Offsite Criticism of the Week

While PolitiFact continues its pattern of ducking public criticism, we will continue to model the opposite approach by dealing with a typical offsite criticism we run across (real example from real discussion board):

First, we'll note that generally citing PolitiFact Bias is the wrong way to defuse criticisms that Trump doesn't tell the truth. There are good reasons to dismiss the use of the PolitiFact statistics that some use to "prove" somebody lies more than somebody else. The liberal use of selection bias lays a poor foundation for reliable statistics. We've been pointing that out for years.

As for the criticism:

The first article in the first link says PolitiFact is wrong because it only used the data that was available to make its ruling. Oh the humanity!

No.

We believe the comment was directed at our evaluation of the Loretta Lynch fact check where Lynch said the LGBT community endures most of the statistically recorded hate crimes. It's true, as Dustin Siggins pointed out, that PolitiFact relied on incomplete statistics. In keeping with PolitiFact's common practice, that should have prevented Lynch from getting a rating any better than "Mostly True." Lynch received a "True" rating.

But our central criticism was quite different from what the writer said.

We noted that Lynch said nothing, in context, about using the per capita measure. If that measure isn't used, then PolitiFact's statistics say blacks experience most of the hate crimes recorded in the statistics. So what Lynch said was false. PolitiFact pretended otherwise to give her the "True" rating.

This is typical of the criticism we receive. We're dismissed based on bias (ad hominem fallacy) and on misrepresentations of our work (straw man fallacy).

We'd love to do "Offsite Criticism of the Week" for a criticism that is not fallacious. Comments are open here and on Facebook if anybody knows of one.

Poynter hides criticism of PolitiFact's transparency?

The Poynter Institute, a journalism school in St. Petersburg, Florida, owns our reason for existence, PolitiFact. I sometimes read articles posted at Poynter, particularly those concerned with fact-checking.

Up through today, I believe Poynter has published every comment I have made.

Poynter re-published the guidelines for fact-checking written by PolitiFact editor Angie Drobnic Holan. I had responded to Holan's PolitiFact article with one of my own published to Medium. At the Poynter website I commented about one of my central points of rebuttal: PolitiFact's claims of accountability to the public can't be taken seriously so long as a story from PolitiFact Missouri from earlier this year sustains its train-wreck status (and it does sustain that status).

I captured the above image right after pushing the button to publish my comment. Note the message that the comment is pending approval. When I visited the page later, I was still able to see evidence I had commented and the comment was pending approval. I was also able to review the content of the yet-to-be-published comment.

Now, when I return to the page, no evidence remains of my comment activity.

Is it possible that the comment remains in moderation and Poynter will publish it later? Sure, that's possible. We will certainly update this item if Poynter publishes the comment.

Meanwhile, here is our message to PolitiFact and the Poynter Institute:

Hiding a problem of transparency does not address that problem of transparency.

 
Update June 30, 2016:

 

Pending for 11 hours. Shall we go for 36?

More PolitiFingers on the scale: Loretta Lynch and LGBT hate crimes

Hat tip to journalist Dustin Siggins for bringing this item to our attention.

U.S. Attorney General Loretta Lynch received a "True" rating in a June 23, 2016 fact check by PolitiFact Florida. Lynch said members of the LGBT community are more often the victims of hate crimes than members of any other recognized community.

The "Truth-O-Meter" rating counts as an out-and-out gift to Lynch.

Our tipster, Dustin Siggins, pointed out that PolitiFact admittedly could only verify Lynch's claim according to a very incomplete data set:

An important caveat: There are several holes in the reporting of hate crimes to the FBI. Local law enforcement agencies voluntarily report their data to a state agency that compiles the information for the FBI. Some local agencies report no hate crimes or don’t submit a report.

A study by the federal Bureau of Justice Statistics found that 60 percent of violent hate crime victimizations were not reported to police in 2012.

A 2016 Associated Press investigation found that more than 2,700 city police and county sheriff's departments have not submitted a single hate crime report to the FBI during the past six years — about 17 percent of all city and county law enforcement agencies nationwide.

