A few days before November’s presidential election, Ann Selzer, then considered the gold standard of polling in politics, released a highly anticipated poll for the Des Moines Register on where the presidential race stood in Iowa.
While the poll came from a state not expected to have a decisive role in the outcome of the election — Iowa is reliably red — its findings were expected to show clues as to how certain demographics would break nationwide.
Selzer’s results sent shock waves through politics. Former Vice President Kamala Harris was up 3 percentage points over President Donald Trump in a state everyone considered to be a walk for Trump.
The poll results immediately had reporters scrambling, with gushing Harris-November surprise storylines in the New York Times, Washington Post, Wall Street Journal, and Politico giving hope where no hope should have been given to Harris supporters looking for something to validate their wish that she would win.
If the gold standard of polling showed Trump was down in ruby-red Iowa, a state he won over then-candidate Joe Biden in 2020 by a whopping 8 points, then the door was open to the possibility that the working class in Pennsylvania, Michigan, and neighboring Wisconsin felt the same way Iowans did.
What was barely noticed that day, exactly one hour before her poll came out, was an Emerson/RealClearDefense poll showing Trump up 10 points over Harris in Iowa. It was a result that ended up being much closer to the 13.2 points Trump eventually won by.
The question is: Why did the reporters and cable news organizations that drive the conversation in U.S. politics give so much oxygen to a poll that was clearly an outlier and barely any coverage to the polling averaging done by RealClearPolitics, whose accuracy has been impeccable for the last two decades?
The answer is that, in the context of politics and news, there is an information war, and over the course of time, polling has become part of that war.
At the center of that war is RealClearPolitics, thanks in part to a New York Times article that accused it of failing to “filter out low-quality pollsters.” The story, written under the headline “How the Right Thinks Trump Is Running Away With the Race,” said by failing to doctor its polling averages, RealClearPolitics was now in the category of partisan polling and was undermining faith in the entire system.
Since its launch in 2002 as a polling aggregator, RealClearPolitics has always been a reliable tool for reporters, campaign professionals, and the general public to access the most comprehensive polling.
The system of averaging all the polls, created by John McIntyre 25 years ago, limited the ability of media outlets to misrepresent where races were. McIntyre and co-founder Tom Bevan started averaging polls after the 2000 presidential election, and the system worked very well until the 2016 election results stunned everyone, and the media started to blame the polls for being wrong.
Interestingly, if you had followed RealClearPolitics’s polling data that year, it showed Clinton with a slight lead within the margin of error. So, no one should have been surprised that 2016 was a very close election. The New York Times polling that year, on the same day, showed Clinton winning Pennsylvania by over 4 percentage points, North Carolina and Florida by over 2 points, and barely winning Ohio.
Its Upshot Forecast modeling, which the New York Times based on its aggregation of polling, showed Clinton with an 84% chance of winning.
One reason RealClearPolitics‘s polling average is accurate is that it doesn’t just aggregate the mainstream legacy polls or left-leaning university polls. It includes a wide variety of polling to get the most accurate reading.
By 2018, the media started using polling as a political weapon. By 2020, there were polls that seemingly gamed RealClearPolitics‘s average. A perfect example of this was when ABC News put out a Battleground Poll in the final days of the race between Trump and Biden, showing Biden with a staggering 17-point lead in the race.
Biden went on to win Wisconsin by just over a half of a percentage point.
Days after that election, the New York Times wrote a piece about RealClearPolitics with the headline “A Popular Political Site Made a Sharp Right Turn. What Steered It?” The piece argued that RealClearPolitics had gone from a “trusted go-to source” for “unbiased polling admired by campaign and news professionals” to a site that had “taken a rightward, aggressively pro-Trump turn” over the past few years.
It is worth noting that RealClearPolitics’s final polling average for the 2020 presidential election had Biden up over Trump nationally by 7 percentage points. Biden won nationally by 4.5 percentage points.
The story also asserted that during the period of counting absentee and mail-in ballots that arrived late, RealClearPolitics took three days longer than other news organizations to call Pennsylvania for Biden.
