Polls are open across the United States, as hundreds of millions of Americans prepare to make their voices heard and elect the next president, alongside countless federal, statewide and town ballot municipal races. According to the popular narrative, the two candidates—former president Donald Trump, a Republican, and Vice President Kamala Harris, a Democrat—are locked in a heated battle describable only by phrases indicative of the narrowest race possible, such as “toss-up”, “coin flip”, “nail-biter” and “too close to call.”
This outlook on the election stems from the polling averages, all of which show Harris and Trump with identical margins in every swing state, with each usually garnering around 47 to 50% of the vote. It seems as though no matter what, every swing state poll ends up with both candidates around these numbers, as unlikely as that might seem given the statistical likelihood of an eventual outlier. This is due to the phenomenon known as herding, where pollsters will adjust their screens to get a result that more closely reflects the pundits’ consensus. After all, outliers (like the infamous Biden +17 poll in Wisconsin in 2020) can severely damage a pollster’s reputation, or bring unwarranted attention to a firm seeking to hedge their bets and retain a good accreditation with aggregate websites. The problem with herding is that it generally presents an overcautious picture of an election cycle that often masks the true dynamics of the race—for example, an unexpected Harris or Trump strength with one or more demographics. Nate Silver calculated that the total odds of every swing poll being so remarkably close were a staggering one in 9.5 trillion. Clearly, something is wrong with polling, and unless the election really does come down to the wire in every swing state, there may be hell to pay on Nov. 6.
Moreover, polling is an industry where even a small firm can gain attention fast if it releases an interesting-enough result. This has led to many unknowns trying their hand at the trade, even without any prior experience. They do so for myriad reasons: an earnest attempt to improve aggregates by adding more sources, to gain money and a reputation, to influence betting markets (illegally, as Data For Progress’ Sean McElwee discovered), or, most disturbingly, to find Republican leads that aren’t there in order to help Trump and MAGA candidates promote claims of election fraud when they end up losing. Whatever the reason, this practice has led to a lot of inaccurate polling over the past few cycles. One particular highlight from the past few days is that AtlasIntel, a Brazilian firm included in the polling averages of Real Clear Politics, 538 and the Silver Bulletin, among others, conducted a whole new round of major swing state polling after the shock result of the latest J. Ann Selzer poll (see below). They claim they fielded over 7,000 respondents in two days, a near-impossibility for a foreign firm of its size (and even a near-impossibility for the very finest American pollsters). I will say this definitively: they are liars, and this did not happen, at least not to a reputable standard. That AtlasIntel’s data is still included in these averages tells you much about the current state of political punditry.
Despite all the herding and the toss-up talk, Democratic fortunes seemed reinvigorated over the weekend when Selzer, Iowa’s gold standard pollster and one of the most remarkably accurate prognosticators in the country, released a shock Des Moines Register poll that saw Harris leading by 3 points in a state where Trump had won by 8 in the 2020 election. The result was so surprising largely because it stood so far apart from polling of much closer states, all of which depicted an incredibly close race decided by maybe tens of thousands of undecided voters. Yet, if Harris is leading in Iowa, or if she has even closed Biden’s gap by a few points, she is overwhelmingly likely to pick up the Blue Wall states and probably the Sun Belt swing states as well.
Selzer gained her sterling reputation for a reason, that being her willingness to stand alone from the popular narrative and trust the data, not her priors. In 2016, her Trump +7 poll was dead on in predicting the margin that decided the state, while in 2020, her poll did not depict the major overestimation of Biden’s strength with white, working-class voters that Election Night proved much closer than expected. Moreover, she simply refuses to herd, and in a polling environment such as this one, even a massive outlier could have more predictive value than 50 herded polls. So, discount Selzer at your own risk. Iowa probably won’t turn blue this time, but if Harris gets anywhere near the 47% of the vote that Selzer projected there, she is all but certain to win the election.
Another reason for the Harris camp to feel confident is her strength in statewide polling compared to the national polling. Glancing at the 538 national poll aggregator, we can see that Harris is often only leading Trump by a point or two in the national popular vote, especially among registered voters. But if this is actually the case, swing state polling should be much stronger for Trump than it has been. Take a look at 538’s state flipper simulator and apply a uniform swing of R+3 to the national popular vote. Trump would win 313 electoral votes, including those of every single swing state and Nebraska’s second district. But with the exception of Arizona, Trump doesn’t have a definite lead in any of the swing states, telling us that something is awry with the screens of national polling.
Indeed, in the swing states, local polling (generally more accurate than outside firms) predicts a decent win for Harris, if only in the Rust Belt. In both Wisconsin’s Marquette Law School Poll and Michigan’s Detroit Free Press poll, Harris has a small but very real lead, breaking the crucial 50% barrier in both states. In Muhlenberg College’s final election survey of Pennsylvania, Harris doesn’t hit 50%, but leads Trump by two points anyways. A close national popular vote result is simply not borne out by all of the more precise swing state polling that we have.
