bigrus

Can you trust surveys? Prediction fever will affect the 2024 elections

Harry October 26, 2024

Who will win 2024 presidential race?

This question haunts the country, as well as forecasters of all stripes – whether pollsters, academics or international bookmakers – advertise their data and intuition to voters who want to peer into a crystal ball that predicts the future Kamala Harris or Donald Trump management.

As tens of millions of Americans know, no one knows what will happen on November 5 they already published ballot papers as part of early voting, and tens of millions of people have yet to vote on their preference for who will lead the country for the next four years.

But the country anxious About the outcome amid an increasingly bitter political divide.

More than 7 in 10 adults in the US say the country’s future, economy and politics are a “significant source of stress” for them personally. report It was published this month by the American Psychological Association.

“It shouldn’t be passive. It shouldn’t be something that people aren’t speculating on because this is a really big event and this election in particular is critical,” said Imani Cherry, a professor of media and public relations at George Washington University. .

“There are a number of issues that are very, very, very important on the minds of millions and millions of people,” Cherry added.

As the unprecedented election between Harris and Trump draws to a close, experts say it is irresponsible for political observers to predict a winner.

“Polling is not predictive. It’s a snapshot in time,” said Republican pollster Robert Blizzard. “My job is not to try and predict an outcome; my job is to use the vote to help my candidate or client’s cause or issue succeed.”

The 50-50 contest is largely unaffected by major news events, including the President Joe Biden’s summer debutPermanent economic difficulties, Harris’ historic offer and two Assassination attempt on Trump.

Those who spoke with USA TODAY said any predictions that Trump or Harris would win the White House would be rife with flaws due to the rush of voting in recent days, partisan polls aimed at influencing the base and sports bettors looking to make money. and forecasting models that use firms with questionable backgrounds.

Others note a noticeable lack of quality swing state polls that often fill the final weekends of general elections, while others say it may be time for newsrooms to reconsider relying entirely on poll stories.

“People need to get off the polling train. They need to step back because people are playing with this and they’re playing it in our faces,” said Cornell Belcher, a Democratic pollster who worked on both of Barack Obama’s presidential campaigns. 2008 and 2012.

“It’s not a poll’s or pollsters’ job to tell the future, because we can’t,” he added.

Predicting the winner of the horse race

The United States hasn’t had a double-digit White House race since 1984, when GOP incumbent President Ronald Reagan cruised to victory by winning 49 of 50 states against Democratic challenger Walter Mondale.

Elections have gotten closer since then, including in the 2000 and 2016 races where Democrats won the popular vote but fell short in the Electoral College. Add to the mix a hypercompetitive media environment and a cottage industry of pollsters, pundits and even gamblers now trying to offer a preview of what will happen in 2024 to both parties and nervous undecided voters.

Chief among them is FiveThirtyEight, which has become the most popular database since 2008, with strategists, news organizations, and average viewers citing horse racing as the most powerful business on the planet. It uses a complex statistical model on potential outcomes and currently supports TrumpThe person who wins 51 out of 100 simulations.

But critics say people shouldn’t read too much into these models, given that they often use polling companies with varying degrees of accuracy.

FiveThirtyEight, for example, reminds its viewers that Harris was more likely to win weeks ago and that could change if there are “some good polls” for the incumbent vice president.

The site notes that when odds of winning hover around the 50s, it’s “little better than flipping a coin for the leading candidate,” and founder Nate Silver wrote in a New York Times op-ed that “it’s 50-50.” sole responsible estimate“During this cycle.

And the models are not always accurate; in the 2012 election, for example, then-President Obama was defying the odds to defeat his Republican rival Mitt Romney. Trump similarly beat predictions that Democratic nominee Hillary Clinton would win in 2016. Many companies also underestimated its performance in 2020.

Of concern to pundits in the final stretch of the 2024 race is how predictive models could be influenced by partisan efforts to boost a candidate’s vote averages; This can be risky in the age of misinformation, with hordes of supporters assuming their candidate should vote. You win based on poll trends or early voting data.

