track' or 'wrong track', and historically when 'right track' falls below 30%, the sitting president is in trouble. The sitting president may not be running, but Harris has not been able to establish herself as the 'change' candidate, which makes sense given she is the sitting Vice-President of the United States. What's more, her opponent has also been President, and serves as a point of comparison for what a change might look like for people.
Perhaps the most concerning sign for Democrats comes from a poll I am yet to mention: the Gallup voter registration rolling average. Gallup remain among the best of the best, always cautious but usually on the money. After a rough start in the 1990s, their voter registration averages have been within 1% of the popular vote at the past four elections, always in favour of the Democratic Party (but with such a small error rate it doesn't really matter).
For the first time since this polling began, Gallup recorded a Republican lead in the national average in their final three-month rolling average. It is only a 1% lead, but that is points ahead of where Republicans have landed at previous elections, and is a shift in affiliation on par with the shift towards the Democratic Party in the 2008 election rolling average. Pew Research has also noted a four-point shift towards Republicans in their own party identification polling. These may indicate another problem with 'voter recall' in polling - if pollsters are weighting according to who voters chose in 2020, but the electorate is four or five points more Republican to begin with than it was in 2020, they may be over-estimating Democratic support by that much.
Furthermore, individual polls that go into that rolling average have swung quite wildly from one side to the other, something not really seen in these polls since the Republican candidate last won the national vote in 2004. The opinion of Gallup more broadly is that the election environment is "favourable to the GOP."
Early voting is another data point, albeit one which requires subjective interpretation figure out what it's pointing to. Historically, the Democratic Party has encouraged early voting to 'run up the board', while the Republicans prefer voting on election day, playing 'catch up'. This year, that's changed, as the Republican Party has responded to the effect of widespread early voting in 2020 by encouraging Trump voters to vote before election day.
As of the time of writing, nearly 82,000,000 early votes have been received, with that number likely to go up further. However, this is below the 101,453,111 early votes received in 2020. What's more, the number of in-person early votes (which favours Republicans) has increased by around 9,000,000 so far, while the number of mail-in ballots (which favours Democrats) has dropped by 28,000,000 to date.
Among states that report party registration data for early voting, there is a distinct pattern emerging: Republicans are running at or above the levels of early voting they had in 2020, and Democrats are a long way down on where they would want to be. In a number of swing states, Republican registered early votes are outpacing Democratic ones, which is a number that should terrify Democrats because poll after poll is showing that Republicans are between 10-20% more likely to be voting on election day.
Early voting from Democrats in urban areas seems especially low, and these are voters that historically vote early, if they vote at all. There is, it seems, an enthusiasm gap emerging in the early vote that tracks with how the campaigns are looking. It seems as though Republicans are seriously enthused, and Democrats just aren't. In fact, Trump may be on track to equal or even better his total votes from 2020, given that he is gaining ground with groups he previously struggled with, and more low-propensity Republicans are voting early. This matters, because previously Republicans voted almost entirely on election day, which discourages low-propensity voters who are more likely to see long lines and go home instead of waiting.
Importantly, this means that Trump does not have one-off vote-by-mail inflation baked into his numbers like his opponents do. First-time voters from 2020 who had a significant lean towards Biden seem to be splitting up their choices more this time, including not voting at all. The enthusiasm gap perhaps explains why Trump is doing better with virtually every demographic group among likely voters than he was in the equivalent polls four years ago. Hispanics are especially notable here, as many left the GOP after the GFC, having voted for the party in 2004 and getting Bush over the line in states like New Mexico. Young voters are also noteworthy, as these are people that only recall recent presidencies (maybe even only Obama's second term, Trump, and Biden), and are most likely to use alternative media that is more favourable to Trump.
All is far from lost for the Democrats - college educated liberals are high propensity and will turn out on election day, but on the other hand, even if the GOP is taking some early votes from election day voting, they are also expanding their votes from previous non-voters, groups which the Democrats need to be winning to help run up that early vote tally.
