The results on election night indicated a success for polling even before former President Donald Trump was declared president-elect. 

In both battleground states and the popular vote, pre-election polls throughout the last few weeks of the campaign consistently predicted a close contest between Trump and Vice President Kamala Harris. Those polls were largely correct, as Trump’s winning margins in swing states were narrow. 

Pre-election polls that had on average given Harris a 1-point lead missed by roughly 2.5 points. Trump’s sweeping of all seven swing states fell well within the range of possible outcomes that the polls had indicated, and that people had forecasted. Elections analyst Nate Silver’s model had Trump winning all seven swing states as the most likely scenario.  

The polls in this election weren’t perfect, but they never are. Polls improved this cycle compared to 2016 and 2020, but for three elections in a row, they’ve had the same problem: underestimation of Trump’s support.  

In 2016, polling averages had Trump at 41.8%, he finished with 45.9%. In 2020, his predicted vote share was 43.4%, and he gained 46.9% of the vote. Polling averages estimated Trump’s support at 46.8%, and Trump earned 50.1% of the vote, although that number may drop a few decimal points as more votes are counted on the West Coast.  

While none of these errors were massive, they were systematically off by 3 to 4 percentage points in the same direction each time, calling into question the overall credibility of the polling industry.  

In fact, while the polls had a good but not great election cycle, certain individuals with near-mythic election predicting status faltered. One of these individuals was American University professor Allan Lichtman.  

Allan Lichtman had successfully predicted every presidential election since 1984, with the exception of 2000. Lichtman maintains his prediction of Al Gore was correct, and he cites the U.S. Supreme Court’s halting of Florida’s recount as evidence. “What a [United States Commission on Civil Rights] report showed is that overwhelmingly, disproportionately, Florida tossed out votes cast by 95 percent African American voters as compared to overwhelmingly white voters,” Litchman said. 

His predictions rely on 13 keys, each key being a long-term factor like economic strength that may favor one candidate or not. This year he found eight of his 13 keys favored Harris, and issued his prediction of a Harris win. 

For the first time in 40 years, his interpretation of the keys was false. In a post-election analysis, Lichtman cited the influence of Elon Musk as a confounding variable he did not anticipate: “Disinformation has exploded to an unprecedented degree,” Lichtman said to Chris Cuomo of NewsNation. “We’ve seen something brand new this time – billionaire Elon Musk putting his thumb on the scales.” 

Pollster Ann Selzer could have been considered the oracle of Iowa for more than two decades. Her polls of Iowa hadn’t falsely predicted the winner of a presidential race in 20 years and that miss was only 3.7%.  

Iowa was a Trump +8 state in 2016, and a Trump +9 state in 2020. So, when Selzer’s poll, released shortly before the 2024 election, showed Harris ahead by 3 points in Iowa, it was a political earthquake. Even a large miss for Selzer would indicate a great night for Harris. Instead, Trump won Iowa 55.8% to 42.6%. Selzer was off by a whopping 16 points, the most significant miss of her career. 

The 2024 election taught us that polling averages are reliable tools, and individual predictions held up as unique and extraordinary may not be gospel.