Monday, March 31, 2014

A classic case of biased story focus

We've pointed out ad nauseam the way story selection bias provides an avenue for ideological bias to find its way into PolitiFact's fact checks.  But bias can also easily find its way into a story through the story's focus.

We've got a classic example.

March 25, 2014
Image from PolitiFact.com

PolitiFact picked up on a Facebook meme (see image to right) noting a radical increase in the number of black children born with no father in the home.  The graphic implies, via its concluding question, that Great Society programs carried responsibility for the increase.

PolitiFact found that statistics backing the claim were sketchy and that the 7 percent claim in particular wasn't supported.  So,  yes there as an increase, but not to the extent the graphic suggests.

PolitiFact reminds readers that correlation isn't causation, and uses the implied claim about the "Great Society" program as part of its justification for ruling the claim flatly "False" on its trademarked "Truth-O-Meter."

March 28, 2014

Skip forward all of three days and we have a similar case, this one published by PolitiFact Rhode Island.  Again we have a would-be claim of fact married to an implied argument that one party's policies are responsible.

Image from PolitiFact.com
PolitiFact found that the facts were not quite as suggested.  Via three measures, nine of 10 of the poorest states were "red," which was determined by how the states voted in the 2012 election.

But when PolitiFact factored in cost-of-living data on top of individual income, the numbers went awry.  Four of 10 of the poorest states were blue states, with deep blue California coming in at No. 1.

And with that, PolitiFact administered a "Mostly True" rating using its trademarked "Truth-O-Meter."

What about the part of the graphic implicitly blaming low incomes on "Republican economic policies"?

PolitiFact chose not to consider that part of the claim when making its ruling.  Does the claim carry any real meaning without the accompanying argument about Republican economic policies?  We don't see it.

In these types of cases, it is PolitiFact that misleads by selectively ignoring the context.  And it's worth noting that the data associated with the "Great Society" argument show a more significant trend relative to the "poorest states" argument.

Celebrate, liberals!  Your Facebook meme is "Mostly True" while the conservatives' attempt is "False"!  PolitiFact says so.




Clarification Dec. 14, 2015: Removed a couple of unneeded words that made the opening paragraph less clear.

Friday, March 28, 2014

Critic says we missed PolitiFact's unfairness to Dick Durbin

Intrepid Twitterer Matthew Chapman tweeted a handful of criticisms of PolitiFact Bias recently.

We found little merit in his criticism, but we credit Chapman with providing a concrete example intended to support his claim that PolitiFact is often tougher on Democrats than Republicans when rated on the same issue.



We don't think Chapman's example serves him well.  PolitiFact was unfair to both Cantor and Durbin, but their statements are so different that it's preposterous to compare these "Truth-O-Meter" ratings as though it's apples-to-apples.

In neither case was confusion between the deficit and the debt the issue.  PolitiFact illicitly tried to make it the issue in Durbin's case.

Cantor said the deficit's going up, and he was obviously talking about the long-term structural deficit posed by entitlement spending.  The administration has talked big on entitlement reform and delivered only by raising entitlement spending (ACA subsidies) and taxes while putting a loose cap on Medicare spending.  PolitiFact dinged Cantor down to "Half True" since the deficit is temporarily dropping as bailout and stimulus spending tails off.

In Durbin's case, PolitiFact simply ignored the simple explanation of Durbin's claim that Democrats are cutting the debt:  Durbin was talking about the cumulative effects of a lower deficit, which is an effect on the debt.  But Durbin was less accurate than Cantor since Durbin used the common politician's trick of referring to slower growth of the deficit/debt as a cut.  That language produces the false impression that Congress has a handle on the growth of the national debt.

Cantor's claim about the U.S. budget was far less misleading than Durbin's, and PolitiFact did a bad job on both fact checks.  Durbin was arguably harmed more than Cantor by PolitiFact's ineptitude, but we repeat that the claims are so different that a comparison of the ratings isn't much use.

When PolitiFact is checking essentially the same statement from Republicans and Democrats, the Democrats fare better.  We've seen it, for example, in claims about raising congressional pay and in claims about the gender wage gap.

In comparison, the Cantor/Durbin case offers extremely poor evidence of Democrats suffering worse than Republicans on the same claim.  This is not the sort of case one should offer as the best example.

We're not second-guessing PolitiFact's specific ratings because we consider the "Truth-O-Meter" system hopelessly subjective.