There's other bad data for the folks in blue. Their registration numbers in total have dropped significantly since 2020 (reflecting the Gallup party affiliation polling), which is often a signal that voters are shifting allegiance.
If you're willing to change your voter registration to that of 'the other side', then mentally you have already had to undergo a paradigm shift in your thinking. This is often no easy feat for people, which is why - combined with low turnout - it can also often precipitate 'crossover' voting, where voters registered for one party nevertheless vote for the other one. This is because not everyone can bring themselves to switch away from their ancestral party, even if they no longer identify with it or vote for it.
If this is happening - and there is good reason to think that Trump is pulling these kinds of voters over, in a way no typical Republican could in places like New York City - it may spell further problems for Harris, as crossover voting is, in turn, usually correlated with an advantage among independents (though this is a weaker signal than others - Romney and John Kerry in 2004 both won with independents, but ran adrift overall because the party turnout wave of their opponents overwhelmed that small victory.)
Take a step back and look at that collection of data points. Trump is ahead of his polling in 2016 and 2020, both of which he out-performed on election day; polling around the direction of the country and party affiliation is good for Republicans; and the Democratic Party is significantly under-performing in early voting, to such an extent that their returns trail Republicans in a majority of the seven swing states.
Objectively, that is one way traffic. Subjectively, that matches with how the campaigns appear to be treating their own performance. Trump and his team seem quite happy. Harris and her team are putting on a brave face, but have been throwing anything they can at him in the past two weeks without much success. There is really little reason for traditional media organisations to insist that it is 'too close to call' who is leading the race, because it's forced them to search for narratives that would justify seeing the race that way, even when they don't fit with the overall data picture. If it was the Democratic Party that had all these signs in their favour, I'm not sure they would be saying that.
Vox populi
Alright, alright. You've waited long enough (or just scrolled down - if so, boo!), so let's get to the meat of this thing.
First, we must figure out the range for the popular vote. While the Electoral College means that the popular vote doesn't decide the winner, in practice we know roughly what kind of vote share either candidate would need to win because we have baselines from previous elections. Trump won in 2016 while trailing Clinton by around 2% on the popular vote, but lost to Biden while trailing by 4.5%. This give some indication as to how individual states will swing, even though, as we always say, swing is never uniform.
However, we have to bear in mind that some states can turn the popular vote into an optical illusion. The margin Biden won by in California and New York is essentially equal to the margin he won by overall, around 7 million votes. If one side is especially enthused, their turnout figures will look good, but the other side's turnout will be depressed. Since 2004, turnout has only once dipped below 60% - that was in 2012, when Barack Obama was able to turn out more of the Democratic base than Mitt Romney was of the unexcited Republicans.
The best case scenario for Harris with the popular vote appears to be something like a Clinton result, around 2% ahead of Trump. This would match with some of polls that seem to have been hedging their bets and have worse records at the past two presidential elections. The worst case scenario for her would be a 4+% defeat, which would assume an underestimation once again of Trump in the polls that have a better record matching his real vote percentage, although once the margins get that large it has more to with the loser's turnout being lower than expected.
Within this range, it is more likely than not that Trump will win the popular vote. You may then realise what this would mean for the Electoral College, but first we need to work out some numbers.
Turnout will not match 2020, as there simply aren't enough early votes to do so, and there is only signs of enthusiasm from one party's turnout to date. This will seriously impact the numbers that Harris can put up, as Biden relied on mail-ins to an almost absurd extent. If we take the 2008 and 2016 elections as potential precedents (in 2008, one side was heavily engaged; in 2016, both sides were reasonably engaged), turnout will be somewhere around 61%, which would be approximately 152 million votes. Around 2% of votes are usually cast for third-party candidates, which would leave us with approximately 149 million between the two main candidates.