The caveat obviously wasn't important enough to drop Lynch's rating below "True."

So Siggins made a solid point about the quality of the data.

We were even more concerned about PolitiFact placing its focus on the "per capita" measurement of hate crimes.

There's simply nothing in Lynch's statement to suggest she was talking about a per capita measurement. PolitiFact reported that Lynch said she relied on a story in The New York Times for her information. And it's true that the Times' story focuses on per capita rates of hate crime.

But how does that excuse Lynch for the words she used?

We don't see how it could, and this type of imprecision frequently counts against the accuracy of a claim.

Compare a recent "False" rating PolitiFact gave to Republican presidential candidate Donald Trump. Trump said Americans pay the highest taxes in the world. He may well have been talking about the U.S. corporate tax rate, which PolitiFact said in 2014 was one of the highest in the world. Trump did not explicitly say he was talking about the corporate tax rate, however. Would it, or should it, make a difference if his campaign had responded to PolitiFact saying he was talking about the corporate tax rate?

We think statements get their degree of truth from their immediate context. Later explanations may help point to a different dimension of that context, even potentially revealing a hidden angle to a fact checker. But the later explanation per se is irrelevant to the truth of the claim.

If Not Per Capita Then Not True

According to the words she used and according to the statistics PolitiFact used, the LGBT community is not the group most victimized by hate crimes. Blacks are more often victimized by hate crimes, as PolitiFact plainly admitted (bold emphasis added):

Another way to look at the data about hate crimes rather than per capita is the sheer number of crimes. By that measure, there were more hate crimes against African-Americans than LGBT residents. But that’s not surprising since African-Americans represent about 13.2 percent of the population according to the U.S. Census, while the LGBT community represents about 2.3 percent of the population, according to a survey done by the Centers for Disease Control (other surveys found a slightly higher rate.)

PolitiFact ignores its own evidence showing Lynch was wrong, implicitly modifies Lynch's comment to refer to the per capita measurement, and awards Lynch a "True" rating.

That's fact-checking?

Power Line: "On crime, Trump's right and PolitiFact is wrong"

We profusely thank Power Line writer Paul Mirengoff for linking to PolitiFact Bias prominently in his post.

Power Line blog slammed PolitiFact today over its slipshod fact check of Donald Trump's claim that crime is increasing. PolitiFact absurdly rated Trump's claim "Pants on Fire" despite not considering crime data more recent than 2014:

How did Politifact err on such a basic question? It erred by looking at no data past 2014. Sean Kennedy at AEI Ideas blows the whistle.

Trump made his statement on June 7, 2016. Thus, his claim that crime is rising can only be fact-checked by analyzing current data. By failing to do so, Politifact confirmed that it is either incompetent, hopelessly biased, or both.

While it's true PolitiFact relied at least partly on a pair of experts it interviewed, Mirengoff and Kennedy make a great point. Where was PolitiFact in January 2016 when the Washington Post was claiming an increase in violent crime for early 2015 compared to early 2014?

The number of violent crimes committed across the country was up in the first half of 2015 compared with the same period a year earlier, with increases seen across the country and spanning different types of crimes, federal authorities said Tuesday.

The numbers of murders, rapes, assaults and robberies were all up over the first six months of 2015. Overall violent crime was up 1.7 percent, an increase that followed two consecutive years of declines, according to the FBI.

"No truth for you!" say the Truth Nazis at PolitiFact.

What About PolitiFact's Neutral Experts?

 

Sometimes we survey PolitiFact's list of experts to see if they have any obvious political leanings.

What have we here?

James Alan Fox: Just one FEC individual donation listed. To Elizabeth Warren, $800. It may not be PolitiFact's expert if there's another James Fox at Northeastern University. Warren's a Democrat.