Afters

Regarding Chapman's other criticisms of PolitiFact Bias, he judged our content based on a survey of our stories.  He didn't do a good job with that survey, and his method apparently caused him to skip the pages that spell out the scope of our case against PolitiFact (About, APoP, Research).

Thursday, March 27, 2014

PolitiFact takes the blue pill on Boxer's contraception claim

In the popular 1999 film "The Matrix," the protagonist is offered a choice between a red pill, which will reveal reality as it is, or a blue pill, which will sustain the illusion he has lived for his entire life.

PolitiFact chooses a blue pill in its evaluation of a recent claim from Sen. Barbara Boxer (D-Calif.).

Boxer registered her objection to the position taken by Hobby Lobby in its case now before the Supreme Court.  Hobby Lobby argued that the government could not force it to provide birth control methods through its insurance plan that offend its owners' religious beliefs, specifically by preventing pregnancy by thwarting the implantation of a fertilized egg.  The owners of Hobby Lobby view that as a form of abortion.

Boxer, as PolitiFact's fact check noted, registered the objection that Hobby Lobby fails to oppose insurance plans that might help pay for Viagra, and elaborated on her dubiously relevant criticism with the following (via PolitiFact):
"I view this as very much an anti-woman position to take," Boxer said. "And 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."
PolitiFact dutifully examined whether more than half of women who take birth control do so as a medication for other conditions.  PolitiFact found that 58 percent of women taking birth control pills identify a non-contraceptive purpose as one of their reasons for using the pill.  But only 14 percent took the pill only for its non-contraceptive benefits.

PolitiFact cited a Guttmacher Institute study of women using hormonal birth control pills.

So PolitiFact rated Boxer's statement "Half True" and missed the point.

Hobby Lobby covered birth control pills before the ACA took effect.

Hobby Lobby balked at the HHS mandate that it provide coverage for "morning after" pills and intrauterine devices as part of its contraception coverage since those methods, particularly the IUD, are associated with preventing implantation of a fertilized egg.

That means that Boxer's reasoning effectively has nothing to do with Hobby Lobby's objection to the contraceptive coverage mandate.  And somehow PolitiFact couldn't be bothered to point out the obvious.  Even if all women taking birth control pills did so to treat cancer, it would have nothing to do with Hobby Lobby's legal challenge to the HHS rule.  To address that point, one would need evidence that morning-after pills and IUDs were prescribed for non-contraceptive purposes.

Boxer's apples-and-oranges argument draws not a peep of protest from PolitiFact.

Saturday, March 8, 2014

Nothing To See Here: The White House's aspirational equal pay greeting cards!

We ridiculed PolitiFact's gender wage gap rating connected to President Obama's State of the Union address.  We ridiculed it because when Obama said women are paid only 77 cents on the dollar compared to men, then said that condition is unacceptable today, and then called for equal pay for equal work, PolitiFact decided that the last point was simply "aspirational."  In other words, Obama wasn't saying that 77 cents on the dollar was for doing equal work.  The claims were distinct, therefore Obama earned a "Mostly True" rating instead of the "Mostly False" rating PolitiFact usually uses when it is claimed women receive 77 cents on the dollar for doing the same work as men.

It turns out the White House has been giving away greeting cards (at least) since mid-2013.  Not just any greeting cards, mind you.  These are aspirational greeting cards.  They mention the disparity of 77 cents on the dollar and call for equal pay for equal work.





And why shouldn't the White House keep right on using this type of misleading presentation when PolitiFact will look the other way?

Have an aspirational day!

Friday, March 7, 2014

A note to PolitiFact Florida

During the 2012 election, Democrat incumbent senator Bill Nelson faced off against former Rep. Connie Mack, the Republican challenger.

Nelson accused Mack of protecting oil giant Chevron from a lawsuit.  Nelson ran an ad that said ""In Congress, Mack is protecting Chevron oil from a multibillion dollar lawsuit over pollution of rivers and rain forests."

That was a powerful message.  Mack wants to protect big businesses who would pollute pristine lands with impunity?

PolitiFact only gave Nelson a "Half True" rating, however:
Nelson’s TV ad hits Mack for "protecting Chevron oil from a multibillion dollar lawsuit over pollution of rivers and rain forests."

Mack’s act of protection is a yet-to-be-heard congressional resolution listing the Chevron case as one of many reasons to withhold trade preferences with Ecuador. Resolutions are symbolic gestures, so it would take a law for Congress to take real action on Ecuadorian trade practices.

