Friday, July 19, 2019

PolitiFact Wisconsin: "Veteran" and "service member" mean the same thing

A funny thing happened when PolitiFact examined Democratic presidential candidate Tulsi Gabbard's claim the Trump administration deports service members.

Instead of ruling on whether the Trump administration was deporting service members, PolitiFact Wisconsin decided to look at whether the Trump administration was deporting non-naturalized service veterans.

Therefore "service members" are the same thing as non-naturalized service veterans?

We wish we were kidding. But read PolitiFact's summary conclusion. PolitiFact equates "service members" with "veterans" as though it's the most natural thing in the world, and doesn't even mention citizenship status:
Our ruling

Gabbard said at the same time Trump talks about supporting veterans, "he is deporting service members who have volunteered to serve this country."

The Trump administration expanded the grounds under which people, including veterans, can be deported, which some blame for more veterans being forced to leave the country. That said, GAO documents make clear the issue existed before Trump took office -- something that wasn’t acknowledged in Gabbard’s claim.

Our definition for Mostly True is "the statement is accurate but needs clarification or additional information." That fits here.
PolitiFact does mention citizenship issues in the body of the story. It opens, for example, with a frame emphasizing military service and illegal immigration:
Military matters and illegal immigration.

Both are hot-button issues for voters in the 2020 presidential election, though for different reasons.

U.S. Rep. Tulsi Gabbard of Hawaii, a Democratic presidential hopeful and major in the Hawaii Army National Guard, linked them when she spoke July 11, 2019 at the League of United Latin American Citizens convention in Milwaukee.
In the quotation PolitiFact Wisconsin provided, Gabbard did nothing to explicitly link military service with illegal immigration. The journalist (or reader) would have to infer the connection. And PolitiFact Wisconsin failed to link to a transcript of Gabbard's speech, linking us instead to the Journal Sentinel's news report that fails to supply any additional context to Gabbard's remarks.

Intentional Spin?

We see evidence suggesting PolitiFact Wisconsin applied intentional spin in its story to minimize the misleading nature of Gabbard's statement.

In context, Gabbard referred to "lip service" Trump offers to "our veterans, to our troops," but PolitiFact lops off "to our troops" in its headline and deck material. That truncated version of Gabbard's statement makes it appear reasonable to assume Gabbard was talking about veterans and not active service members.

Put simply, PolitiFact manipulated Gabbard's statement to help make it match the interpretation PolitiFact's liberal bloggers gave it in the story. PolitiFact not only chose not to deal with the obvious way Gabbard's statement might mislead people, but also chose not to transparently disclose that decision to its readers.

Principles Forsaken

PolitiFact's statement of principles is a sham. Why? Because PolitiFact applies the principles so haphazardly that we might as well call the result situational ethics. The ideology of the claimant appears to serve as one of the situational conditions driving the decision as to which principle to apply in any given case.

In Gabbard's case, she made a statement that could easily be interpreted in a way that makes it false. And PolitiFact often uses that as the justification for a harsh rating. In its statement of principles PolitiFact says it takes into account whether a statement is literally true (or false). It also says PolitiFact takes into account whether the statement is open to interpretation (bold emphasis added).:
The three editors and reporter then review the fact-check by discussing the following questions.
• Is the statement literally true?
• Is there another way to read the statement? Is the statement open to interpretation?
• Did the speaker provide evidence? Did the speaker prove the statement to be true?
• How have we handled similar statements in the past? What is PolitiFact’s jurisprudence?
PolitiFact effectively discarded two of its principles for the Gabbard fact check.

We say that a fact-checking organization that does not apply its principles consistently cannot receive credit for consistent non-partisanship or fairness.

With PolitiFact, "words matter" sometimes.



Afters

We've always been open to legitimate examples showing PolitiFact's inconsistency causing unfair harm to liberals or Democrats.

The examples remain few, in our experience.

