As Ed Next readers may have noticed, I haven’t been blogging lately. That’s because I’m working on a book, which is consuming most of my time and attention. I don’t want to say too much about it quite yet, but my research has brought me into personal quarters with the “peer effects” literature. These studies, by the likes of Caroline Hoxby and Eric Hanushek, examine the impact of having peers who are black or white, rich or poor, low or high achieving.

It’s very hard to do these studies because it’s tough to disentangle peer effects from teacher effects (what if a certain teacher gets all of the highest achieving kids?) or school effects (what if the schools with the poorest or lowest achieving kids are also under-resourced?). Still, Hoxby and Hanushek (and their high-achieving peers) have been using innovative methods and are finding that peer effects do in fact exist. All else being equal, kids are better off if their classmates are high-achieving; poor, black, and low-achieving kids are particularly harmed if surrounded only by other poor, black, and low-achieving peers.

The reason I bring this up is the hullabaloo around Caroline Hoxby’s new charter school study. The news is great for charter schools and well worth celebrating: New York City charter schools are knocking the socks off of the state tests and closing the gap with their wealthy peers in the suburbs. And because it’s a gold-standard random-assignment study, we can be sure that it’s an apples to apples comparison: Hoxby examined the performance of kids who won a lottery to get into the charter schools with kids who applied to the same lottery but lost, and remained in their traditional public schools. Such a design means that the two groups of students are identical in every way, even in the intangibles like the value their parents place on education. (After all, parents of both groups of students tried to get them into charter schools.)

But in an editorial yesterday, the Wall Street Journal took this good news and went a step too far, claiming that “Caroline Hoxby has performed a public service by finally making clear that ‘creaming’ is a crock.”

Hold on, not so fast. It’s true that charter opponents can’t look at this study and claim that it unfairly compares one type of student to another. But it doesn’t prove at all that charter schools aren’t creaming. Of course they are creaming. And good for them for doing it.

Consider this: School A is a high-performing charter school with a long waiting list. School B is a crummy traditional public school with lackluster results. Mr. and Mrs. Smith enter the lottery to get their daughter Susie into the charter school, and win. Hooray! Mr. and Mrs. Jones enter the lottery to get their daughter Jill into the charter school, and lose. Boo hoo.

So what happens next? Susie Smith goes to a school filled with kids whose families entered them into the lottery. So while these parents may be poor, they are motivated, savvy, and informed enough to try to get their kids into a good school. And not surprisingly, these kids are somewhat higher-achieving than other poor kids. So now Susie is surrounded by peers who are more motivated, and possibly higher-achieving, than she had before. And these “peer effects” help her learn more.

But poor Jill Jones goes back to her traditional public school, where most of the kids are from families who didn’t even bother to try to get them into the charter school. And these peers, from less educated/savvy/motivated families, are likely dragging down Jill’s performance.

I can’t prove any of this, but I suspect that many (if not most) high flying charter schools are effectively bringing together high-potential poor and minority kids-particularly those from “striving” families-and creating an environment where they can “act smart” and succeed.

But neither can the Wall Street Journal prove that creaming isn’t happening. It’s time to put a lid on that “crock” comment, at least for now.

Update: As Eric Osberg pointed out on Flypaper yesterday, Jonathan Gyurko raised some of these same concerns too.