Tweezers or tongs?

We've noted before PolitiFact's inconsistency in its treatment of compound statements.  It's time to focus on a specific way that inconsistency can influence PolitiFact's "Truth-O-Meter.

We'll call this problem "tweezers or tongs" and illustrate it with a recent PolitiFact fact check of Phil Gingrey (R-Ga.):

"As a physician for over 30 years, I am well aware of the dangers infectious diseases pose. In fact, infectious diseases remain in the top 10 causes of death in the United States. … Reports of illegal migrants carrying deadly diseases such as swine flu, dengue fever, Ebola virus and tuberculosis are particularly concerning."

[...]

The reality is that Ebola has only been found in Africa -- and experts agree that, given how the disease develops, the likelihood of children from Central America bringing it to the U.S. border is almost nonexistent. But most importantly for our fact-check, Gingrey’s office was unable to point to solid evidence that that Ebola has arrived in Western Hemisphere, much less the U.S. border. To the contrary, the CDC and independent epidemiologists say there is zero evidence that these migrants are carrying the virus to the border.

We rate the claim Pants on Fire.

It's tweezers this time.

Gingrey states that disease crossing the border via migration creates a concern.  He mentions reports of swine flu, dengue fever, Ebola virus and tuberculosis crossing the border as examples of concern.  PolitiFact takes its tweezers and picks out "Ebola virus," and drops from consideration the other diseases in Gingrey's compound statement.

Let's review again PolitiFact's guidelines statement of principles:

We sometimes rate compound statements that contain two or more factual assertions. In these cases, we rate the overall accuracy after looking at the individual pieces.

Or sometimes PolitiFact will just settle on rating one piece of the compound statement.  It's up to PolitiFact, based on the whim of the editors.

Burying Gingrey's underlying point

Though we're focused mainly on PolitiFact's inconsistent handling of compound statements, it's hard to ignore another PolitiShenanigan in the Gingrey fact check.  PolitiFact sometimes takes a subject's underlying point into account when making a ruling.  And sometimes not.  In Gingrey's case, PolitiFact buried Gingrey's underlying point:

As a surge of unaccompanied children from Central America was arriving on the United States’ southern border this month, Rep. Phil Gingrey, R-Ga., expressed concern about the impact they could have on public health.

PolitiFact left out part of the story.  Yes, Gingrey was expressing concern about the potential spread of disease from human migration.  But he wasn't simply airing his concerns to the Centers for Disease Control, to whom he addressed the letter PolitiFact fact checked.  He was asking the CDC to assess the risk:

I request that the CDC take immediate action to assess the public risk posed by the influx of unaccompanied children and their subsequent transfer to different parts of the country.

PolitiFact claims "words matter."  Yet, contrary to PolitiFact's claim, Gingrey did not say migrants may be bringing Ebola virus through the U.S.-Mexico border.  Rather, he said it was troubling to hear reports of diseases, including Ebola virus, coming across the border.

Words matter to PolitiFact, we suppose, since one needs to know exactly how much twisting is needed to arrive at the desired "Truth-O-Meter" rating.

PolitiFact's compound problem

Why PolitiFact's rating of Steve Doocy was unfair

After criticizing PunditFact's failure to own up to its mistakes in post this Wednesday past, we promised an example of how PolitiFact applies its rule for compound claims inconsistently.

What is a compound claim?

 

A compound claim is a claim that asserts more than one truth.  For example:

• The car is a red Chrysler

The statement makes two assertions of truth:  The car is red, and the car is a Chrysler.

In its statement of principles, PolitiFact says it divides compound claims into segments, grades the segments separately, then rates the overall accuracy:

We sometimes rate compound statements that contain two or more factual assertions. In these cases, we rate the overall accuracy after looking at the individual pieces.

As is normal with PolitiFact, these principles are more like guidelines.  We'll look at the Doocy rating and compare it to another recent PolitiFact rating, this one looking at a statement from liberal columnist Sally Kohn.

Doocy:
"NASA scientists fudged the numbers to make 1998 the hottest year to overstate the extent of global warming."

PolitiFact rated Doocy's claim "Pants on Fire."

