Ice, Ice Baby: When Fact-Checking Is Not Fact-Checking

For the past couple weeks, I’ve been blogging about the problems newspaper opinion pages have with science. The example I’ve focused on is two columns on global warming by George Will in the Washington Post (and syndicated to 300 newspapers). Will claims that scientists who point to evidence that global warming is having an effect on the planet and reporters who describe their research are all hysterical doomsayers. To make his point, Will offers a range of evidence, from accounts in the 1970s about global cooling to statistics about the area of global ice cover recorded by satellites.

I have argued that George Will’s claims would have not have passed the standard fact-checking carried out by many magazines. He even manages to add extra errors in his second column, which is just a defense of his first. A number of other bloggers have also criticized the Post on similar grounds. The Washington Post editorial staff has responded on three occasions, most recently and at the greatest length this morning. As I’ll explain below, it’s not much of a response.

The first reaction was reported last week in Talking Points Memo. Andrew Alexander, the new Washington Post ombudsman, checked with the editorial page editors and told TPM that they have a “multi-layered editing process” in which columns are fact-checked to the greatest extent possible. They had, in other words, been satisfied that the information in George Will column factually correct in advance of publishing it, and now saw no reason to print any corrections. Then the editorial page editor Fred Hiatt was interviewed Thursday in the Columbia Journalism Review, where he stated that Will may have made inferences from the data that scientists didn’t agree with, and that it was up to those scientists to debate Will. Again, he saw no need for any corrections, and even suggested that pieces like Will’s column helped the public appreciate the uncertainty on issues including global warming, along with other fields like medicine.

I’m not going to deal in detail with these responses here, having already done so yesterday. Instead, I want to take a look at the latest response that came out this morning: a full-blown column in the Washington Post by the ombudsman Andrew Alexander–in fact, Alexander’s first official piece in his new job. You can read it here.

As I read it, I kept hitting one puzzling statement after another. For example, Alexander starts out the piece by focusing his column on what he calls “a key paragraph” about the global area of ice. As I’ve explained before, that paragraph is indeed in error, both in the specifics of the data, and in the way Will uses it as evidence that global warming has not been occurring. It became all the more striking because the scientists whom Will named as his source for the data rejected his claims, and, as I later showed, neither Will nor any of the fact-checkers bothered to contact the scientists to confirm their information. Instead, they pointed to another statement from the scientists as confirming Will’s claim–while ignoring the parts of the one-page statement that showed why Will was wrong.

But as vivid as that case may be, it was only one of numerous errors in the piece. If Will’s columns had indeed been properly fact-checked, the fact-checkers would have drawn attention to other errors in his columns.

For example, Will misrepresents an article by the late great Walter Sullivan in the New York Times in 1975, pretending that it trumpets an imminent plunge into an Ice Age:

The New York Times was — as it is today in a contrary crusade — a megaphone for the alarmed, as when (May 21, 1975) it reported that “a major cooling of the climate” was “widely considered inevitable” because it was “well established” that the Northern Hemisphere’s climate “has been getting cooler since about 1950.”

Here is how that article actually starts:

The world’s climate is changing. Of that scientists are firmly convinced. But in what direction and why are subjects of deepening debate.

The whole article is here [$]. For more on all this, see here and see “The Myth of the 1970s Global Cooling Scientific Consensus,” (free pdf) published in the Bulletin of the American Meteorological Society. (In the interest of full disclosure, I should point out that I write frequently for the Times, although only once about global warming.)

Here’s another error Alexander doesn’t address: Will tries to use a recent satellite sensor glitch as evidence that skeptical scientists get attacked for questioning global warming. I explained how scientists have dealt with that glitch and corrected the record, and how the scientists themselves state that the glitch doesn’t affect their conclusion that the Arctic has shown a three-decade trend of shrinking ice area–a result that also comes from climate models.

But Alexander never addresses anything beyond Will’s claims about the global area of ice now and in 1979. When fact-checkers write up their reports, they do not just look at one paragraph and call it a day. I don’t understand why that is acceptable for a report from an ombudsman about the accuracy of a newspaper column.

But even within this narrow scope, Alexander’s conclusions puzzle me. He states:

My inquiry shows that there was fact-checking at multiple levels.

What Alexander then describes is not fact-checking.

It began with Will’s own research assistant, Greg Reed. When the column was submitted on Feb. 12 to The Washington Post Writers Group, which edits and syndicates it, Reed sent an accompanying e-mail that provided roughly 20 Internet reference links in support of key assertions in the column. Richard Aldacushion, editorial production manager at the Writers Group, said he reviewed every link. The column was then edited by editorial director Alan Shearer and managing editor James Hill.

