The closest election of the century is getting closer. Two weeks ago, 538’s presidential election prediction gave Vice President Kamala Harris just a 58 in 100 chance of defeating former President Donald Trump on November 5. Now – just 18 days away from Election Day – our forecast gives Trump an absolute zero. lead in the race with a 52 in 100 chance of winning.*

You might be tempted to make a big deal about our prediction that we’ll “flip” to Trump, but it’s important to remember that a 52 in 100 chance for Trump isn’t all that different from a 58 in 100 for Harris – both are little better than a coin flip for the leading candidate. While Trump has undeniably gained some ground in recent weeks, a few good polls for Harris could easily put her back in the lead tomorrow. Our general characterization of the race – that it is a toss-up – remains unchanged.

The reason our forecast is close is that the polls are close – well within the range that even a small polling error could be decisive. According to our polling averages, the margin between Trump and Harris is 2 percentage points or less in all seven major swing states (Arizona, Georgia, Michigan, Nevada, North Carolina, Pennsylvania and Wisconsin). And in our average of national polls, Harris is just 2.0 points ahead of Trump. That’s smaller than the margin between Trump and President Joe Biden on every day of the 2020 campaign, and smaller than the margin between Trump and former Secretary of State Hillary Clinton on every day between October 1 and Election Day 2016. (Due to the Due to the Electoral College’s bias toward Republicans, our model currently estimates that Harris would likely need to win the national popular vote by 2.0 points to be favored to win a majority of electoral votes.)

For the most part, recent high-quality polls have only confirmed the closeness of the race. An ABC News/Ipsos poll from October 4 to 8 showed Harris with a 2-point lead nationally. An Oct. 1-10 poll from Marquette University Law School, one of the most accurate and transparent pollsters in America, showed her ahead by three points. The list goes on: YouGov/CBS News recently gave Harris a three-point lead; Marist College gave her a five-point lead; Fairleigh Dickinson University has a three-point lead. Fewer pollsters have given Trump a lead nationally, but there have been a few: Beacon Research/Shaw & Co. Research/Fox News found Trump ahead by 2 points; Hart Research Associates/Public Opinion Strategies/NBC News ranked him No. 1.

In the national polls, the race is not only close, but also stable. In fact, Harris’ current 2.0-point margin in national polls is virtually the same as it was two months ago, on August 18. The cross tables have also hardly changed. Among black and Hispanic voters, for example, Harris’ support has risen or fallen by less than 2 points since she entered the race. Of course, those two points may be decisive, but it is remarkable how little has changed over the course of the campaign.

However, polls show there is relatively more movement – ​​and most of it is against Harris. Compared to her numbers at the beginning of the month, Harris has lost ground in all seven key swing states. This is why her chances of winning in our prediction have decreased. In the American presidential elections, you don’t get points for winning the national popular vote.

Do partisan pollsters distort our averages?

One question we sometimes get is whether polling averages like 538 are biased in favor of Trump because of the influx of polls conducted by Republican companies. Over the past two weeks, 23 of the 121 polls in the seven key swing states came from a Republican pollster or sponsor.** Only four came from Democratic organizations, and the remaining 93 were nonpartisan.

While there is always uncertainty about how accurate state polls are, partisan polls influencing our averages are not one of my biggest concerns. That’s because we work hard to subtract potential statistical bias from each poll before incorporating it into our averages. As you can see in our poll averaging methodology, we adjust party polls to account for the fact that these polls tend to be a little too good for the sponsoring party.

And even for nonpartisan pollsters, we apply something called a “house effects” adjustment, which accounts for how much more Democratic or Republican-leaning a pollster is than his peers (either because of the partisan leanings of his clients or simply because of methodological choices that produce typically more liberal or conservative samples). For example, if a pollster’s poll numbers have consistently been 2 points better for Trump than the poll average, after controlling for factors such as a poll’s population (likely voters vs. registered voters or all adults) and mode (e.g., live phone, online panel, text, etc.), we adjust those polls 2 points toward Harris.

Finally, we give less weight to polls from pollsters without a rating of 538 polls and pollsters who release a number of polls in a short period of time. This ensures that pollsters “flooding the zone” with polls do not have an outsized influence on our averages.

One test of whether these adjustments are working correctly is to see what the 538 averages would look like if we did not include polls from the companies that are supposedly trying to influence the averages. The table below shows the results of this comparison. The second column shows the averages of 538 as of 4:00 PM Eastern on October 18. The third column is what our averages would have been at that point if we had removed those polls, but otherwise ran the same algorithm as we normally do:

As the table shows, this does not significantly change our averages. In most places, the pollsters in question are indeed more pro-Trump than other pollsters. However, this only has a mild effect on our averages, moving them towards Trump by an average of only 0.3 points. (The biggest difference is in Pennsylvania, where our published average gives Harris a 0.1 point lead over Trump, but the nonpartisan average gives her a 0.9 point lead.) That’s not a significant difference in a world where the average electoral error in the presidential election is 4.3 points, and is so small that it can easily be attributed to sampling error or to some methodological factor other than partisan bias. By comparison, our averages move 0.1-0.3 points daily, and we don’t recommend anyone read these shifts.

Granted, this says nothing about the companies’ motivations (we like to stick to the data here at 538) or what’s happening behind the scenes (we don’t really know). What we can say is that polling averages that are more inclusive tend to perform better in the long run than averages that use a more limited group of pollsters or an inconsistent methodology for weighting and adjustments. One of the reasons we have all the rules and methods we use is so that we have confidence in our work, even when we have doubts; that means we must stick to our rules, even when we are tempted to make an exception.

Importantly, however, this does not mean that we should consume all polls without criticism or that we should look more closely at the way pollsters generate their data. In this case, our empirical view of how our averages work reassures us that things are working as intended.

What if the elections were held today?

In short, the presidential race is close no matter how you look at it. This is both because the polls are extremely tight at the moment, and because there is still uncertainty about how the final two and a half weeks of the campaign will develop. As we wrote, the price can change quickly in recent weeks. For example, in the final few weeks of the 2016 and 2020 campaigns, polls moved toward Trump by about 4 and 2 points, respectively.

The value our forecast adds to our polling averages is that it attempts to quantify how likely such a shift is – and how inaccurate polls can be even on Election Day itself. But this also means that as the time left for people to change their minds decreases, the uncertainty of the race prediction decreases as well – and very quickly.

But because the polls are so tight right now, neither will the predictions if they don’t budge. If you look at our presidential election forecast in ‘nowcast’ mode – where we pretend the election is being held today and remove any volatility that comes from the number of days left until November 5 – Trump still has a chance of 52 out of 100 to win. .

Finally, I must remind you that a close election in terms of odds does not necessarily mean that the final outcome will be close. Historically, if the polls go wrong, either candidate could easily win with 300 Electoral College votes. While we can say that this is one of the closest elections in modern history according to the polls, we cannot guarantee that this will go down as one of the closest in terms of votes.

Mary Radcliffe contributed research.

Footnotes

*All figures in this article are as of October 18 at 4:00 PM Eastern.

**Includes all polls added to our database from October 5 to October 18 at 4:00 PM Eastern. Partisan polling is defined as polling from a pollster or sponsor that is marked as partisan in our database, including internal polling conducted for a campaign.