Kohn:
"Hobby Lobby provided this (birth control) coverage before they decided to drop it to file suit."

No, wait.  The above quotation is the one PolitiFact said it was checking.  But the actual sentence went on a bit longer (bold emphasis added):

"Hobby Lobby provided this (birth control) coverage before they decided to drop it to file suit, which was politically motivated."

PolitiFact rated Kohn's claim "Mostly True."

With the amputated ending restored, it's easy to see the parallel between the two claims.  Both Doocy and Kohn make assertions of fact, followed by judgments of motivation.  Doocy's claim arguably reports the results of the numbers-fudging rather than asserting that the scientist were motivated to achieve a particular end, but that point isn't necessary to show PolitiFact's inconsistency.

Given the similarity between the two claims, why did PolitiFact treat Doocy's compound claim as a unitary claim and Kohn's as a two-part compound claim?

Slanted.

We suggest a two-part theory.  Treating Doocy's statement as a compound claim might result in a "Mostly False" or better rating for a claim skeptical of human-caused climate change.  Liberals wouldn't like that.  Treating Kohn's claim as a unitary claim, or even dealing with her evidence-free claim of a political motivation for the Hobby Lobby lawsuit, harms the narrative liberals prefer on that topic.

In short, PolitiFact acted inconsistently because of political bias.  That's the theory.  If anybody has a better one, feel free to leave a comment.

The failure to consistently apply its principles provides avenues for the biases of PolitiFact's staffers to suffuse its fact checks.  This is just one example among many.


Additional note on the Kohn fact check

I can't figure out why PolitiFact fact checked Kohn if it wasn't intended to implicitly support her charge that the Hobby Lobby suit was not based on a sincere religious objection.  PolitiFact said "We can’t determine if politics motivated the company."  Without that charge, who cares if Hobby Lobby covered morning-after pills before it decided to bring a suit against the administration?  Despite its disclaimer, PolitiFact goes out of its way to make a circumstantial case supporting Kohn's charge:

The Greens re-examined the company’s health insurance policy back in 2012, shortly before filing the lawsuit. A Wall Street Journal story says they looked into their plan after being approached by an attorney from the Becket Fund for Religious Liberty about possible legal action over the federal government’s contraceptives requirement.

That was when, according to the company’s complaint, they were surprised to learn their prescription drug policy included two drugs, Plan B and ella, which are emergency contraceptive pills that reduce the chance of pregnancy in the days after unprotected sex. The government does not consider morning-after pills as abortifacients because they are used to prevent eggs from being fertilized (not to induce abortions once a woman is pregnant). This is not, however, what the Green family believes, which is that life begins at conception and these drugs impede the survival of fertilized eggs.

We can't determine PolitiFact's motivation for doing this fact check, but ... you get the picture.

Additional additional note:

Somehow, PolitiFact neglected to include the following information from its implicit concurrence with Kohn's attack on the Hobby Lobby's owners, the Green family:

54.  Hobby Lobby's insurance policies have long explicitly excluded--consistent with their religious beliefs--contraceptive devices that might cause abortions and pregnancy-termination drugs like RU-486.

This is from a court document PolitiFact cited in its fact check of Kohn.  PolitiFact used the next item from the document, No. 55, out-of-context against Hobby Lobby.  That was Hobby Lobby's admission that it unwittingly covered two morning-after drugs that may cause abortion.  No. 54 just wouldn't have fit Kohn's narrative, would it?

Jeff Adds: (7-5-2014) It's worth noting that both the Doocy and Kohn ratings were edited by Aaron Sharockman, so the inconsistency cannot be explained by the different journalistic styles of two people.


Update 7/8/2014:

Here's another recent case of the same compound problem, this time featuring Hillary Clinton (bold emphasis added):

"It’s very troubling that a salesclerk at Hobby Lobby who needs contraception, which is pretty expensive, is not going to get that service through her employer’s health care plan because her employer doesn’t think she should be using contraception," Clinton said.

No worries, Mrs. Clinton.  PolitiFact will just focus on the first part of the claim.  It's not really a fact checker's job to point out that Clinton's claim conflicts with Hobby Lobby's willingness to cover 16 kinds of contraception, right?  Nor should we consider Hobby Lobby's religious objection to paying for certain types of contraception.


