Friday, January 29, 2021

PolitiFact miscounts American deaths during WW2?

When a PolitiFact fact check's subject matter involves math, we (figuratively!) smell blood in the water.

This item came from the PolitiFact article "Joe Biden's inaguration in extraordinary times, fact-checked," published Jan. 20, 2021. Notably, PolitiFact has only done one Truth-O-Meter rating on claims from President Joe Biden since mid-December. That's assuming PolitiFact's page showing Biden's fact checks is accurate.

As it turned out, PolitiFact was right that Biden was "close to accurate." But PolitiFact made a significant methodological blunder in reaching its conclusion. The mistake appears right away in PolitiFact's explanation for its judgment:

As Biden was speaking, the Johns Hopkins University coronavirus tracker was reporting 402,269 deaths in the United States. That is just shy of the 405,399 U.S. deaths during World War II, according to the Congressional Research Service. With the seven-day moving average of coronavirus deaths reaching 3,015 on Inauguration Day, the four-year World War II total was due to be matched by the coronavirus either on Jan. 20 or 21, less than a year after the virus reached the United States.

PolitiFact reports incorrectly in the second sentence of the above paragraph. The Congressional Research Service source document does not give a total for all the American lives lost in World War II. It gives a total for the number of military personnel lost during the war (bold emphasis added):

This report provides U.S. war casualty statistics. It includes data tables containing the number of casualties among American military personnel who served in principal wars and combat operations from 1775 to the present. It also includes data on those wounded in action and information such as race and ethnicity, gender, branch of service, and cause of death. The tables are compiled from various Department of Defense (DOD) sources.

The total PolitiFact used omits more than 10,000 civilian casualties, including nearly 10,000 from the U.S. civilian merchant marine. We don't see where Biden limited his statement to military personnel.

PolitiFact went on to suggest Biden would be right by extrapolating the numbers forward for a full year since the U.S. started to log covid deaths. But doing that turns Biden's claim into a prediction. PolitiFact supposedly does not fact check predictions. Going on the facts alone, Biden was off by more than 10,000 deaths. PolitiFact made his error appear considerably smaller by using a flawed approach to its fact check.

It's what we call PolitiFact's "Rubberstamps for Democrats" program. We argue that the tendency to award lazy favorable ratings to Democrats (and not Republicans) counts as one evidence of PolitiFact's political bias.

Thursday, January 28, 2021

PolitiFact's Layers of Editors: Biden says he will ensure enough supply to vaccinate 300 Americans by early fall (Updated)

PolitiFact reported President Biden's plans would ensure covid-19 vaccine supplies large enough to vaccinate 300 Americans by early fall.

The Biden administration has envisioned a more prominent federal role, including setting up 100 vaccine centers across the nation by the end of February. Officials say the biggest roadblock is lack of vaccine supply. Biden announced that his administration aims to purchase enough doses to fully vaccinate 300 Americans by the end of the summer or beginning of the fall.

 That's an average of three vaccines per planned vaccine center, by our reckoning.

Of course, PolitiFact pretty obviously omitted the word "million" that belonged between "300" and "Americans." So why do we even bother to mention such a small error? It's not even an example of bias, is it?

There are two related reasons why.

First, because PolitiFact loves to trumpet its rigorous system for reporting accurately. Did you know that PolitiFact's system calls for three editors to review each fact check?

The reporter who researches and writes the fact-check suggests a rating when they turn in the report to an assigning editor. The editor and reporter review the report together, typically making clarifications and adding additional details. They come to agreement on the rating. Then, the assigning editor brings the rated fact-check to two additional editors.

The three editors and reporter then review the fact-check by discussing the following questions.

We hope we're correctly assuming that two editors aside from the assigning editor would not dream of discussing the critical questions without bothering to read the fact check. Under that assumption, three editors and one writer together failed to keep the fact check from including a figure that was off by a factor of one million.

So our first point consists of the observation that PolitiFact's system may allow for considerable error, even obvious error.

So far as we're concerned, that by itself is no big deal. Everybody makes mistakes. So that brings us to our second reason for writing this up.

Thanks to its performance since its inception, we have a very low expectation that PolitiFact will run a correction or update and include this item on its page of "corrections and updates."