Next, it went to The Post’s op-ed editor, Autumn Brewington, who said she also reviewed the sources.

Fact-checking descriptions of scientific research involves a wee bit more than perusing Internet reference links. It is not just a pattern-matching game, where you see if a sequence of words is the same in two places. Anyone who has actually fact-checked for a magazine like Discover (where I fact-checked for a few years) can tell you that you need to get familiar with the scientific research to see if the description is a good representation of the science itself.

And one essential part of getting familiar with it is calling scientists who live day and night with that research (especially if those scientists were cited explicitly in the piece being checked). A call to the scientists would have immediately sent up red flags (as I found when I got in touch with them on February 21 to satisfy my own curiosity and clear up some questions of my own).

This is not a criticism of the people Alexander names in his column. Newspapers and magazines are responsible for establishing procedures for fact-checking, which staff members must then follow. What I don’t understand is how Alexander can offer us this account of what happened and call it fact-checking at multiple levels.

Even more puzzling is Alexander’s account of his own research into the narrow question of the ice.

The editors who checked the Arctic Research Climate Center Web site believe it did not, on balance, run counter to Will’s assertion that global sea ice levels “now equal those of 1979.” I reviewed the same Web citation and reached a different conclusion.

It said that while global sea ice areas are “near or slightly lower than those observed in late 1979,” sea ice area in the Northern Hemisphere is “almost one million sq. km below” the levels of late 1979. That’s roughly the size of Texas and California combined. In my mind, it should have triggered a call for clarification to the center.

But according to Bill Chapman, a climate scientist with the center, there was no call from Will or Post editors before the column appeared. He added that it wasn’t until last Tuesday — nine days after The Post began receiving demands for a correction — that he heard from an editor at the newspaper. It was Brewington who finally e-mailed, offering Chapman the opportunity to write something that might help clear the air.

Readers would have been better served if Post editors, and the new ombudsman, had more quickly addressed the claims of falsehoods.

I know that I may be sounding a bit Talmudic by spending so many blog posts on this one bit of information, but examining how these Post editors have dealt with it has proven to be very revealing. They never bothered to check with scientists about the validity of a statement in a column, and after thousands of people have complained, they recognize that there was something so amiss that should have called the scientists. But they still can’t manage to make a decision about whether the statement requires a correction.

What’s more, they continue to ignore the broader, more important problem with Will’s discussion of sea ice: the facts that picking out two days from a thirty-year time series is not a meaningful way to look at climate trends, and that climate models do not, in fact, lead you to expect a decrease in global ice cover. And they have not even taken any notice of all the other errors in Will’s two columns.

Alexander’s prescription for the Post is this:

On its news pages, it can recommit to reporting on climate change that is authoritative and deep. On the editorial pages, it can present a mix of respected and informed viewpoints. And online, it can encourage dialogue that is robust, even if it becomes bellicose.

I don’t see why the news reporters at the Post have to recommit to anything. They’ve been doing their job. What really has to happen is for people who claim to be fact-checking to really do some fact-checking. It’s that simple.

Update, Sunday 3/1: In my initial version of this post, I sometimes referred to Andrew Alexander as Anderson by mistake. When I first noticed this mistake, I thought I only did it once and fixed that error. But commenters have kindly pointed out I had left several Andersons behind. I’ve now fixed them all. Apologies for the confusion.

Update later Sunday: Via Andy Revkin, I came across what is essentially an independent fact-check. It’s from Walt Meier of NSIDC, responding to a question about Will’s column

Basically, Mr. Will made three mistakes:

1. He was factually incorrect on the date that he reported his “daily global ice” number. However, he was merely out-of-date with his facts (it was true on Jan 1, but wasn’t 6 weeks later). This is somewhat nit-picky, though it illuminates how fast things can change in a relatively short period of time, meaning that one should be very cautious about drawing any conclusions about climate from an isolated event.

2. Related to that, it is easy to cherry-pick one date here and one date there to compare to support most any view. The important thing is to look at things in the context of long-term changes. That is what NSIDC always tries to convey by comparing to long-term averages.

3. “Global sea ice” simply has no meaning in terms of climate change. The Arctic and Antarctic are unique and separated environments that respond differently. It would be like taking a drought in Georgia and torrential rain in Maine, adding those up and claiming that “rainfall is normal” in the eastern U.S.

Update, 4/7/09: Alexander’s use of “Arctic Climate Research Center” is incorrect.