Sunday, July 14, 2019

PolitiFact Texas punches "spin cycle" for Julián Castro

Is it possible left-leaning fact checkers still do not realize how their ideologies affect their work?

Consider PolitiFact Texas.


See what PolitiFact Texas did, there?

It presents an accurate hybrid paraphrase quotation of Castro asserting that Section 1325 of U.S. immigration law was put into place in 1929 by a segregationist. And promptly spins the Castro claim into the innocuous-but-loaded "When did it become a crime to cross the U.S.-Mexico border?"

Hilariously, the fact check spends most of its time examining facts other than when it became a crime to cross the border. Instead, it focuses on whether Section 1325 was enacted in 1929 (finding it was not) and whether the legislator who wrote the legislation was a segregationist.

PolitiFact Texas used nine paragraphs to address the segregationist past of Sen. Coleman Livingston Blease, the man who composed the language of an immigration bill in 1929.

Castro was evidently trying to make the point that he was trying to repeal a racist piece of legislation, racist because it was written by a segregationist. Castro was using the genetic fallacy on his audience. PolitiFact took no note of it, instead playing along by fact-checking whether Blease was a segregationist and finding a politically active expert to opine that the Blease-authored legislation was aimed at immigration from Mexico.

Such background does not help establish when it became a crime (at least under certain conditions) to cross the U.S.-Mexico border. It's background information that just happens (?) to support the subtext of Castro's claim.

As for Castro's implication that it was Blease who implemented the policy--as though a U.S. senator has that kind of power--well, that's just not the sort of thing that interests PolitiFact Texas.

In the end, PolitiFact Texas found it false that Blease authored the section of the immigration law Castro mentioned.

But why should that stand in the way of a favorable "Mostly True" rating?

PolitiFact's summary conclusion (bold emphasis added):
Castro said Section 1325 immigration policy, which makes it a crime to enter the country illegally, was "put into place in 1929, by a segregationist."

Technically Blease — a white supremicist [sic] who advocated for segregationist policies and lynching —  was not the author of the statute on illegal entry into the United States as it exists in today’s immigration code. 

But it was the first policy criminalizing all unlawful entry at the nation's southern border, and is considered the foundation of the 1952 policy that  evolved into today's Section 1325.

We rate this claim Mostly True.
Technically false, therefore "Mostly True."

That's how PolitiFact rolls. That's how PolitiFact Texas rolls.

It's spin, not fact-checking. Castro did not assert that the criminalization policy started in 1929. Castro asserted that Section 1325 was put into place in 1929.

Fact checkers should prove capable of noticing the difference. And keeping the spin out of their fact checks.

Friday, July 12, 2019

PolitiFact Unplugs 'Truth-O-Meter' for Elizabeth Warren

We seem to be seeing an increase of fact check stories from PolitiFact that do not feature any "Truth-O-Meter" rating. One of the latest pleads that it simply did not have enough information to offer a rating of Democratic presidential candidate Elizabeth Warren's claim that the U.S. Women's National Team (soccer) pulls in more revenue while receiving less pay than the men.

But look at the low-hanging fruit!


The women on the USWNT are not doing equal or better work than the men if the women cannot beat the men on the pitch. The level of competition is lower for women's soccer. And Warren's introduction to her argument is not an equal pay for equal work argument. It is an argument based on market valuation aside from the quality of the work.

It's reasonable to argue that if the women's game consistently creates more revenue than the men's game then the women deserve more money than the men.

That's not an equal pay for equal work argument. Not by any stretch of the imagination.

It was ridiculous for Warren to make that stretch in her tweet and typical of left-leaning PolitiFact to ignore it in favor of something it would prefer to report.

Did that principle of burden of proof disappear again?

PolitiFact's statement of principles includes a "burden of proof" principle that PolitiFact uses to hypocritically ding politicians who make claims they don't back up while allowing PolitiFact to give those politicians ratings such as "False" even if PolitiFact has not shown the claim false.