The Chevron case is also particularly complicated. Chevron says it did not get a fair trial in Ecuador and should not be held responsible anyway.

Still, the resolution seems designed to exert political pressure on the country in Chevron’s favor.

We rate Nelson’s claim Half True.
Let's summarize the summary.
  • Mack's resolution would have no effect, by itself, on U.S. policy ("symbolic gesture")
  • The Chevron case was one grievance among many mentioned in the resolution
  • Chevron says it did not get a fair trial
  • The resolution "seems" intended to exert political pressure (see the first bullet above)
So PolitiFact Florida reasons that crafting a toothless resolution "seems" intended to exert political pressure on Ecuador.  And how surprising is it that Chevron thinks it didn't get a fair trial.  Crocodile tears, right?

But let's back up a bit. In fairness to PolitiFact Florida, the article does point out that experts were sympathetic to Chevron's claim about the unfair trial:
Other experts empathized with Chevron’s argument that it did not get a fair trial in Ecuador, and that it overlooks that Petroecuador has been in control of drilling operations for the past 20 years.
Those experts just couldn't find their voices in PolitiFact Florida's summary.  And that's not a big surprise, given that PolitiFact Florida made two experts hostile to Chevron the stars of its analysis:
We found a couple of experts who agree [with Nelson], including Judith Kimerling, a City University of New York-Queens College environmental policy professor who authored the 1991 book Amazon Crude, which is widely credited with inspiring the lawsuit. She represents a group of Huaorani people from Ecuador in a lawsuit against the Amazon Defense Front and plaintiff’s lawyers over how the award would be distributed.
Relying on an expert with a direct interest in the case.  What could be better?  Maybe use a quotation from the other biased expert, Lori Wallach, as the kicker quote?
"[Mack]’s shilling for Chevron but has not yet delivered on actually cutting off Ecuador’s trade preferences," she said.
Wallach heads up Public Citizen's "Global Trade Watch."  She judged the Chevron case in a 2012 memorandum:
Ecuador is still struggling to gain compensation from Chevron for decades of contamination of a Rhode Island-sized swath of the Amazon.
 Okay, so?  So what if PolitiFact shows partiality to Chevron's critics in its story?

We're posting this today because Chevron just got a big win
A federal judge on Tuesday blocked U.S. courts from being used to collect a $9 billion Ecuadorean judgment against Chevron for rainforest damage, saying lawyers poisoned an honorable quest with their illegal and wrongful conduct.
Read ABC News' story about the case in full.  Then review PolitiFact Florida's fact check of Sen. Nelson's ad.  PolitiFact bias.

Monday, March 3, 2014

Meet the New Boss Same as the Old Boss ...

We've wondered since PolitiFact's founding editor Bill Adair moved on to the realm of academia in 2013 how things would shake out at PolitiFact.

An interview of new editor Angie Drognic Holan published on Feb. 27 gives us a hint that some things won't be changing.  Here's an item of special note to us here at PolitiFact Bias (bold emphasis added):
"We try to fact-check approximately the same number of Democrats and Republicans but we don’t keep hard-and-fast count, and one thing that we don’t do is try to balance the ratings. We don’t think about if we get a false on one side, we want to go and get a false on the other side. We do not do that."
We've pointed out before that trying to rate about the same number of statements by party serves to skew the sample of statements PolitiFact rates.

Suppose PolitiFact's editorial ears perk up over 10 suspicious-sounding Republican statements but only four for Democrats.  Trying to keep the numbers approximately even creates pressure to change the editorial criteria in order to move the numbers closer together.

What does this mean in practical terms?  It's unwise to collect PolitiFact data and use it as though it's a scientific sample.  It's not a scientific sample.

And we can't complete our post on the Holan interview without noting how she blatantly/wisely dodged one of the questions:

4. You are working with partners across the country, how does everybody draw consistent conclusions? Can this be entirely objective?

No, it's not entirely objective, and PolitiFact doesn't draw consistent conclusions.  Read the article to see how Holan skirts the question.  Adair used to do the same thing.

Oh, and another thing ...

We don't know when this interview took place, but Holan's citing 10 PolitiFact state affiliates and one international one even though the loss of PolitiFact Ohio apparently brings the number of state affiliates down to nine, and PolitiFact has taken down its link to PolitiFact Australia, which has gone on hiatus.  Perhaps somebody should have fact-checked her claims.