SOMETHING GLOD CAN STAY: Last Saturday, we said something flamboyantly wrong (see THE DAILY HOWLER, 4/4/09). In fact, the Washington Post’s Maria Glod did not say that 3.1 months was “nearly four.” Some parts of her report could have been better. But in the relevant part of her piece, Maria Glod wasn’t wrong. She was basically right.

Background: Glod was reporting a new study of DC’s voucher program. Voucher kids had done better in reading but not in math, the study had basically said:

GLOD (4/4/09): Overall, the study found that students who used the vouchers received reading scores that placed them nearly four months ahead of peers who remained in public school...There was no difference in math performance between the groups.

Note: This difference was observed after three years, an important point Glod should have noted. But we were curious: How many months was “nearly four,” the formulation Glod offered? And uh-oh! The bagels fell from our analysts’ mouths when they read this in an editorial, that very same day:

WASHINGTON POST EDITORIAL (4/4/09): An evaluation of the D.C. Opportunity Scholarship Program released yesterday concluded that, after three years, students offered scholarships earned reading scores equivalent to 3.1 months of additional learning.

Say what? Colbert King used the same “3.1 months” figure in his own Post column that day. And sure enough! When we checked, the study’s “Executive Summary” used that same figure! This was its first finding (click here, see page xvii):

EXECUTIVE SUMMARY: After 3 years, there was a statistically significant positive impact on reading test scores, but not math test scores. Overall, those offered a scholarship were performing at statistically higher levels in reading—equivalent to 3.1 months of additional learning—but at similar levels in math compared to students not offered a scholarship.

Let’s say it again: Glod wasn’t wrong in what she wrote; we were wrong, when we jumped to conclusions. But everyone seemed to be saying “3.1 months”—except Glod, who had said “nearly four.”

What’s the answer to this puzzle? Readers, it all comes down to “used” versus “offered!” After Glod told us where to look, we spent an hour or two wrestling with this semi-impenetrable study. We’re still not completely clear about which kids were compared to which others (details below). But here’s the difference between the Executive Summary’s “3.1 months” and Glod’s quite correct “nearly four:”

Some years ago, a lot of DC kids applied for vouchers and got thrown into a lottery. Some of these kids were offered vouchers, some of these kids were not. But uh-oh! Of all the kids who were offered vouchers, 25 percent never used them. The Executive Summary compares all the applicants who were offered vouchers to all the applicants who weren’t offered vouchers. Somewhat counterintuitively, they thus lumped the kids who were offered vouchers but didn’t use them in with the kids who did use them. In short, you take all the applicants who were offered vouchers (including the kids who never used them): Those kids scored 3.1 months ahead of all the applicants who weren’t offered vouchers. It may seem odd to lump the kids who didn’t use their vouchers in with the kids who did use theirs, but that’s what the study refers to in its Executive Summary. Presumably, that’s the comparison to which the Post editorial and King’s column referred.

But uh-oh! Deeper in the study (page xxvi), Glod spotted more detailed data. If you compare the kids who actually used their vouchers to applicants who didn’t get offered vouchers, the difference in reading is 3.7 months. How many months is “nearly four?” As it turn out, 3.7!

We’re never sure why reporters say “nearly four” instead of “3.7.” But Glod was referring to that comparison—a comparison the study omits from its Executive Summary.

Which comparison is more instructive? We’re not sure. (We can imagine some common-sense reasons for leaving the non-users in. Beyond that, there may be more “technical” reasons.) But then, we still aren’t sure who’s included in the comparison Glod reports. (Not her fault. And she may know.) You see, only 41 percent of the kids who were offered vouchers used them for the whole three years; 34 percent of the kids used them only part of that time. Are the partial users included, or not? If we spent a few more hours curled up with the study’s prose, we could possibly cipher it out.

At any rate, we were grossly wrong in this part of Saturday’s post. How many months is “nearly four?” As matters turn out, 3.7! The editorial and King were discussing one thing; Glod was discussing another. (If the study had explained its procedures more clearly, these confusions might not have occurred.)