That's troublesome.

The best journalists make thorough corrections a priority. Take a gander at the expectation the International Fact-Checking Network lays down for its verified signatory fact-checking organizations:

6.3 Where credible evidence is provided that the applicant has made a mistake worthy of correction, the applicant makes a correction openly and transparently, seeking as far as possible to ensure that users of the original see the correction and the corrected version.
PolitiFact makes many--perhaps most--of its corrections secretly and without transparency.

PolitiFact has a sort of loophole in its corrections policy allowing that lack of transparency. And we've written about it before.

Here's the loophole (bold emphasis from the original):

Typos, grammatical errors, misspellings – We correct typos, grammatical errors, misspellings, transpositions and other small errors without a mark of correction or tag and as soon as they are brought to our attention.

The section on "Typos, grammatical error, misspellings" sneaks in an extra category that PolitiFact feels needs no notice of correction. That is the "other small errors." We think that most likely PolitiFact will treat reporting 300 instead of 300 million as a small error and insert the correction without any notice at all to its readers. Under that scenario, PolitiFact will make no attempt to correct the error openly and transparently, and will likewise make no attempt to bring the corrected number to the attention of those who read the erroneous reporting in the original version.

If PolitiFact surprises us and issues a correction notice, we expect it to follow the form "We reported the wrong number of vaccines Biden promised in the original version of this story." PolitiFact prefers to report vaguely on its mistakes instead of detailing the exact nature of the error as full transparency would demand.

Doing the right thing isn't hard: "We reported Biden would acquire vaccines for 300 Americans when we intended to specify 300 million Americans." Nobody would have a real problem with that.

Whatever one thinks about the significance of the error, the PolitiFact method of treating substantive errors the same as it does typos, grammatical errors and misspellings does not scrupulously follow the IFCN's requirement on corrections.

Supposedly the IFCN requires its signatory organizations like PolitiFact to "scrupulously" follow its guidance on corrections. But the truth is the IFCN has turned a blind eye on PolitiFact's correction shenanigans for years. They know about it because I've informed them about it periodically. The IFCN has never directly addressed any of the reports I've sent detailing PolitiFact's failures to follow policy despite showing no clear evidence that the reports have in any way affected PolitiFact's signatory status.

If the complaints carry no validity, the IFCN should transparently declare the how and why of its judgments.

So this isn't evidence of bias. It's evidence that PolitiFact tries to trick its readers into believing it strives to meet the highest journalistic standards. It advertises a system with failsafes that would prevent the type of error it made. PolitiFact and the IFCN assure readers that PolitiFact performs corrections to appropriately high standards.

It's deception.


We'll update this post to report on whatever actions PolitiFact takes to correct its mistake. A sufficiently pathetic response may result in yet another complaint filed with the IFCN.


Update Jan 28, 2021: We sent an email pointing out PolitiFact's error about 1 p.m. When we checked at about 3:30 p.m., the error was fixed. PolitiFact also appended a correction with the milquetoast-style wording we expected. Oddly, neither the correction nor the tag denoting the correction show up for the archived version of the story. Only the corrected wording in the story shows up.

Here's the correction notice:
CORRECTION, Jan. 28: We corrected the number of Americans the Biden administration hopes to fully vaccinate by the end of summer or early fall.
On the plus side, PolitiFact corrected the error, posted a correction notice and even placed the "corrections and updates" tag.

On the negative side, the correction falls well short of full transparency. PolitiFact told readers on what specific topic it erred but not the specific error it made. That style of correction hides from readers the magnitude of the error. And that was probably the plan.

Tuesday, January 26, 2021

PolitiFact writes up Capitol riot, omits mention of John Sullivan

PolitiFact wants you to know right-wing groups were solely responsible for the Capital riot in early 2021--even if it's not necessarily true.

PolitiFact politisplained it in a Jan. 20, 2021 article by Daniel Funke:

Since Jan. 6, we’ve fact-checked several claims that blame antifa, short for anti-facist, for the breach of the U.S. Capitol Building. Republican lawmakers, conservative pundits and social media users have said the loose coalition of communists, socialists and anarchists infiltrated a crowd of Trump supporters to stoke violence.