Now, here's the question: how will Trump's total vote tally compare to 2020? Given that all the indicators suggest that he is upping his vote share among non-white voters, that he is winning over former Biden voters, and that he is bringing on low-propensity voters to vote early with a lot of room to grow due to winning election day by huge margins rather than relying on mail-ins (and thereby not 'cannibalising' his high-propensity voters), it is entirely plausible that he will actually increase his raw vote tally. Alternatively, his vote tally may stay around the same, with his gains off-setting a shrinking of the votes cast on the whole, even with a larger electorate, while the Democrats drop their turnout significantly.
This is not an era where voters are largely persuadable. Many voters are stuck in their corners, and it's a question of whether parties and candidates can convince their side to come out and support them. This is how Obama won both his elections - he brought out incredibly high numbers of his 'base' in 2008, along with receiving crossover from Republicans, and in 2012 he one again energised the base far more than his opponent.
The past two elections have seen strong bases for both candidates, even though the equation of who makes up that base is shifting. Nevertheless, this means that you can expect a bare minimum of 60 million (that's 40% of the voting electorate) to vote for either candidate at any election held today. The problem for Harris is that she simply has little enthusiasm behind her. Ignore the media and polls falling over themselves to show a wave of support - she isn't winning more voters in any demographic than Biden did.
In fairness, a comparison with 2020 arguably pointless. The available electorate that year - however it was formed - was an unrepeatable event. The best baseline we have for this election is Trump's first go-around in 2016. The problem for Harris from that baseline is that Trump has stronger favourability ratings than he did then, is more trusted on key issues than he was then, has a record in the bank with voters on those key issues, and she has less enthusiasm across the board than Clinton did.
Unlike in 2016, where there was high voting enthusiasm but also a high third-party vote, this election appears to be one where voters have made up their mind, one way or the other - or the other, which is to not vote at all. The decision from Robert F. Kennedy to endorse Trump probably reduced overall turnout, as he was likely to gain a significant protest vote from Democrats unhappy with Biden or Harris, but potentially also sway voters that would otherwise vote Trump. Those unwilling to choose Trump or Harris may just stay home.
Given the data available to us, I believe Donald Trump will win around 51% of the national vote, while Harris will do well to reach more than 47%. This is a slight over-estimate from where Gallup has landed, and a little ahead of where 2020's most accurate polling firms have Trump. If there is a large polling miss, it will almost certainly be in favour of Trump, as it will most likely be a result of Democratic turnout on election day being even further below par than early turnout has been. The best case for Harris on the data would be something like a 49-47 in her favour; the best for Trump would be a blowout 53-45. This is all assuming that third parties don't do better than expected. Given the low enthusiasm for Harris, high third-party turnout would be more likely to hurt her.
State of the states
With this in mind, we'll now examine each state to see what the likely outcome would be, given the above. States are ordered in categories for convenience. Electoral college votes and results at the 2016 and 2020 election are listed. State names will be highlighted in the colour of my predicted winner. States in bold are predicted to change hands from the last election.
Safe Democratic states
California (54)
2016: Clinton 61.73% vs Trump 31.62%
2020: Biden 63.48% vs Trump 34.32%
The home state of Kamala Harris, which last voted Republican in 1988. It has changed a lot since then, and Trump's total in 2016 was the lowest vote share for a Republican candidate in California since 1856.
Colorado (10)
2016: Clinton 48.16% vs Trump 43.25%
2020: Biden 55.40% vs Trump 41.90%
Ever since Barack Obama's victory in 2008, Colorado has rapidly turned from swing state in solid blue, owing largely to a liberalised social policy and high student population in Denver.
Connecticut (7)
2016: Clinton 54.57% vs Trump 40.93%
2020: Biden 59.24% vs Trump 39.21%
Has voted for the Democratic candidate since 1992.
Delaware (3) 2016: Clinton 53.09% vs Trump 41.72%
2020: Biden 58.74% vs Trump 39.77%
Has voted for the Democratic candidate since 1992.
Hawaii (4)
2016: Clinton 62.88% vs Trump 30.36%
2020: Biden 63.73% vs Trump 34.27%
Has voted for the Democratic candidate all but twice since the infamous 1960 election, when they sent two 'slates' of delegates to the Electoral College, as it was uncertain who was going to win the state when delegates were being chosen.