The New Republic reported earlier this month: A flood of polls favorable to Trump are being published In order to influence the election predictions that changed throughout the summer in his favor.

There is concern, especially for Trump supporters, that if the former president loses, Trump’s loss could spark a backlash similar to the riot at the Capitol on January 6, 2021.

GOP pollster Blizzard said it hasn’t conducted any polls for Trump’s campaign this year but is confident there are firms on both sides of the partisan divide that hide bad polls while highlighting only data showing their side doing well.

“If I don’t know the pollster, meaning they don’t have real clients, have a clear track record, or are just doing horse racing polls, I’m less likely to even care about their numbers,” he said.

Blizzard said most campaign polls are over at this point in the race, and legitimate companies are turning to using their data to make decisions about advertising and other resource allocation.

What experts find more alarming than bad actors trying to manipulate the narrative is how few people understand that a one or two point difference is a statistical tie that could go either way.

Cheery, the media and public relations professor, said the press plays an important role in this conversation because polling stories are easier to feed audiences.

He said more emphasis should be placed on the consequences of Harris and Trump’s policy differences.

Instead of making predictions, Cheery said he focused his students on grassroots conversations among voters about the risks of the White House race and ways people can get involved in the democratic process, over entertainment-style coverage of who will win.

“I don’t think polls have a place in this political discourse, but they’re often shortsighted,” he said. “Especially when you’re talking about the many variables that will come into play in people’s choices, and for people who say ‘all the polls said (Trump) was going to win, but he didn’t; it’s rigged.'” .”

‘A questionable and dangerous job’

Nearly 34 million Americans have voted in early 2024, with Democrats holding a slight advantage. According to the University of Florida Election Lab.tracking the numbers daily.

For example, it shows that about 41% of votes cast came from registered Democrats, while 35% came from registered Republicans. But those who have been gauging the mindset of American voters for years say there are dangers in diving too deep into early voting data.

For starters, most early voting data is based on states where people register by party. Another caveat is that the key comparison is the 2020 election, when ballot access was first expanded in response to the global COVID-19 pandemic.

Cathy J. Cohen, a professor of political science at the University of Chicago, has been surveying Black and Hispanic voters for months as part of the research. GenForward, a youth project at the university.

He said the average American is unaware of the limitations or nuances of survey data, including sample size, demographic representation and how survey questions are worded and ranked. All of these affect the outcome of each poll, Cohen said.

“There’s a huge gap, a huge gap, between the people who are responding to the survey and the people who are putting on their jacket, getting in the car, getting on the bus, waiting in line and making sure they’re registered to vote,” Cohen said.

Political observers need to focus on trends in the election cycle rather than paying attention to possible outcomes. But that hasn’t stopped other groups and organizations from filling the void, as more established firms appear to be stepping back in 2024.

Many offshore betting markets, e.g. Polymarket, the world’s largest crypto trading platformIt has been cited by Trump and his allies as predicting that the Republican will win a second non-consecutive term as president. A French merchant It was reported that he bet a total of $28 million In four different accounts of the GOP candidate’s return to power.

BetOnline.ag spokesman Joshua Barton said the popularity of betting on everything from who will win in the U.S. election to turnout levels has skyrocketed in the past decade.

“In terms of the amount of betting, it’s going to eclipse what the Super Bowl did because there are so many big bettors coming in looking to get a stake on who’s going to win this election,” Barton said in an interview.

He said some might not bet on anything else for four years, but participants wanted to take action on the outcome.

It is still unclear how much these and other measures – like the stock marketLevels ahead of the presidential election could predict who might emerge victorious between Trump and Harris. But that concerns Cohen and other academic experts who say polls should be about determining how Americans think rather than fortune-telling or making money.

“These can be used to give a sense of prediction, but I would be cautious about weighting too heavily our perception of what will happen based on polls,” he said.

“When we’re talking about human behavior, it’s questionable and dangerous business.”