Polling errors and unpacking the 2024 campaign
W. Joseph Campbell is an American writer, historian, media critic, and analyst. He is professor emeritus of communication at American University in Washington, DC. Campbell has written seven solo-authored books including, most recently, Lost in a Gallup: Polling Failure in US Presidential Elections. He recently spoke with American Habits editor Ray Nothstine.
You’ve written an entire book on presidential polling, and you’ve been critical of the practice. Can you offer us any insights on polling for this volatile and unusual election we’re in?
W. Joseph Campbell: It’s clearly been a very turbulent and volatile time. And I think the polls are going to take a while to stabilize because so much has happened this summer that’s shaken up the race and polls.
Another close presidential election is likely. We are in an era of close elections and 2020 was no exception. Had there been a well-placed, well-positioned switch of 43,000 votes from Joe Biden to Donald Trump in Wisconsin, Georgia, and Arizona, we would have had a 269–269 Electoral College tie in 2020.
You can imagine what that would have done to the body politic in a year of COVID and urban rioting and everything else. There was some talk before Biden dropped out that Trump was headed for a landslide victory, but I didn’t see that at all. We have not had a landslide in American presidential politics since 1984 (and by landslide, the rule of thumb is essentially a popular vote victory of 10 percentage points or greater).
Ronald Reagan defeated Walter Mondale in 1984 by 18 points, and we’ve not come close to that since. Plus, there are so many Democrats willing and eager to vote against Trump. He motivates the Democratic base to show up and vote against him.
Vice President Harris has baggage of her own. She’s been associated with a very progressive administration and has a lot to account for, particularly being the point person on the open Southern border, which is a top issue among Americans this year.
Her role on immigration, a policy that was quite the failure, risks being a serious drawback for her in this election.
Can Trump outperform the polls again? He’s done that twice. Even in the loss, he overperformed a lot of the national polls in that race. Some polls had Biden up 10 to 12 percent going into the election. Is there anything to glean from that?
Campbell: I would not be surprised if Trump were to outperform the polls. He did it in 2016 and 2020. In 2020, some pollsters were predicting a double-digit Biden win. CNN had Biden up 12 points at the end of the campaign, which is tantamount to a landslide. It was a much closer outcome than that.
Some Trump supporters are probably reluctant to participate in polls, which could be at the root of his overperforming the polls. Many Trump supporters don’t want to take polls. They don’t trust them, particularly polls conducted by news organizations. They’re not inclined to take a survey conducted by CNN or NBC or The Washington Post, and ABC News. And this gives rise to the problem of “nonresponse bias,” as the pollsters call it, meaning there are certain segments of the population that do not participate in polls — no matter how earnestly pollsters have tried to reach them. They still rebuff those overtures.
If you think about it—and this is an observation a British journalist made about 20 years ago—talking to a polling company is an odd way to behave: Strangers ask you to give them time and personal information for nothing, so that they can profit from it. I think that’s telling. It’s also an amusing and a somewhat cynical observation. But nonetheless, I think it helps explain what pollsters have been facing in recent years. There’s just no incentive for people, particularly some Trump supporters, to participate in polls, and therefore they’ve been difficult to reach and interview.
Do you think polling helps or hurts the country? There’s some calling for the dialing back of the political rhetoric. Do you think polling feeds this divisiveness given that the numbers are so tight?
Campbell: Polling responds to a fundamental human interest in wanting to know what lies ahead or what’s going to happen. Polling, in some fashion or another, is inevitable. I doubt whether polling contributes very much to divisions in the country. I think polling results tell us something about those divisions, but polling does not create them.
It’s just astonishing how many polls are conducted nowadays. That leads to a phenomenon of cherry-picking polls. If your candidate seems to be trailing in the polls, there’s probably a poll out there that gives you some hope that your candidate’s going to prevail. Cherry-picking the polls is not uncommon, and it does lend encouragement to people supporting a candidate who is behind in the polls.
What are the limitations of polling? What can’t they tell us reliably? Obviously, response bias is one and there are a lot of people who aren’t driven by politics. They opt out or don’t trust pollsters.
Campbell: The mere fact of sampling a segment of the population, rather than surveying the population in its entirety (which is impractical and way too expensive) injects a measure of error. Pollsters take a slice of the population or the electorate they think are going to vote or say they’re likely to vote rather than just being registered to vote.
