Left Jab: Rachel Maddow and the presidential salute

MSNBC television host Rachel Maddow is probably the highest-profile critic of PolitiFact from the left. We've panned a number of her criticisms of PolitiFact as weak, but her Sept. 25 blog scores a palpable hit:

So, what I wrote is true. Punditfact found it to be true. They published an amusing presidential speechmaking anecdote that not only shows that it’s true, but makes you feel all warm-hearted about its being true.  And then gave their rating:  “Mostly False”.  Ta-daa!

Usually, I ignore these guys.  Yesterday, I made the mistake of responding to their letter, which I regret. Don’t feed the trolls.  They included a line from my response to them in their rating, which I realize now may create the impression that I participated in this enterprise as if it was a real thing.  It’s not a real thing: it’s Politifact.  It’s terrible.

We appreciate the absence in Maddow's post of any partisan whining. She just makes the justifiable assertion that PolitiFact does fact checking badly, and supports it with a pretty good anecdote. PolitiFact uses some sort of Associative Property of Quotations to blame Maddow for the questionable claim of a blogger who cited her book.

We'll repeat our position there's nothing inconsistent between PolitiFact treating liberals or Democrats unfairly and our position that PolitiFact displays an anti-conservative and anti-Republican bias. Maddow has a legitimate example of PolitiFact treating her unfairly.

PFB Smackdown: Rachel Maddow (again) and White House baby screening

It does seem that MSNBC host Rachel Maddow has gone a tad unhinged over PolitiFact.

Maddow's latest complaint about PolitiFact stems from a "Mostly False" rating PolitiFact gave to a pro-life/anti-abortion group over its claim that the White House recognizes the existence of unborn babies for security purposes.

Maddow's problem seems to come from the fact that she doesn't recognize that PolitiFact's rating system weights toward the middle ratings because of compound statements and underlying arguments. 

Watch:



Maddow seems inches from screaming (3:05) "It's got to be either true or false!"

Clearly Maddow believed the "Mostly False" rating from PolitiFact was inflated.  But she made her case with cherry-picked information.  And her choices weren't so surprising when PolitiFact did much the same thing, though not quite to the extreme Maddow took it.

This is the key line from the White House e-mail, which neither Maddow nor PolitiFact saw fit to mention (from the National Right to Life press release):

"We have received a number of calls regarding how to enter security information for a baby that has not yet been born," Shafer wrote.

The PolitiFact telling picks up with Shafer's next sentence (bold emphasis added):

The release was a response to an early-morning email from the White House Visitors Office detailing how to record the personal information of babies still in utero.

"Crazy as it may sound, you MUST include the baby in the overall count of guests in the tour. It’s an easy process," Visitors Office director Ellie Schafer wrote to congressional staffers, specifying that nine zeros should be filled in for the infant’s Social Security Number.

Using just the second sentence, it's very easy to charge National Right to Life with distorting the meaning of the email newsletter.  But with the addition of the preceding sentence it is plain that Shafer is giving instructions on entering security information for a baby that has not yet been born.  The first sentence helps illuminate why Shafer says it sounds crazy to include the (unborn) baby in the overall count.  The PolitiFact version at least introduces the quotation with a helpful paraphrase ("detailing how to record the personal information of babies still in utero").

The White House instruction to provide security information for the unborn baby, clear in the White House email, was the fulcrum on which the NRTL built its implication of White House hypocrisy.

But we need to clarify which fact PolitiFact chose to check.

(clipped from PolitiFact.com)

The top portion of the PolitiFact image is a suitably accurate representation of the NRTL newsletter, and in light of the information already discussed, it can't be entirely false that the White House recognizes unborn babies in terms of its security procedures.  It sent out instructions to help expectant mothers fill out security information for babies not yet born.

The bottom portion of the PolitiFact image tends to mislead.  Saying the White House "screens" unborn babies conveys the impression that the babies have enough history to warrant some type of security threat, and implies that some unborn babies may not make the cut.  The NRTL newsletter doesn't mention anything about that type of screening and does not imply it.  On the contrary, the newsletter appears to take for granted that the White House asks for the information for the purpose of providing security for the visitors, including babies born or otherwise.