Illinois (19)
2016: Clinton 55.24% vs Trump 38.35%
2020: Biden 57.54% vs Trump 40.55%
Has voted for the Democratic candidate since 1992.
Maryland (10)
2016: Clinton 60.33% vs Trump 33.91%
2020: Biden 65.36% vs Trump 32.15%
Has voted for the Democratic candidate since 1992.
Massachusetts (11)
2016: Clinton 60.01% vs Trump 38.21%
2020: Biden 65.60% vs Trump 32.14%
Has voted for the Democratic candidate since 1988, one of only ten states to vote for the "Massachusetts liberal" Michael Dukakis, who was serving as Governor of the state at the time.
New Jersey (14)
2016: Clinton 55.45% vs Trump 41.35%
2020: Biden 57.34% vs Trump 41.40%
Has voted for the Democratic candidate since 1992. The Republican Party is stronger in this state than presidential margins indicate, but not so much so that Trump could do much better than reduce the margin to single digits.
New York (28)
2016: Clinton 59.38% vs Trump 36.75%
2020: Biden 60.87% vs Trump 37.74%
Has voted for the Democratic candidate since 1988. While Donald Trump is no longer officially a resident of the state, everyone knows New York is his real home. He dreams of winning it, but New York City is so heavily Democratic and such a large proportion of the state's population that it is almost a state unto itself, and one that he could never win.
Rhode Island (4)
2016: Clinton 54.41% vs Trump 38.90%
2020: Biden 59.39% vs Trump 38.61%
Has voted for the Democratic candidate since 1988.
Vermont (3)
2016: Clinton 55.72% vs Trump 29.76%
2020: Biden 66.09% vs Trump 30.67%
Has voted for the Democratic candidate since 1992.
Washington (12)
2016: Clinton 52.54% vs Trump 36.83%
2020: Biden 57.97% vs Trump 38.77%
Has voted for the Democratic candidate since 1988.
Washington, District of Columbia (3)
2016: Clinton 90.86% vs Trump 4.09%
2020: Biden 92.15% vs Trump 5.40%
Has never voted for a Republican candidate since being granted electors in 1961.
Safe Republican states
Alabama (9)
2016: Trump 62.80% vs Clinton 34.36%
2020: Trump 62.03% vs Biden 36.57%
Has voted for the Republican candidate since 1980.
Alaska (3)
2016: Trump 51.28% vs Clinton 36.55%
2020: Trump 52.83% vs Biden 42.77%
Has only ever voted for a Democratic candidate once, for Lyndon Johnson in 1964.
Arkansas (6)
2016: Trump 60.57% vs Clinton 33.65%
2020: Trump 62.40% vs Biden 33.78%
Was once a swing state, voting for every president from John F. Kennedy to George W. Bush, but since 2008 has voted for the Republican candidate regardless.
Idaho (4)
2016: Trump 59.25% vs Clinton 27.48%
2020: Trump 63.89% vs Biden 33.09%
Has voted for the Republican candidate since 1968.
Indiana (11)
2016: Trump 56.47% vs Clinton 37.46%
2020: Trump 57.02% vs Biden 40.96%
Has voted for the Republican candidate all but twice since 1948, mostly recently for Obama in 2008.
Kansas (6)
2016: Trump 56.16% vs Clinton 35.74%
2020: Trump 56.14% vs Biden 41.51%
Has voted for the Republican candidate since 1968.
Kentucky (8)
2016: Trump 62.52% vs Clinton 32.68%
2020: Trump 62.09% vs Biden 36.15%
This was previously a swing state like Arkansas, voting for every president from Lyndon Johnson to George W. Bush, but since 2008 has voted for the Republican candidate regardless.
Louisiana (8)
2016: Trump 58.09% vs Clinton 38.45%
2020: Trump 58.46% vs Biden 39.85%
Has the same record as Kentucky, except George Wallace carried the state as a third-party candidate in 1968.