Carl Cannon, RealClearPolitics‘s Washington bureau chief who is broadly respected across the media world for his even hand and journalistic integrity, wrote a rebuttal to the story beginning with the New York Times‘s assertion that RealClearPolitics aggregated stories from news outlets quoting Trump supporters who questioned the election results.
Cannon was having none of it and brought the data to prove it:
“I went back and searched for those two headlines, both of which appeared on RCP on Nov. 16. The first was a link to an opinion piece from a conservative outlet, The Federalist. The second was a news story from the Washington Examiner posted on the left column, the section where we post breaking news. It ran between two other stories, ‘National Security Adviser Vows ‘Professional Transition’ of Power’ and ‘Biden Turns Up Pressure for Administration Recognition,'” he wrote.
In other words, the new aggregation RealClearPolitics runs on their front page was part of a roundup of what was happening in both camps in the aftermath of the election.
A review of the RealClearPolitics front page during that contentious period shows a variety of stories that reveal what was happening in the country from many perspectives, ranging from the Trump campaign launching a series of legal battles to a column with the headline “When Trump Vandalizes Our Country” by Nicholas Kristof of the New York Times.
Cannon noted a few days later that the top story on RealClearPolitics was an op-ed by Charles Blow titled “Trump, the Absolute Worst Loser.”
Cannon said RealClearPolitics‘s assurance to its readers has always been to “present all angles and perspectives of political events, a promise we have kept” and that the New York Times‘s criticism of it “seems to be a classic case of psychological projection. We tolerate diverse voices where I work. We encourage it. It’s our business model and our belief system.”
What has made RealClearPolitics the standard-bearer since 2002 has been its polling averages, which use pollsters from a cross-section of outfits. In short, if it is all the same type of polling, you are going to have mistakes. However, because of the way RealClearPolitics averages, it has an actual track record of correctly predicting elections, particularly when Trump is on the ballot.
Another example of polling warfare is the effort not to use Rasmussen Reports polls, a firm that was much more accurate than many of the other pollsters this cycle. Last March, FiveThirtyEight made that decision.
It is worth noting that only two polls called the popular vote nearly exactly as the Washington Examiner‘s Paul Bedard noted: Rasmussen Reports and the Wall Street Journal. Both had the final vote of 49% for Trump and 46% for Harris. Trump won 50% to 47%.
Also worth noting, FiveThirtyEights’s final forecast showed a nail-biter, giving Harris the edge, noting that its model, which uses polling, economic, and demographic data, showed the race shifted slightly that Monday.
Robert Preuhs, a political science professor at the Metropolitan State University of Denver, said in a story for its online newsroom that the mudslinging at pollsters such as RealClearPolitics leads to a fundamental misunderstanding of how polling works and what it is trying to achieve.
“This year’s statewide poll of polls, from RealClearPolitics, was off by less than 2 percentage points in all but one of the swing states, and that’s genuinely useful information for voters,” he said.
The problem with an information war over polling is that when news organizations engage in it, they are not being honest to their readers about the fact that they are just using the poll results they like and ignoring the results they don’t like.
On Nov. 5, 2024, a significant portion of the country believed Harris was going to win. Oftentimes, they were voters who relied on polling models that excluded polling, such as Rasmussen Reports.
Had these voters read RealClearPolitics, they would have understood Trump was more likely to win than lose. That didn’t mean he was automatically going to win, but it was probable.
CLICK HERE TO READ MORE FROM THE WASHINGTON EXAMINER
In October of last year, RealClearPolitics was removed from Wikipedia’s page listing “Nationwide opinion polling for the 2024 United States presidential election,” with journalist Matt Taibbi noting in his Racket News substack that the crowd-sourced editors had decided RealClearPolitics had a “strong Republican bias” that made it “suspect,” even though it didn’t conduct any polls itself, merely listing surveys and averaging them.
It makes you wonder what kind of pressure is out there in the political left hemisphere to use only the approved pollsters. Fortunately, RealClearPolitics has not caved to the mob. The last thing we need in U.S. politics is an information war on polling.
