Sunday, February 18, 2018

PolitiFact partially unveils spectacularly transparent description of its fact-checking process

"The Week in Fact-Checking," an update on the latest fact-checking news posted at the Poynter website, alerted us to the fact that PolitiFact has updated its statement of principles:
PolitiFact made their methodology more transparent, in keeping with other fact-checkers around the world. (And ICYMI,  PolitiFact has moved its headquarters to Poynter, earning a not-for-profit designation.)
We were surprised we had missed PolitiFact's welcome improvement to its methodological transparency. So we visited to check it out.

So ... where is it?

PolitiFact created multiple pages of transparent new content and apparently neglected to equip its website with internal links leading readers to the new content.

Clicking "About Us>>Our Process" on the main menu takes the reader to PolitiFact's 2013 statement of principles.

Clicking "Our Process" on the footer takes the reader to PolitiFact's 2013 statement of principles

There's no apparent way to use PolitiFact's main page to find the new even-more-transparent(!) statement of principles.

But people can see PolitiFact's latest extreme transparency through the website. Or maybe via links posted to Twitter. We haven't noticed any yet, but it's possible.

So there's that.

The new material published on Feb. 12, 2018. As of Feb. 18, 2018, still funneled readers to its 2013 statement of principles.

We see that as illustrative of the PolitiFact bubble. PolitiFact judges its transparency according to its belief it has published a new statement of principles. Those outside the PolitiFact bubble, unaware of the new statement of principles thanks to PolitiFact's oversight, do not likely take the same view of PolitiFact's transparency.

Why are those outside the bubble so ignorant of PolitiFact's extreme transparency?

Monday, February 12, 2018

Guest columnist (Democrat) critiques PolitiFact

We covered PolitiFact's announcement it had hired Democratic and Republican "reader advocates" to help establish its trustworthiness. And we covered how PolitiFact unpublished that announcement when its choice of Alan Grayson, former Democratic congressman from Florida, blew up in its face.

Another announcement followed on Feb. 9, 2018, naming Republican David Jolly and Democrat Jason Altmire as "guest columnists."

The guest columns appear on PolitiFact's blog page, "Inside the Meters," which should prove sufficient to bury the columns outside the notice of anybody who doesn't either get Twitter or email alerts directly from PolitiFact.

Altmire was the first to have a critique published.

We think it's pointless crap.

Altmire says PolitiFact "generously" rated a Republican "Half True." Then later in the column says the "Half True" rating is the correct rating.

No, seriously. That's what Altmire does.

In the lead paragraph, Altimire says PolitiFact gave a "generous" Half True rating (bold emphasis added):
PolitiFact generously rates Congressman Mullin’s Facebook post "Half True." He got the numbers right, but failed to inform readers of the context. In evaluating claims involving the selective use of statistics, PolitiFact must consider whether the omission was accidental or meant to deceive. Mullin’s omission appears to have been purposeful, because he knows an evaluation of Obama’s entire economic record would present a completely different picture than the one the congressman was trying to paint. Is "Half True" an accurate rating in this case?
 And in his concluding paragraph, Altmire says PolitiFact got the rating right (bold emphasis added):
PolitiFact gave the correct rating; Mullin’s post was indeed "Half True." But from now on, when readers consider a statement that has been rated "Half True" based upon the misuse of statistics, I hope they will remember the less-than-complimentary implication of that rating.
("from now on"?????)

Altmire's point, cleverly disguised in the midst of his self-contradiction, was that Congressman Mullin was lying, and Altmire wishes PolitiFact had been more clear about it.

As critiques go, that's plenty lame. Media Matters could have come up with that one with no problem.

If these columnists don't pick up their game immediately, PolitiFact ought to waste no time at all pulling the plug on this experiment.

Saturday, February 10, 2018

How we made our meme mocking PolitiFact

Earlier this week we noticed PolitiFact making yet another hypocritical declaration. PolitiFact has ruled it misleading to use "cuts" to refer to reductions to a future projected spending baseline. In many cases a budget might increase year by year but the legislature "cuts" spending by slowing its increase.

