Wednesday, July 27, 2016

Hoystory: "Abetting Gavin Newsom's Big Lie"

Reformed journalist Matthew Hoy today delivered an exquisite knockout to PolitiFact California over its gun-related fact checking.

California Lt. Governor Gavin Newsom and an associated gun-control advocacy group, Safety For All, claimed that gun dealers outnumber McDonald's in California. PolitiFact California rated that "True." Hoy covers the problems with that rating (which we also noted), and looks at how PolitiFact has since ignored Newsom's triumphant tweet saying that it's easier to buy a gun in California than it is to buy a Happy Meal.

Hoy notes that PolitiFact seems unconcerned over the leftward shove Newsom gives its already left-leaning reporting:
I wanted to see if Politifact would do anything about Newsom’s tweet. After all, it’s their reporting that’s being misused.

And a week later? Nothing new from Politifact on the topic. I emailed the reporter on the original fact check and the editor Tuesday evening. As of press time I had not received a response to my query of whether they would be fact-checking Newsom’s tweet.
Hoy's experience tracks with mine when I tried to get PolitiFact Missouri to fix a terminally flawed fact check. The fact checkers don't seem all that interested in the facts sometimes.

It's good use of your time to visit Hoystory and read his whole post.

The biased "True" rating for Michelle Obama's claim the White House was built by slaves

Over the years we've built up quite a bit of evidence of PolitiFact's bias, based largely on PolitiFact's inconsistent application of standards.

A PolitiFact fact check of Michelle Obama's speech to the Democratic National Convention supplies yet another strong example.

One could easily read Obama's statement to mean that the White House was built exclusively with slave labor. That was not the case, as the text of the fact check concedes. Not telling a significant part of the story often leads to PolitiFact rating a true claim "Mostly True" or worse.

Not this time (bold emphasis added):
Obama said the White House "was built by slaves." Strictly speaking, the White House was not exclusively built by slaves; it was built by a combination of slaves, free blacks and whites. But slaves were significantly involved in the construction of the White House, so we have no quarrel with the way Obama worded her claim. We rate it True.
Obama's claim was imprecise and people might be misled by it. However, PolitiFact has no problem with the way she worded her claim.

That's tossing principles on the scrap head, not that PolitiFact is consistent enough in applying its principles that they deserve the term "principles."

Need a comparison? There are many. How about this one?

PolitiFact's summary draws the perfect contrast:
Trump has a point here, but he should have used different words to make it. We rate his claim Mostly False.
PolitiFact has no problem with Obama's word choice. But Trump should have used different words to make his valid point.

These two political figures are not being judged according to the same standard.

Fact-checking. This is why so many cannot take PolitiFact's brand of fact-checking seriously. The great mystery is why the folks at PolitiFact think it is okay to check facts this way.

If they know it is not okay and yet do it anyway, well, that puts the problem in a different light.


We often hear the excuse from PolitiFact's defenders that PolitiFact always justifies its ratings.

To those people, we ask if you would have accepted this explanation from PolitiFact:
Trump said Chevrolet in Japan "does not exist." Strictly speaking, there are some Chevrolet vehicles in Japan though the number is relatively small compared to the more popular makes. Since the number of Chevrolets is so small we have no problem with the hyperbolic way Trump worded his claim. We rate it True.
Always justifying the rating does not help if the justifications do not follow consistent principles.

Tuesday, July 26, 2016

NTSH: Michelle Obama flip-flops on greatness?

PolitiFact sometimes rates statements on a "Flip-O-Meter," indicating when a public figure has reversed positions on something.

With the "Flip-O-Meter" in mind, we offer Michelle Obama's Democratic National Convention speech as a potential flip, fit to note for our Nothing To See Here feature. As a number of media outlets have noted, the First Lady said back in 2008 that she was proud of her country for the first time. Now, in 2016 she says "This right now is the greatest country on earth."

Did something change other than Michelle Obama's mind? Did the Obama presidency propel the United States to new heights of greatness?

We're certainly not the first to notice the discrepancy. We first saw it mentioned at Power Line blog.

