Media malpractice by The Washington Post and ABC News

The news orgs conducted a poll that the Post admitted is deeply flawed, showing Trump leading Biden by 10 points. But they published it anyway.

The Signorile Report is reader-supported. If you’ve valued reading The Signorile Report, consider becoming a paid subscriber and supporting independent, ad-free opinion journalism. Thanks!

The Washington Post loves to herald the polling it has conducted with ABC News since 1986—the Post-ABC News poll—with a big splash at the top of its homepage, usually on Sundays.

But yesterday’s story about its poll of presidential preferences in the general election (and let’s say it again: general election polling means little this far out), asking respondents if they would vote for Donald Trump or President Biden, was hidden further down on the page.

It also bizarrely didn’t refer in the headline to the results of the presidential preference poll—which ridiculously showed Trump with a 10-point lead, 52-42, wildly out of sync with every other poll—instead focusing on “criticism” of President Biden.

Then came the subhead: “A finding that shows Trump leading Biden by a wide margin does not match other recent polling, however, suggesting it is an outlier.”

And that was the tip-off that the Post knew the poll, conducted for the Post and ABC News by Langer Research Associates, was deeply flawed and why it was buried on the paper’s homepage and its results obscured in the headline. This is gross media malpractice.

A news organization’s mandate is to provide the public with accurate and reliable information. In a climate of conspiracy theories and false claims of election fraud, reporting on polls needs to be especially rigorous—as do the polls themselves if media organizations are conducting them. In the recent past, respected pollsters, such as Celinda Lake of Lake Associates, have noted that results that are so widely off and indicate a problem with the poll shouldn’t be published.

By 3 p.m. Eastern Time yesterday, after social media backlash, including from respected pollsters and analysts, the story had disappeared entirely from the Post homepage. (It’s still on the site, just buried somewhere.)

Many polls are outliers, and the Post-ABC poll has sometimes been an outlier in past polls of political races up and down the ballot, as have polls by other media organizations and polling firms. There’s nothing out of the ordinary about that. Almost all of the time, polling analysts consider a poll an outlier when they’re comparing it to other polls.

But a pollster or news organization conducting a poll rarely, if ever, has presented a poll as an outlier itself from the get-go, since, we assume, they’re confident in their methodology and believe their poll is accurate. And sometimes the outlier poll is actually the one that gets it right, proving all the others wrong.

But in this case, the Post, whose chief polling analyst, Dan Balz, has the lead byline on the story, not only casts its own poll as an outlier; the paper then offers evidence that its own poll is flawed, going into it in depth:

Trump’s lead in this survey is significantly at odds with other public polls that show the general election contest a virtual dead heat. The difference between this poll and others, as well as the unusual makeup of Trump’s and Biden’s coalitions in this survey, suggest it is probably an outlier….

…When asked whom they would [if there is a government shutdown], 40 percent say Biden and the Democrats while 33 percent say Republicans in Congress.

That finding is at odds with previous Post-ABC polls taken over many years at times when the government was partially shut down due to spending disputes.

Looking at some of the support levels among different demographic and political groups also points to reasons for caution on this finding. For example, in the new poll, men favor Trump by 62 percent to 32 percent, a margin of 30 points. In May, Trump’s margin among men was 16 points…

…Among voters under age 35, Trump leads Biden in the new Post-ABC poll by 20 points. Some other recent public polls show Biden winning this group by between six and 18 points. In 2020, Biden won voters under age 35 by double digits…

…Among non-White voters, the poll finds Biden leads by nine points. In four other public polls, Biden’s lead among non-White voters ranges from 12 points to 24 points.

It’s extraordinary how much space the Post devotes to pointing out how out of whack its poll is from other polls, thereby debunking its own poll. The article also points to another data point that seems off, considering that we know the spike in voter registration and new voters in state after state has been among people concerned about preserving abortion rights, most of them women:

Another group that backs Trump by a big margin in the poll are those who say they did not vote in 2020. They account for about 15 percent of the overall sample of registered voters, and they favor Trump over Biden by 63 percent to 27 percent.

The Post goes on and then actually suggests one reason why the results are so skewed: More Republicans are sampled in the poll:

Outlier results occasionally occur in polls due to random error and nonresponse issues, although the political composition of the poll is typical on other metrics…In the poll, Republicans have a four-point advantage on party identification when including independents who lean toward either party, slightly more Republican than other recent polls.

Pundits, polling analysts, and pollsters responded on social media to the Post’s poll and its spin, and many concluded that the poll should not have been published. Again, it’s one thing if your poll is deemed by others as an outlier but you believe your methodology is solid. It’s another thing if you see out of the gate that things just wildly don’t add up and you’ve actually got evidence that shows it must be flawed. Larry Sabato, respected pollster and analyst at the University of Virginia’s Center for Politics, questioned how the Post ever published the poll.

As did Norm Ornstein, the former American Enterprise Institute scholar and frequent critic of the media’s coverage of Trump, who commented, “If I were running this poll and I got these results, I would stop the presses, not publish the poll, and go back to figure out what went wrong. Check the sample numbers. Look at the question wording and order. If all that is checked out, redo the survey.”

“The Washington Post poll should be an alarm bell for the newsroom, telling them both that their polling is broken and that the country is in terrible danger,” noted media critic Dan Froomkin of Press Watch.

I’ve reported in recent weeks on the problems with all of the recent general election polls and why they tell us little if anything, and certainly on the ueselessness of polls on issues like age and the economy, and how actual evidence—real world elections in state legislative races—is showing opposite results, with Democrats on a winning streak.

The Post indeed embarrassingly tried and failed to make lemonade out of lemons by focusing big on other results of the poll, including voters’ opinions of Biden’s age, a possible government shutdown, the economy, immigration, aid for Ukraine, and a number of other issues. Not only, again, are these largely useless questions since people are really answering whether they liked Biden or not, but none of the results of the poll on any issues can be trusted. The entire poll is flawed since the same people who answered regarding presidential preference are responding on these other issues.

The Post at least admitted that its poll has problems, which is more than can be said for ABC News, which recklessly reported it straight up with no caveats or self-criticism.

But the Post's admission was inadequate. The poll should have been junked, as the critics have said, and a new poll taken. Of course, that means scrapping something that is an enormously costly project and spending still more money. Those who conduct polls—the individuals who make the calls—now spend days, even weeks, just to get a 1% response rate in the age of cell phones when few people ever answer a call. (And nonresponse bias is a major factor causing problems for pollsters, as those likely to answer a call may be skewing the results.)

So you could imagine that the suits at news organizations are loathe to scrap a poll and start from scratch. They’d cleary rather just report it out as news—or admit it's flawed and hope that flies—rather than try to get accurate information.

But in this case, it’s even worse than that. The Post-ABC News poll this cycle already clearly had a methodology problem that it should have fixed by now: The last Post-ABC News poll of the general election, taken back in May, had Trump up over Biden by seven points, causing polling analyst Nate Cohn at The New York Times to call it an “aberration” and label it an outlier.

So whatever was broken back in May doesn’t appear to have been fixed. That’s journalistic malpractice, and it's dangerous in a time like this one, as it feeds disinformation before an election and fuels calls of a “rigged election” when these polls don’t pan out. Corporate media has had a major hand in allowing Trump (and MAGA) to spread lies, distortions and conspiracy theories like wildfire, and we all deserve far better. Our very democracy depends on it.