Raymond Paternoster: The background information on Paternoster was equivocal. Paternoster has authored studies on race and the application of the death penalty. As with what he told PolitiFact, it's hard to confidently pin down his stance (bold emphasis added):

"Mr. Trump is wrong if he is talking about overall crime and even violent crime," agreed University of Maryland criminologist Raymond Paternoster. Any possible upward swing in the past year or so wouldn’t show up in the data currently available, he said.

Paternoster's admission in bold makes us very curious about the context of PolitiFact's interview. What question was Paternoster asked when he answered Trump is wrong? How could Paternoster agree that Trump is wrong without recent data to back the assessment?

More News Reports

 

The Associated Press:

CHICAGO (AP) — Violent crimes — from homicides and rapes to robberies — have been on the rise in many major U.S. cities, yet experts can't point to a single reason why and the jump isn't enough to suggest there's a trend.

Still, it is stumping law enforcement officials, who are seeking a way to combat the problem.

"It's being reported on at local levels, but in my view, it's not getting the attention at the national level it deserves," FBI Director James Comey said recently. "I don't know what the answer is, but holy cow, do we have a problem."

 A bunch of liars?

KUTV (Utah) cited a study by the left-leaning Brennan Center for Justice:

WASHINGTON (Sinclair Broadcast Group) — A new study of crime statistics from major cities across the country reveals a rising number of murders in 2015, with violence in three cities fueling half of that increase.

Crime data for the 30 largest cities in the U.S. released by the Brennan Center for Justice indicates a 13.3 percent rise in murders in 2015, but analysts say it is too soon to determine whether this reflects a broader trend.

Note the Brennan Center study involves a comparison between 2015 and 2014. Violent crime in 2016 has thus far built on the violent crime rate in 2015.

As Power Line noted, a fact checker should check the facts before ruling on the facts.





Edit 6/22/2016: Added link to WaPo story in relevant graph. Changed "in" to "an" same graph-Jeff 0857 PST

Allsides: "PolitiFact.com is a bit left"

We preface this post with the disclaimer that the editors of PolitiFact Bias tend not to trust crowdsourced methods of epistemology.

Back in 2012, we noted the emergence of AllSides, a new project intended to help sift through and identify media bias so news consumers could take it into account.

AllSides now has a preliminary assessment of PolitiFact posted to its site. The evaluation could change as more information comes in. But for now, AllSides says PolitiFact leans slightly left:

In comparison with other fact checking sites, it appears that PolitiFact.com is a bit left of FactCheck.org (bias rating "Center") and the Washington Post's Blog "The Fact Checker" (bias either "Center" or "Lean Left" - borderline).

Expect PolitiFact to pretend this evaluation does not exist, and expect journalists to refrain from asking PolitiFact's representatives about it.

'Cause that would be rude, right?

Visit AllSides to learn more about how it develops its bias ratings.

More word games at PolitiFact: voucher edition

On June 5, 2016, we reviewed PolitiFact's position, contrary to what economists write in published journals, that Social Security's "pay-as-you-go" financing is not a Ponzi game.

With this post, we'll see that PolitiFact adopts a different approach to the use of words when the term is "voucher."

Surprise! The inconsistency works against Republicans in both cases.

PolitiFact Wisconsin fact-checked a claim by the Democratic Party of Wisconsin that Republican Senator Ron Johnson voted to turn Medicare into a voucher program. PolitiFact Wisconsin rated the claim "Mostly True," ignoring the fact that the plans Johnson voted for involved no vouchers. Vouchers are pieces of paper representing value that will be covered by the government.

PolitiFact Wisconsin's explanation is priceless:

Our colleagues concluded that, although there are technical differences between voucher and premium supports that may matter to health policy professionals, the two definitions have become almost indistinguishable and voucher program is a fair description for what Ryan proposed.

It makes perfect sense to us that left-leaning PolitiFact staffers would accept that "voucher" is "almost indistinguishable" from "premium support. After all, who cares what health policy professionals think? What matter is what PolitiFact thinks.

So we move to the natural question: Did PolitiFact conclude that the terms were "almost indistinguishable" based on sound evidence? PolitiFact Wisconsin provided a source list featuring a number of PolitiFact stories, so we looked there for the evidence PolitiFact Wisconsin neglected to include in its story. We will review them for their evidence in chronological order.