There is no evidence to support those claims.

PolitiFact plays an interesting game with antifa. On the one hand, it's a "loose coalition of communists, socialists and anarchists" but on the other hand it's just an idea with no structure of membership, so it's next to impossible to identify people as "antifa" even if they are communists, socialists or anarchists.

Funke's article struck us as fascinating for its failure to mention John Sullivan, the left-winger who took video of the rioting and exhorted the rioters with cries such as "We accomplished this s**t. We did this together." (Newsweek)

Sullivan was among those charged in association with the rioting. But he remained unmentioned when Funke gave the rundown of those who were charged:

FBI Assistant Director Steven D'Antuono said during a Jan. 8 press briefing that there was "no indication" that antifa activists were involved in the insurrection. Since then, the U.S. Attorney’s Office for the District of Columbia has charged more than 100 people with crimes related to the Capitol riot, many who bragged on social media about taking part, and the FBI has arrested more than 40. Several suspects appear to have connections to militia and far-right groups.

 Guess what? Sullivan was one of those charged (Deseret News, via MSN):

John Earle Sullivan, 25, was charged federally on Wednesday with being on restricted property, civil disorder and “violent entry or disorderly conduct,” according to a complaint filed in U.S. District Court in the District of Columbia.
These days the media do not associate Sullivan with antifa or Black Lives Matter.

But let's dwell on the past.

Sullivan's History in the Media

The media have reported that Sullivan organized Insurgence USA. The Insurgence USA website peddles the kind of protest gear the public associates with antifa.

Is Sullivan just a right-wing capitalist trying to take advantage of left-wing rubes? Maybe. Let's look for press reports using "Sullivan" and "Insurgence USA" as our search terms, focusing on 2020 and before.

The Guardian (July 27, 2020)

That same Black Lives Matter protest that inspired Robertson’s fear was originally planned as a pro-police event in downtown Provo. John Sullivan, 26, the founder of Insurgence USA, a group for racial justice and police reform, organized a counter-protest alongside several other organizers. Protesters were to meet at the Provo police station at 6.30pm that night.
Newsweek (July 23, 2020):

John Sullivan, of the Insurgence USA activist group, was armed with a rifle for around two hours during what was promoted online as a "Solo Armed Stance" in protest at the unidentifiable agents who have been seen forcing protesters into unmarked vehicles in Portland, Oregon.

It's entirely worth noting that the left-wing protest community aired concerns about Sullivan in late 2020, as reported by the Daily Dot (links lead to profanity-laced reports). At the same time, the reports acknowledge Sullivan's connection to BLM/antifa activity:

last night he leaked the details of a secret event. luckily spotters OTG were able to shut it down before it started once they began to see an increase in police activity.

The report suggests Sullivan received the details of a secret event involving illegal activity. Without receiving such details, leaking them proves difficult.

"Sullivan denied being affiliated with antifa"

The Daily Dot article also contains an amusing (and common) example of liberal press credulity. It points to a Rolling Stone interview of Sullivan in which he supposedly "denied being affiliated with antifa."

But the supposed denial is a non-denial denial (bold emphasis added):

Sullivan carried a simple setup: a cellphone mounted on an image-stabilizing gimbal. Fitting into the mob, he says, required mirroring its revolutionary sentiments. “I was worried about people recognizing me and thinking that I was Antifa or, like, BLM or whatever,” he says. “The entire time they’re yelling, ‘Fuck Antifa! Fuck, BLM.’ I’m not saying I’m Antifa, by any means. But I definitely believe Black Lives Matter.”

Reported accurately, Sullivan did not deny affiliation with antifa. He denied he was saying he was affiliated with antifa. That's not the same thing. We're in trouble when journalists either can't tell the difference or apply spin to obscure the difference.

And if Sullivan had offered a full-throated denial of an affiliation with antifa, that would still not serve as dispositive evidence that no affiliation existed when he gave the interview. How many members of antifa have been publicly identified based on their own admission?

As things stand, we do not have definitive evidence Sullivan sustained a relationship with the antifa movement. But we do have evidence of at least a temporary connection to a secretive left-wing organization planning illegal activity.

