Polling and missing the hard to reach Trump voter three times
In early September, American Habits featured an interview with W. Joseph Campbell, discussing polling errors and insights on the presidential campaign as part of our elections issue. With Donald Trump now president-elect, revisiting key themes from that conversation helps explain the dynamics that shaped the 2024 race.
Campbell, an accomplished American writer, historian, media critic, and professor emeritus at American University, has authored seven books, including “Lost in a Gallup: Polling Failure in US Presidential Elections.” He spoke with American Habits editor Ray Nothstine just over a week after the election.
What are your general thoughts on the polls now that Donald Trump won? There were several that tended to be perceived as more right-leaning and ended up being very accurate, like AtlasIntel, Trafalgar, and Rasmussen. Most polls were close or tied, Harris up one, Harris up two, Trump up one, Trump up two. Some had Harris winning, but you had some real outliers. The Iowa poll had a lead for Harris, and then I saw a Wisconsin poll that had Harris up seven near the end. Is it safe to say that Trump out performed the polls again?
W. Joseph Campbell: Absolutely, he did overperform his polls yet again for the third successive election, but it was not as dramatic as in 2020 and 2016. In 2020, the polls were way off; this time the polls were a little closer to getting the correct outcome. Overall this year, they persistently and consistently showed a close race. And it was a close race, with a modest Trump win.
His popular vote as of this morning was 50.2% of the popular vote to Harris’ 48.2%. That’s a two-percentage point difference, and that’s fairly small. It’s a smaller margin than the 2020 election or any of the last four presidential elections.
You mentioned AtlasIntel. They turned in probably the best and most accurate performance of any of the pollsters. This is a Brazilian-based company and not many people know much about them. They don’t say a lot about their methodology. They claim its proprietary. And they said that they did a great job polling in 2020. I don’t remember AtlasIntel in 2020, but I certainly noticed them this time. They correctly estimated the outcomes in all seven swing states, by margins that were not dramatically different from the final result.
That’s an impressive performance by a little-known pollster who seemed to come almost out of nowhere to turn in the single best polling performance of this presidential election.
Can you offer any insights on the difference between the internal polling done by campaigns and the public polling put out by the networks and other more notable polling groups?
Trump commented here in North Carolina during a campaign stop right before the election saying, “I’m not supposed to say this, but it’s not really close.” He was likely referencing his internal polling. He ended up being fairly accurate, particularly winning the so-called popular vote, which a lot of people didn’t expect.
Campbell: The internal polls are often discussed, but usually not much is known about them. Candidates tend to keep them closely guarded and release them, sometimes, when the results are favorable for them. It’s hard to make a good assessment about how accurate internal polls are. In the run-up to the election, you did hear a number of news organizations saying, ‘Well, the internals must be pretty good.’
There’s a lot of speculation about internal polling. In the past, candidates have paid a lot of money on internal polling. They may have performed better than expected, but it’s hard to know without having their results fully disclosed.
One poll that shocked a lot of people, and you alluded to this earlier, Ray, was the Selzer poll in Iowa, which was done for The Des Moines Register. This was released by the gold-standard pollster, Ann Selzer, just three days before the election. It showed that Harris had opened a three-point lead in what is fairly deep red Iowa.
The implications were that if Harris had a three-point lead in Iowa, then she was going to do very well elsewhere, certainly in the upper Great Lakes states, clearing a path to the presidency. If Harris is ahead or close in Iowa, that signals real trouble for Trump. That’s one of the reasons I was uncertain as to how this election was going to go when Election Day finally rolled around. I thought it was going to be even closer than it turned out to be. But that Selzer poll — it’s hard to understate how much impact it had on the endgame of this election.
Selzer had a terrific reputation, a stellar reputation for accuracy, and for identifying trends in her state before anybody else was picking them up. It turned out that she was dramatically wrong this time. Her poll was 16 percentage points off, and that is quite a misfire. That’s a dramatic miss for somebody who is a seasoned veteran and a highly respected pollster.
You referenced the betting markets in the last interview we did, and I didn’t pay attention to them much last time.
I did look at them in the last week of the race. They showed Trump winning an Electoral College victory if you averaged out the different betting markets.
Of course, there were some on the left saying they were manipulated but they ended up being accurate. I think in some ways they were more accurate than the polls in terms of signaling who was going to win. Do you have any thoughts on the betting markets in general?
Campbell: They certainly are another option for people to keep in mind and to follow. I think the betting markets are another piece of evidence in terms of understanding how elections are going to turn out. Now, whether they’re going to be the gold standard for prognostication moving forward, I’m not sure. I’m skeptical. There’s still going to be a place for pre-election polls, of which we had so many this time.