But: We were wrong about this part of Glod’s report. The analysts taunted us with their conclusion: Something Glod can stay!

While we’re at it, something worth noting: Some kids didn’t show those reading gains. Here’s Glod’s fuller presentation:

GLOD: Overall, the study found that students who used the vouchers received reading scores that placed them nearly four months ahead of peers who remained in public school. However, as a group, students who had been in the lowest-performing public schools did not show those gains. There was no difference in math performance between the groups.

According to the study, voucher recipients from “schools in need of improvement” didn’t show those reading gains. And another key group didn’t show the gains: Kids in the bottom third academically among the voucher applicants.

These facts may mean little or nothing; this is only one study, after all. But why would the lowest-achieving kids show no gains from using their vouchers? We’ll offer an informed speculation:

Deserving kids who are really far “behind” are very hard to teach. It’s hard to find textbooks (and recreational books) they can read, enjoy and learn from; it’s hard to find an organized curriculum which is geared to their academic profile. (By the way: These are lovely kids.) For this reason, we’ve always been skeptical about the idea of helping these kids with voucher programs. If they’re receiving unhelpful instruction in their public schools, it doesn’t necessarily mean their teachers are lazy or stupid. It’s hard to find appropriate books and instructional programs for kids who are way far behind. And this problem doesn’t disappear when they transfer to a Catholic school—or to Sidwell Friends. Those schools will have a hard time finding appropriate materials too. (Teachers at upper-end schools may not even realize how far behind these kids really are.)

This sort of thing doesn’t get discussed when we focus on frightening public school teachers. Frightening teachers might be helpful too. But you can’t get a bunch of readable textbooks out of a frightened “sow’s ear.”

This is only one study. But in this study, the deserving kids who were farthest behind didn’t gain from using their vouchers. They didn’t gain in reading or math. Our speculation may explain why that is.

A GIANT STEP BACKWARD: We like Salon, and we like Joan Walsh. But good God! Yesterday, the debut of Salon’s “Ask a Wingnut” feature represented a giant step backward.

In this weekly feature, an anonymous former Bush official will explain the world from the conservative point of view. (“Will bridge the cultural divide and answer questions from liberals about why conservatives think and do what they think and do.”) In theory, that’s an OK idea, though the need for anonymity is puzzling. By the way: Don’t a lot of conservatives write columns each day, explaining why conservatives think as they do? It’s a bit hard to spot the break-through in this new feature.

In theory, this is an OK idea. But the intellectual standards in yesterday’s piece represent a ginormous step backward. “Dear Wingnut,” Salon puckishly asked. “Do conservatives really think it's the media's fault that the GOP lost in 2006 and 2008?” The resulting column is so intellectually weak it makes Bernie Goldberg’s early books seem like works of high genius. You know? The ones he wrote when the MSM was kicking the tar out of Dems?

What was wrong with yesterday’s presentation? Here’s the way Salon’s new genius approached this inaugural question:

FORMER BUSH OFFICIAL (4/6/09): It is undeniable that the U.S. elite media—what both liberals and conservatives sometimes call the "mainstream media," or MSM—skews to the left. From the news pages to the editorial pages to the Op-Ed pages—where even the conservatives tend to be statists (we call them "big government conservatives")—the liberal point of view on any issue receives more favorable treatment than the conservative one.

Don't believe me? No, you don't. Let me give you a couple of examples from the 2008 presidential race, one about the media treatment of the Democratic candidate, one about how the media covered the Republican candidate.

According to Salon’s anonymous genius, “it is undeniable that the [mainstream media] skews to the left.” In fact, that claim is denied all the time, quite convincingly—and it’s hard to defend so sweeping a claim about such a wide array of news orgs. The mainstream media is made up of many newspapers, magazines, journals, broadcasts. It’s hard to prove a vast, sweeping claim about so many orgs.