PolitiFact National, Aug. 16, 2012

The differences between vouchers and premium support may matter to health-policy professionals, but not necessarily to a general audience. And while the 1995 Aaron-Reischauer paper may have offered a detailed definition for "premium support," language tends to evolve over nearly two decades.

The above essentially repeats the assertion that the terms have converged in meaning over time for the general audience. The fact check does elaborate on the argument. That elaboration takes the form of finding similarities between vouchers and premium supports, followed by having a pair of experts say it's reasonable to use "voucher" for "premium support." That is an approach PolitiFact could have applied to Ponzi schemes, but did not.

In fact, the fact check overlooks an obvious underlying Democratic argument. The fact check acknowledges that Republicans don't like the term "voucher" applied to the Medicare reform plan. But if the terms are interchangeable to the general audience then why would they care? Why would Democrats care, for that matter? We think people associate "voucher" with receiving a piece of paper and then having to go through the trouble of redeeming it. The perceived inconvenience accounts for the negative connotation the Democrats wish to pin on the Republicans' attempts at Medicare reform.

The Democrats are manipulating words to exaggerate inconveniences from a premium support system.

PolitiFact New Jersey, Sept. 10, 2012

PolitiFact New Jersey accepts and repeats the original argument from PolitiFact National:

As our PolitiFact colleagues noted, there are distinctions between the two terms, dealing with the type of inflation adjustment used and the degree of marketplace regulation imposed. Ryan’s most recent plan more closely reflects a pure premium support, but substantively, it’s somewhere between the two approaches. 

Henry Aaron, a senior fellow of economic studies at the Brookings Institution, a policy think tank, told PolitiFact National "that premium support is a type of voucher."

PolitiFact New Jersey adds nothing substantial to the argument.

PolitiFact National, Oct. 3, 2012

Less than a month later, PolitiFact National again recycled its earlier argument:

In the past, PolitiFact has found Obama’s "voucher" characterization reasonable, though as Obama noted, Republicans prefer "premium support."

Merriam-Webster defines a voucher as "a written affidavit or authorization … a form or check indicating a credit against future purchases or expenditures; a coupon issued by government to a parent or guardian to be used to fund a child's education in either a public or private school."

The plan pushed by Romney’s running mate, U.S. Rep. Paul Ryan, isn’t exactly a coupon, but it’s not so far off.

May it never be that "Social Security isn't exactly a Ponzi scheme, but it's not so far off."


PolitiFact New Jersey, Oct. 5, 2012

PolitiFact New Jersey stuck with the same mantra two days after PolitiFact National's "Mostly True" for President Obama:

Menendez’s claim is mostly accurate.

Ryan has proposed providing "premium support" payments to future Medicare beneficiaries to purchase health insurance. There are some distinctions between the two terms, but the word "voucher" generally describes this approach.

Again, the fact check contains no new reporting. The source list simply features the interview from which PolitiFact New Jersey pulled the claim it checked along with two earlier PolitiFact fact checks. That's it.

PolitiFact National, Nov. 19, 2013

In 2013, PolitiFact National finally added some new reporting. It's worth noting that the new reporting included another example of the "Ignore the Conservative Expert" game PolitiFact plays from time to time (bold emphasis added):

Yuval Levin, a health policy expert at the conservative Ethics and Public Policy Center told us he wouldn’t consider the proposal a voucher system at all.

There’s definitely debate over semantics, but to the average voter (if not the average policy wonk), it seems like the word "voucher" would accurately describe the basics of Ryan’s proposal (which, by the way, doesn’t sound all that different from the healthcare.gov marketplaces for the uninsured). Calling programs like this "voucher systems" has been common in the field for years without negative connotations, Van de Water said.

The conservative expert disagrees with the others (liberals?)? The solution is simple. Ignore the conservative. The Democrat gets a "Mostly True" rating because "'voucher system' is the colloquial way to refer to a program that gives people credit to purchase something."