It was almost overwhelming, the number of polls we had, and not all of them of course were giving us a clear understanding of what was going on. As I said earlier, the polls, collectively speaking, did underestimate Trump’s support, including in the seven swing states where the election turned, and that’s a real concern for pollsters. It’s not like the polling industry is broken, but it needs repair in some important areas, such as being able to tap into methodologies and techniques that are going to allow pollsters ability to reach voters who are difficult to reach, as were some of Trump’s voters.
But I don’t think polls are going to be supplanted by or replaced by the betting markets.
What do you think doomed the Harris campaign, there’s a lot there, but just the big picture, what do you see as somebody who’s just followed presidential politics and politics in general for a long time?
Campbell: I can count at least 14 different reasons why Harris lost, starting with her anointed status as a candidate. She did not win a single vote before she became the nominee of the Democratic Party. That left a lot of people feeling she was anointed to replace President Joe Biden, who was not going to be up for a general election campaign. The whole process of Harris’ selection was deeply undemocratic, and it left her with what commentators have said was a “stature gap,” something from which she never quite recovered.
Also, Harris never really shed the reputation for speaking in word salads. She spoke confusedly quite a bit, and frequently confirmed a reputation for speaking in jumbles of words that didn’t make a whole lot of sense, that were often contradictory or repetitive. These “word salads” were a lingering problem for her.
There was the question of authenticity, too.
Harris did not look comfortable or natural opening a can of beer with one of the late-night TV hosts, nor did she sound very authentic when she was talking about having a Glock at home to protect her house. An air of inauthenticity clung to her campaign. That was compounded by her flip-flopping on a lot of different factors, including fracking in Pennsylvania, which was a key state.
She said, indirectly, that she was for fracking after having been against it for a long time. Her message wasn’t clear, it wasn’t candid, and those two problems, message clarity and message candor, were problems for her campaign.
Another shortcoming that I do want to mention was that her campaign was infused by a lack of specifics. This became an issue quickly, the lack of specific policy details and the tendency to speak in platitudes.
This reminded me so much of the unsuccessful campaign that Thomas E. Dewey ran in 1948 against President Harry Truman. That was the famous “Dewey defeats Harry Truman election” in which Truman vigorously campaigned and was on the stump all the time, crisscrossing the country by train, whereas Dewey was running a lethargic, above-the-fray type campaign.
A glidepath campaign is not a very good strategy, then or now.
Finally, I would say that the selection of Tim Walz as Harris’ running mate didn’t help her at all. He described himself as a “knucklehead” at one point in the campaign. I think he brought embarrassment to her campaign and probably didn’t add much in the way of votes. He certainly didn’t help her in the swing states.
Is there anything we can draw from this new electoral map? It’s the biggest Republican presidential victory since 1988. What do we take away from it?
Campbell: It’s an interesting map, that’s for sure. Trump’s coattails seem to have been strong in some places. He got the Ohio Senate candidate Bernie Moreno across the line, and it looks like Dave McCormick in Pennsylvania benefited from Trump’s coattails, too. But I’m just not sure that there’s a realignment in place that’s dramatically transformed the political map.
It might happen, but realignment must be confirmed over time, election after election. Trump’s last election was this month. He’s not going to be able to go out there again and run again for reelection. Whether JD Vance will be the Republican presidential candidate in 2028 remains to be seen, of course. For the kind of realignment that many people are talking about, it must be confirmed, it needs to be solidified, and it must be built upon, election after election.
The classic example of that process, of course, is Franklin Roosevelt, who won in 1932, ’36, ’40, and ’44. He won four elections in a row and solidified a realignment of the Democratic Party, which was buttressed by different coalitions, including white working class, minorities, and some well-to-do people in the Northeast. It took time to build that coalition, and it lasted quite a while. I just don’t see that there’s the same sort of formula at work here, one that will lead to a quasi-permanent Republican realignment that Trump spearheaded in 2024.
I think 2024 is more likely a one-off election in terms of how coalitions assemble themselves, rather than a realignment that’s going to be place in for the next 20 years or so.
Interesting. Trump’s such a dominating figure by himself, it’s hard to factor too much going forward for either party.
Campbell: Vance probably has the best chance of continuing the Trump legacy, if you will. Vance presented himself well, especially in taking on network news types on Sunday talk shows. He held his own, and then some. He was aggressive and well-spoken, and far more articulate than Trump. He came out of the election with the best favorability ratings of any of the four top-of-the-ticket candidates of the two major parties, according to exit polls.