Tangentially related: A Zebra Fact Check of Will McAvoy's rant on American greatness from the television series "The Newsroom."

Sunday, July 24, 2016

Sweet little PolitiLies

(Looks like the ill-timed and unintended press of a button resulted in a blank post under this title--so the initial publication of this content technically qualifies as an update)

Those who can be trusted in small things gain trust in handling larger things.

Hit play, then resume reading.

Mainstream journalists, including those at PolitiFact, often report things that are not true.

Case in point, as PolitiFact rules "Mostly True" Donald Trump's claim that the median household income has fallen by $4,000 since the year 2000:
While this statistic came amid a series of shots at President Barack Obama, the majority of the period Trump was referring to came under the presidency of George W. Bush, a Republican.
Since Trump was taking "a series of shots" at Obama, PolitiFact adds in missing context that we will shortly address. But PolitiFact's statement is pretext and not fact.

Here's Politico's transcript of the relevant segment of Trump's speech (is it coincidence that PolitiFact provided no hyperlink to the text of the speech?):
What about our economy?
Again, I will tell you the plain facts that have been edited out of your nightly news and your morning newspaper: Nearly Four in 10 African-American children are living in poverty, while 58% of African American youth are not employed. 2 million more Latinos are in poverty today than when the President took his oath of office less than eight years ago. Another 14 million people have left the workforce entirely.

Household incomes are down more than $4,000 since the year 2000. Our manufacturing trade deficit has reached an all-time high – nearly $800 billion in a single year. The budget is no better.

President Obama has doubled our national debt to more than $19 trillion, and growing. Yet, what do we have to show for it? Our roads and bridges are falling apart, our airports are in Third World condition, and forty-three million Americans are on food stamps.
Trump's treatment of the economy does occur after a series of paragraphs related to a rise in violent crime. Trump's first general topic was crime. The second was the economy, represented by the quotation above. After the economy Trump moves on to foreign policy, in which he spreads blame between Obama and Secretary of State Hillary Clinton.

It does not make any sense to suppose Trump's statistic was intended to lay all the blame for the lower media wage on Obama. Most people realize that Obama was not in office until 2009, years after the year Trump used as the median income baseline.

But PolitiFact riffs based on this misleading reading of Trump's claim:
During the first eight years of the span Trump referred to -- 2000 to 2008 -- Bush was president. And during Bush’s tenure, inflation-adjusted median household income declined by $2,411, which is more than half the total amount it fell between 2000 and 2014.

Since Obama took office, it has fallen by $1,656. That’s not a great legacy, but it’s worth remembering that slicing the numbers this way puts all of the Great Recession on Obama’s watch while also denying Obama 2015, when there was a slow but steady recovery, because data is not available.
So it's Bush's fault, and that dastardly Trump is denying Obama a year for which data is not available. PolitiFact often makes its own judgments based simply on available data. But if PolitiFact thinks Trump is doing what PolitiFact does, then Trump is doing something wrong. Apropos of that, the Washington Post Fact Checker gave Trump a harsher rating on this same claim, saying the latest data showed median household income has almost returned to the levels from 2000. But what Trump does wrong is okay when PolitiFact does it.

We would draw attention to two main aspects of PolitiFact's fact check of Trump.

First, PolitiFact's defense of Obama in the midst of a Trump fact check was rationalized. It wasn't really needed, since Trump himself excused Obama from full blame by using the year 2000 as a baseline. Everybody understands from that context that Obama is not responsible for everything that happened to median income since the year 2000.

Second, PolitiFact engaged in the same type of misdirection it blamed on Trump. PolitiFact wrongly says Obama gets all the blame for the Great Recession while blithely saying, and we quote, that "during Bush's tenure, inflation-adjusted  median household income declined by $2,411, which is more than half the total amount it fell between 2000 and 2014."

Here's the story that PolitiFact failed to tell, even though it was illustrated by the Federal Reserve chart embedded in its own story: Bush inherited a declining economy from President Bill Clinton. So Bush was dealing with a recession at the start of his tenure as president. Under Bush, inflation-adjusted income had recovered nearly to 2000 levels by 2007, before the Great Recession hit. So the drop in median income PolitiFact lays on Bush was overwhelmingly tied to the Great Recession.