Why is Sullivan entirely missing from Funke's report?

When PolitiFact Reported About Sullivan

We found Sullivan's absence from Funke's report even more strange given that PolitiFact previously did some reporting on Sullivan. The "fact check," by Bill McCarthy, supposedly covered the same subject as Funke's,the allegation that antifa incited the riot:

(C)laims faulting antifa for the violence at the Capitol keep coming. The latest target is Utah’s John Sullivan, the founder of Insurgence USA, an activist group against police brutality.

McCarthy's article published on Jan. 8, 2021. Did Funke somehow not know about it? It's not linked in Funke's story, though he linked other fact checks about the Capitol riot.

On Jan. 15, 2021, PolitiFact updated McCarthy's story noting Sullivan's arrest over his part in the D.C. riot. Did Funke also not know about that?

To be sure, McCarthy's article tries to downplay Sullivan's involvement in the riot, even assuring readers there's no proof he incited the insurgency. That occurs shortly before including quotations of things Sullivan said during the riot, such as "Let's burn this s--- down."

We can't help but think that if the Democratic managers of the Trump impeachment case had Trump on record telling the D.C. crow "Let's burn this s--- down" it would improve the case for Trump's impeachment. Supposedly Trump incited the Capitol insurrection, requiring his impeachment.

Summary

While there's no unequivocal case, at least for now, that John Sullivan counts as antifa or Black Lives Matter, a competent debunking of antifa/BLM involvement in the Capitol riot should not omit all mention of the strongest evidence of that involvement. 

And PolitiFact may want to look up the definition of "incited." 

It's a bit absurd to claim there's no evidence antifa incited the Capitol riot, quote somebody saying "Let's burn this s--- down" and then take the man at his word that he's not antifa when that wasn't what he said.


Correction Jan. 26, 2021: We misspelled "Capitol" as "Capital"in one of two instances in our summary section. That's fixed with this update.

Friday, January 22, 2021

Why does PolitiFact claim it allows license for hyperbole?

It makes sense for a fact checker to account for figures of speech such as hyperbole. And PolitiFact plainly states in its description of its principles that it allows license for hyperbole (bold emphasis added):

In deciding which statements to check, we consider these questions:

• Is the statement rooted in a fact that is verifiable? We don’t check opinions, and we recognize that in the world of speechmaking and political rhetoric, there is license for hyperbole.

For us, the mystery is why PolitiFact makes this claim and then blatantly fails to honor it.

Today we add another example to the collection.

The $38 Burrito?

Turning Point USA commentator Jordan Rachel tweeted about the $15 minimum wage:


Rachel's tweet fits perfectly the form of hyperbole. She chose an extremely high price for the future burrito and took care to offer no clear sign the statement was intended to be taken literally. One would encourage literal interpretation by plainly stating something like "Increasing the minimum wage to $15 will result in $38 burritos."

PolitiFact's graphic presentation of its fact check misleadingly frames Rachel's statement as exactly the latter type of claim:



PolitiFact unfairly manipulates Rachel's claim. Her claim is an if/then statement emphasizing the truth that the $15 minimum wage increase counts as an inflationary policy.

Rachel gets no credit for that point from PolitiFact. We don't know precisely why she gets no credit for that point, but we can at least say with certainty that PolitiFact claim that it allows license for hyperbole rings false.

Hyperbole is not mere exaggeration

As a figure of speech, hyperbole is not the same thing as mere exaggeration. If it was, then fact checkers shouldn't need to take it into account. They could simply show in factual terms how far off the exaggeration fell from the truth and leave it at that.

Hyperbole counts as common figure of speech. Perhaps the modern journalist simply understands it poorly. In the interest of public education, we recommend to our readers the account of hyperbole at YourDictionary.com:

Hyperbole, from a Greek word meaning “excess,” is a figure of speech that uses extreme exaggeration to make a point or show emphasis. It is the opposite of understatement.

You can find examples of hyperbole in literature and everyday speech. You wouldn’t want to use it in nonfiction works, like reports or research papers, but it’s perfect for creative writing and communication, especially when you want to add color to a character or humor to a story.

