Thursday, February 10, 2011

Grading PolitiFact: O'Reilly v. Obama on health care reform approval (Updated x3)

The issue:

The fact checkers:

Angie Drobnic Holan:  writer, researcher
Martha Hamilton:  editor


The relevant exchange between Fox personality Bill O'Reilly and President Obama:
O'REILLY: But the entitlements that you championed do redistribute wealth in the sense that they provide insurance coverage for 40 million people that don't have it.

OBAMA: What is absolutely true is I think in this country, there's no reason why, if you get sick you should go bankrupt. The notion that that's a radical principle, I don't think the majority of people would agree with you.

O'REILLY: Then why do the majority people in the polls not support Obamacare?

OBAMA: Actually, I think it's pretty evenly divided.

O'REILLY: It's close.

OBAMA: It's evenly divided, Bill.
Politics Daily (yellow highlights indicate portions quoted in PolitiFact story)
On with the fact check:
Here, we wanted to referee the dispute on the poll numbers: Do a majority of people in the polls oppose the health care law? Since O'Reilly brought it up, we decided to check his statement.
Hold it right there.

PolitiFact has flubbed the fact check already.

(W)hy do the majority of people in the polls not support Obamacare?
(implicit assertion:  A majority of people in the polls do not support Obamacare)

PolitiFact version:
Do a majority of people in the polls oppose the health care law?
(implicit assertion:  O'Reilly claims a majority of people in the polls oppose Obamacare)

Granted, President Obama appeared to frame the argument the same way PolitiFact framed it.  But there's a big difference in bounding the group of persons who do not support Obamacare as opposed to the group of persons who oppose it.  A person who couldn't care less about Obamacare does not count as one who opposes Obamacare.  But that same person counts as a person who does not support Obamacare.

This is yet another PolitiFact blunder that is difficult to understand without invoking the blinders of bias as an explanation.  Isn't it basic logic?
The best evidence for O'Reilly's position was the most recent poll from the Kaiser Family Foundation and Harvard School Of Public Health. In its most recent tracking poll, it found that 50 percent of respondents had an unfavorable view of the law, while 41 percent viewed it favorably. But another 9 percent said they didn't know or didn't want to answer. Certainly, the largest category in this poll didn't like health care, but if you combined supporters with the unknowns, you do get an even split.
(blue highlights added)
On what planet does it make sense to group the unknowns with the bill's supporters?  If the unknowns belong in either group they belong with those who do not support the reform bill.  The phrasing O'Reilly used gives him a much better claim on that group.
Things get even trickier when you ask people about repealing the law, because a significant portion of those who say they view the law unfavorably also say they they are opposed to a complete repeal. The most recent Kaiser poll showed that 19 percent said keep the law as it is; 28 percent said expand the law; 23 percent said repeal the law and replace it with a Republican alternative; 20 percent said repeal it and don't replace it. That means 49 percent want the law kept or expanded, and 43 percent want it replaced with a Republican alternative or simply repealed.
(bold emphasis added)
They're not kidding when they say things get trickier.  Things get so tricky that the 19 percent who want to keep the law as it is when added to the 28 percent who want the law expanded gives us a total of 49 percent who want the law.

On this planet 28+19 typically equals 47.  Nothing like a little creative rounding to inspire confidence in a fact check.

Do the math correctly and you get approximately 47 percent in favor of the law in comparison to about 53 percent who do not favor it.  Advantage:  O'Reilly.

Update 3:

Some polls have also found that at least a portion of people who dislike the health care bill dislike it because they wanted to see more dramatic changes to the health care system. A CNN poll in December found that 13 percent of those who opposed the law did so because it wasn't liberal enough. Another 43 percent favored the bill and 37 percent said it was too liberal.
While the 13 percent figure is interesting, it doesn't affect the premise of O'Reilly's question.  It does provide material that might have informed the president's answer to the question if he had elected to use it.  But the numbers break down to 54 to 57 percent not supporting the law compared to 43 percent expressing support.  A more current CNN poll also supported O'Reilly (50-55 percent not supporting compared to 45 percent supporting).

PolitiFact's conclusion:
O'Reilly said that a "majority of the people in the polls" do not support the health care law. If we were rating Obama's statement on its own, that the public is evenly divided, we would rate that True. But O'Reilly isn't entirely wrong, in the sense that support for the bill does not hit 50 percent or higher. Taking a broad picture of the polls reveals a divided public, so we rate his statement Half True.
Obama is correct that the country is evenly divided between those who favor the reform law and those who oppose it.  But Obama's response to O'Reilly did not address the issue, which was the number of persons who did not support the new law.  Obama's response was misleading in context with O'Reilly's on a point of fact ("It's evenly divided, Bill").  Obama was wrong on that point.  O'Reilly's premise was correct.

For PolitiFact, it's the fact-checking equivalent of striking out during batting practice.  O'Reilly's premise was accurate and could have warranted anything from "Half True" all the way to "True."  Based on the flawed reasoning and botched math discussed above, PolitiFact gave him the lowball rating.

The grades:

Angie Drobnic Holan:  F
Martha Hamilton:  F


Feb 11, 2011:  It's time to belatedly recognize PolitiFact's weak admission of the point on which I criticized the story:  "O'Reilly isn't entirely wrong, in the sense that support for the bill does not hit 50 percent or higher."  Of course, it ought to have read "O'Reilly is correct in the sense that support for the bill does not hit 50 percent or higher."  And because that's the sense of O'Reilly's literal words, PolitiFact needs some sort of reason for saying that O'Reilly was wrong.  I don't think PolitiFact can make that case.  The argument might be made that he didn't provide sufficient context, but that approach ought to also apply to Obama--yet PolitiFact was willing to give Obama an unqualified "True" despite the fact that the president did not admit the sense in which O'Reilly was correct.  The grades for Drobnic and Hamilton remain the same.

Update 2:

Very early on Feb. 13, I sent a message to the writer and editor pointing out the likelihood that the 49 percent figure was wrong and the lack of support for charging O'Reilly with error.  At about 11:30 a.m. today I received a message from the editor thanking me for pointing out the error in math and pointing me to a corrected version of the story.  Corrected for the math error, that is.  O'Reilly's still supposedly wrong, for reasons PolitiFact has yet to make clear.

Here's the relevant portion of my message to Drobnic and Hamilton:

Click for larger view

Update 3 (Oct. 30, 2011): 

PolitiFact eventually corrected the 49 percent figure, replacing it with the appropriate 47 percent figure.  PolitiFact did include a correction notice this time.  The logic error in evaluating O'Reilly's statement remains the central thread of the story, unfortunately.

Feb. 12, 2011:  I misidentified the PolitiFact editor throughout the original post and the update.  The post now reflects the fact that Martha Hamilton edited the story, not Bill Adair.  My apologies for the error.  Also deleted a redundant occurrence of "adding" in the paragraph about trickiness.

No comments:

Post a Comment

Please remain on topic and keep coarse language to an absolute minimum. Comments in a language other than English will be assumed off topic.