Some type of averaging would provide a more reliable picture of trends under Bush and Obama than the one PolitiFact painted.

So PolitiFact made up the notion that Trump was pinning the drop in median household income on Obama to justify its own misleading editorial opposing Bush-era economic policy (and defending Obama-era policy). "Fact-checking."

Sweet. Restart the music video if desired.

Friday, July 22, 2016

Why PolitiFact flip-flopped on Clinton

We've dedicated two items to PolitiFact's "Half True" gift to Democratic presidential nominee Hillary Clinton on her claim she never sent or received classified information via her primate email account.

First we argued that PolitiFact's defense of its "Half True" rating made no sense following FBI Director James Comey's statement on July 5.

PolitiFact, whether influenced by our post or not, apparently agreed and reversed itself the next day while erasing nearly all the evidence of its embarrassing decision from the day before.

We, namely Jeff D, responded to PolitiFact's reversal by documenting the evidence that PolitiFact had continued its habit of changing stories without posting correction notices.

We have addressed what happened. Now we will consider why it happened.

A stupid idea whose time has come

It was just plain stupid of PolitiFact to say that it could not change Clinton's "Half True" rating in view of its policy of doing its ratings according to information available at the time (bold emphasis added):
(After this fact-check published, FBI Director James Comey released details of the Feb's investigation into Hillary Clinton's use of a private email server. This claim will remain rated Half True, because we base our rulings on when a statement was made and on the information available at that time. But the FBI investigation clearly undercuts Clinton’s defense if she makes a similar claim again. You can read more about the findings of the FBI investigation here.)
Applying that policy as PolitiFact did as described above could justify avoiding any number of corrections. Did an article misuse a word? Sure, but we're not going to correct it because we did not know any better at the time.

Yes, it's silly. But it beyond likely that a group of PolitiFact's editors agreed, at least for a time, that it was the right thing to do for this Clinton fact check.

Why the do-over?

PolitiFact reversed itself pretty quickly. But what kind of impetus could reverse the considered wisdom of PolitiFact's elite editorial group?

We'll consider some possibilities:
  • A trusted figure condemned PolitiFact's defense of its "Half True" rating
This option seems the most likely. But PolitiFact's lack of transparency about its reversal leaves us in the dark as to whether anybody inside the organization was independent enough to rock the boat.

Alternatively, the editors at PolitiFact may have felt distress that they were out of step with the Washington Post Fact Checker. The Post promptly changed its rating of Clinton's email claim from two Pinocchios to four. Despite their claims of independence, the mainstream fact checkers can't avoid seeing each others' work and doubtless feel pressure to make similar findings of fact.
  • PolitiFact changed because of our criticisms?
We condemned PolitiFact's reasoning and explained what was wrong with it before PolitiFact executed its reversal. However, it's not typical for PolitiFact to agree with and act on our criticisms.
  • "Lie of the Year" implications
PolitiFact horribly embarrassed itself with the 2014 "Lie of the Year." President Obama's promise that people could keep their insurance plans under his health care reform bill took the award, or at least PolitiFact tried to make it look that way, despite the fact that PolitiFact never rated the claim worse than "Half True."

Clinton's email fib easily qualifies as the early leader in the "Lie of the Year" sweepstakes. It's high-profile. It was deeply investigated by the FBI. It carries yuge implications for the 2016 election.

Did PolitiFact belatedly realize that it might have another Democratic claim rated "Half True" winning the Lie of the Year award? Two words: bad optics.


We don't know for sure why PolitiFact acted the way it did. We can only offer some possibilities. But one thing is certain. PolitiFact has not acted like a fact checker in this. It has acted like it loves its own reputation better than it loves the truth.

PolitiFact California: Surprise, our gun-related fact check leans left

We carry little respect for PolitiFact in part because PolitiFact falls into bias traps that professional journalists ought to easily avoid.