More likely, Trump will hit the ceiling he has never been able to breach in any of his runs: 47%. Such a polarizing politician will probably not perform much better than he did when he had the novelty sheen of an outsider in 2016 (46.1%) or the benefit of incumbency and major crises befalling the nation in 2020 (46.8%). In the closing weeks of the campaign, Trump danced for 40 minutes while attendees fainted, spent more time with Elon Musk than his own running mate, and allowed a rally opener to disparage one of the most potent single ethnic voting blocs in the entire nation. There are plenty of valid cases to be made to vote against Harris, but Trump has largely failed to capitalize on any of them.
And while horserace polling refuses to predict anything but a tie, issue polling has presented a rosier picture than it might seem for the Harris campaign. Economic indicators and voter feelings on its status are often a strong predictor of election results. Harris has slashed Trump’s originally decisive lead on the economy down to only six points according to New York Times/Siena polling, better than Democrats enjoyed in 2020. She dominates on abortion, climate change, and has even improved on immigration versus Trump, which has been seen as her weakest point—possibly portending a stronger result in Arizona and Nevada than polls predict.
If we can’t rely on polling as much as we have in previous cycles, what can we trust? One trusted indicator has been the Washington primary, which is a “jungle primary” taking place in August, shortly after Harris’ ascension to the top of the ticket. Typically, these results are analyzed by combining the totals of every Democratic candidate and every Republican candidate, and comparing the two has been remarkably solid at predicting the national popular totals. And here, too, it looks good for Harris.
Not only did Democrats improve upon Biden’s 2020 performance by about 1.5% in the 2024 primary, they did so in the right places. In the less urban areas of the state that resemble the Upper Midwest in culture and demographic composition, Senator Maria Cantwell and other Democrats all saw a modest yet favorable shift in their direction. The Washington primary therefore points toward a slight improvement on Democratic numbers nationwide since 2020, including among the suburban voters who will be critical in states like Arizona, Georgia and North Carolina. But barring any major collapses with urban, non-white male voters, a slight improvement may be all Harris needs.
Much ink has been spilled about this foretold realignment of Hispanic and Black men. Marco Rubio has claimed that the GOP is transforming itself into a multiracial worker’s coalition, although the embrace of Musk has stemmed that rhetoric a bit. Republican gains with Latino voters have been a resilient source of anxiety for Democratic activists, but local polling has not really shown this to be the case, nor have the 2022 midterms, or the Washington primary, or 2024 special elections. What seems likely from polling this cycle is that Trump’s numbers have largely not shifted; while he has gained with men, who appreciate his relentless appeals to traditional masculinity, Latino women have shifted further left thanks to the Democratic focus on reproductive access and healthcare. In the Pew Research survey of Latino voters, Harris and Trump are at almost the exact same level as Biden and Trump had been four years earlier. And while it’s entirely possible that Trump really did win a sizable swath of Latino men, many of his gains there were seemingly washed away by the aforementioned “floating island of garbage” comment made at his Madison Square Garden rally.
It’s also possible that the polls are accurate, at least on the face of things, and that Trump commands a substantially larger percentage of the national popular vote than he had earned in either 2016 and 2020. After all, this would align with the aforementioned national polls that show Harris only narrowly ahead in the national popular vote, and would potentially mean that Democratic margins in the urban centers are eroding significantly. But even if Trump has managed to win over more voters than in his previous two campaigns, his third try for the White House faces an existential threat: they may not even turn out at all.
Since Trump first slid down that golden escalator in 2015, the Republican coalition has only ever gotten younger, more male and less white. Unfortunately for Trump, this means that the Republican coalition has also therefore become uniformly lower-propensity: easier to persuade yet, crucially, harder to convince to vote. The newfound reliance on low-propensity voters, whom they had traded with Democrats for higher-propensity college-educated voters, is hypothesized to have cost MAGA candidates critical margins in states where they narrowly fell short in 2022, including Arizona and Pennsylvania.
In order to ensure that these voters get out to the ballot boxes, a campaign must have a dependable, well-funded and enthusiastic physical turnout machine consisting of paid staff and volunteers. Yet for months now, Republican operatives have been ringing alarm bells about the anemic state of the GOP’s ground game, which has mainly been outsourced to Elon Musk’s ridiculous America PAC operation. Democrats, meanwhile, have a decisive advantage with regards to ground game and get-out-the-vote efforts in the swing states, due to their younger, more educated base and higher enthusiasm.
With all of these factors in mind, Harris is well positioned to become the 47th president of the United States. Not only do I believe that Harris will win, I believe that she will sweep every swing state, for 319 electoral votes compared to Trump’s 219 electoral votes—Joe Biden’s 2020 coalition in addition to North Carolina. Trump is a tired, unstable and bigoted conman running purely on resentment, and his poorly-ran campaign will do little to break his ceiling even with some economic headwinds and the unpopularity of the incumbent president. Americans have rejected him once, and every sign except polling indicates that they will do so again. That Harris’ clear advantage is being obscured as a result of herding and inaccurate narratives is a damning indictment of pollsters and pundits. We now must rely on the voters to prove them wrong.