We encourage readers to click the link and read through it all.

No Sign PolitiFact Considered the Possibility of Hyperbole

As for the fact checkers at PolitiFact, we see no sign at all they considered interpreting Rachel's statement as hyperbole.

The summary section of the fact check fairly displays the rush to judgment:

Our ruling

Rachel said a $15-an-hour minimum wage would raise the price of a Taco Bell burrito to $38.

This claim is countered by available evidence, including the current burrito prices at Taco Bell locations in cities and counties where a $15 minimum wage is in effect. Four economists characterized the claim as a far-off estimate at odds with economic theory.

We rate Rachel’s statement False.

 Misleading and disgraceful.

We expect a conscientious fact-checking organization to make a plausible attempt to achieve consistency between policy and practice. If the policy is A and the practice is B, change one of them to match the other.

In claiming to allow license for hyperbole and failing to allow license for hyperbole, PolitiFact deceives its readers.

Tuesday, January 19, 2021

PolitiFact finds Republican's opinion "Pants on Fire" false

PolitiFact says it doesn't fact check opinions.

Republican Pennsylvania senate Majority Leader Jake Corman says he doesn't believe unproven voter fraud claims had any role in (causing) the Capitol riot.

Who is lying?

PolitiFact would have you believe it's Corman making the false claim. But Corman provided perhaps the clearest possible indication that he was giving an opinion.

"Opinion" means someone expressed what they believe, not necessarily what they proffer as established fact.

Note that PolitiFact's headline/deck makes it look like Corman made a statement of fact. But by the fourth paragraph of its fact check, PolitiFact makes an admission against interest:

"House Democrats yesterday called on state Republicans to accept some responsibility for some of the violence that we saw at the U.S. Capitol last week because of unproven voter fraud claims," Maisel said. "What role do you think that played?" 

"I don’t think it played any role," Corman responded.

"I don't think" counts as perhaps the classic introduction to a person's opinion. "In my opinion" might challenge it. 

It ought to astonish that a fact checker could claim it does not fact check opinion and then ignore such an obvious sign that a statement was offered as an opinion. Unfortunately, we cannot register surprise at this, thanks to PolitiFact's track record.

And PolitiFact does not merely make the mistake of fact-checking an opinion. It also apparently equivocates, transforming the role of election fraud claims in motivating the gathering of protesters into a role in causing the violence.

The fact checker interested in staying worthy of the name steers clear of such murky issues of cause and effect.

Hold PolitiFact's beer:

But there’s plenty of evidence that people who traveled to Washington and stormed the Capitol did so because they believed the election was rigged and unfairly stolen from President Donald Trump. He told them so repeatedly. And Corman, like many Republican officials in Pennsylvania and elsewhere, frequently helped advance Trump’s narrative.

There's plenty of evidence that people who traveled to Washington and stormed the Capitol did so because they believed the election was rigged and unfairly stolen from Mr. Trump? Will the fact checker offer evidence that the violence done in association with storming the Capitol stemmed from that motivation?

 Do not hold your breath, gentle reader.

Just before PolitiFact offers its conclusion, we get this:

And the president’s insistence that the election was stolen but could still be reversed with Congress’ help is what rioters cited when asked by reporters why they were there and what they hoped to achieve.

So PolitiFact tells us rioters said the unsupported election claims motivated their actions. And the good news is that PolitiFact provides an embedded supporting link in the paragraph.

The bad news? The link supports the highlighted text, that being "the president's insistence that the election was stolen." It offers no support for the claim the rioters used the unsupported election claims as their motivation.

And that paragraph encapsulates PolitiFact's approach to the fact check. Its proof that unsupported election fraud claims led to the Capitol riot consists of proof that GOP lawmakers made those claims PolitiFact calls unsupported.

To realistically establish a causal link, a fact checker should show that believing unsupported claims about election fraud meaningfully correlates with violent action such as was exhibited at the Capitol.

We should expect the staff at PolitiFact to know that.

PolitiFact's fact check is a disgrace to fact-checking.


Afters:

Color us surprised that this counts as the first time we have tagged a post with "Jessica Calefati." She's among PolitiFact's worst. And that's really saying something.