It’s hard to establish such a claim. But you can’t do so in the way this egg-head attempts—by cherry-picking a pair of incidents from a two-year White House campaign. In our view, the anonymous mastermind is basically right in one critique; we’d say he’s cosmically wrong in the other. But you can “prove” any thesis you want if you’re allowed to make your case this way. The 1961 Mets were history’s greatest team—if you describe two games they won, then simply ignore all the rest.

This is a truly brain-dead post. Why print this crap in Salon?

A much more intelligent person: It’s “undeniable,” our Einstein opines; “the liberal point of view on any issue receives more favorable treatment than the conservative one.”

Please. Could we get any dumber?

Long ago, a much more intelligent person (Eric Alterman) offered a different framework. He said the press corps tends to favor liberal views on most social issues—and conservative views on most economic matters. (Please note our use of several key terms: tends to favor/on most issues.) That was just an approximation—a starting-point for further discussion. But Alterman, an intelligent person, didn’t offer the type of sweeping claim which makes Salon’s first post so worthless.

You can’t prove a sweeping claim with two cherry-picked examples. This is a very unintelligent post. It dumbs press critique way down. Why publish this junk in Salon?

EILPERIN GETS IT RIGHT: It’s never done—except today! In the Washington Post, Juliet Eilperin reports on the continuing decline of Arctic sea ice. (“The Arctic sea ice cover continues to shrink and become thinner, according to satellite measurements and other data released yesterday...”) And oh our God! It’s never done! Look what Eilperin says:

EILPERIN (4/7/09): The new evidence—including satellite data showing that the average multiyear wintertime sea ice cover in the Arctic in 2005 and 2006 was nine feet thick, a significant decline from the 1980s—contradicts data cited in widely circulated reports by Washington Post columnist George F. Will that sea ice in the Arctic has not significantly declined since 1979.

Scientists have begun debating how soon the Arctic will lose its summer ice altogether, with some saying it could happen as early as 2015. White House science adviser John P. Holdren told the crowd at the State Department that the total disappearance of sea ice in the Arctic "may be far, far closer" than scientists thought just a few years ago.

Yikes! Journalists simply never do that! To her vast credit, Eilperin did.

Will’s two columns about climate change did gain wide circulation. Unfortunately, both columns seemed quite shaky; the second was quite disingenuous. (In responding to criticism, Will omitted the largest errors he had apparently made.) After several weeks, the Post published this excellent bit of rebuttal, an op-ed by science writer Chris Mooney. But Will’s columns appear in hundreds of papers. We’ll assume that Mooney’s did not.

Both Will’s columns seemed under-informed; his second column seemed disingenuous. Post ombudsman Andrew Alexander soon stepped in to opine; his piece was weak and disingenuous too. (Why disingenuous? Because he seemed to understate what he’d said in an earlier e-mail, in which he seemed to support Will’s first column.) Meanwhile, we thought some liberals misstated the problem; they seemed to say that no one has the right to interpret scientific data except the scientists who first presented it. (In this case, those at the University of Illinois’ Arctic Climate Research Center.) Clearly, that just isn’t the case. Clarity itself melted down.

What was the fundamental problem with Will’s columns? Just this: Almost surely, Will doesn’t know enough about the science of climate change to be writing interpretive columns about it. (Neither do we; neither do you.) We’ll assume he believes what he seemed to suggest—that the scientific world has gotten conned by a gang of global warming alarmists. But it seemed that Will himself got conned, by the type of misleading, cherry-picked data the fixers are constantly churning. You can always find a fact or three which seem to suggest the conclusion you like. Almost surely, the Post was unwise to give Will license to rummage around among millions of facts and select those which struck him as most relevant. Simply put, Mooney knows much more science than Will. And unless an editor proceeds which a great deal of care, this topic calls for a specialist.

Will is very influential; whenever he writes a column, his claims travel far and wide. But uh-oh! He doesn’t seem qualified to opine about warming in the way he did. Mainstream journalists never embarrass big scribes; today, to her credit, Eilperin did. Citizens deserve to be told what Eilperin told them today: If you believed what George Will wrote, you should likely rethink what you read.