No matter how many economists refer to pay-as-you-go financing as a "Ponzi game," PolitiFact will not acknowledge that the characterization is anything better than "Mostly False" (thanks for nothing, PolitiFact Wisconsin).

It's just one more example of PolitiFact's inconsistency favoring the liberal point of view.


Afters:

PolitiFact Wisconsin deserves special recognition for dinging Sen. Johnson with a "Mostly False" for comparing Social Security to a Ponzi scheme and giving an attack ad against Johnson a "Mostly True" since vouchers are (allegedly) pretty much the same as premium support.

PolitiFact's inconsistent approach to the proper use of terms is a disgrace to fact-checking.

PolitiFact again goofy on guns (Updated)

There PolitiFact goes again.

PolitiFact Florida did a fact check on a gun-related topic. "Orlando area" Rep. Alan Grayson claimed Orlando shooter Omar Matteen's rifle is able to fire 700 rounds per minute helping to directly lead to the high death toll from the tragedy.

The good news is that PolitiFact Florida did a decent-enough job on reporting the gun facts. Sure, "Glock" is a name brand and should have been capitalized. But PolitiFact found expert sources and appeared to mostly relay their statements accurately.

The fact is the rifle Mateen used cannot fire 700 rounds per minute.

So why is it "Mostly False" instead of "False"?

Let's go to PolitiFact's summary section (bold emphasis added):

Grayson said that the rifle Mateen used "shoots off 700 rounds in a minute."

On CNN, he includes this claim without any clarification. In other forums, he noted that his claim is only true for the hypothetical semiautomatic rifle converted to an automatic weapon.

Even then, however, experts say the 700-round-per-minute figure is not an accurate portrayal of rounds fired. This is true for many reasons, they said, including reloading time and the potential of overheating the gun.

That's right, ladies and gentlemen. If Mateen had an automatic version of the rifle he used, which he did not, he still would not have been able to fire off 700 rounds in a minute. But he would have been able to fire off quite a few rounds.

It looks to us as though PolitiFact is saying it is "Mostly False" that an automatic AR-type rifle can fire 700 rounds per minute. And PolitiFact gives that "Mostly False" rating to Grayson for claiming an automatic version of the rifle can fire 700 rounds per minute. And placing that "Mostly False" rating next to Grayson's separate (false) claim that Mateen's semiautomatic version of the rifle can fire 700 rounds per minute.

But we don't know. Maybe PolitiFact Florida thinks it's "Half True" that the automatic AR-type rifle can fire 700 rounds per minute, even though it cannot. And PolitiFact averages that out with a "False" rating for Grayson's claim about the semi-automatic version to get "Mostly False." And after that PolitiFact places the "Mostly False" rating next to Grayson's "False" claim that Mateen's rifle could have fired 700 rounds in a minute.

Any way you slice it, it comes out goofy.

Would anyone like to wager that it would have received a "False" rating if Donald Trump had said it?

Update June 20, 2016

Subsequent reading (and arguments) on the subject of firing rate inspires an update.

Is PolitiFact's "Mostly False" justified simply based on the fact that Mateen's gun is rated by the manufacturer to have 700 rounds per minute capability in short bursts, even if Mateen could never pull the trigger quickly enough to achieve that rate of fire?

We reject that argument as absurd.

We find it absurd because the same is true of non-automatic weapons. If you pull the trigger quickly enough on a revolver, there is no limitation on its rate of fire that a semiautomatic weapon does not share. The absurdity trebles upon considering part of Rep. Grayson's argument that PolitiFact Florida allowed to pass. Grayson said the high rate of fire for the AR-15 style weapon resulted in a much higher death toll than if Mateen had used his Glock pistol.

The quotation comes from PolitiFact Florida's fact check:

"If somebody like him had nothing worse to deal with than a glock [sic] pistol which was his other weapon today, he might have killed three or four people and not 50," Grayson said.

But the Glock 17/18 pistol has a similar--actually higher--rating for its potential rate of fire.