Mississippi (6)
2016: Trump 57.86% vs Clinton 40.06%
2020: Trump 57.60% vs Biden 41.06%
Has voted for the Republican candidate since 1980.
Missouri (10)
2016: Trump 56.38% vs Clinton 35.87%
2020: Trump 56.80% vs Biden 41.41%
Previously a swing state since 1904 (except in 1956), but since 2008 has been solidly Republican regardless.
Montana (4)
2016: Trump 56.17% vs Clinton 35.75%
2020: Trump 56.92% vs Biden 40.55%
Has voted for the Republican candidate since 1968, except in 1992 due to the effect of Ross Perot being a third-party candidate.
Nebraska (2+1+1+1)
2016: Trump 58.75% vs Clinton 33.70%
2020: Trump 58.22% vs Biden 39.17%
One of two states to divide its electoral votes by district as well as to the overall winner, Nebraska is largely safe for the Republicans. Since 1940, the state has only once voted for the Democratic Party. In 2020, the second district, which is centred around Omaha, swung to the Democrats after years of trying. While the state as a whole is solidly Republican, the shifting demographics of Omaha may make the second district difficult to take back for them - although it wouldn't necessarily surprise me if Trump could squeak it out as he did in 2016.
North Dakota (3)
2016: Trump 62.96% vs Clinton 27.23%
2020: Trump 65.11% vs Biden 31.76%
Has voted for the Republican candidate since 1968.
Oklahoma (7)
2016: Trump 65.32% vs Clinton 28.93%
2020: Trump 65.37% vs Biden 32.29%
Has voted for the Republican candidate since 1968.
South Carolina (9)
2016: Trump 54.94% vs Clinton 40.67%
2020: Trump 55.11% vs Biden 43.43%
Has voted for the Republican candidate since 1980.
South Dakota (3)
2016: Trump 61.53% vs Clinton 31.74%
2020: Trump 61.77% vs Biden 35.61%
Has voted for the Republican candidate since 1968.
Tennessee (11)
2016: Trump 60.72% vs Clinton 34.72%
2020: Trump 60.66% vs Biden 37.45%
Much like Kentucky, this state backed every president from Lyndon Johnson to George W. Bush, but since 2008 has voted for the Republican candidate regardless.
Texas (40)
2016: Trump 52.23% vs Clinton 43.24%
2020: Trump 52.06% vs Biden 46.48%
It has for many years now been the dream of the Democratic Party to turn Texas blue, like it used to be many years ago - though if they did it again it would be for very different reasons, mostly due to the changing culture of their big cities. Nevertheless, it remains a distant dream for now.
Utah (6)
2016: Trump 45.54% vs Clinton 27.46%
2020: Trump 58.13% vs Biden 37.65%
Has voted for the Republican candidate since 1968.
West Virginia (4)
2016: Trump 68.50% vs Clinton 26.43%
2020: Trump 68.62% vs Biden 29.69%
Once a state that would flip between the two parties with little consideration for who would win the election, West Virginia has voted for the Republican candidate since 2000 and is, by margin, one of the safest Republican states overall.
Wyoming (3)
2016: Trump 68.17% vs Clinton 21.88%
2020: Trump 69.94% vs Biden 26.55%
Has voted for the Republican candidate since 1968.
Lean Republican states
Florida (30)
2016: Trump 49.02% vs Clinton 47.82%
2020: Trump 51.12% vs Biden 47.86%
One of the many unusual aspects of the 2020 election was that the two traditional bellwether states - Florida and Ohio - were won by Donald Trump quite comfortably, along with 19 of the top 20 bellwether counties. For many years, Florida has been a state that would only shift a couple of points away from a 50/50 battle, and the last time that both it and Ohio went to the losing candidate was during the suspect 1960 election.
Florida is a bellwether on account of its mix of demographics, depending on which part of the state you live in, which together create a miniature version of the United States. This may be changing, as registrations for the Republicans in Florida have continued climbing while the Democratic Party's registrations are in freefall. Republicans have a nearly 1,000,000 vote lead in early vote share, and Trump could well win it by double-digits, having won over many Hispanic voters in Miami in addition to being popular throughout each region of the state.
