Surprises, outliers, oddities: What to anticipate in the campaign’s closing days 

Much can change in the closing days of a U.S. presidential race.  

Donald Trump, for example, received enough late-campaign support from previously undecided voters in battleground states in 2016 to lift him to the presidency. As another example, the pre-election disclosure in 2000 of a drunken-driving arrest years before may have cost George W. Bush a popular vote victory.

Drawing on those and other lessons of presidential elections past, here are some reminders about what could command attention in the campaign’s final days. This is not to say these prospective developments will all materialize, of course. But if the past is any guide, they might. Some already have. 

• Speculation about an October or November surprise, which is an unanticipated development that bursts across the political landscape with potentially shattering effect. Bush’s drunk-driving case, made public by a television reporter in Maine five days before election, certainly was such a surprise. Bush adviser Karl Rove figured the disclosure kept Bush from carrying states that would have ensured him clear popular vote and Electoral College victories. As it was, the 2000 election pivoted on a prolonged dispute about who carried Florida, a dispute settled in Bush’s favor by the U.S. Supreme Court. 

A more recent October surprise was the FBI’s announcement, just 11 days before the 2016 election, that it was reopening an inquiry into the private email server Hillary Clinton used during her time as U.S. secretary of State. Clinton later said the FBI’s renewed investigation was “the determining factor” in her loss to Trump. 

• Persistent discussion about the campaign’s great known unknown, which is whether pollsters have effectively modified their methodologies so that surveys this year accurately measure Trump’s support. Polls in 2016 and 2020 underestimated his backing, resulting in back-to-back polling embarrassments. 

Some pollsters thought adjustments made after the 2016 election would serve them well in 2020. That wasn’t the case. The collective performance of the polls four years ago was the worst in 40 years. If polls underperform as they did in 2020, Trump could be well-positioned to win reelection.  

Polling’s known-unknown won’t be resolved till after the election. Even so, it has been a leading topic of discussion among pollsters and pundits for months. 

• Conjecture about an unanticipated state-level outcome, such as Trump winning Virginia. Or Minnesota. Or Texas falling to the Democrats. Every presidential election cycle seems to produce such speculation, and it often goes unfulfilled.

In 2012, for example, the Republican presidential nominee, Mitt Romney, sensed a chance to carry Pennsylvania and took his campaign to suburban Philadelphia two days before the election. Romney’s hopes proved illusory. He lost the national vote to Barack Obama by almost 4 percentage points; he lost Pennsylvania by 5 points. 

• Scattered talk, however improbable, about a landslide — even though polls have consistently signaled a tight race between Trump and Vice President Kamala Harris, and even though the last landslide in a presidential election was 40 years ago.  

Filmmaker Michael Moore wrote in early October about his “gut” feeling that Kamala Harris will defeat Trump by 13 percentage points. The “only way to guarantee” Trump’s “permanent removal from the public eye,” Moore added, is in a crushing landslide. “We should settle for nothing less.”  

About the same time, Dick Morris, who served as adviser to former President Bill Clinton, speculated Trump will win in a rout. “Landslides take time to build and they are by no means evident weeks before an election,” Morris wrote. “And so it is with Trump v. Harris.” 

A plausible analog to the 2024 race, Morris wrote, was the surprise outcome in 1980, when Republican Ronald Reagan defeated President Jimmy Carter in a rout that no pollster saw coming. In fact, most late-campaign polls in 1980 anticipated a close election. Reagan won by nearly 10 points.  

• Outlier polls that report findings so unlikely as to invite derision. An outlier poll offers a puzzling, off-target result arising from flaws such as sampling error or built-in biases in a pollster’s methodology, which are known as house effects. A memorable outlier was the Washington Post/ABC News late-campaign poll in Wisconsin in 2020, which reported Joe Biden was 17 points ahead in the state. Biden carried Wisconsin by less than 1 point. 

“Once in a blue moon, you see a poll that makes you blink twice to make sure you’re not seeing things,” a polling analyst wrote, adding the Post/ABC survey “was just such a poll.” 

• Pretend polls that make news, like the “cookie poll” conducted by a bakery in Hatboro, Pa. Such undertakings suggest a gimmicky, entertaining side to polling but are by no means scientific. Or reliable.  

The “cookie poll” was begun quietly enough during the 2008 election campaign. In 2020, Trump’s son Eric visited the bakery, which brought unprecedented attention to the poll. During that campaign, the bakery sold nearly 32,000 cookies bearing Donald Trump’s name and 5,750 with Biden’s, according to the Philadelphia Inquirer. Biden won Pennsylvania by fewer than 2 percentage points, confirming that the poll offered little more than quirky amusement value. 

“I tell everybody this is not scientific,” the bakery’s owner, Kathleen Lochel, was quoted by the Inquirer as saying. “This is a cookie with someone’s name on it.” 

As Election Day nears, we’ll learn soon enough for whom the cookie crumbles in 2024. 

W. Joseph Campbell is a professor emeritus at American University in Washington, D.C., and the author of seven books including, most recently, “Lost in a Gallup: Polling Failure in U.S. Presidential Elections.” His handle on X is @wjosephcampbell.