In the past, we've pointed out how PolitiFact tended to rate Republicans "Mostly False" for claiming the Affordable Care Act cut Medicare by hundreds of millions of dollars. When President Donald R. Trump and the Republican Congress tried the same thing with Medicaid in 2017, PolitiFact discovered that the claim was "Half True" on the few occasion(s?) it noticed the Democrats' ubiquitous claim and then quickly lost interest.

Fast forward to 2018, and PolitiFact published a fact check of a Trump statement about protests over the United Kingdom's National Health Service, its universal care program. PolitiFact treated Trump unfairly by rating him on something he did not say, but what really knocked our socks off was a sentence PolitiFact reeled off in its summary:
While the NHS has lost funding over the years, the march that took place was not in opposition to the service, but a call to increase funding and stop austerity cuts towards health and social care.
The problem? You guessed it! Spending has gone up for the NHS pretty consistently. The fact checkers at Britain's Full Fact even did a fact check in January 2018 relating to NHS funding. It only reported spending going up.
Spending on the NHS in England has increased in real terms by an average of around 1% a year since 2010. Since the NHS was established spending increases have averaged 4% per year.
So the NHS hasn't "lost funding" except against baseline future spending. The austerity "cuts" PolitiFact reports are a decrease of the rate of future spending.

PolitiFact is making a claim it has rated "Half True" and worse in the past.

We don't appreciate that type of hypocrisy from a supposedly non-partisan and objective fact checker. So we went to work on meme.

First, we looked at PolitiFact's list of stories with the "Medicare" tag. We knew we'd find stories reporting on budget cuts to a baseline. And from those stories we looked for one with a summary that would fit the present case. It didn't take long. We found a "Half True" rating from PolitiFact Ohio that fit the bill:

"So-called cut reflects savings from slowing growth in spending." Doesn't that sound much better than "cutting Medicare"? Hurrah! It's savings!

Our next step was to replace the text and image to the left of the "Truth-O-Meter" graphic. We decided to pin the blame on PolitiFact Editor Angie Drobnic Holan instead of on the intern who wrote and researched the fact check. Holan had good reason to know PolitiFact's history on rating cuts to a future baseline.

We took Holan's image from her Twitter account.

We replaced the text with Holan's name and the outrageous quotation from the Trump fact check.

We credited our faux fact check to "PolitiFact National" on the day the Trump fact check came out. We skipped the em-dash this time since it takes a few extra steps.

And we put a big "PARODY" watermark on the whole thing to make clear we're not trying to trick anybody. The point is to mock PolitiFact for its inconsistency.

Our finished product:

Seriously: It's ridiculous for a national fact-checking service to do such a poor job of reporting consistently. Holan is the chief editor, and she doesn't notice this clear problem? She let the intern down by not catching it. And how long will it take to correct the problem? Eternity?

PolitiFact's past work on budget cuts is already so chaotic that one more miss hardly matters. We don't expect anything to change. PolitiFact will go right on giving readers a slanted view of budget cuts.

For that matter, we expect the other two of America's "elite three" fact checkers to independently follow the same misleading pattern PolitiFact uses. That's what happens when all three lean left.

Thursday, February 8, 2018

Are fact checkers fact-checking opinions more? Blame Trump! (Updated)

The fact checkers at PolitiFact apparently can't keep themselves from allowing their opinions to seep into their work.

Fortunately, we can all blame President Trump. That way, the fact checkers need not acknowledge any error.

A Feb. 6, 2018  PolitiFact fact check took as an assertion of fact Trump's apparent opinion that the word "treason" might apply to Democrats who failed to applaud good news about the United States during Trump's State of the Union Address.

PolitiFact, in classic straw man fashion, insisted that "treason" had to refer to the type codified in law, and so rated Trump's claim "Pants on Fire" (bold emphasis added):
Trump said that at the State of the Union address, Democrats, "even on positive news … were like death and un-American. Un-American. "even on positive news … were like death and un-American. Un-American. Somebody said, ‘treasonous.’ I mean, yeah, I guess, why not? Can we call that treason? Why not?"