Iowa (6)
2016: Trump 51.15% vs Clinton 41.74%
2020: Trump 53.09% vs Biden 44.89%
We've already mentioned the Selzer/Des Moines Register poll which had Harris up by three points. Put simply, it doesn't add up. Iowa was a largely rural Democratic state for many years, but Trump's message has had a lot of cut-through in previous elections and there is nothing to indicate that has changed. One poll, way out on its own, doesn't change that.
Ohio (17)
2016: Trump 51.31% vs Clinton 43.24%
2020: Trump 53.27% vs Biden 45.24%
2020 marked only the third time since 1896 that Ohio was not won by the president. One of the others was 1960, and the other was 1944, in the middle of the Second World War. Republican candidate Thomas Dewey beat Franklin Roosevelt by only 12,000 votes, and his running mate was the Governor of Ohio. Needless to say, Ohio has a very good record, owing to its status as the 'median' of the United States as a whole. Trump's message resonates here strongly, and he may well win the state by double digits just as with Florida. If so, it is hard to see how he loses the election, because it's hard to see how both Florida and Ohio could lose their bellwether status after what was, to be honest, a freak election in 2020.
Swing states
Arizona (11)
2016: Trump 48.08% vs Clinton 44.58%
2020: Biden 49.36% vs Trump 49.06%
One of the most contested results of 2020, Arizona's numbers look much more promising for Trump this time around. For starters, Republican registration rates are way above where they were in 2020, especially in the all-important Maricopa County, and are even better than in 2016. Republicans are leading early voting margins by a considerable distance, and it seems quite conceivable that Trump wins the state comfortably.
Georgia (16)
2016: Trump 50.38% vs Clinton 45.29%
2020: Biden 49.47% vs Trump 49.24%
The most contested result of 2020 - so contested that it's still going through legal battles - was the Democrats finally winning Georgia after talking about it for many years. It's still a bit of a head-scratcher even today, and the Republican grassroots seem to have been working overtime to take it back. All indications are that it will return to a margin similar to 2016.
Michigan (15)
2016: Trump 47.50% vs Clinton 47.27%
2020: Biden 50.62% vs Trump 47.84%
Because the three 'Rust Belt' states have moved together since 1988, it can be forgotten that they each have distinct characteristics. Of the three, this is the one that the Harris campaign can feel most positive about, though that doesn't necessarily translate to a greater than 50/50 chance of winning. Indeed, if Trump is anywhere near winning the popular vote, Michigan will almost certainly get swept up in that. If Harris can get out enough of the base and hope enough third party votes come through, maybe there's a chance. Of the seven swing states, it is the one where early voting data appears to be weakest for Republicans, but that's a high hurdle to pass. This might be the election where the three have a break-up, though I think they're more likely than not to stick together. Robert Kennedy is still on the ballot here, after the Secretary of State broke state law to keep him on. By the time the courts ruled on it, it was too late to change.
Nevada (6)
2016: Clinton 47.92% vs Trump 45.50%
2020: Biden 50.06% vs Trump 47.67%
Just as Democrats have dreamed about some states falling their way, so too have Republicans. Pennsylvania used to be one of them, and today it's Nevada. Under former Sen. Harry Reid, the Clark County machine would always find the votes needed to keep the state blue. But since his death, the grip of the Democratic Party on the state has weakened, and it now has a Republican governor. Republican registrations have been catching up as well, and Republicans actually lead by some 40,000 votes on early voting, which is a massive deal in a state that has around 75% vote early. Also, the Culinary Union believes Trump will win, and they have an excellent record when it comes to telling the pollsters whether they're right or wrong.