So polling error is inevitable. There are many ways that errors creep in, from minor things like the order of questions, or the way in which pollsters decide who is most likely to vote, or not. These are known as likely-voter screens and they tend to be more art than science.
Polling’s record of accuracy in anticipating election results, particularly in presidential elections, is checkered at best. It’s a flawed and uneven way to try to get a sense of what the country is thinking, how the country is moving, what the country is going to do in an upcoming election.
Even though it is flawed and prone to error, polling is better than any other technique available now. One could probably go and count yard signs in a neighborhood or in a swing precinct, and determine which candidate has more yard signs. Or one could go and count noses at a rally to see whether Trump, say, is attracting more or fewer people to his rallies. These are very ineffective and unreliable ways of getting a reading on popular opinion.
Same for shoe-leather journalism: You could send out reporters to conduct interviews with people they encounter at random on the streets. Even that is going to be distorted and open to all kinds of bias: How do you get to the “right” people and know that that is a representative sample? You really don’t know.
Others point to betting markets as a prognosticator. I don’t know much about the betting markets, frankly. I don’t think they’re a substitute, at least in the short term, for election polling. The French have an expression, “faute de mieux,” which means there is no better alternative. Unfortunately, we don’t have anything better than polling to tap and get a sense of the country’s views and its preferences in upcoming elections. We’re stuck with polling and its deficiencies, drawbacks, and sources of error.
Polls are often wrong, but not always. Maybe that’s some comfort we can take from pre-election polling.
Our first theme this year was on the role of media in a free society, and the importance of the Fourth Estate as a watchdog. They are an important entity when it comes uncovering truth and informing the public. What grade do you give to the media in covering this election?
Campbell: I would not give the news media high marks at all, especially this year. They failed miserably to pierce the White House veil thrown up around Biden’s cognitive decline and other age-related infirmities. His infirmities were evident back in 2020 or 2019, when he was gearing up to run for the presidency.
Any attempts this year to question whether Biden was up to another term in office — whether he was fit, whether he had the mental acuity to be president until 2029 — those kinds of attempts to pierce the veil were often harshly attacked by the White House and by supporters in the media. Think back to early June: The Wall Street Journal had a lengthy article about Biden seeming to slow down. It was a searching article, and it was no Biden hit piece.
What did Joe Scarborough on MSNBC say? He called it a “Trump hit job.” Other media reactions were similar, that The Wall Street Journal was just cherry-picking comments from Republican sources and not getting the full picture about Biden and how he was sharp as a tack, and so forth. Trying to do some accountability journalism about whether Biden was up for the job, or was fit for another term as president, were met with a ferocious response. The New York Times wrote an article that scoffed at those offering video evidence, clips that revealed a disoriented or confused Biden in public settings.
Another example, the Associated Press earlier this year credulously quoted Jill Biden as saying that her husband’s age is an asset, that he was up for the job and the right man for this moment in history. The AP went out with that story and didn’t push back. These credulous media accounts blew up with Biden’s performance at the debate on June 27.
That was a disaster. And the veil around Biden’s cognitive decline soon fell. Everyone who watched the debate could see that Biden had a lot of trouble in forming replies; they could see he had infirmities that were unlikely to improve. His withdrawal from the race became inevitable. Media outlets such as The New York Times and The Washington Post and others almost immediately called for his departure from the campaign. It was an almost-breathtaking swing of opinion.
I think those low levels of trust, while concerning for the news media, are well deserved.
The Gallup polling has found that trust in the media is at an all-time low. Fewer than a third of the respondents to the Gallup survey said they had a great deal or a fair amount of trust in the media’s reporting. Almost 40 percent in that poll said they had no confidence or trust at all. I think those low levels of trust, while concerning for the news media, are well deserved.
I just don’t know if there’s any concern with them about it—those numbers would send me into a panic. I’m somebody who wants to go after the truth. I want to know what the truth is in all phases regardless of my political beliefs.
Campbell: You would think that those poll results would send the signal to the news media and say “it’s time to wake up. It’s time to take these trust levels into account and respond to them.” A lot of journalists seem terrified by the notion of another Trump presidency. They are eager to jettison the normative value in American journalism of being detached and impartial observers.
It used to be the norm for journalists to be skeptical, or suspicious, of politicians of all stripes. The norm has been transmuted in recent years. Skepticism-based reporting is not how political journalism is usually practiced nowadays. It has become much more partisan, openly so. Journalists try to find a way to justify this, saying, “Well, these are unusual times, and they call for unusual tactics. Therefore, it is okay to abandon impartiality or detachment in reporting.” But once discarded, impartiality and detachment and trust are difficult to retrieve or restore.