Edit 7/5/2014: Added links to PolitiFact's Doocy and Kohn ratings - Jeff
Edit 7/5/2014:  Corrected some misspellings, including Mr. Doocy's name.

PolitiFact's Subjective-O-Meter

PolitiFact Florida provides us with yet another sterling example of subjective "Truth-O-Meter" ratings.  PolitiFact "uses" the "Truth-O-Meter" to find the truth in politics.  Translated, that means PolitiFact does its version of fact checking and then picks one of the available "Truth-O-Meter" ratings, ranging from "Pants on Fire" to "True."

PolitiFact Florida gave DNC Chair Debbie Wasserman Schultz a "Mostly True" rating for her June 30 claim that nearly 60 percent of women who use birth control use it for purposes other than contraception.  That's supposed to remind us of the awful effect on women from the Supreme Court's Hobby Lobby decision.

Wasserman Schultz's claim is false, as I show with an item at Zebra Fact Check.  But there's another aspect to this case that serves especially well to show the subjectivity of PolitiFact's trademarked "Truth-O-Meter."

PolitiFact rated "Half True" a very similar claim from Barbara Boxer (D-Calif.).  We took note of that rating with an item of our own here at PolitiFact Bias.  And PolitiFact Florida noted the Boxer rating as well when it fact checked Wasserman Schultz:

Previously, we fact-checked a similar claim based on the same report. Sen. Barbara Boxer, D-Calif., said on MSNBC in March that "it’s important to note that women take birth control, more than half of them, as a medication for other conditions, so it is an attack on women."

We rated that claim Half True because it required additional clarification about the type of birth control and the survey results. Wasserman Schultz's statement was somewhat more carefully worded.

In what way was Wasserman Schultz's claim more carefully worded?  PolitiFact Florida doesn't say.  But it's true that PolitiFact rated Boxer "Half True" owing to the need for clarification about the type of birth control and the survey results:

Boxer said more than half of women use birth control  "as a medication for other conditions." She was referring specifically to the pill. Guttmacher says 58 percent of women on the pill take it for noncontraceptive reasons. But that number drops to 14 percent if you look at women who take the pill only for noncontraceptive reasons.

Her claim is based in fact, but requires additional clarification about the type of birth control and the survey results. We rate it Half True.

Both of these fact checks by PolitiFact are ridiculously generous, but let's focus on the differences between the Boxer rating and the one for Wasserman Schultz.  The fact is, there's not much of a difference:

Wasserman Schultz said that "nearly 60 percent of women who use birth control do so for more than just family planning."

This claim gets support from a Guttmacher Institute report that found 58 percent of pill users citing at least one non-contraceptive reason. However, Wasserman Schultz’s comment glosses over two important caveats.

Both Boxer and Wasserman Schultz said "birth control" while referring to data on birth control pills.  That accounts for PolitiFact's "clarification about the type of birth control."

Subjective.

The clarification about "survey results" from the Boxer fact check appears to match the second caveat in the Wasserman Schultz fact check.  Both fact checks point out the number of women in the survey who used the pill only for non-contraceptive purposes.

The stated reasons for the ratings are effectively identical, yet the ratings are different.

So let's return to the lone difference PolitiFact identifies, however unclearly: "Wasserman Schultz's statement was somewhat more carefully worded."

We can identify one respect in which Wasserman Schultz was more accurate than Boxer.  Boxer's wording strongly suggested women were using birth control pills for clinical non-contraceptive reasons.  Wasserman Schultz did the same thing, but with greater subtlety.   It's not much of a difference when it comes to justifying a "Mostly True" rating as opposed to a "Half True" rating, especially when the statement is obviously literally false.

It's another case that helps confirm that the "Truth-O-Meter" is subjective.

The Doocy-Goddard update

On June 30 we posted about a PunditFact fact check that drew a response from Steven Goddard, the pseudonymous climate science blogger who runs "Real Science."  Goddard pointed out a number of problems with PunditFact's fact check.  We considered it unlikely PunditFact would revisit the issue.

We were only half right.