North Carolina (16)
2016: Trump 49.83% vs Clinton 46.17%
2020: Trump 49.93% vs Biden 48.59%
For years now, the Democrats have been waiting for North Carolina to turn blue thanks to changing demographics, particularly around the Research Triangle. It hasn't happened, with the state remaining stubbornly red by small margins. Polling can be quite difficult in this state, but early voting signs, as in the rest of the Sun Belt, are extremely positive for Republicans, who may even have a virtually unheard of lead coming into election day.
Pennsylvania (19)
2016: Trump 48.18% vs Clinton 47.46%
2020: Biden 50.01% vs Trump 48.84%
Once the pipe dream for Republicans, Trump changed the equation in 2016, and both campaigns are aware that the strongest (and, on early voting numbers, really only) chance Harris has of become the next President of the United States runs through Philadelphia, towards Detroit and into Milwaukee. She has spent a lot of her time here, hoping to hang on to the 'Scranton Joe' vote that swayed things towards Biden in 2020, as well as bringing out high numbers in urban Philadelphia as her predecessors did. The early numbers are extremely discouraging for the Democrats, with turnout down massively in Philadelphia. The size of the population means it is unlikely that either candidate will win by more than a few percent, but things look extremely positive for Trump in this state, possibly the strongest of the three neighbours.
Wisconsin (10)
2016: Trump 47.22% vs Clinton 46.45%
2020: Biden 49.45% vs Trump 48.82%
Wisconsin is a notoriously difficult state to poll, and is the only one of these seven states where Harris has a (very small) lead in every major aggregate. But that is pretty much entirely down to some outliers which have her 5 and 6 points ahead, while the rest have Trump up by smaller margins. The early vote data is harder to gauge than some other states, but based on which counties are turning out, the best signs are again for Republicans from Republican counties, and more significantly, there are signs of weakness in Democratic counties.
Lean Democratic states
Maine (2+1+1)
2016: Clinton 47.83% vs Trump 44.87%
2020: Biden 53.09% vs Trump 44.02%
Easily forgotten in the far north-east corner of the country, Maine is possibly also the hardest state to poll (that isn't incredibly safe for either party). The difficulty is raised further by being one of two states to divide Electoral College delegates, with the two districts being quite distinct from each other. Trump has won the second district on both occasions, and his chances of winning the majority of state votes has been somewhat of a sleeper possibility, owing to the state's small population. He came close to beating Clinton outright in 2016 without really trying, so any success in New Hampshire may end up tilting things over the northern border as well.
Minnesota (10)
2016: Clinton 46.44% vs Trump 44.93%
2020: Biden 52.40% vs Trump 45.28%
Donald Trump came ever so close to winning this state in 2016, something that even Ronald Reagan couldn't accomplish. Not since Richard Nixon in 1972 has a Republican carried the state. Were it not for Tim Walz, I would have with confidence said Trump would be able to tip things over the line here, running stronger than in 2016 and the most likely state to tip of this whole group. If Virginia is moving big enough, Minnesota is going to move in the same direction. But sitting governors have a noticeable effect on votes in that state when they're on the ticket.
Nevertheless, I think it is better to be bold and go where the direction of travel leads than hedge my bets like everyone else. I would rather someone be logically, reasonably and earnestly wrong than irrationally stumble on the right answer. There is every chance that someone like Jill Stein racks up third party votes as happened in 2016, which lowers the margin Trumps needs to overcome. Trump is winning over Muslims in the north of the country, which is significant. Whoever wins will win it by a slim margin. I think it is a real toss-up. But I've said that before, and this is Minnesota. Fool me once...
New Hampshire (4)
2016: Clinton 47.62% vs Trump 47.25%
2020: Biden 52.71% vs Trump 45.36%
Shame on me, New Hampshire has fooled me twice. The only state to defy me in 2016, by the tiniest of margins, that alone almost put me off saying that they would flip this time, but New Hampshire voters have historically been more willing to shift towards candidates they genuinely prefer (or away from candidates they dislike) and the libertarian vote may well be more comfortable with Trump this time around, particularly with the noises around making government more efficient. The movement of registrations is also much more kind to Republicans than it was in 2020. On the other hand, there is a not inconsiderable intake of Massachusetts liberals now living there that may be able to keep it in Harris hands.