Monday, January 18, 2021

PolitiFact's meritless fact check about merit

Is PolitiFact entering a golden age of Joe Biden rubberstamp fact checks?

Consider this Jan 8, 2021 item from PolitiFact with bylines for Amy Sherman and Miriam Valverde:

PolitiFact cited an expert who counted 63 cases that were dismissed either on the merits or for other reasons such as lack of standing. That's an okay thing for fact checkers to do:

Marc Elias, a lawyer who has filed and defended cases on behalf of Democrats, keeps a tally on the outcome of the election cases.

"It is 63 losses by Trump and his allies," Elias told PolitiFact the morning of Jan. 8. "We treat each case separately — so if there is a federal case and a state case, we treat them as two cases. We only ever count a case one time — so if there is an appeal or remand, we do not treat that as a separate loss."

But PolitiFact ended up having a hard time distinguishing between lack of merit and lack of standing. Instead of distinguishing between the two, PolitiFact appeared to assume that a case dismissed for lack of standing also lacked merit. This type of wording was typical of the fact check:

More than 60 lawsuits brought by Trump and his allies failed because they were unable to prove their allegations. Some lawsuits were dismissed due to errors in the filings and other procedural issues.

A proper fact check of Biden's claim would look specifically at the cases dismissed on the merits, putting a number on it and then looking at whether that number exceeded 60.

That never happens in PolitiFact's fact check. Instead, PolitiFact tells us that more than 60 cases were dismissed for a variety of reasons. And then concludes on that basis that Biden was correct. The "True" rating is supposed to mean that there's no missing context.

But PolitiFact itself omits critical context. The law distinguishes between the reasoning judges use to determine lack merit and the reasoning used to determine lack of standing. When a judge rules the plaintiff lacks standing to sue, the court need not examine the merits of the case.

Glenn G. Lammi sketched the essence of the doctrine for Forbes:

A plaintiff’s lack of standing to sue is about as close to a silver-bullet defense as civil-litigation defendants have at their disposal in federal court. The doctrine is based in Article III of the U.S. Constitution, which limits federal courts to hearing only "cases and controversies." The doctrine puts the onus on a plaintiff to prove, among other factors, that she suffered an actual harm, and if she can't, the court has no jurisdiction over the case.

And when a court has no jurisdiction over a case, it need not consider the case on its merits (Wikipedia):

In the United States, the current doctrine is that a person cannot bring a suit challenging the constitutionality of a law unless they can demonstrate that they are or will "imminently" be harmed by the law. Otherwise, the court will rule that the plaintiff "lacks standing" to bring the suit, and will dismiss the case without considering the merits of the claim of unconstitutionality.

It follows that a suit dismissed for lack of standing was not dismissed as meritless, for the court had no need to decide the case on the merits.

Without counting the number of cases dismissed for reasons other than lack of merit, a fact check cannot make a determination that more than 60 cases were dismissed as meritless. Yet PolitiFact's so-called fact check does just that.

PolitiFact used 63 as the total number of relevant cases. Based on that number, if just three cases were dismissed for reasons other than merit, then Biden's claim is false. But PolitiFact decided not to put that kind of effort into its fact check.

As happens so often, PolitiFact was only willing to put enough effort into its fact check of a Democrat to find the claim true. In this case, PolitiFact appeared to assume that any dismissal indicated lack of merit.

In other words, it was another PolitiFact rubberstamp.

Thursday, January 14, 2021

PolitiFact takes Devin Nunes out of context to give "False" rating

PolitiFact says a "Half True" claim is one that is "partially accurate but leaves out important details or takes things out of context."

 Also PolitiFact:

Did Rep. Nunes self-stultify when he, a Republican, communicated the notion that Republicans have no way to communicate with the demise of Parler?

PolititiFact's treatment of his claim is almost that simple, and it could have been as simple as that given PolitiFact's failure to consider the context. As a general rule of literary interpretation, the interpreter ought to look for evidence in context that might explain an apparently self-contradictory statement.

To help illustrate that point, we'll look for evidence PolitiFact followed its own published statement of principles in fact-checking Nunes' claim.

What did Nunes say?