The principle pops out of existence at times. Note what PolitiFact says about its evidence touching Warren's claim:
Ultimately, the compensation formulas are too variable — and too little is known about the governing documents — for us to put Warren’s claim on the Truth-O-Meter.
 So instead of the lack of evidence leading to a harsh rating for Warren, in this case it leads to no "Truth-O-Meter" rating at all.

Color us skeptical that PolitiFact could clear up the discrepancy if it bothered to try.


Afters

Given Warren's clear reference to "equal pay for equal work," we should expect a fact checker to note that women who compete professionally in soccer cannot currently field a team that would beat a professional men's team.

Not a peep from PolitiFact.

Women's national teams do compete against men on occasion. That is, they do practice scrimmages against young men on under-17 and under-15 teams. And the boys tend to win.

But PolitiFact is content if you don't know that. Nor does its audience need to know that the U.S. Women's National Team's success makes no kind of coherent argument for equal pay for equal work.

Thursday, June 27, 2019

Selection Bias, Magnified

How PolitiFact uses inconsistent application of principles to help Democrats, starring Beto O'Rourke


PolitiFact Bias has repeatedly pointed out how PolitiFact's selection bias problem serves as a trap for its left-leaning journalists (that likely means somewhere between most and all of them). Left-leaning journalists are likely to fact check suspicious claims that look suspicious to left-leaning journalists.

But beyond that left-leaning journalists may suffer the temptation of looking at statements through a left-leaning lens. Fact-checking a Democrat may lead to confirmation bias favoring the Democrat's statement. The journalist may, perhaps unconsciously, emphasize evidence confirming claims coming from liberal sources. Or cutting the fact-finding process short after finding enough to supposedly confirm what the Democrat said.

When Democratic presidential hopeful Beto O'Rourke claimed to have received more votes than any Democrat in the history of Texas, PolitiFact Texas fact-checked the claim and found it "True."

Note that the fact check was written by long-time PolitiFact staffer Louis Jacobson. PolitiFact National employs Jacobson.

It is literally true that O'Rourke received the most votes for a Democrat ever received in the state of Texas. But literal truth is rarely the benchmark for fact checkers. In this case, we immediately noticed a problem with O'Rourke's claim that typically causes fact-checkers to find fault: As the number of voters in Texas grows, the number of raw votes received shrinks in significance. Measuring the percentage of the total vote (48.3 percent for O'Rourke) or the percentage of registered voters (about 25.6 percent) offers a more complete picture of a candidate's electoral strength in a given state.

For comparison, President Jimmy Carter won Texas in 1976 with 2,082,319 votes. Carter's percentage of the vote was 51.1 percent. His percentage of registered voters was 31.2 percent. It follows that Carter's performance in Texas was stronger than O'Rourke's even though Carter received about half as many votes as O'Rourke received.

We pointed out the problem to a PolitiFact Texas employee on Twitter. PolitiFact elected not to update the story to address O'Rourke's potentially misleading point about his electoral strength.

But it's justified resisting the efforts of conservatives to "work the refs," right? Who would think of trying to put the number of votes in context like we did other than right wing zealots?

Try the BBC, for starters. BBC noted that Hillary Clinton received the most presidential votes in history, then promptly tempered that statement of fact with a caveat:
So the proportion of Clinton votes might be more illuminating than simply how many votes she earned.
Indeed. And even PolitiFact Texas devoted more than one paragraph to the context O'Rourke had left out. Yet PolitiFact had the left-leaning sense not to let that missing information interfere with the "True" rating it bestowed on O'Rourke.
Our ruling

O’Rourke said that in 2018 when he ran for senator, "young voter turnout in early voting was up 500%. We won more votes than any Democrat has in the history of the state of Texas."

His assertion about young voter turnout is backed up by an analysis of state election data by the firm TargetSmart. And he’s correct that no Democrat has ever won more raw votes in a Texas statewide election than he has, an accomplishment achieved through a combination of his own electoral success, a pro-Democratic environment in 2018, and Texas’ rapid population growth in recent years.