Grayson did not know what he was talking about. If the cyclic rate is the relevant factor (it isn't) then the Glock pistol is the more dangerous weapon. Did anybody learn that from PolitiFact's fact check?

The fact goes whoosh past PolitiFact Florida. No worries, Rep. Grayson. PolitiFact Florida's got your back.

Correction 6/15/2016: In the first paragraph, originally had "700 rounds per second" where "minute" was intended instead of "second." The wording now matches the intent.
Clarification 6/20/2016: Belatedly included a link to PolitiFact Florida's fact check of Grayson.

WUWT: "Note to Politifact: Obama DID say there is No Greater Threat than Climate Change"

The climate skeptic site "Watts Up With That" posted an item critical of PolitiFact on June 6, 2016. Contributor Eric Worrall begged to differ with the "Mostly False" rating PolitiFact Arizona gave to Pinal County Sheriff Paul Babeu. Babeu said President Obama has said climate change is the No. 1 security threat facing the United States.

Worrell responded with his post "Note to PolitiFact: Obama DID say there is No Greater Threat than Climate Change."

We think Worrell ends up striking a glancing blow by not addressing Babeu's use of the term "security," but even the glancing blow does a good bit of damage (bold emphasis added):

President Obama may have made other statements which contradict some of his statements on Climate Change – he is after all a politician. But Politifact’s assertion that it is a “mostly false” exaggeration, to say that President Obama thinks Climate Change is the greatest threat to national security, is clearly unreasonable – unless you think that suggesting Climate is the “greatest threat” to future generations, suggesting climate, unlike terrorism, might be an “existential threat” to the entire world, suggesting “we need to act now”, could not reasonably be interpreted as being a suggestion that climate is the nation’s number one priority.

The point in bold could have used more emphasis in Waddell's critique. PolitiFact literally used Obama's claim of prioritizing the fight against terrorists to pooh-pooh Babeu's claim (bold emphasis added):

Obama continues to cite climate change as a great threat to the world, but framing the issue as the country’s top national security threat is an exaggeration. Obama has said fighting terrorism is his most urgent priority.

The Arizona sheriff ignores important context, so we rate his claim as Mostly False.

The truth is that if Obama has said climate change is the top national security priority, he cannot undo the statement by claiming a different top priority.

PolitiFact Arizona's fact check shows its bias by failing to provide the most obvious counterbalance to its key evidence against Babeu.

PolitiFact:

Contrary to Babeu’s claim, the president’s top national security threat appears to be terrorism.

Out of context

In March, after the terrorist attacks in Brussels, Obama said, "I’ve got a lot of things of my plate, but my top priority is to defeat ISIL."

The president's remarks in the Atlantic downplaying the threat of ISIS compared to the threat of climate change failed to find their way into PolitiFact's fact check (we note that Worrell used the quotation to good effect):

“ISIS is not an existential threat to the United States,” he told me in one of these conversations. “Climate change is a potential existential threat to the entire world if we don’t do something about it.”

Though it is common sense to suppose that a president will respond to the greatest existential threat by making it the highest priority, it does not follow as a matter of logic that the greatest threat is always the highest priority. In his the Atlantic interview, Obama went on at length about the political difficulty of addressing climate change.

Maybe that is a big part of the reason he does not call it his top priority?

Once again, PolitiFact substitutes opinion journalism for fact-checking. Babeu did not claim he was giving Obama's exact words. So PolitiFact Arizona arrogates to itself the privilege of cooking up, on the spot, a set of standards that result in the "Mostly False" rating.

There is no solid epistemological backing for the rating. In PolitiFact Arizona's opinion, what Babeu said was "Mostly False."

The truth is that President Obama has said climate change is an immediate and growing threat to U.S. national security. And his administration acts as though border security counts very low in terms of national security.

And wasn't that last point really Babeu's point?

The difference between a true underlying point that matters and a true underlying point that doesn't matter, we suppose, is that sometimes it matters and sometimes it doesn't.