I'll go out on a limb and assume that if the White House security team has information about a pregnant woman visiting the White House on a day the White House is attacked, rescue efforts will take into account the fetus and take special action to help ensure its survival.  That's in line with the purpose of the NLRB press release.  There's a touch of hypocrisy in the policy and the NLRB newsletter doesn't overplay that angle.  If "Mostly False" isn't the correct rating then it should be higher than that.  Indeed, the NLRB's newsletter handled the truth more carefully than either Maddow or PolitiFact in this case.

Maddow's rant fits with a pattern of low-quality criticism of PolitiFact from the Left.

PFB Smackdown: Lawrence O'Donnell, Rachel Maddow and Tommy Christopher

Uh-oh!  Liberals are once again scandalized by a PolitiFact fact check!

MSNBC's Lawrence O'Donnell appeared in the following political ad:




PolitiFact looked into O'Donnell's claim about critics calling the GI Bill "welfare" and ruled it "Mostly False."   The fact check does have some problems.

PolitiFact went easy on O'Donnell

The fact check contains a huge error.  PolitiFact overlooks the fact that O'Donnell is making an equivocal argument.  O'Donnell stresses that the GI Bill was an education program.  But when PolitiFact pressed MSNBC to support O'Donnell's claim, the latter responded by providing criticisms that almost exclusively aimed at unemployment benefits that were part of the bill.  O'Donnell's argument is a bait-and-switch.

PolitiFact claims to take such abuse of context into account.  Some of the "Truth-O-Meter" grades, in fact, carry clear signs of the perils of making a claim with limited context.

Rachel Maddow

PFB Smackdown: Rachel Maddow (Updated)

We agree with Rachel Maddow up through about the 55 second mark.  Yes, PolitiFact is bad, and PolitiFact is so bad at fact checking that it doesn't deserve frequent citations as a trustworthy source. 

After that, our level of agreement starts to drop.

Sen. Rubio (R-Fla.) stated that most Americans are conservative and went on to argue the point based on attitudes toward the labels "conservative" and "liberal."

Maddow ignores the context of Rubio's remarks and attacks it using survey data about the way Americans self-identify politically.

Maddow is supposed to be ultra smart.  So how come she can't figure out that Rubio's statement isn't properly measured against self-identification numbers?

It appears that Maddow uncritically followed PolitiFact's approach to judging Rubio's accuracy.  The self-identification numbers serve as interesting context, but it's perfectly possible for 100 percent of Americans to self-identify as "liberal" yet reasonably classify as majority conservative.  That's because people can have inaccurate perceptions of their location on the political spectrum.

So, was Rubio correct that the majority of Americans are conservative?  That depends on his argument.  Rubio didn't cite surveys about self-identification.  He used a method concerned with attitudes toward the respective labels.  One can argue with the method or the application of the method, but using an inappropriate benchmark doesn't cut it.

When you ask people which party they lean toward, the independents split up so that the country is almost evenly divided. For the year of 2011, Gallup reported that 45 percent of Americans identified as Republicans or leaned that way, while 45 percent identified as Democrats or leaned that way.

Is "Republican" the same label as "conservative"?  No, of course not.

PolitiFact came close to addressing Rubio's point by looking at the political leanings of moderates, but fell short by relying on the wrong label along with the self-identification standard.  Maddow's approach was even worse, as she took Rubio's comment out of context and apparently expected PolitiFact to do the same thing.

Meanwhile, PolitiFact defends itself with the usual banalities:

“Our goal at PolitiFact is to use the Truth-O-Meter to show the relative accuracy of a political claim,” Adair explained. “In this case, we rated it Mostly True because we felt that while the number was short of a majority, it was still a plurality. Forty percent of Americans consider themselves conservative, 35 percent moderate and 21 percent liberal. It wasn’t quite a majority, but was close.”

“We don’t expect our readers to agree with every ruling we make,” he continued.

Pretty weak, isn't it?


Update 2/19/2012:

With a hat tip to Kevin Drum of Mother Jones (liberal mag), we have survey data that help lend support to Marco Rubio (as well as to my argument in his defense):

(click image to enlarge)

1)  The survey, from Politico and George Washington University, is limited to likely voters.
2)  The poll essentially forces likely voters to choose between "liberal" and "conservative."
3)  A plurality of those surveyed (43 percent) lean Democrat or self-identify as Democrat.
4)  Despite the plurality of Democrats in the survey sample, 61 percent identify as conservative ("Very conservative" or "Somewhat conservative").