There’s a good reason why not: Declining to applaud the president doesn’t come anywhere near meeting the constitutionally defined threshold of treason, which in any case can’t occur except in wartime. Rather, legal experts agree that it is a clear case of constitutionally protected free speech. We rate the statement Pants on Fire.
In fact, "treason" has a broader definition than PolitiFact allowed:

  1. the offense of acting to overthrow one's government or to harm or kill its sovereign.
  2. a violation of allegiance to one's sovereign or to one's state.
  3. the betrayal of a trust or confidence; breach of faith; treachery. 
Failing to applaud good news about one's state would, in a sense, violate allegiance to one's state. And, more to the point, one can define words as one likes. One could, for example, choose to define the word "Rump" to refer exclusively to President Trump. One can do such things because words are ultimately just symbols representing ideas, and people can choose what idea to associate with what symbol.

Is it a good idea to use words in ways that run against their commonly understood meanings? That's a different issue.

Trump afforded his critics another marvelous opportunity to criticize his temperament and wisdom, but that criticism belongs in op-eds, not fact checks.

The dastardly Trump forced helpless journalists to abandon their objectivity.

How dare he.

Update Feb. 8, 2018

We weren't going to make a big deal of PolitiFact saying that Trump was suggesting that not applauding for him (Trump) might qualify as treason.

But then PolitiFact started emphasizing that misleading headline on Twitter:
That's just bad reporting, and it's a classic example of a biased headline. Trump says failing to applaud good news about the United States might pass as treason, not the failure to applaud President Trump.

Nonpartisan and objective journalists should be able to distinguish between the two.

Tuesday, February 6, 2018

PolitiFact: One standard for me, and another for thee

On Feb. 5, 2018, PolitiFact published an article on cherry picking from one of its veteran writers, Louis Jacobson. Titled, "The Age of Cherry-picking," it led with a claim of fact as its main hook:
These days, it isn’t just that Republicans are from Mars and Democrats are from Venus. Increasingly, politicians on either side are cherry-picking evidence to support their version of reality.
With cherry-picking on the increase, and with both sides using it more, certainly readers would want to see what PolitiFact has to say about it.

But is it true? Is cherry-picking on the increase?

One had to read far down the column to reach Jacobson's evidence (bold emphasis added):
So is there more cherry-picking today in political rhetoric than in the past? That’s hard to say -- we couldn’t find anyone who measures it. But several political scientists and historians said that even if it’s not more common, the use of the tactic may have turned a corner.

If a writer tries to hook me into reading a story based on the claim that cherry-picking is on the increase, then takes over 20 paragraphs before getting around to telling me that no good evidence supports the claim, I want my money back.

This isn't hard, fact checkers. If it's hard to say if there is more cherry-picking today in political rhetoric than in the past, don't say "Increasingly, politicians on either side are cherry-picking evidence to support their version of reality."

Don't do it.

Even a Democrat probably couldn't entirely get away with a claim so poorly supported by the evidence, thanks to PolitiFact's occasionally-applied principle of the burden of proof:
Burden of proof – People who make factual claims are accountable for their words and should be able to provide evidence to back them up. We will try to verify their statements, but we believe the burden of proof is on the person making the statement.
We used Twitter to needle PolitiFact over this issue, surprisingly drawing some response (nothing of substance). But the exchange ended up productive when co-editor Jeff D, who runs the PFB Twitter account, contributed this summary:
That about sums it up. One standard for me, and another for thee.

Update Feb. 7, 2018: Supplied URL to PolitiFact's article on cherry picking, added tag labels.

Monday, February 5, 2018

Does "lowest" mean something different in Georgia than it does in Texas?

Today PolitiFact National, posing as PolitiFact Georgia, called it "Mostly True" that Georgia has the lowest minimum wage in the United States.