We rate his statement True.
PolitiFact does not count the missing information significant, even though it was apparently significant enough to mention in the story.

Partial review of PolitiFact's rating system:
MOSTLY TRUE – The statement is accurate but needs clarification or additional information.

HALF TRUE – The statement is partially accurate but leaves out important details or takes things out of context.
If O'Rourke's statement did not need clarification or additional information, such as the growing number of voters in Texas, then why did PolitiFact provide that clarifying information?

These gray area "coin flips" between ratings offer yet another avenue for left-leaning fact-checkers to express their bias.

PolitiFact has never revealed any mechanism in its methodology that would address this weakness.

Thursday, June 13, 2019

Transparency: How to access PolitiFact's page of corrected or updated fact checks

It has long amused us here at PolitiFact Bias how difficult PolitiFact makes it for readers to navigate to its page of corrections and updates. There are pretty much three ways to navigate to the page.


Someone could link to it by hotlinking using the page URL.

This is the method PolitiFact uses to make finding the page seem easy-peasy in tweets or other messages. Works great!



The reader could use a search engine to find it

No, not the search function at the PolitiFact website. That will not get you there.

We're talking about a search engine like Google or DuckDuckGo. Search politifact + corrections + and + updates and reaching the page is a snap.


The reader could navigate to the page from PolitiFact's homepage. Maybe. 

This is the amusing part. We've already noted that using the "search" function at the PolitiFact website won't reach its dedicated page of corrected and updated fact checks (other corrections and updates do not yet end up there, unfortunately).

And without a guide such as the one that follows, most people browsing PolitiFact's website would probably never stumble over the page.

How To Do It

Step 1: On the homepage, move the cursor to the top menu bar and hover over "Truth-O-Meter" to trigger the drop-down menu
Step 2: Move the cursor down that menu to "By Subject," click on "By Subject"
Step 3: On the "Subjects" page, move the cursor to the alphabet menu below the main menu, hover over "c," click "c"
Step 4: Move the cursor to the subjects listed under "c," move cursor to hover over "Corrections and Updates," click "Corrections and Updates"

Done! What could be easier?

The key? Knowing that PolitiFact counts "Corrections and Updates" as a category of "statements" defined by PolitiFact as Truth-O-Meter stories. The list of corrections and updates consists only of fact checks. Corrections or updates of explainer articles, promise ratings and flip-flop ratings (etc.) do not end up on PolitiFact's page of corrections and updates.

What you'll find under "c" at PolitiFact.com



Afters


When I (Bryan) designed the Zebra Fact Check website, I put the "Corrections" link on the main menu.



It's not all about criticizing PolitiFact. It's also about showing better and more transparent ways to do fact-checking.

This isn't exactly rocket science. Anybody can figure out that putting an item on the main menu makes it easy to find.

There is reason to suspect that PolitiFact is less than gung-ho about publicizing its corrections and updates.

Wednesday, May 29, 2019

More Deceptive "Principles" from PolitiFact

PolitiFact supposedly has a "burden of proof" that it uses to help judge Political claims. If a politician makes a claim and supporting evidence doesn't turn up, PolitiFact considers the claim false.

PolitiFact Executive Director Aaron Sharockman expounded on the "burden of proof" principle on May 15, 2019 while addressing a gathering at the U.S. Embassy in Ethiopia:
If you say something, if you make a factual claim, online, on television, in the newspaper, you should be able to support it with evidence. And if you cannot or will not support that claim with evidence we say you're guilty.

We'll, we'll rate that claim negatively. Right? Especially if you're a person in power. You make a claim about the economy, or health, or development, you should make the claim with the information in your back pocket and say "Here. Here's why it's true." And if you can't, well, you probably shouldn't be making the claim.
As with its other supposed principles, PolitiFact applies "burden of proof" inconsistently. PolitiFact often telegraphs its inconsistency by publishing a 'Splainer or "In Context" article like this May 24, 2019 item:


PolitiFact refrains from putting Milano's statement on its cheesy "Truth-O-Meter" because PolitiFact could not figure out if her statement was true.

