Monday, August 18, 2014

PunditFact gives Tucker Carlson biased fact check

PolitiFact's PunditFact, the fact-checking arm that rates the statements of pundits, gave Fox News' Tucker Carlson a hilariously slanted "Pants on Fire" rating last week. PunditFact's August 15, 2014 fact check of Carlson was chock full of the baloney we're used to finding in PolitiFact's fact checks.

Carlson objected to the tone of a Fox News segment playing up the dangers of unsecured firearms.  Carlson tried to add perspective to the story by claiming accidental bathtub drownings claimed the lives of far more children last year than did accidental gun deaths.

That looked like a job for ... PunditFact!

The PunditFact writer, Jon Greenberg, used search tools at the Centers for Disease Control website to research the numbers.  The CDC tracks the causes of death logged on death certificates.  That information is logged using ICD-9 or ICD-10 codes like those hospitals use in medical billing.

Greenberg flubbed up his work in a number of ways.

First, Greenberg used 2011 to check Carlson's claim about 2013.  It's fair to ding Carlson for trotting out an unverified claim, but the fact of the matter is that 2011 is not 2013.  Any fact checker should know this.  PunditFact rated Carlson "Pants on Fire" on a fact that PunditFact could not check for lack of data.

Second, Greenberg assumed 2011 was a suitably representative year to substitute for 2013.  That's not fact checking.  It's possible to estimate whether Carson was in the ballpark with his claim by looking at trends for the two types of accidental death.  But 2011 isn't a trend.  It's just one year.

Third, Greenberg used three ICD-10 coded deaths to represent accidental gun deaths.  The third code accounted for the largest number, and it's a catch-all code that includes deaths from flare guns and airguns.

Fourth, Greenberg excluded from the other side of the ledger accidental bathtub drownings that also included a fall into the bathtub.  Yes, there's an ICD-10 code just for deaths caused by a fall into a bathtub with a subsequent drowning.

None of these problems should occur in a competent fact check, at least not unexplained.  Yet Greenberg doesn't explain how any one of these problems affects PunditFact's ability to verify Carlson's claim.

The bogus chart for "Drowned in a bathtub" vs. "Accidental gunfire"

Here's PunditFact's bogus chart:

We looked at the numbers affecting 0-14 years.  The total for the first column on PunditFact's chart, with the deaths from a bathtub drowning associated with a fall, rises to 95.  Additional deaths from accidental bathtub drownings may have ended up under another ICD-10 catch-all code: "Unspecified cause of accidental drowning and submersion."  So the number may be higher than 95.

Also for the 0-14 age group, we checked the numbers for accidental gunfire deaths.  Of the 74 deaths in the that age range in 2011, 56 were documented using the catch-all code for "Accidental discharge and malfunction from other and unspecified firearms and guns."  That code is one of three alternatives.  The first code, W32, specifies handguns.  The second code, W33, specifies rifles and shotguns.  The third code, W34, covers everything else, including BB guns and paintball guns.

We don't know how many W34 deaths were caused by guns in the commonly understood sense of the term.  Neither does PunditFact.

One might argue that it makes sense to lump in deaths caused by every type of gun, including flare guns and paintball guns.  But if that's the case, shouldn't bathtub drowning get the same broad treatment?  Isn't a swimming pool essentially a large bathtub?

We're not saying Carlson was right that many more children died in 2013 from accidentally drowning in a bathtub than from accidental gunshots.  But in the 0-14 age range more children died from accidental bathtub drowning than from accidental gunshots (W34's included) each year we checked, from 2009-2011.

That's enough to show that Carlson has something to his point about perspective.  And it's enough to show that PunditFact stacked the deck against Carlson.

The Howler

We can't end this review without mentioning one particularly hilarious line from PunditFact's fact check.  PunditFact tried to artificially narrow the definition of "child" to make Carlson's claim look worse (bold emphasis added):
Carlson didn’t say what age children he had in mind, but in the context of the story he was responding to — and his rhetorical question about something "I want to know before I let my child go over to your house" — this is not about children under 4 years old. Parents don't let toddlers "go over" to a friend’s house.
"Go over" gets the scare quotes, we suppose, to show that the term refers only to children sufficiently autonomous to safely go unattended to a friend's house.  That's assuming the neighbor isn't in the adjacent duplex apartment.  More importantly, it assumes no danger to toddlers posed by older children, such as babysitters, playing with unsecured guns.

How ridiculous.

That's PunditFact/PolitiFact fact checking for you.

No comments:

Post a Comment

Thanks to commenters who refuse to honor various requests from the blog administrators, all comments are now moderated. Pseudonymous commenters who do not choose distinctive pseudonyms will not be published, period. No "Anonymous." No "Unknown." Etc.