The headlines were inevitable: “Fact-Checking Site Finds Fox News Only Tells the Truth 18 Percent of the Time,” “Analysis: Over Half of ALL Statements Made On Fox News Are False,” “Fox News wins battle for most-false cable network.” One particularly dim-witted account, written by Jameson Parker, reads: “A new analysis by PunditFact found that of every statement made by a Fox News host or guest, over half of them were flat-out false. What’s more, only a measly 8 percent could be considered completely ‘true.’ In other words, a fancy review of hundreds of hours of video confirmed what many who watch Fox News with any regularity already know: Fox News lies. A lot. Like all the time.”
The “study” from PunditFact (a subdivision of PolitiFact), which is not really a study, says no such thing. You do not have to rely on my word for that: PunditFact itself warns against using its figures “to draw broad conclusions,” e.g. that Fox News lies “like all the time.” (Like, is this like a 1980s Valley-girl movie, or are you just, like, functionally illiterate?) That is because the study is an exercise in drawing nonsensical conclusions from arbitrary data.
The most obvious problem — though certainly not the only problem, not even close — is selection bias: PolitiFact is a readership-driven online publication, and thus it exercises a great deal of discretion about which statements it chooses to evaluate and why. The most obvious factor is that it evaluates only statements that are disputed. Specifically, it evaluates only statements that are disputed and that its editors believe will be of some interest or benefit to its readers.
Fox News is a personalities-driven opinion network with occasional news reports; it is inevitable that its broadcast hours will be more rapidly punctuated by controversial statements than those of, for example, ABC. Unsurprisingly, the opinion-heavy Fox News and MSNBC both have relatively high falsehood scores on the PolitiFact report card, while CNN doesn’t. It’s not as though Michaela Pereira never says anything that might be disputed — she simply never says anything that is interesting, true or false, so nobody cares. Or, as PolitiFact puts it: “We use our news judgment to pick the facts we’re going to check, so we certainly don’t fact-check everything. And we don’t fact-check the five network groups evenly.”
No kidding. Say what you will about Bill O’Reilly, nobody ever made a living out of pretending to be Rachel Maddow. (No, Chris Hayes doesn’t count.)
PolitiFact’s kindergarten-level methodology here is to take the total number of statements it evaluates, tally up the “mostly false,” “false,” and “pants on fire” ratings, and then do a little division. Given the underlying selection issues, this amounts to nothing more than doing meaningless arithmetic on meaningless data. If PunditFact editor Aaron Sharockman spent more than 20 minutes on this so-called research, he should demand a refund from his university. (Given that he has a B.A. in journalism, he should demand a refund on general principles.)
By the same measure, approximately 100 percent of statements made by Paul Krugman evaluated in National Review are 1. mostly false; 2. false; 3. pants-on-fire; or, my own favorite designation, 4. wearing-full-Wayne-Newton-makeup-while-singing-“Danke Schoen”-at-4-a.m.-under-a-bridge-in-Cleveland crazy. But that does not mean that the sum of what comes out of Professor Krugman’s mouth is 99 and 44/100 percent pure B.S., like some Bizarro World version of Ivory Soap — it just means that we mostly tend to take notice of him when he’s wrong. If he says you should try the cheese plate at Il Bambino, give the claim due consideration.
The deeper problem with PunditFact is the bias in how it evaluates statements. Consider two structurally identical questions: In the first, it considered Chris Wallace’s claim that Hillary Clinton had “defended Syria’s President Assad as a possible reformer at the start of that country’s civil war.” That statement, the editors decided, was only half-true, because that was “not expressly her opinion.” Rather, she had said that members of Congress of both parties who had visited Syria had suggested that Assad was a possible reformer. (Never mind that Mrs. Clinton’s claim is itself untrue, a three-Pinocchio offender in the Washington Post’s judgment.)
In the second instance, PunditFact considered a claim from Bill O’Reilly, made during an interview with President Barack Obama, that he had not accused the administration of obscuring the motive behind the Benghazi attack for political reasons. O’Reilly had in fact interviewed people who said that, but he himself had not made that claim. PunditFact nonetheless rates it “mostly false,” because O’Reilly had, in its view, “nurtured suspicion.” Mr. O’Reilly and Mrs. Clinton were engaged in precisely the same rhetorical strategy: the time-honored Washington dodge of using others to suggest indirectly what you think or suspect yourself, e.g. “it’s a serious charge,” “some have said,” “it has been suggested that,” etc. In both cases, the statement was made on Fox News, but Mrs. Clinton gets a pass (“not expressly her opinion”) while Mr. O’Reilly gets labeled a liar — for precisely the same thing. This is what simple bias looks like.
It’s mostly a matter of sympathy — not with different news networks, but with different political figures. When it considered Eric Bolling’s claim that the great majority of new jobs created during the Obama administration have gone to men, PunditFact labeled that a half-truth, too, even though, in its own words, the “numbers are right.” Why? Because “as an attack on Obama, it rings hollow.” Not exactly Karl Popper’s intellectual standards at work.
But the fact is that unsupportable, boneheaded claims such as “over half of ALL” — thanks for that all-caps attack, Einstein von Brainstorm — “statements made on Fox News Are False” will live forever, because people are mostly interested in having their biases confirmed and their values affirmed rather than learning new things about the world and how it works. True, much as I like yelling at people on television, it is pretty hard to feel too bad for Fox News and MSNBC over an exercise in confirmation bias, but this sort of sloppy thinking and malicious manipulation does have the effect of leaving the polity a little dumber than it absolutely has to be. And that is an unforgivable sin.
— Kevin D. Williamson is roving correspondent for National Review.