For starters, we put together a transcript from the Fox News clip PolitiFact used as its source. Fox News is poor for preserving its videos and also poor for publishing transcripts. Our transcript may end up the best available before much time has passed. We note that regretfully.

We highlight in bold the parts of Nunes' statement that PolitiFact used in its story (punctuation may vary).

The effect of this is that there is no longer a free, and open social media company, or site, for any American to get on any longer. Because these big companies, Apple, Amazon, Google, they have just destroyed a, what was likely, Parler's likely a billion dollar company. Poof, it's gone. 

But it's more than just the financial aspect of that. Republicans have no way to communicate. If, it doesn't even matter if you're Republican or conservative. If you don't want to be regulated by left wingers, that are at Twitter and Facebook and Instagram, where you get shadowbanned, nobody gets to see you, nobody gets to see you, they get to decide what's violent or not violent. It's preposterous. 

So, I don't know where the hell the Department of Justice is at right now, or the FBI, uh, this is clearly a violation of antitrust, civil rights, RICO statutes, there should be a racketeering investigation on all the people that coordinated this attack on not only a company, but on all of those, of those, like us, like me, like you, Maria; I have three million followers on Parler. Tonight, I will no longer be able to communicate with those people. And they're Americans. 

Context, Context, Context

Is it hard at all to see Nunes point in the above? He's saying with the loss of Parler, people (including Republicans) have no social media option that isn't regulated by left wingers. That's the context for "Republicans have no way to communicate." PolitiFact follows that with "And" in its telling. We are confident our transcript is correct. Nunes followed with "If." So the idea he was trying to communicate was "Republicans have no way to communicate if (they) don't want to be regulated by left wingers."

PolitiFact's misquotation of Nunes counts as minor, but as we have noted, when such mistakes help PolitiFact in setting a misleading narrative they count as significant indicators of bias.

He added: "Republicans have no way to communicate. And it doesn't even matter if you're a Republican or conservative. If you don't want to be regulated by left-wingers that are at Twitter and Facebook and Instagram, where you get shadowbanned, nobody gets to see you."

If a fact checker has no interest in seeing the contextual clues explaining a statement, then the fact checker will more likely overlook contextual clues explaining a statement.

Nunes said "If" not "And" and his statement about it not mattering if you're a Republican or conservative was an interruption of his idea that hamstringing Parler took from Republicans the social media company not regulated by left wingers.

How did PolitiFact miss it?

Careless Interpretation Leads to Faulty Premise

When PolitiFact overlooked the context of Nunes' remarks, it led to the false premise that Nunes was literally saying Republicans have no way at all to communicate. Though PolitiFact did not use Nunes own ability to speak to contradict his claim, the evidence the fact checkers used was not far removed from that:

Some users on Twitter found it ironic that Nunes was making his claim on Fox News, the most watched cable network, and a favorite medium for Republicans to communicate.

We may forgive Twittter randos for badly missing Nunes point. But isn't PolitiFact supposed to be better than that?

More:

And long before social media, members of Congress had several venues to communicate with constituents, including some unfiltered ones.

Long before social media (the thing Nunes was talking about), Republicans could communicate with constituents a number of ways! 

One would think that context simply does not exist for PolitiFact.

Want more? PolitiFact has more:

House members can communicate via press releases, send out mass emails or paper mail to constituents, or host phone-in or online town halls. Capitol Hill has places where members can hold in-person press conferences or satellite interviews with media outlets back home or national outlets, or record a video message to email to their constituents.

What does that have to do with social media companies having the power to moderate conservatives who try to use social media? Nothing, really. It's just PolitiFact ignoring the context and fact-checking a straw man version of Nunes' claim. It's the type of thing fact-checkers do routinely, and we've been pointing it out for years and years.

PolitiFact continues on and on in the same vein. But by now we hope our readers get the idea.

 It's just another claim PolitiFact invented in order to find it false.


Afters:

PolitiFact used a number of (scare quote) "expert" sources in this fact check. The quotations likely reflect that the experts were asked whether Republicans can communicate without Parler. PolitiFact does not share with its readers the text of its inquiries. But it's a good bet those inquiries used PolitiFact's false premise as their basis.