There’s some talk, even media chatter about dialing back the divisiveness. Does the media have a responsibility to do that with their role as the Fourth Estate and their task as a watchdog. Do they have any role in dialing back the divisiveness or are they not to blame?
Campbell: I don’t see the media taking seriously the importance of dialing back the rhetoric. Again, they seem terrified by the prospect of a Trump presidency. There have been attempts to rationalize this by saying that when politicians like Trump emerge, and when their words go way beyond the pale, journalists have an obligation to shed their impartiality and respond accordingly.
It’s a solipsistic argument, but it does have a certain appeal among journalists these days — that Trump is an existential threat. I don’t see them saying, “Let’s be a little more even handed.” It’s not going to happen, not this year.
I like your book on polling. At the end you talked about some media going back and saying we covered this 2016 election wrong. We missed it and it’s on us. There was a small cadre of journalists who were owning up to some of their inadequacies in terms of covering that race. I didn’t predict a Trump win, but I remember thinking he had a shot because I used to live in Grand Rapids, Michigan, and I saw the line forming to see him that midnight before the election. It felt like his support wasn’t reflected in polls and I’d see homemade signs in rural America.
Do you think the media is doing a good job of going back and talking to these people? Folks that are concerned about the vanishing American dream or in Rust Belt states that have had their affordability and opportunity taken away from them and not just Rust Belt, but all over. Is the East Coast–dominated media giving a voice to middle America?
Campbell: I don’t see it. There was an abortive attempt in the immediate aftermath of the 2016 election to do just that, to go out and say, “Okay, how did we fail in picking up the signals?” Trump’s winning the presidency was a huge miss for the news media, as well as for pollsters. Michigan, Wisconsin, and Pennsylvania were supposedly going to be the protective or impenetrable blue wall for Hillary Clinton. Trump won each of those states narrowly and surprisingly, because most polls in those states and elsewhere did not indicate that Trump had a chance.
In fact, the Huffington Post’s poll-based prediction model said Clinton had a 98.2 percent chance of winning the election, that Trump had essentially no path to an Electoral College victory. He won all the 2016 battleground states narrowly. And with them, the presidency.
But you’re right, there was a small cadre of journalists in the aftermath of the 2016 election who thought that the news media should do a better job of talking to people, getting a better sense of what is going on in those states. The then-executive editors of The New York Times and The Washington Post both said that their newspapers should commit more effort to getting a better sense of the country between the two coasts.
That attention to middle America, and the grievances and the problems there, didn’t last long, because the purported Trump-Russia scandal sucked up a lot of media attention almost immediately after he became president in 2017.
I don’t think the media really understood Trump, and when to treat his comments literally and when to treat them figuratively. I think it was in 2016 when a reporter said Trump supporters take him seriously, but not literally. The media on the other hand, tend to take Trump literally, but not seriously. It’s an interesting distinction.
You’ve been around for a lot of presidential races. There’s a lot of discouragement out there. There’s incessant political noise, the hyperpoliticization of society and political dysfunctions. Is there anything about our elections or political trends that you find encouraging?
Campbell: That’s an interesting question. I do think there are grounds for encouragement, in a roundabout way. The hyperbolic criticisms, the incessant noise of politics, are exasperating, but they are a price for democratic governance. After all, democratic governance is messy. It’s unpredictable. It’s prone to hyperbolic claims by politicians. I think the noise and the apparent dysfunction of the American political system are totems of democratic rule.
Democratic rule offers a form of government that is superior to any other system. Maybe I’m putting too fine a point on this, but messiness is what democratic governance produces and that’s not just new to 2024, or new to Trump. We’ve had convulsive elections in the past and this is certainly not going to be the most important election we’ve ever had in the United States, despite a temptation to say so.
The most important election was probably in 1860 when Abraham Lincoln was voted into office on the eve of what became the American Civil War.
Or 1864.
Campbell: Or 1864. Yes, one can make a good case for 1864. Civil War–era elections were far more vital and far more profound tests of a democratic system than what we’re seeing now. That’s not to minimize this era. But I don’t see Trump as a threat to democracy. I don’t think the American system is going to collapse if he’s elected again. Abandoning that line of argument would be one way to begin to ratchet down the heat and the hyperbole we see these days in American politics.