Now doesn't that sound exactly like a potential application of the "burden of proof" criterion Sharockman discussed?

Why isn't Milano "guilty"?

In this case PolitiFact found evidence Milano was wrong about what the bill said. But the objective and neutral fact-checkers still could not bring themselves to rate Milano's claim negatively.

PolitiFact (bold emphasis added):
Our conclusion

Milano and others are claiming that a new abortion law in Georgia states that women will be subject to prosecution. It actually doesn’t say that, but that doesn’t mean the opposite — that women can’t be prosecuted for an abortion — is true, either. We’ll have to wait and see how prosecutors and courts interpret the laws before we know which claim is accurate. 
What's so hard about applying principles consistently? If somebody says the bill states something and "It actually doesn't say that" then the claim is false. Right? It's not even a burden of proof issue.

And if somebody says the bill will not allow women to be prosecuted, and PolitiFact wants to use its "burden of proof" criterion to fallaciously reach the conclusion that the statement was false, then go right ahead.

Spare us the lilly-livered inconsistency.

Friday, May 17, 2019

PolitiFact gives "policy trajectory" a "False" rating

Earlier this week PolitiFact Executive Director Aaron Sharockman said PolitiFact does not rate opinions or predictions.

Also this week, PolitiFact Health Check, the PolitiFact partnership with Kaiser Health News, apparently contradicted Sharockman's claim.

Behold:


If it looks like PolitiFact is fact-checking a prediction, that's because PolitiFact is fact-checking a prediction.

More than one of PolitiFact's pool of four experts apparently saw it exactly that way (bold emphasis added):
But, Adler said, the structure of Trump’s claim — promising what his administration "will" do, rather than commenting on what it has done — leaves open the possibility of taking other steps to keep preexisting condition protections in place.

That’s true, other experts acknowledged. So far, the White House has postponed a legislative push until after the 2020 election — leaving a vacuum if the courts do wipe out the health law.
History shows that PolitiFact will not allow mere expert opinion to stand in the way of the hoped-for narrative. When experts offer opinions that do not fit comfortably with PolitiFact's conclusion, PolitiFact ignores them.

Hilariously, PolitiFact's conclusion uses language strongly suggesting an awareness that it is rating a prediction:
Our ruling

Trump said his administration will "always protect patients with preexisting conditions."

The White House’s policy trajectory does exactly the opposite.
What is "policy trajectory" if it is not a projection of what will happen tomorrow based on Trump administration policy today?

PolitiFact Health Check is not fact-checking Trump. It is rating a pledged policy position. PolitiFact could potentially address such cases with an expansion of its ratings of executive promises ("Trump-O-Meter" etc.).

But this so-called "fact check" makes Sharockman a liar if it isn't corrected somehow.

Monday, May 6, 2019

PolitiFact unfairly harms Joe Biden

On May 6, 2019, PolitiFact fact-checked a claim from Democratic Party presidential hopeful (and frontrunner) Joe Biden.

Biden said he was "always" labeled as one of the most liberal Democrats in Congress.

PolitiFact rated Biden's claim "False." Perhaps the rating is fair. But PolitiFact's would-be paraphrase of Biden's claim, below, treats Biden unfairly.


We think there's room for one to count as a "staunch liberal" without always counting as a one of the most liberal.

PolitiFact, for purposes of its headline, changed Biden's claim from one to the other. In terms of its messaging, PolitiFact offers the opinion that Biden does not count as a staunch liberal.

We think fact checks should stick to the facts and not make headlines out of their opinions. PolitiFact's opinion, trumpeted above its fact check, unfairly harmed Biden.


Note: We have always said that PolitiFact's problems go beyond left-leaning bias. PolitiFact represents fact-checking done poorly. The bad fact-checking unfairly harms right and left, with the right getting the worst of it.