


US election polls: With one week to go, the race for the White House is still too close to call
In DepthKamala Harris's lead has shrunk nationally, while the two candidates remain neck-and-neck in the majority of the seven key swing states. The election could well come down to certain fractions of the American electorate.
The 2024 presidential election could well be the closest in modern American history. With just one week to go before the election, the polls measuring voting intentions for Kamala Harris and for Donald Trump are drawing closer and closer, with the difference lying in the margin of error.
This comes despite Harris having long enjoyed a decisive lead over Trump in the national polls. After a strong start to her campaign, supported by the Democratic party establishment and the party's national convention in mid-August, the vice president had succeeded in bringing in a breath of fresh air, which many voters, who were put off by the prospect of another Biden-Trump duel, had been hoping for. Yet the Democratic candidate's lead has eroded since mid-October, despite her efforts to appear as unifying as possible in the face of her divisive Republican rival, whom many voters have said they do not like.
Trump and Harris are neck and neck in the swing states
In the seven swing states that will decide the outcome of the election, the situation isn't much clearer, with polls showing extremely close results between Trump and Harris.
The former president nevertheless appears to be in a favorable position in Georgia and Arizona, two swing states that have historically been very favorable to the Republicans, which Joe Biden took from them in 2020. According to the aggregated voting intentions collated by the specialized polling website FiveThirtyEight, Trump leads in those states by 2 and 2.2 points, respectively. These would seem to be comfortable margins for the Republican candidate when compared to other swing states, but still slim ones.
In the Midwestern states (Wisconsin, Michigan and Pennsylvania), opinion polls have shown the gap closing between the two candidates, making the race more uncertain than ever in these three important states.
The erosion of the Democratic vote among Black and Hispanic voters, a risk for Harris
In a scenario in which the election could come down to a few tenths of a percentage point in the swing states, the election's outcome may hinge on fractions of the American electorate. In this respect, the relative decline in support for the Democrats among Black and Hispanic voters, which has documented over several months by several opinion surveys (i.e. for the The New York Times or on the website Split Ticket), constitutes a risk for Harris, and not an inconsiderable one.
There was, for example, a clear drop-off in Democratic votes among Black voters in 2024. While 92% of them voted blue in 2016, and 90% in 2020, they were allegedly only 78% to do so in 2024. Conversely, support for Trump could significantly increase among these segments of the electorate, rising from 9%, in 2020, to 15%, in 2024. Similar observations have been made regarding Hispanic voters: 68% of them had supported Hillary Clinton in 2016, but the latest New York Times poll now puts that level of support at 56% (-12 points).
The decline has been especially pronounced among young male voters. For example, 45% of male Hispanic voters under the age of 30 intend to vote for Trump, compared with 31% of women voters of the same age. Trump's racist outbursts would seem to have had no effect on this electorate, at least on those who are citizens born in the US, only 30% of whom feel targeted by the Republican candidate (43% for those who were born abroad). Economic conditions also play a role among the Hispanic electorate, which have been particularly vulnerable to the record rise in inflation in recent years, and Trump is often seen as being more convincing on economic issues.
This decline is, admittedly, a relative one, as it only concerns a fraction of Hispanic and Black voters, who still overwhelmingly support the Democrats. Yet, with the possibility of a very close election, this erosion could represent a considerable risk for Harris, who would then become more dependent on support from white, middle-class suburbs such as around Philadelphia, Pennsylvania, where the Democrats need to maintain Joe Biden's 2020 lead if they hope to win.
By way of illustration, the last Democrat to garner less than 60% support from Hispanic voters was John Kerry, in 2004, who lost that year's election to George W. Bush.
A campaign that has been more divided by gender than ever before
The "gender gap" in the American electorate is not a new phenomenon, having been continuously observed since the 1980s. Yet the 2024 election could be the most gendered one in contemporary US history.
A New York Times poll, conducted from October 20 to 23, shows that Harris was supported by 53% of women surveyed, compared with 39% of men. Conversely, 53% of men said they would vote for Trump, versus 41% of women. If it were to be confirmed in the election, this gender gap, which many surveys conducted since August have put at around 13 points, would at least be the largest since Ronald Reagan's 1980 election.
In addition to the fact that the Democratic candidate is a woman – the second ever, after Hillary Clinton – the women's electorate's enthusiasm for the party also comes against a backdrop of renewed struggles for reproductive rights in the US. The Supreme Court's repeal of federal abortion rights in June 2022, after they had been on the books for 49 years, caused a massive stir in the mid-term elections, and continues to mobilize and motivate American women (and, to a lesser extent, men).
Yet the widening gap between men and women in their electoral choices may also be the result of growing gender polarization, particularly among younger voters. Young women voters are increasingly left-leaning, while a growing proportion of young men are turning to more conservative ideas.
Such a gap could have very real consequences for the presidential election, in which women voters have had a higher turnout than men for over 40 years. In 2020, they accounted for 51.8% of voters, well ahead of men (44%). This gap, moreover, correlates with overall voter turnout: The higher it rises, the greater the proportion of women voters. Women are also more likely to be registered to vote. In 2020, there were 89 million women voters, compared with 79 million men. Harris's lead among the women's electorate could therefore prove decisive in winning enough key states.
The polls are close, but the election might not be
Polls, whether on the national or state level, very often differ from election results by several points. According to The New York Times, since 1988, the final estimate from national polls conducted before each election has, on average, differed from the election result by about 2.3 points. This is significantly better than estimates from polls conducted in swing states, which, since 2000, have "missed" the actual result by an average of 3.1 points. In fact, the newspaper has calculated that these polls have almost systematically underestimated the Trump vote since 2016, sometimes quite heavily.
In recent years, pollsters have been trying to identify and remedy the causes of the 2016 and 2020 errors, in order to better account for Trump's voters, the profiles of which are generally those of people who are further removed from politics, and who, therefore, would be more reluctant to respond to polls. This is at least one of the hypotheses that would explain the polls' difficulty in understanding the vote in the billionaire candidate's favor, since his appearance on the national political scene.
Despite these precautions, as well as the methodological changes introduced in recent years (contact methods, rectification techniques, etc.), it remains entirely possible that the polls might underestimate one candidate or the other. A victory could turn out to be a far more clear-cut and decisive affair than the current state of public opinion would suggest. Even an error of one percentage point would be enough to swing the election. In recent history, the polls have underestimated the Democrats as often as the Republicans.