Georgia law sets the minimum wage at $5.15 per hour, the same rate Wyoming uses, and the federal minimum wage of $7.25 applies to all but a very few Georgians. PolitiFact National Georgia hit Democrat Stacey Evans with a paltry "Mostly True" rating:
Evans said Georgia "has the lowest minimum wage in the country."

Georgia’s minimum wage of $5.15 per hour is the lowest in the nation, but Wyoming also has the same minimum wage.

Also, most of Georgia’s workers paid hourly rates earn the federal minimum of $7.25.

Evans’ statement is accurate but needs clarification or additional information. We rate it Mostly True.
Sounds good. No problem. Right?

Eh. Not so fast.

Why is it okay in Georgia for "lowest" to reasonably reflect a two-way tie with Wyoming, but in Texas using "lowest" where there's a three-way tie earns the speaker a "False" rating?

How did PolitiFact Texas justify the "False" rating it gave the Republican governor (bold emphasis added)?
Abbott tweeted: "The Texas unemployment rate is now the lowest it’s been in 40 years & Texas led the nation last month in new job creation."

The latest unemployment data posted when Abbott spoke showed Texas with a 4 percent unemployment rate in September 2017 though that didn't set a 40-year record. Rather, it tied the previous 40-year low set in two months of 2000.

Abbott didn’t provide nor did we find data showing jobs created in each state in October 2017.

Federal data otherwise indicate that Texas experienced a slight decrease in jobs from August to September 2017 though the state also was home to more jobs than a year earlier.

We rate this claim False.
 A tie goes to the Democrat, apparently.

We do not understand why it is not universally recognized that PolitiFact leans left.

Correction/clarification Feb. 5, 2018:
Removed unneeded "to" from the second paragraph. And added a needed "to" to the next-to-last sentence.

Thursday, February 1, 2018

The bird's-eye lowdown on PolitiFact's partisan reader advocates (Updated)

Today PolitiFact announced it will publish content from two reader advocates at its website.

The announcement didn't go so well. Shortly after making the announcement PolitiFact nixed Democrat Alan Grayson's planned involvement.

The Hill reports:
Fact-checking website PolitiFact on Thursday announced that it had hired former Florida Reps. David Jolly (R) and Alan Grayson (D) as “reader advocates” before hours later nixing Grayson's hire after fierce backlash.
The dumping of Grayson aside, does this represent a sincere effort from PolitiFact to help improve its product?

Probably not, despite the we're-so-humble sales job from PolitiFact Executive Director Aaron Sharockman.

Huh. Well, we intended to link directly to PolitiFact's announcement about its new reader representatives, but it looks like PolitiFact unpublished it. We'll go with the reporting from the Hill instead, which has coincidentally also been altered since its publication: (an earlier version carried a hyperlink to PolitiFact's announcement)
The two former lawmakers had been set to critique the website’s fact-checks and provide additional insight on political issues as part of a pilot program that will run through the end of April, according to a Thursday PolitiFact post announcing the hires. The post has since been deleted.

“David and Alan are both particularly qualified, we think, to critique the work of PolitiFact, because they’ve been subject to our fact-checks as members of Congress,” PolitiFact Executive Director Aaron Sharockman wrote an his initial statement.

Well-qualified to critique the work of PolitiFact! That implies that the critiques may prove valid, right?

PolitiFact said it would learn from Jolly and Grayson, though it's hard for us to prove that now that PolitiFact is disposing of the evidence. Like a good fact-checker should, I guess.

We think PolitiFact's flirtation with reader advocates fits well with its old narrative about how it receives criticism from both sides. Getting criticized from both sides, using mumbo-jumbo logic, shows the reliability of the entity getting criticized. That's a superb narrative for PolitiFact to promote. It's certainly far better than the alternative narrative, that getting criticized by both sides means you're probably doing something wrong.

PolitiFact's founding editor, Bill Adair, did research suggesting that PolitiFact receives most of its substantive criticism from the right. Will PolitiFact allow such research to impact its choice of which narrative to promote? We doubt it.

