Why A Tiny Polling Error Could Lead To A Trump—Or Harris—Electoral Landslide Tomorrow

04 Nov 2024 by Sara Dorn · Forbes

Topline

Former President Donald Trump and Vice President Kamala Harris are in a dead-heat race for the White House, according to polls of the national electorate and every battleground state—a scenario that could lead to a sweep for either candidate if polling errors from previous elections repeat.

Former President Donald Trump greets supporters during a campaign rally at Lee’s Family Forum on ... [+] October 31, 2024 in Henderson, Nevada. (Photo by Chip Somodevilla/Getty Images)Getty Images

Key Facts

Polling averages by FiveThirtyEight currently show Harris leading Trump by 1.2 points nationally, while Pennsylvania is tied, Trump is up in the swing states of North Carolina, Georgia, Arizona and Nevada, and Harris is ahead in Wisconsin and Michigan.

All seven swing states have average polling margins of 2.2 points or less, meaning that if polls are off by just two to three points in favor of one candidate, they could win by an electoral college landslide.

If polls are off by as much as they were in 2016, Trump could win every battleground state except Nevada by anywhere from two to seven points more than polling averages estimate, while Harris could win Nevada by two points more, according to a Washington Post analysis.

Trump would win by anywhere from one to three points more with 2020’s polling errors in all battleground states except Georgia, which he would still claim, but by one point less than survey averages predict.

On the other hand, Harris could win most of the swing states if this year’s polling miss resembles 2012’s errors, which underestimated the Democratic candidate.

It’s unclear what direction any error could point, and while polls have underestimated Trump in the past, some pollsters predict surveys overly favor Trump this year—political analyst and pollster Evan Roth Smith told Bloomberg he expects any polling error to favor Harris, citing polls’ inability to predict potentially high turnout among Black voters, referencing 2012 polls that underestimated former President Barack Obama’s win by about six points.

In the 2020 election, most national polls accurately predicted Biden would win the popular vote, but overestimated by how much to “an unusual magnitude,” according to a study by the American Association for Public Opinion Research, which found polls in the final two weeks before the election were wrong in either direction by an average of 4.5 points for national popular vote polls and 5.1 points for state-level presidential polls.

There’s no consensus on why the polling errors occurred, the report found, since there’s no way to compare survey respondents to those who didn’t respond, because there’s no information on non-respondents since they didn’t participate, though the group theorizes that Trump supporters were less likely to participate, in part because of distrust in polling promoted by Trump.

What To Watch For

There’s a chance that pollsters overcompensate to make up for the belief that Trump supporters are under-represented.

Surprising Fact

Statistician Nate Silver predicts either candidate has a 60% chance of winning at least six of several battlegrounds, he wrote in a piece for The New York Times, in which he predicts a win for Trump based on his gut instinct, but warns “50-50 is the only responsible forecast.”

Big Number

93%. That’s the share of national polls that overstated Biden’s support in 2020, according to Pew Research.

Key Background

Pollsters have taken steps to correct errors made in the 2016 and 2020 elections, including expanding polling methods to text and mail and making statistical adjustments for under-represented groups, The New York Times reported. They also benefit from the availability of data on voting trends in previous elections, including who voted, how and party registration changes. The pandemic was also thought to be a factor in 2020 polling errors, as more Democrats followed protocols and stayed home than Republicans, making them available to answer polls, The New York Times noted.

Contra

Election-year polling isn’t always wrong—in 2008, the national polling average underestimated Obama’s win by less than one point, according to The Times. Surveys on the 2022 midterms were more accurate than any cycle since at least 1998, according to Five Thirty Eight.