Saturday, January 2, 2021

PolitiGnatStraining No. 573: Is natural immunity really a vaccination?

PolitiFact truly never ceases to amaze us with its forays into free interpretation of political claims. Perhaps President Trump deserves some of the credit for driving PolitiFact mad via Trump Derangement Syndrome.

Would you believe that Mr. Trump believes that immunity acquired through recovery from the coronavirus counts as a "vaccine"?

That's where PolitiFact went on Dec. 15, 2020:

This analysis will take some doing, because PolitiFact's fact check contains multiple layers of fact check lunacy.

First, we need to see what Trump actually said. PolitiFact covers that adequately in the fact check:

"You develop immunity over a period of time, and I hear we’re close to 15%. I’m hearing that, and that is terrific. That’s a very powerful vaccine in itself," said Trump, who was responding to a reporter’s question about what his message to the American people was as the holidays approach and levels of COVID cases in the U.S. continue to rise.

Immediately after, the fact check runs off the rails at about a 90 degree angle:

It wasn’t the first time Trump had given credence to the idea that if enough people in a population gain immunity to a disease by being exposed to it, the illness won’t be able to spread through the remainder of the population — a concept known as "herd immunity."

Assuming PolitiFact and fact checker Victoria Knight did not seek to deliberately mislead their readers, this looks like a likely case of confirmation bias. Knight and PolitiFact were looking for Trump to say something false and interpreted Trump's claim to create the falsehood they sought. Otherwise, it's a mystery how anyone could interpret Trump's claim as they did.

First, Trump is pretty obviously talking about natural immunity as a metaphorical vaccination. Natural immunity is not entirely unlike immunity acquired via a human-devised vaccination. We would hope that even liberal bloggers working as non-partisan fact-checking journalists would know about figures of speech such as metaphors.

Second, why take the 15 percent plateau as the notion Trump was comparing to a vaccine instead of the concept of naturally acquired immunity? We can't imagine a reason apart from either confirmation bias or premeditated deceit. Naturally acquired immunity has approximately the same role in achieving herd immunity as a vaccine. Both confer immunity, and both would contribute toward a population's potential for herd immunity. If that was Trump's point, as seems to reasonably be the case, he's right.

How does a mainstream fact checker get so far off course so quickly? And why isn't everybody blowing the whistle on this kind of fact-check shenanigan?

As for PolitiFact, it keeps chugging along on one of the sharpest tangents ever devised:

However, experts have warned that attempting to achieve herd immunity naturally, by allowing people to get sick with COVID-19, could result in more than a million deaths and potentially long-term health problems for many. A better way to achieve protection across the population, experts say, is through widespread vaccination.

Why do Knight and PolitiFact think this paragraph has any relevant relationship to Trump's claim? Do they suppose Trump calling natural immunity a "powerful vaccine in itself" represents an intent to change policy to forgo the use of man-made vaccines and seek to combat the virus with natural immunity alone?

This is fact-checking gone insane. The paragraph doesn't belong. PolitiFact could have talked about the role natural immunity plays in achieving herd immunity. Readers could benefit from that explanation. But they get hardly any of that explanation from PolitiFact. Instead, they get railroaded down PolitiFact's branching tree of tangent:

So, we thought it was important to check whether 15% is anywhere close to the herd immunity threshold, and whether this level of natural immunity could be considered "as powerful as a vaccine."

So PolitiFact fact checks whether Trump is right that 15 percent natural immunity could confer herd immunity, even though that really has nothing to do with what Trump said. And where does that "as powerful as a vaccine" line come from? PolitiFact puts it within quotation marks, as though it is quoting Mr. Trump. Is it supposed to be something Trump said?

It's not in the transcript of Trump's remarks that PolitiFact linked. And our Google search using "Trump" AND "as powerful as a vaccine" did nothing to encourage us to believe the quotation came from Trump.

Remember, a team of editors supposedly reviews each PolitiFact fact check. Apparently each of them gave some sort of okay to this dumpster fire of a fact check.

The PolitiFact team overlooked an obvious use of metaphor and then went circus clown on a balloon to turn Trump's claim into something it wanted to fact check.