By Shawn Cochran 
By Shawn Cochran
Photo Credit: Gage Skidmore/Flickr
On November 5, former President Donald Trump defeated Vice President Kamala Harris in the 2024 Presidential Election, becoming only the second president to win a non-consecutive term. While this sent shockwaves throughout the nation, Trump’s victory stunned Americans for a different reason: The Economist, FiveThirtyEight, and Silver Bulletin–all leading election forecasting models–were wrong.
How They Work
Forecasting models have been used for nearly a century and are renowned for their comprehensive methodologies. Each uses a variety of polling, economic, and demographic trends to predict a winner. On Election Day, all leading models predicted Vice President Harris would win. The Economist, which utilized polling data and economic indicators, gave Harris a 56% chance of victory. FiveThirtyEight, meanwhile, used a polling aggregate and demographic trends to run 1000 simulations, determining the Vice President’s likelihood of victory was 50.3%. Silver Bulletin ran 80,000 simulations using a comparable methodology, assigning Harris a 50.015% chance of winning.
However, Trump ultimately swept all seven swing states, winning the Electoral College and the popular vote. So why were all three forecasting models incorrect despite their rigorous methodologies? Amid increasing political polarization, models may have overestimated fundamental metrics while failing to account for three defining events of the 2024 Presidential Election: Democratic infighting, voting shifts, and political unrest.
Democratic Infighting
Following his controversial presidential debate performance against former President Trump on June 27, President Biden faced calls to step down as his party’s nominee from congressional Democrats, spearheaded by Speaker Emerita Nancy Pelosi and Senate Majority Leader Chuck Schumer. After insisting he would remain the nominee, the President eventually dropped out and endorsed Harris as his replacement on July 21.
After President Biden’s withdrawal, prominent congressional Democrats and Democratic National Committee officials endorsed Vice President Harris including Association of State Democratic Committees (ASDC) President Ken Martin who declared “we have to unify quickly.” However, many moderate Democrats criticized Harris’ policies, including Representatives Marie Gluesenkamp Perez, Jared Golden, and Mary Peltola, as well as Senators Joe Manchin, Kyrsten Sinema, and Jon Tester. These members, who all faced competitive reelection campaigns, argued Harris’ support for restrictions on oil and abolishing the filibuster would alienate swing voters. Still, the Vice President won the nomination with 4,563 votes, about 99% of the delegates. While Democrats hoped her nomination would illustrate unity, their prior criticism of President Biden’s debate performance and disagreements over key policy positions damaged Harris’ perception among independent voters.
Forecasting models give presidential candidates a “convention bounce” to their likelihood of victory in the weeks following their respective conventions to emulate voter enthusiasm. After the 2024 DNC, however, Vice President Harris’ polling numbers did not noticeably increase. Still, forecasting models granted her a convention bounce, incorrectly suggesting the Vice President’s chance of victory had increased amid a contentious primary, party infighting, and stagnant polling.
Voting Shifts
Nevertheless, Vice President Harris enjoyed a national polling advantage throughout the campaign, though the polls narrowed as the election approached. Two weeks before Election Day, the New York Times, Ipsos-Reuters, and Daily Mail found President Trump tied with or leading Harris. Once early voting began, Republicans outperformed expectations; whereas Democrats led the early vote by 14% in 2020, their advantage was reduced to 3% in 2024.
Furthermore, returns on Election Day showed notable gains for President Trump within key demographics compared to 2020, including 13% among Gen Z and 10% among Latinos. The President-elect ultimately received 3.08 million more votes than his 2020 margin, whereas Vice President Harris received 6.27 million fewer votes than President Biden’s 2020 margin.
Political scientists use historical trends, voter registration data, and exit polling to gauge these kinds of voting shifts before Election Day. Despite the importance of these shifts, all three forecasting models may have underestimated the role and magnitude of demographic shifts in the 2024 Presidential Election, producing incorrect predictions for Vice President Harris.
Political Unrest
Political unrest also shook the 2024 campaign. Namely, former President Trump survived an assassination attempt during a campaign rally in Butler, Pennsylvania, on July 13–the most significant attack on an American president since President Ronald Reagan in 1981. Trump was shot in the ear by Thomas Matthew Crooks, who also struck three attendees, killing one. Doug Mills of the New York Times photographed the bullet hitting Trump, and Evan Vucci of the Associated Press photographed the bloodied Republican nominee raising his fist moments later while chanting “Fight! Fight! Fight!” against the backdrop of the American flag. The images went viral on social media and prompted many Republicans to purchase t-shirts depicting the attack and wear ear bandages as an act of solidarity with their nominee.
Two months later, on September 15, former President Trump survived a second assassination attempt in West Palm Beach, Florida. While the Republican nominee was golfing, a Secret Service agent opened fire at Ryan Wesley Routh, who was hiding in shrubbery with a rifle aimed at Trump’s security detail. The former President escaped uninjured as Routh fled before being taken into custody half an hour later. Upon his arrest, the Department of Justice concluded Routh intended to assassinate President Trump which was supported by inflammatory social media posts and a manifesto stating “this was an assassination attempt.” The Republican nominee quickly accused Democrats of “hatred, abuse, and distrust,” alleging divisive political rhetoric had prompted the attempted assassinations.
Meanwhile, the Uncommitted National Movement diminished Vice President Harris’ support among core Democratic constituencies, including Gen Z and Arab American voters. The movement, started by progressives, criticized the Israel-Hamas war and demanded an immediate ceasefire in Gaza and an arms embargo on Israel. After encouraging Democrats to vote “uncommitted” throughout the primaries as a pressure tactic, the Uncommitted National Movement garnered 706,591 votes, winning 37 delegates to the Democratic National Convention. Consequently, after failing to meet with the Vice President and being denied a speaking slot at the DNC, the Uncommitted Movement urged its supporters to withhold their support from Harris.
Overall, the assassination attempt on former President Trump and the Uncommitted National Movement created a landscape where political forecasting was exceedingly challenging; well-established forecasting methodologies struggled to adjust to these unprecedented events.
Implications
Overlooking the defining events of the 2024 campaign season may have been a key factor leading to inaccurate predictions. The historic withdrawal of President Joe Biden, major demographic shifts in voting, two assassination attempts against the Republican nominee, and the Uncommitted National Movement created significant barriers to accurate political forecasting. Thus, the leading models incorrectly predicted the 2024 Presidential Election for Vice President Kamala Harris. Looking toward 2028, political scientists, analysts, and pundits may benefit by reflecting on these missteps to refine their models by accounting for the new style of presidential elections and placing less emphasis on historical trends.