ANALYSIS

Are the polls even reliable? Experts examine election polling predictions from 1980 to 2024

After Biden's dismal debate against Trump, pundits turn to polls to learn America's future; what do they say?

By Matthew Rozsa

Staff Writer

Published July 13, 2024 5:45AM (EDT)

Joe Biden (Photo illustration by Salon/Getty Images)
Joe Biden (Photo illustration by Salon/Getty Images)

During the first debate of the 2024 election cycle, President Biden infamously flubbed his words, trailed off mid-sentence, stared off into space and otherwise acted every bit like a mentally incompetent 81-year-old man. With Biden's fellow Democrats determined to save democracy from an election-denying Republican — former President Trump — who was recently empowered by the Supreme Court to be a dictator, their experts are now turning to one place for information about the future: Polls

"The fact that [Biden's] behind in all of them, albeit by a small amount, reaffirm that it isn’t a complete toss up"

Yet can they be trusted?

If the polls are to be believed, Trump had held an almost uninterrupted lead over Biden for more than nine months before the debate; in its aftermath, both of the highest quality polls (CNN/SSRS and Wall Street Journal, New York Times/Siena College) show Trump ahead by six points among likely voters and with solid leads in the states key to winning the Electoral College. Nevertheless, in the same New York Times article where he issues a glum forecast for Biden based on polls, chief political analyst Nate Cohn acknowledges the polls from the 2020 election were "terrible," producing "the worst year for the polls in a competitive presidential race since 1980." One could fairly speculate that, since those polls had been skewed in Biden's favor and the pollsters no doubt compensated by increasing their share of Republicans, it is possible the polls are now inaccurately skewed to Trump.

Then again, the polls could also be wrong for myriad other reasons: Perhaps they under sample or oversample disengaged voters, or undecided voters, or some other under-appreciated bloc. Maybe — as statistician Nate Silver incorporates into his mathematical model predicting election outcomes — analysts should focus less on day-to-day survey results and more on how that data intersects with so-called "fundamentals," or variables "such as economic conditions, state partisanship and incumbency." (Unlike Cohn, who is quite bullish for Trump, Silver's model currently only gives Trump a 51-to-49 chance of winning.)

The bottom line is that, even though all sides rely on polls to understand political reality, there is no surefire way to know that that reality is actually showing up correctly in those polls. To understand why that is the case and how ordinary voters can best inform themselves, Salon reached out to J. Miles Coleman, the associate editor of Sabato’s Crystal Ball, the University of Virginia Center for Politics’ authoritative, nonpartisan newsletter on American campaigns and elections. Salon asked Coleman about the accuracy of polls going back to the 1980 election, when President Jimmy Carter lost in an upset to former California governor Ronald Reagan in an event that transformed political history. Have they gotten any more reliable in the subsequent 44 years?

"Maybe not the most satisfying answer, but it is a bit of a mixed bag," Coleman said, referring Salon to an archive of Gallup polls going back to 1936. "In 1980, for instance, Carter was still quite competitive with Reagan, and was generally leading — Reagan was helped by a strong performance in the only debate of that year and sort of surged late." While the polls accurately showed Reagan with a comfortable lead for the duration of the 1984 election cycle "in 1988, George H. W. Bush was famously down 54%-37%" as of July (the same month which 36 years later shows Biden behind Trump by a much smaller margin) "but picked up steam later in the ​campaign." By the time of the 2020 election, things had not gotten much better. "According to FiveThirtyEight's aggregates, Biden had a roughly 51%-41% lead over Trump from May 2020 onwards. Biden got 51% of the vote, but Trump beat expectations, ultimately getting 47%. Another case that I often cite is 2008: it really wasn't until September (when the stock market crashed) that Obama really started to pull away from McCain—with that, Bush's disapproval numbers were going to be hard for any Republican to overcome."

If there is any consistent theme from recent political history, it is that third-party candidates like this year's Robert Kennedy Jr. always walk home disappointed. It is not simply that America has not elected a third-party candidate as president since Abraham Lincoln was elected as the first Republican president in 1860; more significantly, the people who flirt with third-party candidates in polls tend to return to one of the two major parties by Election Day.