New Mexico (5)
2016: Clinton 48.26% vs Trump 40.04%
2020: Biden 54.29% vs Trump 43.50%
Trump is making some serious gains with Hispanics, especially working class Hispanics, so much so that he made a stop in Albuquerque near the end of the campaign. New Mexico has a small population, and Democratic registrations are still quite high. Like other states here, we really don't know much about how things are moving. There is a sense that Democratic votes are embedded here in a way they aren't in some other states. But it is definitely conceivable that a big movement happens here that gets the margin much closer to even odds than the Democrats would be comfortable with.
Oregon (8)
2016: Clinton 50.07% vs Trump 39.09%
2020: Biden 56.45% vs Trump 40.37%
Originally, I had this as a safe Democratic state - and why wouldn't I? This state has voted for the Democratic candidate since 1988. But Oregon has only a middling population as far as states go, which makes it easier to shift from one side to the other, and it walks to its own drumbeat like some of the other states in this group. The only poll done here since July gave Harris a twelve point lead, but was from a Democratic Party partisan group, while a July poll only gave Harris a five point lead. There are often mutterings from the rest of the state that it would be solidly Republican if not for Portland.
But here's the reason I've shifted it: the early voting is really bad for the Democrats. They are ahead on party registrations, but only by 120,000 as of election eve. If the non-aligned registrants break 2-to-1 to Trump (assuming each party's voters vote along party lines) and election day voting favours him as it does in most places, that would put him ahead of Harris. I don't quite trust the maths to work out that way, but there is so little data coming out of the Beaver State that I feel it's worth putting as something to watch out for, just in case.
Virginia (13)
2016: Clinton 49.73% vs Trump 44.41%
2020: Biden 54.11% vs Trump 44.00%
While it's got somewhat lost among traditional media coverage, there is a lot of noise in Virginia at the moment. Republicans in the state are feeling confident of turning the tide following the victory of Glenn Youngkin in the gubernatorial race two years ago. While not necessarily a natural ally of Trump, Youngkin has whole-heartedly thrown his support behind him, wanting to motivate the same voters that got him elected in 2022 out again, as well as hoping that the Democrats are as demotivated as they were that year too.
Polling has it as a 2-5 point race in favour of Harris, which would be closer than 2016. The one thing that I think works against Trump here is that I'm not sure voters will be as demotivated as in 2022. He can get the Republican turnout on his side, especially with Youngkin there to allay the fears of the kinds of suburban mothers that were aligned with his policies but might not like Trump, but Virginia's results relies a great deal on swings in the suburbs of DC, and around Fairfax County in particular. While government workers there may not have been worked up about Youngkin, I can see them being much more concerned about a Trump administration that may well send in Elon Musk and Ron Paul to give them the chop. Even if Trump has a good night, I think this will be a tight race that Harris can probably hold on to, but it may be quite uncomfortable for the Democrats in the downballot races.
According to these estimates, the final tally will look like this:
This gives Trump a clean sweep of the seven swing states, as well as three surprise victories in the neighbouring New England states of Maine and New Hampshire, and Minnesota. These are, obviously, his least likely wins, and if the national popular vote is closer than I estimate, he will not win any of them - but he will sweep the swing states regardless. If Harris wins a swing state, it will most likely be Michigan, and the only way she can conceivably pull off a victory is through the Rust Belt. Her best bet outside there is probably Nevada, followed by North Carolina. If Trump has a great night, Virginia is up for grabs, and New Mexico would be in touching distance along with potentially Oregon.
The third act is about to reach its climax. It has been quite the ride, and this will likely be the last American election preview I do - famous last words, I know - because there will probably never be a political figure like Donald Trump again in my lifetime.
Enjoy the election.
Enjoy reading our work? Why not buy us a metaphorical drink on Ko-Fi? Rest assured that such support is treated as real, even if the drink is not.
Comments