PunditFact did not change its original article, as we expected.  Instead, it added an article ("After the Fact") underneath the original article under the title "The man behind the science Fox quoted responds."  The article links to Goddard's response but does not quote from it at all.  Goddard lambasted the original fact check pretty thoroughly, but PunditFact just teased out one of the criticisms via paraphrase:

He noted that one of the experts we quoted in our initial piece has since revised his views of Heller’s research.

It's true Goddard (Heller) pointed out statement of reversal from Anthony Watt, the expert in question.  But that was the least of Goddard's criticisms.  It was the least of Goddard's criticisms because PunditFact quoted Watt on a different subject than the one it was fact checking.  Yet PunditFact gives its "After the Fact" update without addressing or even explicitly acknowledging Goddard's other criticisms.  There's no admission that PunditFact conflated two different issues in its fact check.

PolitiFact's response represents a pretense that there was nothing wrong with its original article.

Goddard made his central point the fact that the raw temperature data make the 1930s the hottest decade.  Fox's Steve Doocy turned that fact into a compound claim, attaching the adjustments to the data into a willful deception.  In terms of logic, there's some foundation for rating Doocy's entire compound statement as one claim--but that's not the usual practice we see from PolitiFact/PunditFact.  As PolitiFact says in its statement of principles:

Statements can be right and wrong – We sometimes rate compound statements that contain two or more factual assertions. In these cases, we rate the overall accuracy after looking at the individual pieces.

Or not. It depends on what PolitiFact feels like doing on any given day.

Consistent with PunditFact's principles, Doocy is more correct than not that the raw data need an adjustment to make 1998 hotter than peak temperatures from the 1930s.  And after that PunditFact might have considered what it would take to fact check whether the adjustments were done to intentionally overstate global warming.

PunditFact's response to Goddard lacks the needed correction. Nor does it do much to clarify the misleading aspects of the original fact check.


Addendum:

We've run across another PolitiFact story that helps illustrate how PolitiFact treated Doocy's claim unfairly compared to the alternative practice PolitiFact applies for compound statements.  Look for a story on that later this week. That post can be found here.


Edit 7/5/2014: Added links to PFB posts in first and last sentences and one link to original PolitiFact article- Jeff

Real Science: 'My Rebuttal to PolitiFact'

A climate science blogger who writes under the pseudonym Steven Goddard, sent out a blistering response to last week's PunditFact piece on Fox News Channel show host Stephen Doocy.

Goddard made public the email he sent to PolitiFact writer Jon Greenberg.  Here's the first part:

Politifact accused Steve Doocy of being a liar,  for accurately reporting on a blog post made on my blog (stevengoddard.wordpress.com) which showed how NASA has altered the US temperature record over time.

Politifact’s claim is the result of a failure to understand the topic, for the following reasons.

Sour grapes, right?

Not so fast.  It appears PunditFact conflated two different issues, and one of the persons quoted going against Goddard's conclusions, Anthony Watts, has since reversed himself.  Plus the basic criticism, that NASA has adjusted the temperature record, isn't contested despite the "Pants on Fire" rating.  Read Goddard's letter for the details.

Clearly, it's appropriate for PunditFact to revisit the issue and amend its fact check.  The question is whether PunditFact will bother.

Place us firmly in the skeptics' camp on that one.


Jeff Adds:

It's worth noting Watt's reversal was unequivocal (emphasis in original):

All of that added up to a big heap of confirmation bias, I was so used to Goddard being wrong, I expected it again, but this time Steve Goddard was right and my confirmation bias prevented me from seeing that there was in fact a real issue in the data and that NCDC has dead stations that are reporting data that isn’t real: mea culpa.

It's dishonest for PolitiFact to present Watt's original position without updating their story to reflect his further investigation into the matter. It would be reasonable to expect an update from an outfit that claims to help you "sort out the truth" of an issue. But PolitiFact appears to have less interest in uncovering the truth than they do in advancing an agenda.

Update July 3, 2014
PunditFact responds, sort of, and we clear away PolitiFact's smoke.


Clarification 4:02 a.m. EDT, July 1, 2014:  Substituted "an update" for "that" in the last paragraph.