One classic example of that genre comes from PolitiFact's treatment of illustrated numbers comparisons. It makes sense to PolitiFact to compare the frequency of voter fraud to shark attack. But comparing the frequency of gun crimes by concealed-carry permit holders to the frequency of alligator attacks doesn't make sense because gun crimes and alligator attacks are so different from each other.

Got it?

Predictably, PolitiFact's treatment of the former case proved a benefit to liberalism. PolitiFact's treatment of the latter case unfairly harmed conservatism.

On July 19, 2016, PolitiFact California approved the comparison of the number of licensed gun dealers to the number of McDonald's restaurants.

Of course the fact check contains no discussion at all whether McDonald's restaurants are similar enough to gun dealers to justify the comparison. And we're okay with that, aside from the inconsistency it shows from PolitiFact, because sometimes a numbers comparison is just that. One may compare the number of ants at a picnic to the number of stars in the galaxy without needing to show another similarity between ants and stars.

PolitiFact California went the extra mile for liberalism on this fact check, however.
PolitiFact Georgia rated a similar claim in 2013 as Mostly True. A civil rights activist said: "There are twice as many gun shops as McDonald’s in the United States." Our California claim deals with licenses in this state only, and not necessarily brick-and-mortar gun shops.
Above, PolitiFact California starts to make a mockery of PolitiFact's principled assurance that "words matter."

The claim was that licensed gun dealers more than double the number of McDonald's restaurants. Counting those who have licenses to deal guns but who do not sell guns is like adding closed or re-purposed McDonald's locations to the total number of McDonald's restaurants. And the distinction matters in this case:
The ATF keeps a monthly log on its website of how many dealers and pawnbrokers are licensed to sell guns in each state.

This month’s tally shows 2,315 dealers and pawnbrokers licensed to sell guns in California. That’s a shade less, 15 in fact, than twice as many McDonald’s.
PolitiFact's fact-finding found the claim from the gun control group "Safety For All" was not quite true using those numbers. So Safety for All pointed out that adding in the number of licenses not necessarily connected to a gun seller would provide the numbers needed to make the claim true.

PolitiFact California was fine with that approach and rated the claim "True."

The problem? That approach fails to jibe with the spirit of the comparison. Safety For All was trying to emphasize that licensed vendors from whom one might buy a gun more than double the number of McDonald's restaurants. Would the comparison work if half the licensed gun dealers did not sell guns? Of course not. Having a license to sell a gun does not necessarily make one a "gun seller" in the sense implied by the Safety For All claim.

It is not appropriate to shoehorn in persons licensed to sell guns who do not ordinarily act as gun sellers.

One could go further and nitpick the comparison to death (example here). We will stick with the point that a competent and neutral fact-checker does not illegitimately maximize a number to the benefit a political point of view. We expect that type of behavior from biased fact checkers.

Update: Afters

PolitiFact California went the extra-extra mile by working up a meme-worthy graphic to promote Safety For All's not-quite-true factoid:

Update July 22, 2016: Added link to PolitiFact California fact check

Tuesday, July 19, 2016

When Critics Axe I

We don't have quite enough brave antagonistic souls to have a Criticism of the Week, so we're settling on a "When Critics Axe" for occasions where we find a critic grinding an axe against us.

Without further ado, the criticism (from a message board):

Supposedly the first article on our site at the time said PolitiFact was wrong because it only used the available data to make its ruling. Sounds like our critic made a snap judgment.

Our article, about Clinton's defense of the "Half True" rating it at the time gave Clinton over her claims about never sending or receiving information marked as classified, actually just pointed out that PolitiFact's defense was nonsense. We explained how using the information available at the time makes sense where the facts can change with time, as with changes in poll data. But the same justification doesn't work when it's a matter of the fact-checker not having the information at the time. If Clinton was the one sending and receiving her emails, then she had the information available at the time she made the claim, even if PolitiFact was left in the dark.

PolitiFact came around to our way of thinking in a matter of hours, quietly dropping its defense of the "Half True" rating and changing the rating to "False."

So if we're wrong on that point, so is PolitiFact. But we're not wrong.