We captured images from Sharockman's Twitter feed--ones he had so far elected not to delete. The one at the top pretty much shows PolitiFact's thinking behind this experiment. It's not about improving the product. It's about encouraging the public to trust in the existing product.

Note that PolitiFact's experiment was only slated to run through April 2018. Three months. If PolitiFact detects signs that people are trusting it less during the experiment, it will terminate the experiment.

Does that sound like a sincere effort to improve the product? Or more like a cynical ploy intended to trick people into trusting PolitiFact?

How many times do we have to say it? One gains trust by proving trustworthy. One proves one's trustworthiness through accuracy and transparency.

Making published works entirely disappear is not transparency.

Jeff adds:

We wonder what it says about PolitiFact that they considered Grayson representative of a conventional Democrat voice.

We'd also like to congratulate the fact checkers on selecting conservative powerhouse David ... uh ... *checks notes* ... Jolly.

We're not surprised that PolitiFact's ill-advised attempt to gain credibility resulted in more of the same deception we've come to expect. For example, un-publishing articles is somehow a sign of trustworthiness?

For years we've said PolitiFact would benefit from an inside critic of their work and suggested most of their obvious blunders would have been prevented with a heterodox voice on staff. On its face the notion of "reader advocates" seems like a step in that direction, until you realize it's just another click-seeking gimmick (Is Grayson really the top pick for any serious endeavor?).

If PolitiFact were actually sincere about gaining reader trust there's more effective ways than adding sideshow acts performed by clowns and cranks. For instance, they could unequivocally disavow their longtime use of stealth edits. That might help bring them into compliance with the International Fact Checking Network code of principles that they currently violate.

But the most obvious thing they could do to improve their image is to credibly rebut the volumes of legitimate, earnest criticism of their work. So far, PolitiFact's response to charges of bias has been to call critics "mental" or to ignore them altogether. For some reason PolitiFact does not view an honest defense of their work as a viable remedy for reader distrust.

PolitiFact's incompetent and biased editorializing has never earned credibility. Adding more clowns to the car won't change that. This "readers advocate" stunt only shows how unserious PolitiFact is about providing readers the truth.

Edit 0835PST 2/2/2018: Added word "PolitiFact" in penultimate graph -Jeff

Imperious PolitiFact

PolitiFact decides who built what, and it's ridiculous

Back in 2016 we reviewed the "True" claim PolitiFact awarded First Lady Michelle Obama for her claim the White House was built by slaves.

Slaves definitely helped with the labor of constructing the White House, but to an unknown degree. Regardless of that, PolitiFact awarded Obama a "True" rating on its subjective "Truth-O-Meter."

We thought PolitiFact went too easy on the claim, given that one could use the same standard to claim it "True" that European immigrants built the White House. Including a word like "helped" allows either claim to rise to credibility.

Fast forward to 2018 and the State of the Union Address response from U.S. Rep. Joe Kennedy III (D-Mass.).

Kennedy said immigrants built Fall River, Massachusetts.

Breitbart, a right-leaning news outlet, judged Kennedy's statement "Mostly False," reasoning that the establishment of Fall River by native-born descendants of English settlers made it reasonable to say the city was built, at least in part, by those native-born people. Breitbart added that the native-born population has always outnumbered immigrants in the county that contains Fall River.

PolitiFact apparently doesn't care for sharing credit. If one group helped build something then that group gets credit and other groups that helped do not get credit.

PolitiFact rated Breitbart's claim "False." Yes, that implies that immigrants who helped "build" Fall River by coming to work at factories established by the native residents were the ones who exclusively built Fall River.

Call us radical right-wingers if you like, but we think if the facts show that credit for building something should be shared, then a fact checker should acknowledge shared credit in its ratings.

Breitbart's "Mostly False" rating hints at an ability to make that type of acknowledgement.

PolitiFact's "True" and "False" ratings make it look more partisan than Breitbart.