"In 1980 and 1992, John Anderson and Ross Perot, respectively, were competitive with, or even leading, the major party nominees," Coleman said. "But in the end, neither actually carried any states."


Want more health and science stories in your inbox? Subscribe to Salon's weekly newsletter Lab Notes.


"Because these voters aren't actively switching over to Trump, the goal of the Biden campaign will be to bring these voters home."

David Barker, director of the Center for Congressional and Presidential Studies and a professor of government at American University, reminded Salon that in the 1992 election, Perot's third-party candidacy was performing so well that Democratic nominee Bill Clinton — then Arkansas governor but, after the election was over, America's future president — was placing third. This helps illustrate why he has a healthy skepticism of polls.

"Traditionally, the polls are irrelevant until after both conventions," Barker said. "Typically, the polls one wants to pay the most attention to are the ones that come out right after Labor Day." While this may augur well for Biden, Barker added that one reason polls have little predictive power until the national conventions is that voters are usually unfamiliar with the two nominees... which is not a factor in 2024.

"In this case, we essentially have two incumbents, both of whose name recognition is 100%, who have been in the public eye forever and everyone knows them," Barker said. "So in this case, the polls may be more like those we would traditionally see in September."

Even so, Barker views Biden as merely an underdog at this point in the election cycle, not as a doomed candidate.

"They are all well within the margin of error," Barker said. "So it could go either way. The fact that he’s behind in all of them, albeit by a small amount, reaffirm that it isn’t a complete toss up (otherwise, in some he’d be up and in some he’d be down), but it’s impossible to predict turnout or how the third party candidates will ultimately do, etc., so really anything could still happen." Biden's biggest problem is not so much that he is behind in the polls but that, because of the Electoral College, the bar for him to win is higher than for a Republican.

"One problem for Biden, though, is that he needs to win the popular vote by a fairly significant amount to win the electoral college—as we’ve seen," Barker said. "He won the popular vote by like 4 or 4.5 points last time, and still nearly lost the [Electoral College]. So if we take at face value his standing right now as being about 2-3 points behind, he really has about 6 or 7 points to make up. That’s quite a bit, and we know that he isn’t going to change the dynamic with some future dynamite debate performance or something."

Coleman also told Salon that Biden "to say the least has his work cut out for him." Although Biden is still "in the game" in the states that he absolutely must win like Pennsylvania, Michigan and Wisconsin, he is farther behind in other states that he flipped in 2020 like Arizona and Georgia. These dismal showings in the Electoral College are reflected in Biden's weaknesses with key Democratic subgroups.

"Even before the 2024 election began in earnest, Biden was showing some weakness with core Democratic subgroups, mainly younger voters and minorities," Coleman said. "However, in some surveys, these groups have been 'parked' in the 'Undecided' or third-party categories. So because these voters aren't actively switching over to Trump, the goal of the Biden campaign will be to bring these voters home." Elaborating on this point, Coleman said that "the ground game, and general campaign operations, of either side could matter. For instance, Trump usually tells his voters that they should wait until Election Day to vote (because, you know, early voting = fraud!), while Democrats are, generally, more likely to try to run up the score in the early/absentee vote."

Barker also fingered turnout as "really the thing" that, along with economic indicators like employment and inflation, will determine who wins the election.

"Jobs need to not nosedive and inflation needs to continue to abate (a nice drop in gas prices would be really helpful to him, as would an interest rate cut by the Fed)," Barker said. "The Gaza situation really needs to get resolved too. Otherwise, Biden can’t make any more significant gaffes on TV (a tall order), Trump needs to make a couple (which he surely will)."

Read more from Matthew Rozsa on American history:


By Matthew Rozsa

Matthew Rozsa is a staff writer at Salon. He received a Master's Degree in History from Rutgers-Newark in 2012 and was awarded a science journalism fellowship from the Metcalf Institute in 2022.

MORE FROM Matthew Rozsa


Related Topics ------------------------------------------

2024 Election Analysis Donald Trump Joe Biden Poll Science Polling Polls Reporting